rss
Gut 57:A1-A172
  • Abstracts

Abstracts

Inflammatory bowel disease

THE USE OF ADALIMUMAB IN THE MANAGEMENT OF CROHN’S DISEASE: SCOTTISH SOCIETY OF GASTROENTEROLOGY AUDIT 2007

G. Ho1, L. Smith1, S. Aitken1, H. Lee1, D. Ting1, J. Fennell1, C. Lees1, K. Palmer1, I. Penman1, A. Shand1, I. Arnott1, M. Priest2, R. Boulton-Jones3, R. Russell4, C. Mowat5, D. Wilson6, J. Satsangi1. 1Gastrointestinal Unit, Western General Hospital, Edinburgh, 2Gastrointestinal Unit, Gartnavel General Hospital, 3Gastrointestinal Unit, Victoria Infirmary, 4Gastrointestinal Unit, Yorkhill Hospital, Glasgow, 5Gastrointestinal Unit, Ninewells Hospital, Dundee, 6Gastrointestinal Unit, Royal Hospital for Sick Children, Glasgow, UK

Introduction: Adalimumab is a humanised monoclonal antibody targeting tumour necrosis factor-α. Recent clinical trials have demonstrated its efficacy in Crohn’s disease, however experience in clinical practice remains limited.

Aims & Methods: We present the experience on the efficacy and safety of Adalimumab in a Crohn’s disease (CD) cohort in Scotland over a 4-year period (2003–7). All members of the Scottish Society of Gastroenterology were approached and data extraction was performed from case notes using a standard proforma form.

Results: 42 (26 females; age at therapy: 28.6 years, IQR 19.5–40.2 including 5 paediatric cases) patients were treated with Adalimumab from 7 hospitals. 34 (80.9%) patients were treated using an 80/40 mg induction regime followed by fortnightly 40 mg treatment (6 and 2 patients had 160/80 mg and 24/24 mg induction respectively). 38 (90.5%) patients had had previous infliximab infusions and 11 (26.2%) were primary infliximab non-responders. Over a median period of 0.57 year (IQR 0.26–1.71), 23 (54.7%) patients achieved and were maintained in steroid-free remission (survival curve analyses 56.3% at 1-year follow-up). Two (40%) paediatric CD cases achieved steroid-free remission (range 0.44–0.49 years). 19 of 34 (55.9%) patients on fortnightly therapy required dose escalation to 40 mg weekly (0.65 years IQR 0.13–1.44). 7 (63.6%) primary non-responders to infliximab achieved and maintained steroid-free remission (median 0.80 years follow-up IQR 0.44–1.87). There were 4 (9.5%) serious complications requiring permanent discontinuation of Adalimumab therapy: 2 patients developed serious infective complications (severe facial cellulitis and psoas abscess secondary to colonic perforation, after 8 weeks and 2.6 years of therapy), one patient developed lung cancer and one with central demyelinating illness (after 1.7 years and 9 weeks of therapy respectively). No patients developed infusion reactions.

Conclusion: Adalimumab is efficacious in the treatment of patients with refractory CD, with particular benefit in patients who have lost response or developed adverse reaction to infliximab. In this audit, we also present data for the first time in paediatric Crohn’s disease. However, many patients require escalation of dosing regime. Moreover, the treatment carries definite risks of serious adverse events, and careful patient selection, and continuous monitoring is needed.

ONCE- VERSUS TWICE-DAILY MESALAZINE (PENTASA) GRANULES FOR THE MAINTENANCE OF REMISSION IN ULCERATIVE COLITIS: RESULTS FROM A MULTINATIONAL RANDOMISED CONTROLLED TRIAL

A. Dignass1, H. Veerman2. 1Head Department of Medicine I, Markus-Krankenhaus, Frankfurter Diakonie-Kliniken, Frankfurt-Main, Germany, 2Medical, Ferring Pharmaceuticals, Hoofddorp, The Netherlands

Introduction: Patients with quiescent ulcerative colitis (UC) do not frequently adhere to prescribed mesalazine treatment. This is, in part, due to the low acceptability of currently available mesalazine formulations, which often require multiple-daily dosing regimens and a high number of tablets. Non-compliance is directly linked to relapse and disease exacerbation. This study investigated remission and relapse rates in patients with UC in remission receiving once daily (od) a single 2 g mesalazine granules sachet or twice daily (bd) a 1 g mesalazine granules sachet.

Aims & Methods: A 12-month, multicentre investigator-blinded randomised controlled trial. Patients with mild to moderate UC in remission who had had a relapse within the past year were randomised to receive either mesalazine granules 2 g/day given od or bd. Clinical evaluations were undertaken at 0, 4, 8 and 12 months. Primary objective was to demonstrate non-inferiority (defined as not more than 10% less efficacy below the od treatment) of the two regimes in terms of probability to remain in clinical and endoscopic remission during one year defined by an UCDAI score of <2, or an UCDAI of 2 without any adjustment of therapy. Treatment difference was calculated using Kaplan-Meier methodology.

Results: 362 patients were randomised. At end-of-study, 73.8% and 63.6% of patients in the od and bd groups were in clinical and endoscopic remission, respectively. The difference between the two treatment groups after one year was significantly above the non-inferiority limit with a treatment difference of 11.9% in favour of od (p = 0.024). Once daily dosing had a higher compliance than bd dosing in terms of % sachets used. Patient questionnaires yielded higher compliance scores for od than bd (p<0.05) and patient acceptability of treatment was significantly better for od (96.3% vs 85.6%, p<0.001) at end-of-study.

Conclusion: Once-daily 2 g mesalazine granules sachets (PENTASA) are an effective maintenance therapy in patients with mild to moderate UC. A once daily dosing regimen using a single 2 g sachet is not only non-inferior but even superior to twice daily dosing. Mesalazine granules taken once daily (2 g) ensure better remission rates, patient compliance and acceptability than twice-daily (2×1 g) administration.

Endoscopy free papers

CAN DEATH FOLLOWING ERCP BE PREDICTED? RESULTS OF A LARGE SCALE PROSPECTIVE STUDY

E. J. Williams. On behalf of the BSG audit of ERCP Steering Committee, British Society of Gastroenterology, London, UK

Introduction: The incidence of early mortality (from all causes) following endoscopic retrograde cholangiopancreatography (ERCP) is rarely described in prospective studies, though retrospective analyses of deaths suggest patient selection could be improved.

Aims & Methods: The aims are to examine 30-day mortality following ERCP, and identify independent predictors of early death. Results are derived from a prospective multicentre study of ERCP complications, based in 5 English regions. Where patients consented to follow-up, death within 30 days of ERCP was identified through contact with carers and interrogation of hospital electronic records. Where patients did not consent to follow-up mortality was calculated from death registrations. Binary regression analysis of participating patients’ first recorded procedures identified independent risk factors for early mortality.

Results: A total of 6910 ERCPs were performed in 66 centres during the study period. Consent for detailed follow-up was acquired for 5264 procedures, performed on 4561 patients. A therapeutic intervention was attempted in 3447/4561 (76%) of these patients as part of their first recorded ERCP. In total 121 participating patients (2.7%) died within 30 days of first recorded ERCP. Thirty-day mortality among non-participating patients was calculated to be higher (92/1646; 5.6%). Closer examination of participating patients revealed 18/121 deaths were directly attributable to the patient’s first recorded ERCP, with a further 2/121 deaths attributable to a second ERCP performed within 30 days of the first procedure. The remaining 101/121 deaths resulted from progression of the underlying disease process (which was carcinoma in 82 cases). Risk factors for death (OR, 95% CI) among participating patients were: ASA score (120.1, 24.5 to 588, when ASA 5 patients were compared to ASA 1 patients); a malignant indication for ERCP (2.94, 1.67 to 5.15); lower albumin (1.05, 1.02 to 1.08 for every g/l decrease); ln bilirubin (1.49, 1.18 to 1.88) and ln white cell count (1.54, 1.09 to 4).

Conclusion: The extrapolated 30-day mortality following ERCP, based on participating and non-participating patients is 3.4%. While detailed analysis was limited to 5264/6910 procedures, results suggest that most deaths arise from progression of underlying disease, rather than as a result of ERCP itself. ASA score and pre-ERCP diagnosis are identified as important predictors of outcome.

VALIDITY AND RELIABILITY OF AN ACCREDITATION ASSESSMENT FOR COLONOSCOPY

R. Barton. For the Joint Advisory Group on Gastrointestinal Endoscopy and NHS Bowel Cancer Screening Programme, Newcastle University and Northumbria Healthcare NHS Trust, Newcastle upon Tyne, UK

Introduction: After national audit demonstrated very variable quality of colonoscopy, the introduction of the NHS Bowel Cancer Screening Programme prompted the introduction of a searching assessment to select the screening programme’s future colonoscopists. Leading endoscopists, trainers and the JAG collaborated to develop an appropriate assessment tool after consultation with clinical educationalists in 3 countries. This comprised performance data, a knowledge test (MCQ), and a DOPS (Direct Observation of Procedural Skills) assessment.

Aims & Methods: The aims of this study are to determine the validity and reliability of the assessment. A true/false 30 item × 5 answer MCQ was devised covering key issues. To ensure both professional and educational input, DOPS forms from two centres were reviewed by the Trainers Group appointed under the NHS National Endoscopy Training Programme, and important colonoscopy skills were identified and amalgamated into domains. Detailed descriptors for all skill levels were written for each domain. An outline curriculum was published, and assessors underwent both the assessment and further assessor training. Self-reported, verified, performance and demographic data, as well as assessment data both from the MCQ and DOPS over two cases by two assessors were collected prospectively from 76 candidates. Semi-structured questionnaires were collected from candidates and 17 assessors on the validity of the assessment.

Results: Colonoscopists submitted performance data and attempted the assessment. In 2284 paired judgments over 76 candidates during a total of 151 cases, overall there was 96% congruence of agreement across the pass/fail divide (98% for major domains), and high absolute congruence between assessors in many domains, for example 72.4% for “communication”, 69% “maintains luminal views”, 73% “uses torque steering” and 81% “adequate mucosal visualisation”. The expert global opinion agreed with the grading system in 74/76 (97%) of cases. Gradings correlated weakly with self-reported caecal intubation rates and MCQ scores (r = 0.24 and 0.27, p<0.01) respectively. No correlation was seen between the grading and colonoscopies in previous year. Overall 27/30 candidates felt the DOPS assessment was fair/very fair, while 27/32 felt the MCQ was fair/very fair. Of the assessors, 12/16 felt the DOPS was valid/very valid, while 17/17 overall felt the process was fair/very fair.

Conclusion: The DOPS accreditation has good face and content validity, and very high reliability across the crucial pass/fail divide. There is ongoing work improving the MCQ.

BOWEL CANCER SCREENING PROGRAMME ONE YEAR IN: EXPERIENCE OF A SINGLE FIRST WAVE CENTRE

M. R. Tighe1, M. G. Phillips1, W. S. L. Stebbings2, B. Brett1, A. Cook1. 1Department of Gastroenterology, 2Department of Surgery, Norfolk and Norwich University Hospital, Norwich, UK

Introduction: The Norfolk & Norwich University Hospital was one of the first centres to launch the national bowel cancer screening programme, commencing in July 2006. The NNUH started covering a population of 550 000 based in mid Norfolk, with an estimated 68 000 falling into the 60–70-year-old age group. In June 2007, the population covered by screening in Norwich was expanded to the whole county of Norfolk with a population of 900 000.

Aims & Methods: A year on, we have reviewed our work to date to see if our results reflect the experience drawn for the large scale FOB trials and the pilot sites.

Results: Uptake for the programme has been high at 67%. In the first year, 53 711 invitation FOB kits were sent out, with 2933 over 70-year-olds requesting additional kits. 35 592 FOB kits have been returned with a positivity rate of 1.65% (584). Among these 584, 71 FOBT +ve were over the age of 70. 99% of patients attended pre-assessment and only 4 were deemed unsuitable for colonoscopy and underwent CT colonography instead. Colonoscopy was complete in 98.9% and 6 cases went on to have a CT colonography to complete the examination. Pathology detected: In year one 84 cancers have been detected (14.8% of colonoscopies)—23 were polyp cancers, an additional 10 cases were staged as Dukes A, 15 cases as Dukes B and 19 cases as Dukes C. 17 patients are awaiting final staging and surgery. All 23 polyp cancers were removed endoscopically. Six went on to have a segmental resection for poor prognostic features with no evidence of residual disease being found. Overall, 56% of cancer detected in screening population were Dukes A, 21% Dukes B and 23% Dukes C. Among over 70-year-old self-referrers, 14.6% with FOBT +ve had cancer, staged as Dukes A 18%, Dukes B 27% and Dukes C 55%, mirroring the symptomatic service. 49% of patients had adenomas at colonoscopy, subdividing into 13% high risk adenomas, 54% intermediate risk and 34% low risk adenomas. 35% of colonoscopies were normal. Complications: One patient has died following surgical removal of a caecal adenoma. Two patients have bled post large polypectomy, one of whom required surgery—the other settled spontaneously.

Conclusion: The first year experience of the screening programme supports the trial and pilot study data. The detection of predominant early stage disease amongst the screened population infers that the trial benefits in terms of mortality from colorectal cancer will be seen in our population. The programme has been very well received by the public. We are targeting publicity to those geographical areas of lower take-up.

COMPARISON OF HIGH RESOLUTION MAGNIFICATION ENDOSCOPY, NARROW BAND IMAGING WITH MAGNIFICATION AND CHROMOENDOSCOPY WITH MAGNIFICATION IN THE DIAGNOSIS OF COLORECTAL NEOPLASIA: A FEASIBILITY STUDY

R. Singh1, A. Shonde1, P. Kaye2, C. J. Hawkey1, K. Ragunath1. 1Wolfson Digestive Diseases Centre, 2Department of Histopathology, Nottingham University Hospitals Trust, Nottingham, UK

Introduction: With the emergence of colorectal cancer screening, the distinction between non-neoplastic from neoplastic colorectal polyps during colonoscopy is paramount. A suitable technique which allows this differentiation could increase the efficiency of treatment by eliminating additional cost associated with unnecessary biopsies. Although chromoendoscopy (CHR) has been propagated as a technique that improves prediction of polyp histology, it has failed to be incorporated into routine practice. Narrow band imaging (NBI) is a novel endoscopic imaging technique which is relatively simple to use and simulates CHR during the procedure.

Aims & Methods: We evaluated the positive predictive value (PPV), negative predictive value (NPV) and accuracy (Acc) of three different techniques: conventional high resolution white light endoscopy (WLE), NBI and CHR, all with magnification in differentiating adenocarcinomas, adenomatous and hyperplastic colorectal polyps. Each lesion was sequentially assessed first by WLE, followed by NBI and finally by CHR. Digital images of each polyp with each modality were taken. Biopsies or polypectomies were then performed followed by blinded histopathological analysis. Each image was then blindly graded based on the Kudo’s pit pattern (KPP). In the assessment with NBI, the meshbrown capillary network pattern (MBCN) of each polyp was also described. The PPV, NPV and Acc of differentiating hyperplastic (Type I & II - KPP, Type I - MBCN) adenomatous (Types III, IV - KPP, Type II - MBCN) and carcinomatous polyps (Type V- KPP, Type III- MCBN) were then compared with reference to the final histopathological diagnosis.

Abstract 006 PPV, NPV and Acc of WLE, NBI (KPP), NBI (MCBN), NBI (KPP + MCBN), CHR in colon polyps

Results: A total of 50 colorectal polyps (5 adenocarcinomas, 38 adenomas, 7 hyperplastic) in 37 patients (25 men, mean age 68.5 years) were assessed. Morphologically, 19 were classified as type Is, 6 type Ip, 24 type IIa and 1 type IIa-c polyps. The PPV, NPV and Acc of the various modalities are depicted in the table.

Conclusion: NBI was superior to WLE and CHR if both the KPP and MBCN criteria were combined and applied in the prediction of non-neoplastic and neoplastic colorectal polyps. This feasibility study demonstrated that this combined classification for NBI could potentially be useful in routine clinical practice, allowing the endoscopist to predict histology with higher accuracies.

CONFOCAL ENDOMICROSCOPIC LASER SCANNING IN VIVO CELLULAR IMAGING OF BARRETT’S OESOPHAGUS: FIRST PROOF OF ENDOSCOPIC CONCEPT IN THE UK

D. P. Hurlstone, N. Tiffin, S. S. Cross. Endoscopy, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Confocal laser scanning endomicroscopy permits cellular resolution imaging in vivo during ongoing video endoscopy. Microvascular architecture and connective tissue stroma can be visualised to a depth of 250 μm. In Barrett’s oesophagus, specialised columnar epithelium presents in a “patchy” distribution. Previous studies assessing magnification +/− chromoscopy have failed to show a definitive clinical advantage compared to quadrantic biopsies alone.

Aims & Methods: To evaluate the efficacy of live cellular endomicroscopic imaging for the in vivo diagnosis of Barrett’s epithelium during ongoing video endoscopy. Consecutive patients with long-term reflux, known Barrett’s oesophagus or patients with known complicating IN were recruited. Endomicroscopy following methylene blue 0.1% chromoscopy was performed using topical acroflavin for surface mucosal nuclear imaging in addition to 5 ml of iv 10% sodium fluorescene for deep z-axis vascular imaging. Endomicroscopy was performed at 10 mm quadrantic segmental intervals and targeted to focal mucosal defects unmasked by chromoscopy—classified according Paris guidelines. Confocal greyscale macro data were collected at 1.6 frames/sec/optical z-axis sectioning 0.7 μm—field view 500 μm×500 μm, minimum 12 macros per segment. The in vivo endomicroscopic diagnosis (according to Mainz criteria) was then compared to targeted biopsies equating to the raster scanned mucosal segment taken as the gold standard.

Results: 48 patients completed the endomicroscopy protocol-median duration of procedure 18 min (range 10–32)/median Barrett’s length 3.4 cm (range 1–12). A total of 9216 endomicroscopy images from 243 target sites were analysed. The in vivo diagnosis of Barrett’s oesophagus using endomicroscopic vascular and cell architecture criteria was predicted with a sensitivity 94.3%/specificity 91.8%/PPV 94.8% and NPV of 96.4%. Basement membrane integrity loss, sub-epithelial vascular tortousity +/− leakage and irregular dark epithelial cell architecture were highly predictive of high-grade dysplasia in vivo (sensitivity 88.1%/specificity 94.1%)/K = 0.88.

Conclusion: Endomicroscopy permits live cellular, sub-cellular and deep vascular imaging of high quality during ongoing video endoscopy. In vivo cellular-vascular architecture has a high correlation with ex vivo histopathology and may prove a novel tool in the rapid diagnosis of Barrett’s oesophagus and associated neoplasia. Multiple optical biopsy sites can be obtained in vivo without white light “field loss” due to biopsy site bleeding. Targeted high quality specimens can improve histopathological processing and patient “flow” in endoscopy.

CATCHING THE BIT IN BETWEEN: ENDOSCOPIC ULTRASOUND-GUIDED MURAL TRUCUT BIOPSIES IN THE DIAGNOSIS OF THICKENED OESOPHAGOGASTRIC WALL

T. Thomas1, P. Kaye2, K. Ragunath1, G. Aithal1. 1Gastroenterology, 2Histopathology, Nottingham University Hospitals NHS Trust, QMC Campus, Nottingham, UK

Introduction: A thickened oesophagogastric wall (OGW) in patients with alarm symptoms raise the possibility of malignancy, but establishing a tissue diagnosis may be difficult as mucosal biopsies are usually negative. Using linear array endoscopic ultrasound (EUS) it is possible to target and biopsy the thickened part of the gut wall.1

Aims & Methods: We aimed to evaluate the efficacy and safety of EUS-guided transmural biopsies in detecting underlying malignancy in patients with thickened oesophagogastric wall and negative mucosal biopsies. Patients were referred for EUS guided sampling as a part of their clinical management. Linear array echoendoscope (GF-UCT240-AL5, Olympus- KeyMed) and 19 G Trucut needle (Wilson Cook, Quickcore) were used for sampling. A previously described technique was used for mural biopsy.1 Clinical and investigatory data were collected prospectively from 2004 to date.

Results: Thirty one patients (19 men) aged 60–74 years (median 68) were included. All patients had thickened oesophageal wall (n = 10), gastric wall (n = 20) or both on CT scan. Prior to EUS, patients had 1–5 (median 1.2) endoscopies and 2–8 mucosal biopsies. The mean oesophageal and gastric wall thickness was 12 and 18 mm respectively. During sampling 1–5 (median 3) needle punctures were made. On EUS trucut biopsy, adequate specimen was obtained for histology in 25/31 (80.6%) although biopsies were fragmented in 16/31 (51.6%) patients. The size of the samples were 4–10 (median 6) mm. Malignancy was confirmed in 17/31 patients (54%) on histology and in 11/31 patients (35.5%) an underlying malignancy was excluded (confirmed at a median follow-up 12 months). There was no significant correlation between wall thickness and biopsy size (Rho, 0.11, 95% CI −0.25 to 0.45, two sided p = 0.53). Transmural biopsies have a sensitivity and specificity of 85% and 94% respectively. The positive and negative predictive value was 94% and 77% respectively. There were no major immediate or late complications

Conclusion: EUS-guided trucut biopsy is a safe and useful technique in the investigation of thickened oesophagogastric wall with a high sensitivity and specificity for the diagnosis of cancer.

DEDICATED BLEED UNITS: SHOULD THEY BE ADVOCATED?

R. Sidhu, P. V. Sakellariou, M. E. McAlindon, D. P. Hurlstone, S. A. Riley, M. T. Donnelly, S. Johal, J. M. Hebden, K. K. Basu, A. J. Lobo, D. S. Sanders. Department of Gastroenterology, Sheffield Teaching Hospitals NHS Trust, Sheffield, UK

Introduction: Upper gastrointestinal haemorrhage (UGH) is a common emergency, however dedicated bleed units with 24-hour endoscopy are still only available in selected hospitals. A previous study from this unit supported a lower case mix adjusted mortality rate in a specialised bleed unit.

Aims & Methods: To evaluate the re-bleeding and mortality rate of patients admitted with UGH to the Sheffield Bleed Unit in comparison to the National Audit and previous performance. Notes of patients admitted between January 2005 and December 2006 were reviewed retrospectively for demographics, Rockall and Child Pugh scores and outcome. Multiple logistic regression was performed.

Abstract 009

Results: 66% were emergency admissions, while 23% were tertiary referrals and 11% were current inpatients. 29% of gastroscopies were done out of hours while 82% were within 24 hours. The table tabulates results with a comparison to the previous study and the National Audit. In the 2005–6 cohort, pre and post-endoscopy Rockall scores were predictive of re-bleeding (p = 0.013, p = 0.045) and mortality (p = 0.003, p = 0.01) while only the Child Pugh was predictive of mortality in liver patients (p = 0.002). Compared to the previous cohort, a reduction in the number of bleeding points identified at endoscopy and the number of patients requiring surgery may reflect the vigorous use of intravenous proton pump inhibitors pre-endoscopy and the aggressive endoscopic intervention performed in our unit.

Conclusion: This study has shown that in comparison to the National Audit, the re-bleeding rate and mortality is consistently lower in both cohorts of patients admitted to our bleed unit over time. This further supports the evidence to encourage the development of dedicated bleed units across the UK.

PATIENT POSITION CHANGES DURING COLONOSCOPE WITHDRAWAL INCREASE ADENOMA DETECTION: A RANDOMISED, CROSSOVER TRIAL

J. E. East1, P. Bassett1, M. Stavrinidis1, N. Arebi1, S. Thomas-Gibson1, T. Guenther2, B. P. Saunders1. 1Wolfson Unit for Endoscopy, 2Academic Department of Cellular Pathology, St Mark’s Hospital, London, UK

Introduction: Colonoscopy has a miss rate for adenomas which may partly relate to poor visualisation of the colonic surface. Position changes during colonoscope withdrawal can improve luminal distension.1

Aims & Methods: We aimed to assess whether position changes during colonoscope withdrawal improve adenoma and polyp detection. 130 patients attending for routine colonoscopy underwent back-to-back segmental examination of the colon proximal to the sigmoid-descending junction, randomised to have the first examination in either left lateral position alone or with position changes (cecum to hepatic flexure, left lateral; transverse, supine; splenic flexure and descending colon, right lateral) followed by examination in the other position. After both examinations, polyps were removed for histopathology. Luminal distension was measured in each position on a scale of 1–5: 1, total collapse; 5, fully distended.

Abstract 010 Patients with at least one adenoma detected by colonic area

Results: At least one adenoma was detected in 34% of patients in colonic areas where position change differed from left lateral (transverse, splenic, descending) compared to 23% examined in left lateral alone, p = 0.01. At least one polyp was detected with position changes in 52% versus 34% respectively, p<0.001. Both adenoma and polyp detection were positively correlated with improved distension score, correlation coefficient 0.12, p<0.001. Adenomas were detected in 16% of colonic areas with an adequate distension score (4 and 5) compared to 7% of those with a borderline or non-diagnostic score (1–3), p<0.001.

Conclusion: Position changes during colonoscope withdrawal, a cost-neutral intervention, significantly improved adenoma detection between hepatic flexure and sigmoid-descending junction. The effect appears to relate in part to improved luminal distension. This intervention could improve efficacy of colonoscopy in cancer prevention without additional resources.

ClinicalTrials.gov Identifier: NCT00234650.

PREDICTING OUTCOMES FOLLOWING GASTROSTOMY INSERTION USING THE SHEFFIELD GASTROSTOMY SCORE. A PROSPECTIVELY DEVISED SCORING SYSTEM WITH A VALIDATION COHORT

J. S. Leeds1, M. E. McAlindon1, S. R. Morley2, R. Marr2, H. E. Robson1, J. Grant1, F. K. T. Lee3, M. T. Donnelly4, D. S. Sanders1. 1Gastroenterology, 2Clinical Chemistry, 3Radiology, Royal Hallamshire Hospital, 4Gastroenterology, Northern General Hospital, Sheffield, UK

Introduction: Several scoring systems are used in the field of Gastroenterology. Although previous studies have demonstrated the substantial risk of death following gastrostomy insertion and some risk factors have been identified, no previous investigators have described a scoring system for gastrostomy insertion. We undertook a prospective, unselected, dual centre study in order to establish the relative importance of risk factors for mortality after gastrostomy insertion. We then formulated a simple numerical scoring system to categorise patients’ risk of death. We then sought to validate our scoring system on an independent second cohort of patients.

Aims & Methods: All patients referred for gastrostomy are prospectively included in a database along with demographic, biochemical and outcome data. We analysed gastrostomy insertions from two teaching hospitals from February 2004 to February 2007. There were 887 referrals resulting in 837 gastrostomy insertions. The largest cohort was at site A and was used to construct a risk stratification scoring system. Site B was used to validate the scoring system.

Results: Site A received 552 referrals and 403 new gastrostomies were inserted (median age 64, 268 males). Overall 30-day mortality rate was 51/403 (12.7%) with the highest risk in those with dementia (40%) followed by stroke (22.2%). Univariate analysis identified that 30 day mortality was associated with age (OR 3.4), albumin (OR 5.6), cardiac comorbidity (OR 2.0) and neurological comorbidity (OR 1.7). On multivariate analysis only age and albumin remained significant (both p<0.001) and were then modelled and attributed scores with age scoring 0 or 1 and albumin scoring 0, 1 or 2 giving composite scores from 0 to 3. Scores of 0, 1, 2 and 3 gave 30-day mortalities (95% CI) of 0% (0 to 2.1), 7% (2.9 to 13.9), 21.3% (13.5 to 30.9) and 37.3% (24.1 to 51.9) respectively. Kaplan-Meier curves stratified by total score showed a significantly increased mortality at 7 (p = 0.0003), 30 (p<0.0001) and 90 (p<0.0001) days. Site B (validation cohort) received 335 referrals and inserted 153 new gastrostomies (median age 77, 64 males) with a 30-day mortality of 24/153 (15.7%). Application of the scoring system in this validation cohort gave comparable 30-day mortality figures of 0%, 7.7%, 15.6% and 29.3% for scores 0, 1, 2 and 3 respectively.

Conclusion: The gastrostomy score can be used to categorise patients being considered for gastrostomy insertion and to calculate risk of death at 30 days. Further external validation is required.

NATURAL ORIFICE TRANSLUMINAL ENDOSCOPIC SURGERY (NOTES) VERSUS LAPAROSCOPY FOR COLON PERFORATION AND REPAIR: FEASIBILITY AND RISK OF ADHESIONS

J. Morris, J. Romagnuolo, K. Morgan, S. Palesch, R. Hawes. Digestive Disease Center, Medical University of South Carolina, Charleston, USA

Introduction: Natural orifice transluminal endoscopic surgery (NOTES) has evolved as a novel endoscopic approach to surgery with the potential to offer scarless procedures and minimise adhesion formation. Postoperative adhesions are an important cause of morbidity in the USA each year. Case reports suggest that NOTES procedures are feasible but little is known of the pathophysiological response to transluminal procedures.

Aims & Methods: To determine if colon perforation repair is feasible with NOTES and to assess adhesion severity with NOTES transgastric colon perforation repair (TGCR) compared with laparoscopic colon repair (LCR) 40 25 kg male pigs were randomly assigned to either TGCR or LCR. TGCR involved a trangastric needleknife puncture with a double channel therapeutic endoscope, peritoneal cavity entry after balloon dilatation, pneumoperitoneum with CO2 and performance of a 2 cm perforation of the spiral colon with a needleknife, and repair of the perforation with clips. The gastric perforation was then closed with a detachable snare and clips using a purse string technique. Pigs were given colon prep preoperatively and antibiotics during the procedure followed by standard chow the next day. Necropsy was performed at 21 days. Adhesion score was a 9 point composite score of (a) density and vascularity of adhesions (1 = filmy, avascular, 2 = dense or vascular, or 3 = dense and vascular) (b) breadth of adhesions (0 = <4 cm, 3 = >4 cm) and (c) extent of adhesions (1 = 1×2 organ pairs, 2 = 2×2 organ pairs, 3 = 3 or more 2 organ pairs). Final score was a mean of the necropsy surgeon and 2 independent reviewers of necropsy video. Mann-Whitney and Fisher’s exact test were used.

Results: Preliminary composite adhesion scores were analysed for 13 TGCR and 14 LCR pigs. Mean scores were 4.8 (TGCR) and 4.5 (LCR) p = 0.7. Subscores were similar. No pigs had significant procedure-related adverse events. Weight gain and postoperative behaviour were normal in both groups. All colon perforation sites were well healed.

Conclusion: NOTES colonic perforation repair is feasible. Preliminary data show similar adhesion score with TGCR compared with LCR.

Service development free papers

PROVIDING A REGIONAL SMALL BOWEL ENDOSCOPIC SERVICE: THE SHEFFIELD EXPERIENCE

R. Sidhu1, M. E. McAlindon1, D. P. Hurlstone1, M. Thomson2, I. C. Cameron3, D. S. Sanders1. 1Department of Gastroenterology, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Trust, 2Department of Gastroenterology, Sheffield Children’s Hospital, 3Department of Surgery, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Trust, Sheffield, UK

Introduction: The small bowel is an important area for diagnosis, but endoscopic access has historically been limited. Our centre provides push enteroscopy (PE), capsule endoscopy (CE) and double balloon enteroscopy (DBE). However the clinical utility and impact of these modalities on service provision (in the UK) is unknown.

Aims & Methods: To evaluate the diagnostic rates of PE, CE and DBE as well as the impact on patient management, in a tertiary setting in the UK. Data were collected on patients that underwent PE (since January 2002), CE (since June 2002) and DBE (since July 2006). Case notes were reviewed for age, sex, indication of referral, comorbidity, previous investigations and diagnosis obtained including follow-up data. The initial choice of investigation performed depended on the referral indication and the suspected anatomical location of the pathology. The demand for each modality was also assessed since the inception of the service.

Results: Eight hundred CEs, 182 PEs and 30 DBEs have been performed over 70 months. Forty per cent of the referrals originated from outside the region. The commonest indication for all three procedures was obscure gastrointestinal bleeding (OGB). Patients underwent a mean of 4.3 tests prior to CE or DBE. The overall diagnostic yield was 39% for CE, 33% for PE and 43% for DBE respectively. The diagnostic yield was highest for the indication of overt bleeding for both PE (p<0.01) and CE (p<0.001) when compared to other referral indications. Management was altered in 26% of patients that underwent CE, 22% for PE and 43% for DBE respectively. Importantly there were no cases where PE or DBE recognised a lesion that had not already been detected at CE. When comparing CE and DBE, complete small bowel examination was more likely with CE; p<0.001, however, there was no significant difference in the diagnostic yields of the two groups (p = 0.41). DBE allowed intervention in 38% of patients who had a lesion identified by CE. Referrals for either a therapeutic DBE or PE have occurred in 11% of all CE undertaken. There has been a year by year increase in demand for CE. The increasing demand for CE has resulted in a relative decline in the number of diagnostic PEs performed while the demand for DBE is rising particularly for lesions considered beyond the reach of PE.

Conclusion: As part of the regional service, CE should be used as a first-line investigation for the small bowel. PE and DBE should be used as an adjuvant to CE for therapeutic intervention, depending on the location of pathology seen at CE. This ensures efficient use of the small bowel service.

TRANSFUSION IN ACUTE UPPER GASTROINTESTINAL BLEEDING: PRELIMINARY RESULTS FROM THE UK COMPARATIVE AUDIT

S. A. Hearnshaw1, D. Lowe2, R. F. Logan3, S. P. L. Travis1, K. R. Palmer4, M. F. Murphy5. 1Gastroenterology, John Radcliffe Hospital, Oxford, 2CEEU, Royal College of Physicians, London, 3Epidemiology and Public Health, University of Nottingham, Nottingham, 4Gastroenterology, Western General Hospital, Edinburgh, 5NHS Blood and Transplant, John Radcliffe Hospital, Oxford, UK

Introduction: More than 60% of red blood cells (RBC) are now transfused to medical patients. Patients with acute upper gastrointestinal bleeding (AUGIB) are one of the main groups of medical patients using blood (∼14% of all RBC transfusions).1 However, there are no large randomised controlled trials showing transfusion in AUGIB improves outcome. Every transfusion carries risk, and inappropriate transfusion is also a waste of a scarce and costly resource. The BSG guidelines recommend transfusion in patients with a haemoglobin (Hb) <10 g/dl with acute bleeding, or ongoing haematemesis with shock.2 UK transfusion guidelines exist for the use of RBC, platelets and FFP in bleeding patients.35 This comparative audit of AUBIG provides the opportunity to review current practice in these patients where transfusion is common. The audit data will be provided to each participating hospital with recommendations for avoidance of inappropriate transfusion.

Aims & Methods: To audit current transfusion practice in AUGIB against published guidelines, and to investigate the relationships between transfusion and patient outcomes including re-bleeding and death. All UK acute NHS hospitals were invited to participate. Enrolled hospitals identified all new admissions and inpatients presenting with suspected AUGIB between 1/5/07 and 30/6/07. Endoscopy, laboratory and transfusion data were entered online.

Results: Data were received from 6750 patients from 208 UK hospitals treating patients with AUGIB. 43% received RBC transfusion for AUGIB. 10% of all initial transfusions were to haemodynamically stable patients with Hb>10 g/dl, and 5% were given to stable patients with no recorded Hb. FFP was transfused to 503 (7%) patients, 29% of whom had normal coagulation studies. 189 patients received a platelet transfusion, 39% being given to patients with a platelet count >50 ×10*9/l. 23% of patients who received RBC transfusion had ongoing or further bleeding following transfusion. For all Rockall scores, the rate of re-bleeding was lower in patients who did not receive RBC transfusion.

Conclusion: RBC transfusion remains common in patients with AUGIB, with the majority of RBC transfusions being given according to published guidelines. 15% did not appear to be justified. Inappropriate transfusion of platelets and FFP was higher.

30-DAY MORTALITY AUDIT USING A LATE QUALITY AND SAFETY OUTCOME IT SYSTEM

S. Hebbar, E. Nowell, K. Dowler, R. Valori, P. Dunckley. Gastroenterology, Gloucester Royal Hospital, Gloucester, UK

Introduction: Following an endoscopic procedure, the capture of immediate outcomes is relatively easy compared to that of late outcomes. Locally, we have set up an automated late quality and safety outcome IT system to identify patients who have died within 30 days of an endoscopic procedure, thus facilitating the systematic audit of these procedures and their appropriateness.

Aims & Methods: The aim of the audit was to identify the number of cases assigned to the clinical governance review, the number of days from the endoscopic procedure to death and the percentage of deaths in community. The average time to review each case note was also identified. The late quality and safety outcome IT system obtains 30-day mortality data by linking the PAS and Patient Administration System’s (PAS) theatre module, SwiftOp. An endoscopist was assigned to review the notes who, on signing into the system, can open a data entry screen for each case which displays full details of the endoscopic procedures, into which the outcomes and comments from the notes review are entered. Date of procedure and death and any adverse events are also displayed in the system. Cases where it was determined that the procedure was inappropriate or contributed to death, or where the reviewer was unsure about either of these, were assigned to a clinical governance review.

Results: Between January 2006 and October 2007, 17 744 endoscopic procedures were performed and 341 patients died within 30 days. 138 notes have been reviewed to date. The average time to review each notes was 3 min. The average number of days to death from the time of endoscopic procedure was 14.53. 25 patients (18.1%) died in the community. Nine (0.05% of total endoscopic procedures) cases are assigned to clinical governance review, of which 7 cases possibly and 2 (0.01% of total endoscopic procedures) cases definitely contributed to death. The deaths in all the 9 cases have been within 8 days of procedure and all died within the hospital.

Conclusion: The late quality and safety outcome IT system is user-friendly and quick. The audit has identified 2 cases whereby the endoscopy directly led to patient death. These cases will be reviewed as part of the endoscopic clinical governance framework in order to further enhance the quality and safety of our endoscopic practice. The 9 cases where the endoscopic procedure possibly or definitely contributed towards death all died within 8 days of the procedure and within the hospital. Although this is only a small number of cases, these initial data suggest that reducing the late outcome mortality audit data from 30 to 10 days, in patients who have died in hospital only, may still capture the majority of late deaths and reduce the workload associated with collecting this data.

CAN WE ACHIEVE THE 18 WEEK PATHWAY TARGET? EFFECT OF DIFFERING PATIENT PRESENTATIONS

P. Sakellariou, D. A. Elphick, G. Toy, D. S. Sanders, D. P. Hurlstone, M. E. McAlindon, A. J. Lobo. Gastroenterology Unit, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Investigation of out-patients with a wide range of presenting features is a major part of a gastroenterologist’s workload. Few data exist on the extent to which different units investigate similar presentations. This is of critical importance in provision of an 18 week (126 day; from referral to initiation of treatment) patient pathway—a key objective for the NHS.

Aims & Methods: This study aimed to quantify the investigative load of newly referred patients to the Gastroenterology clinic of a single University Teaching Hospital and to evaluate performance against the local Trust’s 18 week “clock rules”. 306 new out-patients were reviewed by 4 consultant teams over a 62-day period (16/08/0616/10/06). All case notes were reviewed. Referral category, main and additional symptoms, time to first appointment and outcome at 18 weeks were recorded. Time and reason for stopping the clock were identified.

Results: Median age (range) was 56 (18–94) years and 61% were female. Referrals were from GPs (240, 78%) or another consultant (66, 22%). 77 (25%) referrals were through the “suspected cancer” system. Median time to first appointment was 56 (1–192) days. The commonest main symptoms were diarrhoea (74, 24%), dyspepsia (43, 14%), iron deficiency anaemia (IDA) (53, 17%). At 18 weeks the clock had been stopped in 164 (54%) patients while 60 (19%) were undergoing further investigation, 53 (16.7%) awaiting treatment to be agreed, 22 (8%) awaiting first appointment, 6 (2%) referred to another specialty and 1 (0.3%) had a new clock started. The clock had been stopped for 268 (87%) patients with a median time 111 (7–499) days (68 (25%) offered treatment; 112 (43%) decision not to treat; 20 (7%) declined treatment or investigation; 68 (25%) commenced active monitoring). With administrative changes, the pathway could have been accelerated to stop the clock at median 77 (7–499) days compared to actual time (p<0.01, Wilcoxon test). Median time to stopping the clock was 41 (11–249) days for the “suspected cancer” pathway, and 130 (7–499) days for the remainder. Those with diarrhoea had median 2 (1–6) further investigations (dyspepsia 2 (0–5); IDA 3 (0–5)). Median time to stopping the clock for patients with diarrhoea was 120 (13–491) days (dyspepsia 78(7–499); IDA 103 (12–494)) and proportion of patients with completed investigation by 18 weeks was 42% (diarrhoea), 70% (dyspepsia) and 47% (IDA).

Conclusion: The investigative practice of one hospital unit is described allowing comparison with others and benchmarks to be set. To achieve the March 2008 milestone with 90% of patients having received initial treatment by 18 weeks from referral, our audit highlights the requirement for a transformation of services and for the complexity of investigative pathways to be considered in delivering that objective.

USING THE 13C-UBT TO PROMOTE NICE GUIDELINES FOR SIMPLE DYSPEPSIA: AN 8-YEAR FOLLOW-UP COHORT STUDY

A. J. Norman, S. Din, N. J. Taylor, S. Swain, R. P. H. Logan. Department of Gastroenterology, King’s Mill Hospital, Mansfield, UK

Introduction: The 2004 NICE guidelines advise that in patients with dyspepsia, gastroscopy (OGD) should be reserved for patients with alarm symptoms. However there is continued demand for OGD, often as a means of providing reassurance. This study evaluates the long-term effectiveness and safety of an alternative management pathway in minimising the use of OGD in patients with uncomplicated dyspepsia.

Aims & Methods: Observational cohort study of patients 65 years old and under with uncomplicated dyspepsia routinely referred to a district general hospital between 11/08/98 and 1/11/03 with follow-up to 1/11/07. Hospital activity and pathology database analysis for death, cancer diagnosis, OGD (out-patient and in-patient) and case note review. Ascertainment bias (for OGDs performed elsewhere) was evaluated by review of GP records in a subset of referred patients (n = 30). The two-step pathway was a 13C-UBT and a subsequent 20 min out-patient consultation in all patients with a negative result.

Results: Between 11/08/1998 and 28/10/2003, 242 patients (median age 41 years, range 14–65 years, 130 male, 112 female) with uncomplicated dyspepsia were referred, had a negative 13C-UBT and were reviewed as an out-patient. The clinical diagnoses were GORD (74%) and non-ulcer dyspepsia (14%). During the period of follow-up (median 78 months, range 45 to 111 months), 39 patients had an OGD either within one year (n = 15; indication for OGD: severe/atypical symptoms (n = 7), pre-fundoplication (n = 4)) or subsequently (n = 24, 24/227 (11%), median follow-up 33 months, range 14–92 months). In the latter group, a normal OGD was the most common endoscopic diagnosis (n = 20) followed by oesophagitis (n = 6). During a total follow-up of 1593 patient years, there were 3 deaths (lung cancer and LVF. One patient’s cause of death could not be found). From a subset of GP records (n = 30), only one patient had an OGD performed elsewhere.

Conclusion: These data show that uncomplicated dyspepsia can be safely managed without performing OGD in up to 90% of cases referred to secondary care. The alternative pathway exploits the reassurance value of a negative 13C-UBT reinforced by out-patient clinic review. This strategy may allow endoscopy resources to be used more appropriately.

A TEMPLATE PROFORMA AND OPERATIVE INFORMATION IMPROVES THE QUALITY OF HISTOPATHOLOGY REPORTING IN COLORECTAL CANCER

P. N. Siriwardana1, A. Pathmeswaran2, J. S. D. E. Hewavisenthi3, K. I. Deen1. 1University Surgical Unit, North Colombo Teaching Hospital, 2Department of Community Medicine, 3Department of pathology, Faculty of Medicine, Ragama, Sri Lanka

Introduction: The histopathology report is vital to determine the need for adjuvant therapy and prognosis in colorectal cancer (CRC).1 We and others have highlighted the inadequacy of histopathology reports in paragraph prose.2 3 To improve the completeness and quality of histopathology reports, the Royal College of Pathologists (RCP), UK has formulated a proforma to include a minimum dataset required for accurate assessment of a colorectal cancer specimen.4

Aims & Methods: The aim of this prospective study is to assess the robustness of the histopathology report following introduction of a template proforma based on RCP, UK guidelines, which had 19 items for rectal cancer (RC) and 15 items for colon cancer (CC). A total of 150 reports were evaluated; 68 consecutive histopathology reports in proforma format were prospectively analysed and compared with 82 consecutive histopathology reports in text format. The percentage of reports containing a statement for each data item in the dataset in template proforma format was obtained and compared with the series in paragraph format. Completeness (%) of each report classified <80% or >80%.

Abstract 018 At least 80% complete

Results: Eleven data items common to CC and RC had a comment in all proformas compared to only 2 in text format reports. The completeness of reports in text and template format is shown in table 1. Except for comments on presence of “histologically confirmed liver metastases” in CC and RC, “distance from dentate line” (information provided by surgeon) and “distance to circumferential margin” in RC, all other items were reported in more than 90%, where 71% of the items based on the minimum dataset were present in all. Mean percentage completeness of all reports in template format for RC was 91% (SD±4%) and 99% (SD±1%) for CC, whilst a text report had completion rates of 74% (SD±8%) and 81% (SD±5%) respectively.

Conclusion: A checklist of report items improves the quality of the histopathology report in CRC which can be further enhanced by the surgeon’s contribution of operative findings.

PRELIMINARY RESULTS OF THE UK COMPARATIVE AUDIT OF ACUTE UPPER GASTROINTESTINAL BLEEDING: ORGANISATION OF CARE

S. A. Hearnshaw1, D. Lowe2, R. F. Logan3, M. F. Murphy4, S. P. L. Travis1, K. R. Palmer5. 1Gastroenterology, John Radcliffe Hospital, Oxford, 2CEEU, Royal College of Physicians, London, 3Epidemiology and Public Health, University of Nottingham, Nottingham, 4NHS Blood and Transplant, John Radcliffe Hospital, Oxford, 5Gastroenterology, Western General Hospital, Edinburgh, UK

Introduction: Prompt assessment, investigation, resuscitation (including appropriate transfusion), and endoscopic therapy are required to manage optimally patients with acute upper gastrointestinal bleeding (AUGIB). BSG guidelines state that patients with major bleeding should undergo urgent endoscopy, and that therapeutic endoscopy should be provided by skilled endoscopists 24/7.1 2 A dedicated space on routine endoscopy lists should be provided daily for patients presenting with AUGIB.3

Aims & Methods: To survey the current organisation of care for patients admitted with AUGIB against published guidelines. To examine the availability of emergency endoscopy, endoscopic therapies, and skilled medical and nursing staff in hospitals admitting patients with AUGIB. Organisational questionnaires were sent to the clinical lead for endoscopy in the 217 hospitals that participated in the audit.

Results: 190 hospitals provided organisational data. 91% provide level 2 care beds on site, with 15 hospitals having a dedicated AUGIB unit. 99% provide endoscopy on site, and 89% have out of hours endoscopy. 99% have facilities for endoscopic therapy in patients with AUGIB, with 57% ensuring a dedicated slot for emergency endoscopy. Only 55% of hospitals have a consultant on call rota for out of hours emergency endoscopy; 75% of these have 3–8 consultants on the rota. 41% of hospitals have an endoscopy nurse assistance on-call rota. In 14% of sites, out of hours emergency endoscopy is provided by unsupervised junior doctors. 5% of hospitals treating patients with AUGIB do not have a transfusion laboratory on site. 82% of hospitals have written guidelines for the management of AUGIB in the hospital, and 85% audit their practice.

Conclusion: There is wide variation in service provision for patients with AUGIB in the UK, with many hospitals not having a named consultant on call for endoscopy out of hours. The use of out of hours emergency endoscopy and its appropriateness needs to be reviewed both at national and local levels, and will be evaluated in this audit. The impact of the current service provision on patient outcomes needs to be assessed, before recommendations of “best practice” can be made.

“STRAIGHT TO TEST” UPPER GASTROINTESTINAL CANCER SERVICE: EFFICIENCY, SAFETY AND COST EFFECTIVENESS

R. Madhotra1, P. Steer2, A. Ignjatovic2, C. Akubuine2. 1Gastroenterology, Milton Keynes Hospital Foundation Trust, Milton Keynes, 2Gastroenterology, Milton Keynes General Hospital, Milton Keynes, UK

Introduction: Rapid access to diagnosticvs is essential for early diagnosis of upper gastrointestinal (GI) cancer. In the UK, Cancer Plan 2000 requires all suspected cancer patients to be seen in two weeks and diagnostics must be completed within 31 days. These targets put huge service delivery challenges for most NHS hospitals with limited resources. We, therefore, started a “Straight to Test” STT service to reduce access times, and rapid diagnosis and staging suspected cancer patients. There were no extra cost for setting up the service. The patients were followed up for 6 months after endoscopy by telephonic interview and/or review of medical records.

Aims & Methods: We conducted a prospective audit of the new STT service over the first 12 months to study the efficiency, safety and cost effectiveness. All consecutive patients referred by the GPs for urgent cancer assessment were studied. Data were collected to include the demographics, presenting symptoms, endoscopy findings, follow-up plan and 6-month outcome. Cost savings were calculated using Outpatient Mandatory Tariff Coding system.

Results: A total of 241 patients (116 males, 125 females) were studied. Ninety five per cent patients had endoscopy within two weeks of referral. Majority had multiple symptoms. the commonest presenting symptoms were dyspepsia (68%), dysphagia (48%) and weight loss (40%). The endoscopy diagnoses were gastritis (32%), hiatus hernia (26.9%), oesophagitis (16.8%), duodenitis (12%) and gastric ulcers (2.4%). Cancers were detected in 27 (11.2%) patients. Cancers included 11 oesophageal, 4 gastric, 3 pancreas, 2 liver and 3 colon. One hundred and seven (44.3%) patients were discharged back to the GP after endoscopy, 134 (55.7%) were followed up in hospital. Excluding cancer patients there were no significant difference in presenting symptoms, endoscopy findings and clinical outcomes at 6 months in the two groups. No missed cancers were detected at 6-month follow-up. STT service saved 107 new patient appointments (£188 each) and reduced 134 follow-up appointments (£88 each) for the remaining. The net saving was £32,576.

Conclusion: Our audit shows that STT service for upper GI cancer assessment is efficient, safe and highly cost effective. The audit also shows that more patients could be safely discharged to their GPs, hence there is scope more cost saving. This model of cancer assessment service in the NHS has potential for widespread use in upper GI as well as other cancers.

LIVER HISTOPATHOLOGY IN THE YORKSHIRE REGION: A NETWORK MODEL

J. I. Wyatt1, G. Reall1, R. Jones2. 1Histopathology Department, 2Hepatology, St James’s University Hospital, Leeds, UK

Introduction: The national plan for liver services UK 2004 envisages that “each hepatology centre should have access to a specialised liver pathologist. Ideally this would be in the site of the lead centre providing expertise and training for the local pathology departments within the managed clinical network.”1 This arrangement has evolved informally in West and North Yorkshire.

Aims & Methods: The aim of this study was to investigate the volume and provision of liver histopathology in our region originating outside Leeds, as a potential model for a hepatopathology network. A questionnaire was sent to all pathologists in west and north Yorkshire, covering 11 hospital trusts, referring into the main centre at Leeds. It covered departmental statistical data on liver biopsy workload, and attitudes of individual pathologists to reporting liver biopsies. Individual attitudinal responses to statements were on a scale of 1 (strongly disagree), 3 (neutral), to 5 (strongly agree).

Results: Replies were received from all 11 hospitals: 23/31 (74%) pathologists in non-specialised DGH departments and 3/4 liver pathologists in larger specialised departments (Bradford, Hull). The DGH departments receive on average 70 biopsies/year (range 30–150; 67% medical, 33% for tumour) representing 0.55% (0.33–1.1%) histology requests. Reporting is shared by all 2–6 pathologists/department. On average estimated 35% (5–100%) cases are discussed with the clinician during a CPC (5/9) and/or by phone (5/9). About 10% biopsies are reviewed in Leeds. The two larger departments receive 120 and 170 biopsies/year reported by 2 named subspecialist pathologists out of the 10–11 consultants/department. <5% biopsies are reviewed in Leeds. Of 23 DGH pathologists expressing a view (>16 not neutral for each question), >90% enjoyed liver work, often read up about biopsies, and discussed them with their colleagues. 94% said they would like to do liver CPD, 62% had done liver CPD in the last 3 years. 7/23 would like to refer more biopsies centrally. Only one said they would prefer not to do any liver work.

Conclusion: In Yorkshire, liver biopsies account for 0.5% DGH histopathology specimens. Most pathologists enjoy the challenge of this work, and would like more opportunity for CPD. A minority would like a more centralised service. This indicates support for a more formalised network, providing support, training and guidelines to pathologists to work with their local clinical hepatologist.

A DECADE AFTER ROCKALL: A 12-MONTH SINGLE-CENTRE STUDY OF CLINICAL OUTCOME IN NON-VARICEAL UPPER GASTROINTESTINAL BLEED

R. Arulraj1, I. Khan1, S. Pathmakanthan1, K. Kane1, R. Boulton1, M. Hallissey2, J. Goh1. 1GI Unit, 2Surgery, University hospital Birmingham NHS Trust, Birmingham, UK

Introduction: A decade following the publication of the Rockall risk score have seen significant medical and endoscopic advances in treatment of non-variceal upper gastrointestinal (GI) bleed (NVGIB) and also a change in the epidemiological pattern of NVGIB. Many gastroenterologists feel there has been a progressive fall in the incidence of bleeding and endoscopic therapy is useful in control of bleeding.

Aims & Methods: We report a 12-month (2006) study of clinical outcome of patients presenting to the emergency department in a single centre with a 24/7 consultant-led GI bleed service. All patients admitted with acute NVGIB were included (GI bleed in our tertiary services including transplant units were excluded). Case/electronic record of all patients were reviewed and discussed at weekly bleed audit meeting.

Results: 202 patients were identified as true NVGIB (mean age 67±18; M:F = 125:77). Medication: Aspirin (n = 54), Clopidogrel (2), NSAID (22) and Warfarin (15). Endoscopy was performed within 24 h in 90 (45%; mean Rockall: pre 2.9 post 4.6); 24–48 h in 46 (23%; mean Rockall pre 2.3 post 3.7) and >48 h in 59 (29%; mean Rockall pre 2.3 post 3.5). 23 (11%) were performed out-of-hours (Rockall pre 3.2; post 5.8). The top 4 diagnoses were duodenal ulcer (56; 28%), gastric ulcer (32; 16%), oesophagitis (32; 16%) and gastritis (32; 16%). In total, 15 (7%) underwent emergency surgery. The overall mortality was 9% (18/202). Endoscopic interventions were carried out in 35 (18%; Rockall pre 3.4 post 6.1). Mortality in the intervention and non-intervention groups were 5.7% (2/35) versus 8.9% (15/169) respectively (p = NS; χ2 test).

Conclusion: NVGIB remains a significant workload for gastroenterologists both in and out of hours and is still associated with significant morbidity and mortality. The Rockall score continues to be useful in risk stratification despite the shift in epidemiological pattern from Helicobacter to increasing anti-platelet therapy usage.

BSG pancreatic/radiology section joint symposium

A COMPARISON OF CAPSULE ENDOSCOPY WITH SMALL BOWEL MRI IN THE MANAGEMENT OF PEUTZ-JEGHER’S SYNDROME

A. J. Postgate1, A. Fitzpatrick1, D. Burling2, A. Gupta2, C. Fraser1. 1Wolfson Unit for Endoscopy, 2Intestinal Imaging Department, St Mark’s Hospital, London, UK

Introduction: Capsule endoscopy (CE) is more accurate and better tolerated than barium follow through x ray for small bowel polyp surveillance in patients with Peutz-Jegher’s Syndrome (PJS).1 2 However polyp size estimation and localisation is difficult at CE. Small bowel MRI is an evolving modality which may have a role for this indication.

Aims & Methods: We compared the accuracy of CE with MRI follow through for the detection of significant (>10 mm) small bowel polyps in adults with PJS. Study patients underwent an MRI and a subsequent CE. Each test was reported by different reviewers, blinded to the results of the other modality. A visual-analogue scale documenting patient comfort was completed. The results were confirmed at double balloon enteroscopy (DBE) or operative enteroscopy (IOE) when clinically appropriate.

Abstract 023 Number of polyps identified by CE and MRI according to size

Results: An interim analysis of 9 patients was performed. 705 polyps of all sizes were identified by both modalities: 681 by CE and 24 by MRI (table). CE identified significantly more diminutive polyps (6 mm or less) than MRI (643 vs 2, p = 0.008), and more 6–10 mm polyps (24 vs 8, p = 0.07). Six patients had significant (>10 mm) polyps: 5 were identified at MRI and CE, and 1 at MRI only. Each modality identified 14 significant polyps. To date confirmation of the results by DBE has been performed in 1 patient (no 1, table 1), at which 3 significant polyps were noted: two 12 mm polyps in the jejunum, and one 22 mm polyp in the proximal ileum. CE identified a 15 mm and 20 mm polyp in the jejunum and two 12 mm polyps in the ileum, while MRI found a 16 mm polyp in the jejunum and a 13 mm polyp in the ileum. Patients found CE more comfortable than MRI (78% vs 52%) but of equal convenience (77% vs 77%). Patient preference was split between the two tests.

Conclusion: CE is superior for identifying small (10 mm or less) polyps but MRI is comparable in performance for clinically significant (>10 mm) polyps. Further data from DBE or IOE are required to examine which modality most accurately localises and sizes significant polyps. Although patients rate CE as more comfortable than MRI, neither test was preferable overall. In conclusion, MRI follow through is a promising alternative to CE for small bowel polyp surveillance, and it is likely that the two modalities will complement each other in the management of difficult cases prior to organising invasive polypectomy.

Neurogastroenterology/motility free papers

SYMPATHETIC AND PARASYMPATHETIC SYSTEM CO-ACTIVATION UNDERLIES HEART RATE INCREASE IN BOTH HUMAN VISCERAL AND SOMATIC PAIN

P. Paine1, J. Kishor2, L. Gregory1, Q. Aziz2. 1Department of Gastroenterology and Translational Imaging Unit, Manchester University, Manchester, 2Wingate Institute of Neurogastroenterology, Queen Mary University of London, London, UK

Introduction: Somatic pain animal models have recently identified novel, seemingly paradoxical, sympathetic (SNS) and parasympathetic (PNS) co-activation in pain evoked heart rate (HR) increase. It is not known whether this can be seen in humans or for visceral pain.

Aims & Methods: To compare brainstem selective autonomic nuclei outputs for visceral and somatic pain. Methods: 18 healthy volunteers (2 males), mean age 35.4 (range 18–59 years) underwent serial nail bed (NB) pressure stimuli; 16 of whom also had serial oesophageal balloon (O) distension. Both stimuli were delivered to pain tolerance threshold. In addition to HR, selective cardiac PNS (cardiac vagal tone (CVT)) and SNS responses including cardiomotor (Cardiac Sympathetic Index (CSI)); sudomotor (skin conductance (SCR)) and vasomotor (mean blood pressure (MBP)) were acquired pre- and post-stimulation using a novel beat to beat technique.

Results: There were comparable ratings between sites of pain intensity (O = 7.9 + 0.2, NB = 8.1 + 0.2; p>0.05) and unpleasantness (O = 7.9 + 0.2; NB = 7.8 + 0.3; p>0.05). HR increased during both somatic (71.02 + 2.1 bpm pre-stimulus; 73.9 + 2.05 post-stimulus, p<0.05) and visceral pain (72.3 + 2.05 bpm pre-stimulus, 76.9 + 2 bpm post-stimulus, p<0.05). This was accompanied by CVT increases for both somatic (pre-stimulus 8.11 + 1.06, post-stim 8.50 + 1.12, p<0.05) and visceral pain (pre-stimulus 8.15 + 1.4, post-stimulus 8.84 + 1.32, p<0.05). CSI also increased for visceral (pre-stimulus 2.38 + 0.19; post-stimulus 3.33 + 0.31, p<0.05); p<0.01) and somatic (pre-stimulus 2.36 + 0.15, post-stimulus 2.91 + 0.26) pain. SCR increased for somatic (pre-stim 6.86 + 0.97 uS, post-stim 8.35 + 1.05, p<0.05) and visceral (pre-stimulus 6.25 uS + 0.96, post-stimulus 7.32 uS+1.08, p<0.05) pain. MBP fell however: somatic (pre-stim 99.1 + 2.5 mmHg, post-stim 95.5 + 2.24, p<0.05) visceral (pres-stim 95.8 + 3.5 mmHg post-stim 91.2 + 4.3, p<0.05). The pain related change was greater for SCR for NB than O (p<0.05) but there were no differences otherwise in magnitude of pain-related changes between sites.

Conclusion: For the first time, co-activation of SNS and PNS has been shown to accompany pain evoked HR increases in humans for somatic and visceral pain. This confirms recent similar findings in animal studies. The ANS is not a simple reciprocal system but its subcomponents are capable of nuanced patterned responses in pain. Visceral and somatic pain responses were qualitatively similar.

ANXIETY INDUCTION IS ASSOCIATED WITH GREATER SENSITISATION TO OESOPHAGEAL ACID IN A HUMAN MODEL OF VISCERAL PAIN HYPERSENSITIVITY

A. Sharma1, L. Van Oudenhove2, P. Paine1, Q. Aziz3. 1GI Sciences, Hope Hospital, Salford, UK, 2Department of Psychiatry, University Hospital Gasthuisberg, Catholic University of Leuven, Leuven, Belgium, 3Wingate Institute of Neurogastroenterology, Barts and the Royal London, Queen Mary University, London, UK

Introduction: Patients with stress or anxiety at the time of gastrointestinal injury/inflammation (for example, gastroenteritis) have a higher risk of developing subsequent functional gastrointestinal disorders (FGID). Patients with these disorders often display a heightened sensitivity to experimental gut stimulation, termed visceral pain hypersensitivity (VPH). We have previously shown that distal oesophageal acidification induces sensitisation of spinal dorsal horn neurones leading to the development of VPH in the non-acid exposed proximal oesophagus (PO).1 How anxiety modulates this VPH is unknown.

Aims & Methods: To determine whether anxiety induction affects the magnitude of acid-induced oesophageal pain hypersensitivity. 16 healthy volunteers (10 female, age 22–57 years, mean 35.7 years), were studied on two occasions. Pain thresholds (PT) to electrical stimulation (in milliamperes, mA) were determined in the PO and foot (somatic control) pre- and post a 30-min infusion of 0.15M HCl. During the infusion, all subjects randomly received a 10 min cognitive mood induction paradigm combining autobiographical recall of an anxiety-inducing life event with presentation of fearful faces or a neutral life event with neutral faces in a cross-over design. A Spielberger state anxiety inventory (SSAI) was recorded before and after the acid infusion. Subjects also completed unpleasantness ratings on a 10 cm Likert scale before, during and after the infusion.

Results: The mean change in SSAI score with infusion was significantly greater for anxiety versus neutral conditions (21.7 (12.6 SD) vs 3.6 (6.4 SD), p<0.0001 Wilcoxon). Anxiety induction increased unpleasantness scores compared to neutral (2.5 (2.0 SD) vs 0.63 (1.2 SD), p = 0.008 Wilcoxon). There was a greater reduction in PT in the PO during anxiety compared to neutral condition (−4.5 mA (6.4 SD) vs −2.1 mA (6.2 SD), p = 0.05 Wilcoxon). The mean maximum change in PT was also greater with anxiety than neutral mood induction (−7.6 mA (6.7SD) vs −3.8 mA (5.8SD), p = 0.02 Wilcoxon). There was no difference in the mean change in foot PT between the conditions (p = 0.14, Wilcoxon).

Conclusion: Combining anxiogenic autobiographical recall with fearful faces increases subjective anxiety, and unpleasantness scores during oesophageal acidification and the magnitude of subsequent visceral sensitisation. These findings provide objective evidence for the role of anxiety in the development of prolonged VPH, and provides a model for studying the biological reasons for the effect observed and will also allow testing of therapeutic strategies.

FAECAL LACTOFERRIN: A NOVEL TEST TO DIFFERENTIATE BETWEEN IRRITABLE OR INFLAMED BOWEL?

R. Sidhu1, L. Foye2, S. Morley2, P. Wilson2, A. C. Wright1, A. J. Lobo1, M. E. McAlindon1, D. S. Sanders1. 1Department of Gastroenterology, 2Department of Clinical Chemistry, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Trust, Sheffield, UK

Introduction: Irritable bowel syndrome (IBS) and inflammatory bowel disease (IBD) may not always be discriminated by using symptoms alone. Thus invasive tests may be performed. Lactoferrin is a glycoprotein expressed by activated neutrophils. Previous studies have demonstrated the presence of faecal lactoferrin (FEL) in patients with IBD, coeliac disease and infective diarrhoea. It has been proposed as a non invasive test to screen patients with gastrointestinal symptoms for the presence of inflammation. However there is paucity of data on the use of FEL in IBS.

Aims & Methods: To investigate the clinical utility of faecal lactoferrin (FEL) as a marker of GI inflammation when assessing patients with Rome II IBS, IBD and healthy controls. Patients with Rome II IBS and IBD were invited to participate in this study. Disease activity in IBD patients were assessed using the modified Harvey-Bradshaw Activity Index (HBI). Healthy volunteers acted as controls. Stool samples were analysed using an ELISA assay (Techlab). A two-tailed t test was used to compare means between groups and Kendall-Tau correlation calculations were performed between lactoferrin concentrations and disease activity.

Results: Eighty seven patients with IBS, 55 patients with IBD: 33 ulcerative colitis (UC) and 22 Crohn’s disease (CD) and 98 healthy volunteers were recruited. The mean +/− SE FEL concentration (μg/g faecal weight) was 1.63 +/− 0.44 for IBS patients, 39.3 +/− 16.6 for UC patients, 35.4 +/− 10.7 for CD patients and 2.43 +/− 0.72 for healthy controls. FEL levels were significantly higher in IBD patients compared to IBS patients (p = 0.001) and healthy controls (p = 0.002). There was no significant difference in mean FEL concentration between UC and CD patients (p = 0.8) and between IBS and healthy controls (p = 0.34). However, the mean FEL concentrations were significantly higher in active UC patents compared to inactive UC patients (mean FEL: 100 +/− 40.9 vs 4.67 +/−1.2, p = 0.04). This comparison was not significant for Crohn’s patients (p = 0.1). The correlation coefficient between disease activity and FEL concentrations for UC patients were 0.54 (p = 0.0001) and for CD patients was 0.4 (p = 0.02).The sensitivity and specificity of the FEL assay for distinguishing active IBD from IBS/healthy controls were 90.5% and 95.7% respectively. The positive and negative predictive values were 70.4% and 98.9%.

Conclusion: FEL is an inexpensive and non invasive test that can provide the clinician with a marker to differentiate between IBD and IBS. These preliminary data suggest that FEL could be used as a non-invasive marker for determining whether patients who have IBS should have endoscopic investigations.

REVERSING A VIRTUAL LESION IN HUMAN SWALLOWING MOTOR CORTEX WITH CORTICAL STIMULATION: A MODEL FOR DYSPHAGIA AFTER STROKE?

S. Jefferson1, S. Mistry1, L. Hancock1, J. Rothwell2, S. Hamdy1. 1GI Sciences, University of Manchester, Salford, 2Sobell Department of Neurophysiology, Institute of Neurology, UCL, London, UK

Introduction: We have previously shown that transcranial magnetic stimulation (TMS) delivered at 1 Hz to pharyngeal motor cortex can unilaterally suppress cortical excitability (virtual lesion) which can transiently alter swallowing behaviour.1 Here we examine if an excitatory intervention (5 Hz repetitive TMS (rTMS)) applied to the contralesional hemisphere can reverse a virtual lesion as a prelude to applying such neurostimulation in the treatment of dysphagia after stroke.

Aims & Methods: Healthy volunteers (n = 12, 5 male, mean age 39 years) underwent single pulse bi-hemispheric TMS measurements of cortico-pharyngeal excitability via a swallowed intraluminal catheter, before and for 60 min after 10 min of 1 Hz rTMS delivered to unilateral pharyngeal motor cortex. Five Hz rTMS (250 pulses over 90 seconds) or sham was then applied to the contralesional hemisphere randomised to different sessions, 1 week apart. Post-lesional cortical excitability following active intervention vs. sham was compared using ANOVA.

Results: As expected, sham rTMS had no effect on the suppression of excitability induced by the virtual lesion, with both lesional and contralesional hemispheres decreasing by a maximum of −35% and −16% respectively. By contrast, active 5 Hz rTMS completely abolished the virtual lesion, reversing the direction of excitability both in the lesional and contralesional hemispheres by +35% and +25%, respectively (p = 0.035, fig).

Abstract 027

Conclusion: Fast rTMS applied to the contralesional hemisphere completely reverses a unilaterally induced virtual lesion in pharyngeal motor cortex. These data support the notion that rTMS might be usefully applied in stroke patients, as a therapeutic intervention for dysphagia.

AUTONOMIC EFFECTS OF OESOPHAGEAL ACIDIFICATION AND THEIR RELATIONSHIP TO THE DEVELOPMENT OF HUMAN VISCERAL PAIN HYPERSENSITIVITY

A. Sharma1, P. Paine1, Q. Aziz2. 1GI Sciences, Hope Hospital, Salford, 2Wingate Institute of Neurogastroenterology, Barts and the Royal London, School of Medicine and Dentistry, Queen Mary University, London, UK

Introduction: Visceral pain hypersensitivity (VPH) is a common feature of functional gastrointestinal disorders (FGD). We have shown previously that distal oesophageal acidification induces sensitisation of spinal dorsal horn neurones leading to the development of VPH in the non-acid exposed proximal oesophagus (PO).1 Variability exists in the degree to which individuals sensitise but factors that modulate this are unknown. As the parasympathetic (PNS) and sympathetic nervous systems (SNS) have been proposed as pro-nociceptive and anti-nociceptive respectively we decided to study their influence on sensitisation in our model.

Aims & Methods: To determine whether differential autonomic reactivity predicts the degree of VPH to oesophageal acidification. Methods In 25 healthy volunteers, pain thresholds (PT) to electrical stimulation (in milliamperes, mA) were determined in the PO pre- and post a randomised double-blind 30-min distal oesophageal infusion of either 0.15M HCl or saline. All subjects had autonomic monitoring before and during the infusion. Selective PNS (cardiac vagal tone (CVT) and cardiac sensitivity to baroreflex (CSB)) and SNS (cardiac sympathetic index (CSI)) measures were derived.

Results: There was a significant difference between the mean change in PT in the PO between acid (−3.9 mA (5.5 SD)) and saline 2.6 (4.9 SD)) suggesting the development of VPH with acid (p<0.0001 Wilcoxon, n = 25). Of the 25, 16 had a fall in PO PT of ⩾5 mA with acid and thus sensitised. End-infusion heart rate (HR) and mean arterial pressure (MAP) were significantly higher with acid than saline (73.4 (9.9 SD) vs 67.7 (7.6 SD), p = 0.0001 for HR, paired t test) and (100.4 (13.4 SD) vs 95.0 (13.5 SD), p = 0.0007 for MAP, paired t test). End-infusion CVT and CSB were significantly lower with acid than saline ((p = 0.01, Wilcoxon; CVT) and (p = 0.0038, Wilcoxon; CSB). End-infusion CSI was greater for acid than saline (p = 0.0016, paired t-test). In the 16 sensitisers, the degree of sensitisation was related to increasing heart rate (r = −0.66, 95% CI −0.87 to −0.25, p = 0.005), and fall in CVT (r = 0.54, 95% CI 0.06 to 0.82, p = 0.031) and CSB (r = 0.54, 95% CI 0.06 to 0.82, p = 0.030). A trend was also apparent that increasing CSI might be associated with increasing sensitisation (r = −0.43, 95% CI −0.76 to 0.08, p = 0.09).

Conclusion: Oesophageal acidification is associated with SNS activation and PNS withdrawal. In turn greater PNS withdrawal is associated with greater sensitisation. These data suggest the PNS may have anti-hyperalgesic properties in the human viscera. Future studies using pharmacological modulation of the autonomic nervous system may be beneficial in modulating VPH.

Plenary session

PRELIMINARY RESULTS OF THE UK COMPARATIVE AUDIT OF OVER 6500 PATIENTS WITH ACUTE UPPER GASTROINTESTINAL BLEEDING

S. A. Hearnshaw1, D. Lowe2, R. F. Logan3, M. F. Murphy4, S. P. L. Travis1, K. R. Palmer5. 1Gastroenterology, John Radcliffe Hospital, Oxford, 2CEEU, Royal College of Physicians, London, 3Epidemiology and Public Health, University of Nottingham, Nottingham, 4NHS Blood and Transplant, John Radcliffe Hospital, Oxford, 5Gastroenterology, Western General Hospital, Edinburgh, UK

Introduction: In 1993, Rockall et al reported a large audit of patients presenting with acute upper gastrointestinal bleeding (AUGIB) to 4 health regions in England.1 Mortality overall was 14%, with in-patient bleeding mortality 33%. A risk-score developed from this audit (Rockall score) has been shown to predict re-bleeding and death.2 Guidelines recommend early endoscopy for those at high risk.3 Considerable changes have occurred in the presentation and management of AUGIB in the 14 years since Rockall. These include differences in demography, resuscitation, endoscopic therapy and use of specific drugs. At the same time there is concern that inadequacies in service provision may have adversely affected outcomes in some hospitals.

Aims & Methods: To audit current care in UK hospitals for managing AUGIB against published recommendations.3 To look at aetiological factors, diagnoses and mortality in AUGIB, and to audit the use of risk stratification and early endoscopic therapy in high risk patients. All UK acute NHS hospitals were invited to participate. Enrolled hospitals identified new admissions and in-patients presenting with suspected AUGIB between 1/5/7 and 30/6/7. Endoscopy, laboratory and transfusion data were entered online.

Results: 6750 patients were included for analysis from 208 hospitals; 82% new admissions and 16% in-patients. 37% of patients were aged <60 yrs and 28% >80 years with mean (SD) age 64.4(19.7) years. 50% had at least one comorbidity, with 9% having documented cirrhosis. 29% were taking aspirin, 12% an NSAID. 26% had a history of alcohol excess. Risk assessment scores were calculated by hospitals for 19% of cases. Scores calculated from submitted data showed 28% had final Rockall score 0–2, 44% score 3–5, 24% score 6–8 and 3% score >8. 50% had endoscopy within 24 h of presentation, including 50% of those with a pre-endoscopy Rockall score of >2. Varices or portal hypertensive gastropathy were found in 12%; peptic ulcer disease was the commonest diagnosis (36%). Endoscopic therapy was undertaken in 19% of non-variceal and 65% of variceal cases. 64% received iv PPI after endoscopic therapy (excluding varices). 2% had surgery for AUGIB. Postoperative mortality was 30%. Unadjusted mortality was 10% overall—7% in new admissions and 26% in in-patients.

Conclusion: Despite a population with increasing medical comorbidity and a marked increase in the proportion of patients with varices, the unadjusted mortality of patients with AUGIB has significantly improved in the UK. This could be due to improvements in general health, supportive care, endoscopic haemostasis and acid suppression.

DEFINING THE IMPACT OF ORGAN DYSFUNCTION IN CIRRHOSIS: SURVIVAL AT A COST?

D. L. Shawcross, M. J. Austin, R. D. Abeles, M. McPhail, A. Yeoman, A. J. Portal, W. Bernal, G. Auzinger, E. Sizer, J. A. Wendon. Institute of Liver Studies, King’s College Hospital, London, UK

Introduction: Cirrhosis kills more men than Parkinson’s disease and more women than cervical cancer. The incidence of cirrhosis is increasing exponentially and it is projected that there will be a fivefold increase in demand for liver transplantation in the next 6–10 years. Historically, the perception of cirrhosis with organ dysfunction as having a poor prognosis has fuelled a self-fulfilling prophecy with iniquitous access to intensive care (ICU). However, recent data to support this view is lacking. We report the 7-year experience of the outcome, physiological disturbance and resource utilisation of 658 emergency admissions with cirrhosis and organ dysfunction to a specialist liver ICU at King’s College Hospital.

Aims & Methods: We prospectively collected and analysed physiological and biochemical variables on day 1 and 3 of admission. Outcome variables, organ scores (Child Pugh, MELD, SOFA and APACHE II) and number of days requiring vasopressors, ventilation and renal replacement therapy (RRT) were recorded. The Therapeutic Intervention Scoring System (TISS) score, validated as a tool for estimating cost in ICU, was calculated.

Results: Alcohol was the most common aetiology (47%) and variceal bleeding the most common reason for admission (35%). 51% required inotropes, 72% invasive ventilatory support and 49% RRT during their ICU stay. Despite this, ICU admission for many of these patients with cirrhosis was not futile, with 55% surviving their ICU stay and 41% surviving to discharge. Variceal bleeders had a 30-day survival of 53% versus 33% with a non-variceal indication for admission (p<0.0001). 19% of survivors subsequently underwent liver transplantation. Survival was at a significant cost to the NHS (median cost per patient: £12,403 (4636–27,283) for a median 7-day admission (3–15)). Admissions with multi-organ failure (with and without requirement for RRT) cost significantly more (multi-organ failure p<0.001; RRT p<0.0001) and had more prolonged lengths of stay compared to those that did not (p<0.0001). There was no difference in cost, length of stay or outcome (p = 0.2) between patients with alcoholic liver disease and other aetiologies. RRT was well tolerated and facilitated stability and/or recovery in 23/59 (39%) patients who were subsequently transplanted.

Conclusion: More than 50% of patients admitted as emergencies will survive their ICU stay with the majority of these surviving to hospital discharge. These data challenge the widely held prejudice that patients with cirrhosis requiring emergency admission to ICU inevitably die. This endorses the establishment of managed clinical care networks and demands the engagement of all levels of critical care and ward based care models in the treatment of organ dysfunction in cirrhosis.

Gastrointestinal physiology associates group

THE EFFECT OF ARGON PLASMA COAGULATION ABLATION ON OESOPHAGEAL MOTILITY AND CHEMORECEPTOR SENSITIVITY

K. Elghellal, A. Smythe, R. Ackroyd. Surgery, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Barrett’s oesophagus is of clinical importance because of the strong association with oesophageal adenocarcinoma. The use of endoscopic argon plasma coagulation (APC) ablation to eliminate intestinal metaplasia combined with oesophageal acid control has been shown to minimise if not remove the cancer risk.

Aims & Methods: Oesophageal motility and sensitivity appear to be disturbed in Barrett’s oesophagus. As ablation treatment is a thermal technique causing superficial damage, the aim of this study therefore was to look at any further effect of APC on both oesophageal motility and mucosal sensitivity. APC was used to treat 20 patients with Barrett’s oesophagus (16 males and 4 females, median age 56 (range 51–81) years) confirmed at endoscopy and histologically, median Barrett’s segment length 4 cm (range 2–10 cm). Station pull through manometry was carried out 1 week before APC treatment. The lower oesophageal sphincter pressure (LOSP), proximal and distal wave amplitude, primary peristalsis with 5 ml water swallows and secondary peristalsis with air and water bolus was measured. To assess oesophageal sensitivity, acid (0.1NHCl) and alkaline (0.1N NaOH ) provocation was carried out. Lag time to initial discomfort (t), sensory intensity rating (i) (no discomfort = 0, intense pain = 20) and acid perfusion sensory score (t × i/100) were obtained. These tests were repeated 4–6 weeks following successful APC.

Results: Using the Wilcoxon Rank Sum test for statistical evaluation, no significant differences were seen in any of the motility parameters comparing before and after APC. Sensitivity intensity rating and acid perfusion sensory score were significantly higher after APC treatment (pre-APC sensory score for acid = 1.6 (1.8), post-APC sensory score for acid = 0.5 (0.3), p<0.01. Pre-APC sensory score for alkali = 1.5 (1.7), post-APC sensory score for alkali = 0.9 (1.0).

Conclusion: Following reversal of Barrett’s epithelium using APC, no changes in oesophageal motility are seen but a decrease in oesophageal sensitivity occurs. Longer term studies are needed to see if sensitivity improves.

THE EFFECTS OF GLUCOSE, FRUCTOSE AND TAGATOSE ON GASTRIC EMPTYING

A. McGlone, T. J. Little, D. G. Thompson. Section of GI Sciences, University of Manchester, Salford Royal NHS Foundation Trust, Salford, UK

Introduction: The mechanisms by which nutrients, in the small intestine, signal to the CNS to alter gastrointestinal function are poorly understood. It has been suggested that sugars slow gastric emptying (GE) (a proxy measure of brainstem activation) by stimulating “osmoreceptors” in the small intestine; however, at a low osmolality, solutions of glucose and fructose appear to empty differently,1 suggesting that the potency of the feedback may also depend on the sugar molecule per se. It is known that the length of intestine exposed to nutrient is important for the regulation of GE. In dogs, slowing of GE by glucose is dependent on exposure of at least 15 cm of the small intestine to nutrient, and the intensity of the feedback increases as the length of exposure is increased.2 Differences in the absorption of hexose sugars may, therefore, be an important determinant of their effects on GE.

Aims & Methods: This study therefore aimed to determine whether tagatose, a poorly absorbed hexose sugar (used as a non-caloric sweetener), would slow gastric emptying to a greater extent than glucose or fructose. 12 non-obese subjects (7F, 5M; BMI 23 (0.76) kg/m2) drank 500 ml solutions of 250 mOsmol glucose, fructose or tagatose, and water (non-osmotic control for volume), each labelled with 100 mg of 13C-acetate. Gastric emptying was assessed over 45 min (breath samples collected every 5 min) using a 13C-acetate breath test; the lower the cumulative 13CO2:12CO2, the slower the rate of gastric emptying.

Results: There was a significant effect of hexose type on gastric emptying (p<0.05). The mean cumulative 13CO2:12CO2 values for tagatose were 139.5 (18.2) units lower than those for water, 123.8 (23.4) units lower than those for glucose, and 99.4 (17.5) units lower than those for fructose (p<0.05, for all), whereas there was no difference between glucose, fructose and water. There was, however, substantial interindividual variation in the gastric emptying response, particularly for tagatose and fructose. In general, those individuals who were most able to slow their GE in response to tagatose, had a tendency to slow the GE of fructose.

Conclusion: The poorly absorbed hexose sugar, tagatose, slows GE more potently than fructose and glucose, both of which would be expected to be absorbed more completely in the proximal small intestine, thus limiting the length, and duration, of exposure of the small intestine. However, there was substantial phenotypic variation in the gastric emptying responses to hexose sugars. Since the rate of GE closely correlates with ingested meal volume, this observation may have implications for the regulation of food intake.

Oesophageal free papers

AN IMPROVED STAGING SYSTEM FOR OESOPHAGEAL ADENOCARCINOMA BASED ON THE NUMBER OF INVOLVED LYMPH NODES

C. J. Peters1, R. H. Hardwick2, J. R. E. Rees3, S. L. Vowler4, P. M. Safranek2, V. Save5, N. Carroll6, S. Dwerryhouse2, N. Imrit7, N. Maynard7, G. W. B. Clark8, R. C. Fitzgerald1. 1MRC Cancer Cell Unit, Hutchison MRC Research Centre, Cambridge, 2Cambridge Oesophago-Gastric Centre, Addenbrooke’s Hospital, Cambridge, 3Department of Surgery, Bristol Royal Infirmary, Bristol, 4Cancer Research UK, Li Ka Shing Centre, Cambridge, 5Department of Histopathology, Addenbrooke’s Hospital, Cambridge, 6Department of Radiology, Addenbrooke’s Hospital, Cambridge, 7Oxford Oeosphago-Gastric Unit, Oxford Radcliffe Hospital, Oxford, 8Department of Surgery, University Hospital of Wales, Cardiff, UK

Introduction: The incidence of oesophageal adenocarcinoma has quadrupled in the last 30 years and outcomes remain poor. Oesophageal adenocarcinoma is now the commonest oesophageal malignancy yet is staged using a system originally developed for squamous cell carcinomas. This system does not take into account the number of involved lymph nodes despite increasing evidence this is prognostic. There is a growing consensus in the literature that a new staging system is required.

Aims & Methods: The aim of this study was to develop a new staging system for GOJ adenocarcinoma based on the number of involved lymph nodes. A systematic retrospective database was created of 280 patients diagnosed with GOJ adenocarcinoma in whom outcome data were available from 3 UK tertiary referral centres. We collected detailed information on lymph node status and found the association with outcome using SPSS and Prism. We developed a new N-stage using recursive partitioning with the R Statistical package.

Results: T-stage (p<0.0001), current N-stage (p<0.0001), differentiation (p = 0.0002), presence of neurovascular invasion (p<0.0001) and resection margin status (p<0.0001) were all prognostically significant on univariate analysis. Recursive partitioning of nodal positive patients generated the following revised N-stage: N0 = No involved nodes (from present system), N1 = 1–2 involved lymph nodes, N2 = 3–5 involved lymph nodes, and N3 = 6 or greater involved lymph nodes. This new system was highly prognostically significant on univariate analysis (p<0.0001) and outperformed the current N-stage on multiple regression analysis. Furthermore it identified a subgroup of nodal positive patients (N3) with a dismal prognosis (0% 5-year survival), equivalent to that of M1 disease.

Conclusion: We have developed a new N-stage for GOJ adenocarcinoma, based on the number of involved lymph nodes, which outperforms the existing system on multiple regression analysis and provides the ability to discriminate between nodal positive patients. We are in the process of validating our findings on an external dataset but this system has the potential to vastly improve our ability to make management decisions. We feel it is time to change the way we stage GOJ adenocarcinoma.

PRELIMINARY RESULTS OF A RANDOMISED CONTROLLED TRIAL INTO THE SAFETY AND EFFICACY OF ALA VERSUS PHOTOFRIN PHOTODYNAMIC THERAPY FOR HIGH GRADE DYSPLASIA IN BARRETT’S OESOPHAGUS

G. D. Mackenzie, J. M. Dunn, M. R. Novelli, S. Mosse, S. M. Thorpe, S. G. Bown, L. B. Lovat. National Medical Laser Centre, University College London, London, UK

Introduction: Photodynamic therapy (PDT) using Photofrin is a licensed minimally invasive treatment for Barrett’s high grade dysplasia (HGD). Photofrin PDT affects the entire oesophageal wall and causes strictures in 18–50% of patients. Patients must also avoid bright light for 3 months to prevent sunburn. 5 aminolaevulinic acid (ALA) is preferentially concentrated in the oesophageal mucosa. The risk of stricture is therefore very low. Light sensitivity for ALA lasts for only 36 hours. The aim of current RCT is to determine if PDT for HGD using ALA is safer and more effective than Photofrin.

Aims & Methods: Patients with HGD in Barrett’s oesophagus confirmed on at least two occasions by at least two independent pathologists were offered participation in the above RCT. Any visible nodules of HGD were removed and patients only treated if residual HGD was proven to still be present. To ensure length of Barrett’s and extent of HGD were similar in each treatment group patients were stratified for these two factors prior to randomisation. Patients received either Photofrin PDT with the standard licensed protocol or as previously shown to be the most effective 60 mg/kg of ALA activated by 1178 J/cm of red laser light. Follow-up was quadrantic biopsies every 2 cm throughout the treated area of Barrett’s at 6 weeks, 4 months and 1-year post therapy. All side effect data were collected.

Results: Forty patients (out of a planned 66) have so far been recruited and 32 treated; 16 with ALA and 16 with Photofrin. Five patients are undergoing repeat therapy (3 with photofrin and 2 with ALA) and 27 patients are in follow-up. Remission rates are 14/14 (100%) in the ALA group and 9/14 (64%) in the photofrin group (Fisher’s exact test, two-sided p value<0.05). Strictures developed in 6/16 patients treated with Photofrin and 1/16 with ALA, the latter is thought to relate to uncontrolled and significant acid reflux at the time of therapy measured on 24-h pH monitoring (Fisher’s exact test, one-sided p value<0.05). Skin photosensitivity developed in 7/16 patients treated with Photofrin, one of whom was briefly admitted to hospital. There were none with ALA (Fisher’s exact test, one-sided p value<0.05). There are no significant differences to date between groups with respect to other side effects.

Conclusion: The preliminary data are encouraging and suggest that ALA PDT is both safer and potentially more effective than Photofrin PDT but follow-up is short and only half the planned number of patients have been treated to date.

NARROW BAND IMAGING WITH MAGNIFICATION IS SUPERIOR TO WHITE LIGHT ENDOSCOPY WITH MAGNIFICATION IN STUDYING THE MUCOSAL MORPHOLOGY IN BARRETT’S OESOPHAGUS

R. Singh1, V. Owen2, H. Karageorgiou1, G. K. Anagnostopoulos1, K. Garsed1, P. Fortun1, E. Fogden1, V. Subramaniam1, A. Shonde1, P. Kaye1, C. J. Hawkey1, K. Ragunath1. 1Wolfson Digestive Diseases Centre, Nottingham University Hospitals Trust, 2Trent Research and Development Support Unit, University of Nottingham, Nottingham, UK

Introduction: Narrow band imaging with magnification (NBI-Z) can reliably predict histology in Barrett’s oesophagus (BE). Although available in a single unit, the utility of white light endoscopy with magnification (WLE-Z) in BE has not been assessed in prior studies.

Aims & Methods: The aim of the study was to see if there is any appreciated difference in imaging characteristics between WLE-Z and NBI-Z in BE and if this translated into clinically relevant benefit in the prediction of histology, specifically looking at dysplasia detection. This was a prospective cohort study involving 21 patients (75 areas with corresponding NBI-Z and WLE-Z images) with BE. Mucosal patterns were evaluated for its image quality on a visual analogue scale (VAS) of 1–10 by 5 expert endoscopists specifically looking at the pit pattern and microvascular morphology. The endoscopists then predicted the histology based on four subtypes which can be visualised in BE: Type A: round pits with regular microvasculature (columnar mucosa without intestinal metaplasia); Type B: villous/ridge pits with regular microvasculature (specialised intestinal metaplasia); Type C: absent pits with regular microvasculature (specialised intestinal metaplasia); Type D: distorted pits with irregular microvasculature (high grade dysplasia). The sensitivity (Sn), specificity (Sp), positive (PPV), negative predictive values (NPV) and accuracy (Acc) compared to the final histopathological analysis and the interobserver variability among the 5 assessors were then calculated.

Abstract 035 Mean values of Sn, Sp, PPV, NPV and Acc for all five assessors

Results: The overall pit quality was significantly higher for NBI-Z compared to WLE-Z (VAS: NBI-Z = 6, WLE-Z = 4.5; z = 11.0, p<0.001). The overall vasculature quality was also significantly better for NBI-Z (VAS: NBI-Z = 7.3, WLE-Z = 4.9, z = 14.1, p<0.001). This translated into a superior prediction of histology in BE with higher Sn, Sp, PPV, NPV and Ac for NBI-Z compared to WLE-Z (table). Looking specifically at the prediction of dysplasia, NBI-Z was superior to WLE-Z (χ2 = 10.3, df = 2, p = 0.006). The overall kappa agreement amongst the 5 endoscopists for NBI-Z and WLE-Z were 0.59 and 0.31 respectively (p<0.001).

Conclusion: NBI-Z is superior to WLE-Z in the prediction of histology in BE with good reproducibility. This novel imaging modality could be an important tool for surveillance in patients with BE.

IMPROVING OUTCOME IN PATIENTS WITH OESOPHAGEAL CANCER

A. K. Saha, C. Sutton, S. P. L. Dexter, A. I. Sarela, H. M. Sue-Ling. Academic Division of Surgery, The General Infirmary at Leeds, Leeds, UK

Introduction: The management of oesophageal cancer has changed significantly over the past 30 years. Improved preoperative and postoperative care, sub-specialisation and increasing use of chemotherapy and radiotherapy have all contributed to improved survival. Our unit has previously reported how these changes had begun to improve overall outcomes1 and suggested that they could be further improved by continuing these changes.

Aims & Methods: Using a prospectively maintained database, all patients who underwent oesophageal resection between January 2000 and December 2006 were reviewed. The results were analysed and compared with previously published and presented audits of oesophagectomy from our unit between 197588 and 19909.

Abstract 036 Improving outcomes after oesophagectomy

Results: Results are summarised in the table. In-hospital mortality has improved from 27% to 2.9% and 5-year survival has improved from 7% to 52%.

Conclusion: These results are encouraging and suggest that the outlook for patients with oesophageal cancer can be greatly improved by earlier diagnosis, greater specialisation and a multi-disciplinary approach to the management of oesophageal cancer.

EOSINOPHILIC OESOPHAGITIS: RARE DISEASE OR A DIAGNOSIS FREQUENTLY MISSED?

V. Kanakala1, C. Lamb2, R. W. Stirling3, S. E. Attwood1. 1General Surgery, 2Gastroenterology, 3Pathology, North Tyneside General Hospital, North Shields, UK

Introduction: The diagnosis of eosinophilic oesophagitis (EO) is relatively rare although there has been a dramatic rise in the frequency of case reports and case series over the past 10 years.

Aims & Methods: This study aimed to identify the frequency of diagnosis of EO in our community, how the diagnosis was made or suspected and the outcome of specific interventions once the diagnosis was made. The pathology records of gastroscopies performed (62 437) in all the hospitals in Northumberland during years 2001–Oct 2007 were reviewed for patients with a diagnosis of EO (>20 per HPF) on any biopsy sample. The clinical details were obtained from the case notes and hospital electronic database.

Results: Eighteen patients who were diagnosed as having EO were all male with a median age of 46 years (range 17–85 years). There was a median time delay of 2 years between onset of symptoms and diagnosis (range 1–>10 years). Our results show that the endoscopic findings rarely directed the decision to biopsy as most patients had the suspicion of the disease raised beforehand (13/18). In 5 the endoscopic appearances prompted a biopsy not previously planned. 17 patients presented with intermittent dysphagia which is the characteristic symptom of EO. Only two patients had definite response to PPI therapy. In three we felt that treatment was not necessary due to mild symptoms. Treatment outcomes >6 months are available in 13 with good response to oral systemic steroids, ingested topical steroids and oral montelukast. Stricture dilatation was indicated in two patients with good results.

Abstract 037

Conclusion: It is very likely the infrequent diagnosis of eosinophilic oesophagitis is due to a lack of clinical suspicion and failure to biopsy rather than there being a true low incidence. The presentation of frequent bolus obstruction, intermittent or continuous dysphagia refractory to routine PPI therapy should prompt oesophageal biopsies to look for eosinophils, as this condition can be effectively treated by medical management.

RISK FACTORS FOR THE DEVELOPMENT OF OESOPHAGEAL CANCER IN BARRETT’S OESOPHAGUS: A PRIMARY CARE NESTED CASE-CONTROL STUDY

S. C. Cooper1, P. Graham2, N. J. Trudgill1. 1Department of Gastroenterology, Sandwell General Hospital, West Bromwich, 2Information Department, Worcester Acute Hopsitals, Worcester, UK

Introduction: The annual risk of Barrett’s oesophagus (BO) progressing to oesophageal cancer (OC) is estimated between 0.5–2%. Previous studies examining risk factors for OC in BO have been limited to surveillance patients, with inherent selection biases.

Aims & Methods: The Health Improvement Network (THIN) and General Practice Research Database (GPRD) collects information from UK primary care patients. Patients with BO with at least two years of follow-up were identified. The influence of age, gender and drugs potentially associated with OC development was examined. Cox regression analysis, adjusted for age and gender, generated hazard ratios for association with OC development.

Results: 4571 BO patients were eligible for the study providing 20 703 years of follow-up, median 4 (IQR 3–6) years. 70 developed OC (0.3% pa). Median age (IQR) in OC patients was 67 (62–72) and in BO without OC 63 (52–72) and, hazard ratio (95% CI) 1.03 (1.01–1.05) p = 0.001, for each increasing year. Male gender was associated with developing OC (OC 81% male, BO 63% male); 2.64 (1.44–4.8) p = 0.002. After adjusting for age and gender, inhaled steroids (OC 30% vs BO 19%), 1.79 (1.07–2.99) p = 0.025 were the only drug associated with progression to cancer. Use of PPIs was negatively associated with progression to cancer 0.42 (0.18–0.98) p = 0.045, while aspirin showed a non-significant trend towards a negative association 0.68 (0.4–1.15). Cumulative addition of drugs for COPD/asthma led to increasing association with OC (1 drug 0.83 (0.37–1.84), 2 drugs 1.38 (0.74–2.57), 3 drugs 2.84 (1.27–6.34) p = 0.011).

Conclusion: Male gender and increasing age in BO patients were associated with progression to OC. Inhaled steroid use, and cumulative use of respiratory drugs was associated with the development of OC, suggesting a potential role for COPD or asthma in the aetiology of oesophageal cancer in BO.

ANALYSIS OF SUPINE GASTRO-OESOPHAGEAL REFLUX VARIABILITY DURING 96 HOURS MONITORING USING BRAVO CAPSULE

K. P. Basavaraju1, A. Anggiansah2, M. Fox1, T. Wong1. 1Gastroenterology, 2Oesophageal laboratory, St Thomas Hospital, London, UK

Introduction: Bravo catheter-free oesophageal pH monitoring allows prolonged monitoring of gastro-oesophageal acidic reflux (GOR) up to 96 hours. The day-to-day concordance of upright GOR data during 48-hour recordings has been reported as fair, but poor for supine data. There are very few data available regarding the variability of supine GOR in patients with 96-h recording.

Aims & Methods: This study aimed to evaluate inter- and intrapatient variability for supine GOR over 96 h and determine the effect of meal to bedtime (MBT) on the supine GOR.

We analysed 80 consecutive oesophageal pH studies that completed 96-h recordings between October 2004 and October 2007 in our unit. 34 (42.5 %) of these patients with abnormal supine GOR (pH>3.45%) were included in the study. The main recumbent period for each 24-h period was divided into two halves (H1 and H2). The two halves were compared on reflux time (%) and total number of refluxes (TNR) across the 96-h period. Cumulative diagnostic yield using prolonged pH monitoring for each 24 h was analysed. Reflux time (%) was compared for patients with MBT cut-off of 2 and 3 h during both H1 and H2 across the 96-h period.

Abstract 039

Results: Within-individual variability for supine GOR for H1 and H2 across the 96-h period was high (intraclass correlations only 0.091, 0.146 respectively). Median values for reflux time (%) and TNR for each of the halves across the 96-h period are shown below (table 1). Measurements of % GOR for H1 were significantly higher than those for H2 (Wilcoxon signed-rank test, p<0.001). There was a similar decrease for TNR (p<0.001). The cumulative diagnostic yield for pathological supine GOR was 73.5%, 88.23%, 88.23% and 100% at 24 h, 48 h, 72 h and 96 h respectively. The median for MBT was 3.53 h. During H1, MBT of <2 h (n = 29.4%) was significantly associated with increased supine reflux compared to MBT >2 h (n = 70.6%) (Wilcoxon rank sum test, p = 0.02), but not during H2 (p = 0.21). There was no significant difference in supine reflux with patients retiring to bed <3 h (n = 47.8 %) vs >3 h (n = 52.2%) in either H1 (p = 0.38) or H2 (p = 0.79).

Conclusion: Variability in the supine position showed little day-to-day diagnostic reproducibility with resulting increased diagnostic yield with prolonged recording up to 96 h. The supine GOR is greatly reduced in the second half of the night compared to the first half. Patients going to bed within 2 h, but not 3 h, of a meal are at a higher risk of experiencing increased supine GOR.

Health service research and IT free papers

DEEP CLEAN: THE WAY FORWARD IN CONTROLLING C DIFFICILE INFECTION?

A. Malik1, D. W. Bullimore2. 1Medicine, Nevill Hall Hospital, Abergavenny, 2Medicine, Barnsley NHS FT, Barnsley, United Kingdom

Introduction: Because of concerns about hospital acquired infection the Government has proposed a “Deep Clean” programme of wards in all NHS hospitals. Alan Johnson1 has expressed this as: “This deep-clean programme will give hospitals a one-off blitz so walls, patient equipment and ventilation ducts are disinfected and scrubbed clean, a ward at a time.” It is noted the total cost could be £50m. Our Trust undertook a similar process in 2006. We have examined the results of the Deep Clean on subsequent infection rates with C difficile on eight medical wards to see what benefits resulted.

Aims & Methods: A microbiology database of all C difficile positive samples recorded during 2006 was analysed. The file was filtered for in-patients only, ranked by date, duplicates removed and further sub-filtered by ward. The Deep Clean took approximately 4 weeks/ward. The return date for the ward and the subsequent 3-month period was plotted on the spreadsheet against the year’s cases of C difficile, so that for each ward the number of cases in this period could be compared against the number for the whole year. Figures for 8 medical wards were analysed. The Medical Admissions Unit and a decanting/overspill ward were excluded from analysis. Ward infection rates were similarly derived for 2005 and January–April 2007.

Results: A total of 209 c.difficile infections were identified on the 8 wards in 2006, this was an increase from 81 infections in 2005. Of the 209 infections 53 (observed 25% vs expected 25%, no significant difference) occurred during the 3-month period immediately following return of patients to the cleaned ward. The number of cases/ward was highly variable, despite 7 of the wards being broadly similar in size. The ranking of number of cases between wards also varied considerably for the periods 2005, 2006 and January–April 2007. In the longer term 70 cases of C difficile occurred during the first 4 months of 2007. This gave a projected annualised rate for 2007 of 210 cases, compared with 209 cases for 2006, demonstrating no significant difference. (The closeness of observed and expected figures is itself a statistical aberration.)

Conclusion: The Deep Clean process did not alter subsequent infection rates in the short term or longer term. This perhaps suggests that infection via spore transmission is not common in the hospital setting. Patient to patient transmission may occur and/or use of antibiotics in debilitated patients may lead to auto-infection in a suppressed bowel flora. The use of communal toilet facilities needs to be investigated as a factor in patient to patient transmission. Deep Clean policies are unlikely to control C difficile infections on hospital wards.

A RETROSPECTIVE RECORD-LINKAGE STUDY TO FOLLOW UP PATIENTS WITH NO CLINICALLY OBVIOUS LIVER DISEASE FROM THEIR FIRST LIVER FUNCTION TESTS IN PRIMARY CARE

D. J. McLernon1, P. T. Donnan1, J. F. Dillon2. 1Tayside Centre for General Practice, 2Pathology and Neurosciences, University of Dundee, Dundee, UK

Introduction: Liver function tests are routinely performed in primary and secondary care and often lead to further invasive and expensive investigations. The aim of this study was to follow up individuals with LFTs initiated in primary care with no obvious liver disease and to ascertain their outcomes.

Aims & Methods: A population-based retrospective cohort study followed up these patients in Tayside from 1989 to 2003 to subsequent diagnosed liver disease, mortality from liver disease and all-cause mortality. Record linkage of datasets including biochemistry, hospital admissions, psychiatric admissions, death registry and prescriptions enabled ascertainment of risk factors and outcomes. Sensitivity, specificity, positive predictive value and negative predictive value analyses were carried out and Weibull regression analysis from initial LFT date was performed on outcomes of mortality and liver disease by level of LFT abnormality.

Results: 95 977 patients had 364 194 incident initial LFTs taken during the 15-year period 1989 to 2003, with median follow-up 3.7 years. At initial tests, 21.7% had at least one abnormal LFT and 1213 (1.3%) developed liver disease. Elevated transaminases were strongly associated with diagnosed liver disease, HR = 4.20 (3.54, 4.98) for mild levels and HR = 11.99 (9.26, 15.52) for severe levels vs. normal. For GGT these hazards were 2.46 (2.12, 2.86) and 13.21 (10.60, 16.46) respectively. Low albumin was strongly associated with all cause mortality, 2.65 (95% CI 2.47 to 2.85) for mild levels and 4.99 (95% CI 4.26 to 5.84) for severe levels. Sensitivity for predicting events over 5 years was low and specificity high. For liver disease the sensitivity and specificity values were: albumin (9%, 98%), alkaline phosphatase (32%, 90%), transaminase (35%, 93%) and GGT (62%, 85%).

Conclusion: All LFTs were predictive markers for liver disease as well as general ill health, although sensitivity was poor. Many patients with abnormal LFTs detected in this study had no later formal diagnosis of liver disease but may be alcohol abusers on the road to liver problems. That liver disease takes time to develop illustrates a window of opportunity to intervene in these patients with lifestyle advice, alcohol therapies or drugs.

DIVERTICULAR DISEASE HOSPITALISATIONS HAVE A RAPIDLY INCREASING IMPACT ON THE HEALTH SERVICE AND AFFECT YOUNGER PEOPLE AND DEPRIVED SOCIOECONOMIC GROUPS

S. Jeyarajah1, P. P. Tekkis2, P. Aylin3, I. Bjarnason4, S. Papagrigoriadis1. 1Department of Colorectal Surgery, King’s College Hospital, 2Department of Biosurgery and Surgical Technology, St Mary’s Hospital, 3Department of Epidemiology, Imperial College, 4Department of Gastroenterology, King’s College Hospital, London, UK

Introduction: Diverticulosis of the colon is common in the Western world with a prevalence of approximately 33% in patients over 60 years of age. In the UK, there are few data on the epidemiology of diverticular disease (DD) particularly with regards to the effect of social deprivation on disease presentation and clinical management.

Aims & Methods: This was a population-based study across the UK of all hospital episodes reporting on clinical outcomes of patients admitted with diverticular disease as the primary or secondary diagnosis between 19962006. The Hospital Episode Statistics and the general population 2001 Census databases were used for data extraction. Social deprivation was evaluated using quintiles of socioeconomic deprivation based on the Carstair’s classification. Logistic regression analysis was used to evaluate the study end points.

Results: Over the study period there were 1 219 480 admissions with the diagnosis of DD; 567 423 patients with DD as a primary cause for admission and 652 057 as a secondary diagnosis (comorbidity). Primary admissions increased from 0.6 to 1.2/1000 population (p<0.001). The mean age of primary DD decreased from 67.8 to 66.8 years (p<0.001). 732 493 (60.1%) of all admissions with DD were female patients. The median in-patient length of stay (LOS) was 6 days (range 0–2,283, 75th percentile 12 days). 179 426 (31.62%) patients were emergency admissions. 38 796 (6.8%) patients with primary DD underwent a laparotomy and 560 (0.1%) required admission to ITU. There were 29 168 (5.1%) in-hospital deaths. The admission rate within 28 days and at one year was 4.9% and 46.8% respectively. 101 688 (17.9%) patients were in the fifth Carstairs quintile (most deprived). In-hospital mortality following admission for DD was highest in the lowest socioeconomic group (OR 1.2, p<0.01). Emergency re-admission within 28 days (OR 1.4, p<0.01) and at 1 year (OR 1.7, p<0.01) were more likely in the lowest socioeconomic groups. In-hospital LOS >12 days was more likely in the least deprived patient group (OR 1.4, p<0.01). The need for surgery was most likely in the most affluent patients.

Conclusion: The study suggests an increasing incidence of diverticular disease hospital admissions across all socioeconomic groups with disease migration to younger patients. In-hospital mortality, length of stay and re-admissions were more likely to occur in the most deprived patients. Diverticular disease has a large financial and social burden to the NHS. New research and management strategies are required for prevention in the community thus minimising in-hospital complications.

AN OPEN-SOURCE BROWSER-BASED ENDOSCOPY MANAGEMENT SYSTEM THAT MAPS TO SNOMED CT

J. L. Brown1, A. MacDonald2, J. Davies2, P. Ryder2. 1Department of Gastroenterology, Gloucestershire Royal Hospital, Gloucester, 2Beaumont-Colson Ltd, Software Developers, Newcastle, UK

Introduction: European, American and Japanese Gastrointestinal Endoscopy Societies have worked collaboratively to devise an international lexicon of terms and attributes (with definitions) to be used for endoscopy reporting (Minimal Standard Terminology for Digestive Endoscopy V2.0 – OMED 2002). Since April 2003, SNOMED CT was identified as the preferred clinical terminology for NHS information systems. A project to map MST to SNOMED CT was commissioned in 2002 by the former NHS Information Authority and this was undertaken with representation from the BSG IT subcommittee. The dataset is freely available for BSG members to review on the society’s website and may also be obtained from Connecting for Health. With the exception of the National Bowel Cancer Screening software, we are not aware of any clinical implementations of the endoscopy subset.

Aims & Methods: In order to demonstrate that this dataset is fit for purpose we have built a web application (Endoplex) that allows clinicians to aggregate terms in a browser to generate a clinical report. Open Source materials (comprising an Apache web server, MySQL database server and PHP scripts) have been used throughout. The application, supported by AJAX (asymmetric javascript XML) methods, permits client sided processing of the SNOMED CT dataset from within the browser cache.

Results: The result is a remarkably fast reporting tool which can write coded clinical data to multiple server-sided database tables without the delay of sequential post operations and screen refreshes. Reports can be printed onto paper in portable document format or can be viewed as a web page containing embedded multimedia.

Conclusion: We believe our platform-independent solution to be scalable for use within large organisations such as the NHS because the only client-sided requirement is for a web browser and the server PHP scripts allow SQL connections to all the major commercial database engines. Furthermore, inevitable developments in endoscopic technique and reporting syntax merely require a modification to the database reference tables with new SNOMED codes which automatically update the client browser cache. The architecture of Endoplex has been designed to support tariff price calculations, quality assurance monitoring, accessory stock control, HL7 messaging and translation of reports into foreign languages that map to SNOMED CT.

AN INTERACTIVE POLYP DETECTION TRAINING TOOL: DEVELOPMENT AND INITIAL VALIDATION

P. L. Youd, A. Haycock, S. Preston, S. Thomas-Gibson. Wolfson Unit for Endoscopy, Imperial College London, London, UK

Introduction: Cancer prevention depends on the detection, safe removal and post-procedure surveillance of colonic polyps. NHS Bowel Cancer Screening accreditation requires colonoscopists to have an adenoma detection rate (ADR) of at least 20%. Adenoma miss rates of up to 22% have been reported and vary considerably amongst colonoscopists. Colonoscopy trainees receive little specific training in polyp detection and lesion recognition. Self-directed learning material in this area is limited. We describe the first interactive e-learning tool designed specifically for the training and assessment of polyp detection (PDTT).

Aims & Methods: The objectives of this study were: (1) To determine construct validity of a video-based polyp detection assessment tool, and (2) To determine whether a self-directed learning package improves polyp detection. During the initial self-assessment part of the PDTT, trainees watch eleven 10–15 second video clips and are scored based on their ability to accurately detect colonic polyps. The second part is a self-directed, computer-based learning package covering topics such as the importance of polyp detection, improving ADR and newer optical modalities for polyp detection. To confirm construct validity, the self-assessment exercise was undertaken by expert colonoscopists, colonoscopy trainees, endoscopy nurses and clerical staff. Secondly, as part of a pilot study, candidates at the St Mark’s Basic Therapeutic Endoscopy Training Course (BTEC) performed the self-assessment exercise before and after using the learning package. Endoscopy nurses without access to the learning tool were used as a control. Candidates completed feedback questionnaires.

Results: There was a statistically significant difference between the scores on the self-assessment exercise for the four groups (p<0.001, Kruskal-Wallis test). The median score for expert colonoscopists (⩾500 procedures) was 88% (n = 6, range 82–100); trainees (<500 procedures) 45% (n = 18, range 27–64); endoscopy nurses 45% (n = 12, range 18–77) and clerical staff 36% (n = 6, range 23–41). The 14 BTEC candidates demonstrated an average improvement in score of 8.1% after using the learning tool. The 8 endoscopy nurses who undertook the self-assessment exercise twice, without access to the learning material, demonstrated an average improvement of 2.2%, although the difference was not statistically significant (p = 0.18). Candidate feedback described the training module as excellent, fun and useful.

Conclusion: This first dedicated training and assessment tool for polyp detection, employing specifically-designed e-learning technology, has been very well received by trainees. The interactive, self-assessment tool has been shown to have construct validity. The results from the initial pilot study in the use of the teaching tool show promise and require further validation.

Small bowel and nutrition free papers

OUTCOMES FOLLOWING GASTROSTOMY: RADIOLOGICALLY INSERTED VERSUS PERCUTANEOUS ENDOSCOPIC GASTROSTOMY

J. S. Leeds1, M. E. McAlindon1, J. Grant1, H. E. Robson1, S. R. Morley2, F. K. T. Lee3, D. S. Sanders1. 1Gastroenterology, 2Clinical Chemistry, 3Radiology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Gastrostomy insertion has been demonstrated to be of benefit in selected patients. Percutaneous endoscopic gastrostomy (PEG) using the pull through technique is the most widely used insertion method, but it is recognised to have significant complications particularly in patients with respiratory risk factors. An alternative is a radiologically inserted gastrostomy (RIG). It has been suggested that RIG may be advantageous in patients who are potentially ‘high risk’ from respiratory complications. However there are no large studies comparing PEG against RIG.

Aims & Methods: All patients referred for a gastrostomy are prospectively included in a database along with demographic, biochemical and outcome data. Analysis of gastrostomy insertions over the period February 2004 to February 2007 was performed with reference to method of insertion and outcome at 30 days. Selection for method of insertion is left to the discretion of the referring clinician. Patients were subgrouped into cognitive impairment (n = 5), dysphagic stroke (n = 36), nasopharygeal cancer (n = 175), neurological (n = 116) and other (n = 71).

Results: Over the study period 170 RIGs and 233 PEGs were inserted (mean age 62, 268 males). There were no differences in age between the RIG and PEG group and case mix was comparable except in the nasopharyngeal cancer group (proportionally more RIGs). The RIG 30 day mortality was 26/170 (15.3%) and the PEG 30-day mortality was 25/233 (10.7%) (p = 0.17). One year mortality was 92/170 (54.1%) for RIG and 131/233 (56.7%) for PEG (p = 0.60). Within subgroups the only significant difference in 30-day mortality was in those patients with nasopharyngeal cancer; 14/106 (13.2%) for RIG and 1/69 (1.4%) for PEG (p = 0.005). However, patients referred for RIG were significantly older than those referred for PEG (mean age 59.7 vs 64, p = 0.019) and had a higher prevalence of significant comorbidities (21.1% in the PEG group and 37.7% in the RIG group).

Conclusion: Overall RIG and PEG appear to have similar 30-day mortality rates. In patients with nasopharyngeal cancer there was a higher mortality in those referred for RIG however pre-selection by the referring clinician due to perceived risk of endoscopic insertion may have biased the outcome. A randomised trial comparing both methods in this subgroup is needed.

TWO-YEAR PROSPECTIVE AUDIT OF USE OF LIQUID DIETS AS “BRIDGE” THERAPY FOR ACTIVE CROHN’S DISEASE

T. Demetriou1, C. Gawad2, C. Kok2, A. W. Harris1. 1Gastroenterology, 2Nutrition, Kent and Sussex Hospital, Tunbridge Wells, UK

Introduction: Liquid diets may be used to induce remission as an alternative to corticosteroids in active Crohn’s disease.13 In our practice liquid diets, high dose mesalazine or antibiotics may be used as “bridge” therapy to allow thiopurine or methotrexate to work. Steroids are used rarely, if at all. The tolerability of such diets may be a significant limiting factor.

Aims & Methods: To compare the efficacy and tolerability of 28 days’ therapy with either E028 or Modulen IBD in patients with active Crohn’s disease while starting treatment with mercaptopurine or methotrexate. Prospective study of all patients treated with liquid diets between April 2005 and August 2007. E028 was prescribed for the first 8 months of the audit and thereafter Modulen IBD was used. Liquid diets were started following dietetic advice; no other form of nutrition was allowed and patients received weekly dietician input. Clinical response (defined as decrease in stool frequency, pain and lethargy) was assessed in clinic around day 28 (or sooner if liquid diet failed).

Results: E028 was used in 13 patients (10 females; median age 41 (range 21–78) years); 7 (54%) completed 28 days treatment. Modulen IBD was used in 13 patients (9 females; median age 33 (range 16–63) years); 9 (69%) completed 28 days treatment. Clinical response occurred in 6 of 7 patients and in all 9 patients (p>0.05) who completed treatment with E028 or Modulen IBD, respectively. Overall, 15 of 16 patients who completed 28 days treatment with liquid diet achieved clinical response. By contrast, by intention-to-treat analysis, only 6 of 13 (46%) patients treated with E028 & 9 of 13 (69%) patients treated with Modulen IBD achieved a clinical response (p = 0.15; Fisher’s exact test).

Conclusion: In our cohort of well-motivated and supported patients 28 days treatment with liquid diet was completed in only 16 of 23 (65%) patients. Clinical response was achieved in 15 of 23 patients—a similar outcome after treatment with steroids where 35–40% achieve a partial or complete response to 30 days treatment.4 This audit has shown that the use of liquid diet to avoid the use of steroids and to act as a “bridge” to allow the onset of action of thiopurine or methotrexate appears practical and effective.

Acknowledgement: Dr R Loke for allowing us to access data on his patients.

OPTIMAL LEVELS OF PERIOPERATIVE PARENTERAL NUTRITION SUPPORT: A DOUBLE-BLIND RANDOMISED CONTROLLED TRIAL

T. R. Smith, P. Austin, S. Harding, Z. Leach, S. A. Wootton, M. A. Stroud. Institute of Human Nutrition, Southampton University Hospitals NHS Trust, Southampton, UK

Introduction: Optimal levels of energy and protein provision in perioperative patients requiring parenteral nutrition (PN) support are not known. There is considerable debate as to whether PN should be started at rates likely to meet all a patient’s metabolic needs immediately, or whether slower provision during the first few days is safer in terms of metabolic upset and refeeding risks. NICE guidelines (2006) favoured the latter approach, recommending the introduction of PN at 50% of estimated requirements. We have therefore investigated the effect of higher or lower rates of PN provision on biochemical parameters and clinical outcomes.

Aims & Methods: Patients referred for peri-operative PN were randomised to a lower (17.5 non-protein kcal/kg and 20.3 total kcal/kg, 0.12 gN/kg) versus higher (29.9 non-protein kcal/kg and 35.0 total kcal/kg, 0.22 gN/kg) regimen for the first 5 days of their nutrition support, with recording of both biochemical and clinical outcomes. Patients at very high risk of refeeding syndrome were excluded (BMI <16, >20% weight loss, >2 weeks with no nutritional intake).

Abstract 047

Results: 60 patients (57% male, mean age 66 years (sd 15)) were recruited to the trial. 28 received the lower and 32 received the higher feeding regimens. Clinical outcome and glucose data are shown in the table. There were no significant differences between groups in the need for insulin infusions or biochemical measures indicative of refeeding syndrome (potassium, magnesium or phosphate). 13% of patients developed worsening oedema during PN support but with no significant difference between the two feeding regimens.

Conclusion: Although our data show clinical outcome trends that appear to favour introduction of PN feeding at lower levels, possibly related to significantly improved glycaemic control, standard deviations were extremely variable in both groups and hence differences in clinical outcomes were not significant. The introduction of nutrition support at lower levels may therefore be unnecessary if patients are not at very high refeeding risk, although a much larger, probably multicentre, trial is needed to confirm this. Our data permit the power calculations needed to plan such a trial.

THE “CUT AND PUSH” METHOD OF PERCUTANEOUS ENDOSCOPIC GASTROSTOMY TUBE REMOVAL IN ADULT PATIENTS

D. Kejariwal, D. Bromley, Y. Miao. Gastroenterology, Ipswich Hospital NHS Trust, Ipswich, UK

Introduction: The standard method of removing percutaneous endoscopic gastrostomy (PEG) tubes is by repeat gastroscopy and retrieval of the PEG bumper by the oral route. Most PEG manufacturers recommend the endoscopic method of removal except in those tubes with balloon or flexible ends designed for removal by external traction. Modern PEG design has made the ‘bumpers’ of the tubes softer and more malleable. We believe that the endoscopic removal method is not always necessary and use a non-endoscopic approach. “Cut and push”, is a simpler and cheaper alternative to endoscopic PEG removal and has been reported in 19911 and 20002 for tubes of 15 Fr or less. This involves cutting the tube at skin level and allowing the internal bumper to pass spontaneously.

Aims & Methods: The files of PEG tube removals by the cut and push method over a 5-year period at Ipswich Hospital were reviewed.

Results: Between February 2002 and May 2007, 80 Fresenius Freka 15Fr tubes were removed by the cut and push method. The age range of patients was 19–98 years (mean of 62, median of 63). The average duration of the tube in-situ before removal was 6.7 months (range 1–28). No reported hospital re-admissions or mortality occurred within 30 days of removal. No reports of minor complications occurred either. The original indications for the PEG tubes that were removed were head and neck malignancy (44), cerebrovascular accident (14), other neurological illness (15) and supplementary feeding (7).

Conclusion: The necessity to endoscopically remove PEG tubes is based on reported complications in a few case reports.3 4 The complications were principally small bowel perforation and obstruction. However these reported complications occurred in either children or those with tube sizes of more than 15 Fr. In our hospital, none of our patients with the “cut and push” method of tube removal suffered any complications. We believe that this provides further evidence that PEG tubes can be safely removed using this method. Elderly and neurologically impaired patients are spared an often-long and tedious journey to the Endoscopy unit as well as the hazards of an endoscopy. The saving in resources, in an already overworked system, by reducing endoscopies is also considerable.

BILE ACID MALABSORPTION: TWO DECADES ON

P. Basumani1, M. J. Smith2, M. Srinivas1, B. F. Dawson2, P. J. A. Willemse1, B. Höeroldt1, B. A. Mason1, C. Royston1, K. D. Bardhan1. 1Gastroenterology, 2Medical Physics, Rotherham General Hospital, Rotherham, UK

Introduction: Chronic diarrhoea, a common problem, often results in protracted negative and frequently repeated investigations and unsatisfactory outcome. While bile acid malabsorption (BAM) as a cause of diarrhoea is well known in specific settings, such as ileal resection, it is now recognised more often in patients with diarrhoea associated with other conditions.

Aims & Methods: We wish to draw attention to the simplicity of BAM diagnosis by the SeHCAT retention test and the benefits this brings. The procedure is simple, non-invasive and measures retention of an isotope-tagged synthetic bile acid (75-Seleno homo-taurocholic acid). BAM was diagnosed when retention was <10% at day-7. 1076 patients with chronic diarrhoea underwent SeHCAT test 1988–2007. Of these, 1028 could be grouped into five broad categories (see table) based on information on the structured referral form. BAM patients who failed treatment with conventional anti-diarrhoeals were treated with bile acid sequestrants (BAS), principally colestipol or cholestyramine. We reviewed the response in a subset of 56 patients treated in 2004.

Abstract 049

Results: Treatment response (n = 56; 2004): number treated = 44; 18/25 (72%) D-IBS patients and 11/19 (58%) of the other groups showed good response to therapy with BAS (consistent with previous experience). BAM is commonly encountered in expected conditions such as ileal resection but is also common in post-cholecystectomy diarrhoea. Surprisingly, one third of patients with D-IBS have BAM; this proportion has remained constant over two decades. SeHCAT in our centre is now done among initial investigations in patients with chronic watery diarrhoea but with otherwise good general health, particularly where D-IBS is suspected, a common condition and our largest patient group. The early and definitive diagnosis of BAM by the readily available SeHCAT test frequently concludes investigations speedily; treatment with bile acid sequestrants is often more effective than conventional anti-diarrhoeals.

Conclusion: (1) BAM is common in many diarrhoeal conditions; the largest group is D-IBS. (2) The SeHCAT retention test allows rapid and definitive diagnosis. (3) Treatment with BAS often improves the patient’s quality of life.

MIDGUT NEUROENDOCRINE TUMOURS WITH LIVER METASTASES. RESULTS OF THE UKI NETS SURVEY

A. Ahmed1, G. Turner2, B. King1, L. Jones3, C. Bouvier4, D. McCance2, J. Ardill2, B. Johnston5, M. Rees6, G. Poston3, M. Buxton-Thomas7, M. Caplin4, J. Ramage8. 1Gastroenterology Department, North Hampshire Hospital, Basingstoke, 2Northern Ireland Neuroendocrine Tumour Group, Royal Victoria Hospital, Belfast, 3Department of Surgery, Royal Liverpool University Hospital, Liverpool, 4Neuroendocrine Tumour Unit, Royal Free Hospital, London, 5Neuroendocrine Tumour Unit, Royal Victoria Hospital, Belfast, 6Department of Surgery, North Hampshire Hospital, Basingstoke, 7Department of Nuclear Medicine, 8Institute of Liver Studies, King’s College Hospital, London, UK

Introduction: Midgut neuroendocrine tumours (NET) encompass tumours arising in the jejunum, ileum and proximal colon. Prediction of prognosis and guiding patients about the most appropriate course of action is difficult. This is partly due to the variation in the natural course of the disease but also due to the lack of robust evidence on the many available treatment options

Aims & Methods: We intended to identify the prognostic factors and the effect of interventions on patients with liver metastastic midgut carcinoids across UK and Ireland. Five institutions which are part of UK and Ireland NET group took part in this study. Clinical and biochemical data were collected from hospital charts, pathology reports, radiology reports and biochemistry records for each patient retrospectively, with entries onto a database for each year of each patient under follow-up.

Results: Data was collected on a total of 360 patients. The mean duration of follow-up from diagnosis was 5.29 years. 159 patients died during follow-up. The median survival from time of diagnosis of liver metastases was 5.95 years (SE 0.476). Increasing age at diagnosis, raised urinary HIAA levels, raised plasma Chromogranin A levels, high Ki67 and size of largest liver metastases >4 cm at presentation were identified as factors associated with poor outcome. Resection of liver metastases, Resection of small bowel primary, treatment with Somatostatin analogue therapy and treatment with targetted radiotherapy was associated with improved prognosis. Multivariate analysis using a Cox’s proportional hazards model revealed that age at diagnosis (sig 0.014), Ki67 level (sig 0.039) and resection of primary (sig 0.015) were independent predictors of survival.

Conclusion: This is the largest study to our knowledge specifically looking at the prognosis and clinical course of patients with liver metastatic midgut NETs. For the first time we have shown that Ki67 and resection of primary are independent predictors of survival. In the absence of randomised controlled trials for established therapies this gives evidence that will be useful in the management of these patients.

FIBROBLAST GROWTH FACTOR 19 MAY HAVE A ROLE IN THE PATHOGENESIS OF IDIOPATHIC BILE ACID MALABSORPTION

A. M. Tasleem1, G. W. Brydon2, T. Dew3, S. Balesaria1, C. W. LeRoux4, J. R. F. Walters5. 1Gastroenterology, Imperial College London, London, 2Clinical Biochemistry, Western General Hospital, Edinburgh, 3Clinical Biochemistry, Kings College Hospital, 4Metabolic Medicine, 5Gastroenterology, Imperial College Healthcare, London, UK

Introduction: Idiopathic bile acid malabsorption (IBAM) is an under-recognised syndrome where excess colonic bile acids produce chronic diarrhoea, which responds to sequestrants such as colestyramine. Bile acids are absorbed in the ileum and undergo enterohepatic circulation. Unlike the situation in secondary bile acid malabsorption, evidence suggests that the ileal uptake of bile acids is normal in IBAM. Recently Fibroblast Growth Factor 19 (FGF19), produced by the ileum, has been suggested to have a role in regulating bile acid synthesis.

Aims & Methods: We hypothesised that disordered signalling of FGF19 from the ileum to the liver could be a factor leading to IBAM. FGF19 transcripts in mucosal biopsies of human intestine were investigated in archived samples by quantitative RTPCR. Fasting serum levels of FGF19 and 7α-hydroxy-4-cholesten-3-one (C4), to quantify bile acid synthesis, were measured in 14 patients with chronic diarrhoea, where IBAM had been diagnosed by SeHCAT testing and who had been screened for secondary causes of BAM, and also in 8 controls without diarrhoea, also screened for symptoms of malabsorption. C4 was determined using HPLC and FGF19 by commercial ELISA.

Results: FGF19 mRNA was abundantly expressed in human ileum with wide variation in IBAM patients and in controls, but was undetectable in duodenum. In the IBAM patients, the mean bowel frequency was 3.1 vs 1.6/24 h in controls (p = 0.04, Mann-Whitney U test). The mean C4 was significantly higher in patients than in controls (59.0±9.5 vs 18.1±3.7 ng/ml; p = 0.0006). FGF19 was easily detected in human serum and median fasting values were lower in patients then in controls, with borderline significance (128 vs 222 pg/ml, p = 0.056). FGF19 and C4 showed a significant inverse relation overall (r = −0.415; p = 0.027, Spearman rank).

Conclusion: This is the first study of FGF19 in IBAM. The results obtained indicate that disordered signalling from the ileum to the liver by FGF19 could play a role in the pathogenesis of IBAM.

PRELIMINARY DOUBLE BALLOON ENTEROSCOPY EXPERIENCE FROM A TERTIARY CENTRE IN THE UK

R. Sidhu1, M. E. McAlindon1, D. P. Hurlstone1, M. Thomson2, I. C. Cameron3, S. Hardcastle1, D. S. Sanders1. 1Department of Gastroenterology, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Trust, 2Department of Gastroenterology, Sheffeld Children’s Hospital, 3Department of Surgery, Royal Hallamshire Hospital, Sheffield Teaching Hospitals NHS Trust, Sheffield, UK

Introduction: Double balloon enteroscopy (DBE) is a novel technique that has revolutionised small bowel (SB) examination. This technique has the potential to allow pan-endoscopic visualisation of the SB, with the ability to perform therapeutic intervention. Since its introduction in 2003, the use of DBE has surged internationally. Despite this, there is still no published literature from the UK

Aims & Methods: To assess the diagnostic yield of DBE and impact on patient management in a tertiary UK setting. Data were collected prospectively on all patients referred for DBE over 15 months. Case notes were reviewed for patient demographics and referral indications. All patients had capsule endoscopy (CE) prior to DBE and the route for DBE was based on location of abnormality on CE (that is, capsule reader described findings as proximal or distal small bowel). DBE findings, therapeutic intervention performed and subsequent change in management were noted.

Results: Thirty DBE procedures were performed in 21 patients between July 2006 and October 2007. 66.7% of the referrals originated from outside the region. Sixteen patients had DBE via the oral route and 14 patients had DBE via the anal route. 62% were female with a mean age of 55 years (range 20–77 years). The indications were obscure gastrointestinal bleeding - overt bleeding (n = 6) and iron deficiency anaemia (n = 12), suspected Crohn’s disease (n = 2) and assessment of polyps (n = 1). Four patients were transfusion dependent whilst 19% had significant cardiovascular co-morbidity. The mean procedure time was 70 min (range 2–120) with no reported complications. The overall diagnostic yield was 43%. The diagnoses were angiodysplasia (n = 4/19%), ulcers (n = 4/19%) and tumour (n = 1/4.8%). DBE was normal in 57% and unsuccessful in 4 patients (3 anal route, 1 oral route) due to poor patient tolerance. Intervention was performed in 38% (biopsy 50%, argon plasma coagulation 50%) with a subsequent change in management in 33% of patients. The diagnostic yield was not significantly different when comparing referral indications (p = 1) or whether referred from outside the region (p = 0.4).

Conclusion: Our preliminary experience from the UK has shown that DBE is a useful modality for diagnostic and therapeutic intervention in patients with suspected small bowel pathology. DBE is feasible within the UK setting, with a high diagnostic yield, low complication rate and a positive impact on patient management. Larger studies are required particularly in high-risk patient groups before advocating its use in routine clinical practice.

WHAT ARE THE IMPLICATIONS OF NEWLY IDENTIFIED COELIAC DISEASE IN PATIENTS WITH TYPE 1 DIABETES MELLITUS? EFFECT UPON GLYCAEMIC CONTROL, QUALITY OF LIFE, CARDIAC RISK FACTORS AND PERIPHERAL NERVE FUNCTION

J. S. Leeds1, A. D. Hopper1, M. Hadjivassiliou2, S. Tesfaye3, D. S. Sanders1. 1Gastroenterology, 2Neurology, 3Diabetes, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Previous work from our group has shown that the prevalence of coeliac disease (CD) in Type 1 diabetes mellitus is 33/1000 (3.3%). Those with newly identified CD had worse glycaemic control compared to age, sex-matched controls but similar quality of life scores. CD has also been associated with peripheral neuropathy but there are no data examining peripheral nerve function in patients with both CD and Type 1 diabetes mellitus. The effect of a gluten-free diet (GFD) on glycaemic control, quality of life, cardiac risk factors and peripheral nerve function are unknown.

Aims & Methods: From the initial 1000 patient cohort 4 specific groups were identified and studied; Group 1 Type 1 diabetes and new CD (n = 12), Group 2 Type 1 diabetes and established CD (n = 12), Group 3 Type 1 diabetes alone (n = 20), Group 4 Type 1 diabetes with positive coeliac antibodies but normal duodenal biopsy (n = 17). Patients with CD alone were recruited to comprise group 5 (n = 15). HbA1c was used to assess glycaemic control and SF-36v2 was used to assess quality of life. Peripheral nerve function was assessed by nerve conduction velocities in the right radial, sural, tibial and common peroneal nerves. Cold, vibration and heat-pain detection threshold was determined using the CASE IV machine and cardiac autonomic function was assessed using the O’Brien protocol.

Results: There were no differences in age, gender distribution or body mass index between the groups. In group 1 median HbA1c was 8.2% prior to starting a GFD and 8.6% after 1 year (p = 0.45) and SF-36v2 scores did not change significantly. There were no changes in total cholesterol or triglyceride levels but cholesterol: HDL ratio increased from 3.7 to 3.2 (p = 0.033). On cardiac autonomic testing, mean resting heart rate was significantly lower in group 5 vs the other groups (65.5 vs 75.5, p = 0.028) but mean systolic and diastolic blood pressures were comparable. Radial and common peroneal nerve conduction velocities were significantly reduced in group 1 and 2 compared to the other groups (p = 0.007 and p = 0.0001 respectively). However, there were no differences in cold, vibration and heat-pain detection thresholds between the groups.

Conclusion: Following identification of CD in Type 1 diabetes and subsequent treatment with a GFD glycaemic control and quality of life are not altered however there appears to be improvement in HDL cholesterol levels. Resting heart rate was lower in non-diabetic coeliac patients but blood pressure was similar. Patients with Type 1 diabetes and CD have a higher prevalence of peripheral nerve dysfunction compared to controls.

CHANGES IN LIPID PROFILE WITH EXPOSURE TO A GLUTEN-FREE DIET IN INCIDENT CASES OF COELIAC DISEASE: A PROSPECTIVE COHORT STUDY

N. R. Lewis1, D. S. Sanders2, R. B. Hubbard1, R. F. A. Logan1, J. West1. 1Division of Epidemiology and Public Health, University of Nottingham, Nottingham, 2Gastroenterology and Liver Unit, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Undetected coeliac disease is associated with 0.5 mmol/l reduction in mean serum cholesterol in comparison to the general population.1 A review of incident coeliac disease patients that had happened to have their lipid profile measured observed a significant increase in total cholesterol and HDL cholesterol with exposure to a gluten-free diet.2 Since each 0.03 mmol/l increase in HDL cholesterol is associated with a 23% reduced risk of future coronary heart disease,3 we prospectively observed the effect of a gluten-free diet on the lipid profile of a contemporary cohort of incident coeliac disease patients.

Abstract 054

Aims & Methods: Lipid profile has been prospectively collected on incident cases of coeliac disease starting a gluten-free diet at the Royal Hallamshire Hospital since October 2004, at diagnosis and followed up for 12 months (n = 51). We compared measures at diagnosis and 12 months after using paired t tests

Results: There were twice as many female as male incident cases but the mean age at diagnosis was 47 years (SD 15) and similar between men and women (p = 0.45). 28% had partial villous atrophy; 40% had subtotal villous atrophy and 23% had total villous atrophy. The median tTG at diagnosis was 223 IU (range 5300) and 90% (n = 80) were EMA positive. The mean value of HDL cholesterol and total cholesterol: HDL significantly increased with exposure to a gluten-free diet (table) as did folate. These changes did not vary by gender or by age. However, on stratifying according to severity of disease as measured by tTG titre and Marsh grading, HDL cholesterol and total cholesterol: HDL only significantly increased in those people with more severe coeliac disease.

Conclusion: Our data confirm HDL-cholesterol increases in people with celiac disease after starting a gluten-free diet. Improved intestinal absorption or some other reason may be responsible. We conclude that gluten-free diet treatment does not have an adverse effect on lipid profile and may be beneficial with regard to risk of coronary artery disease.

Joint IBD/paediatric/nutrition symposium

SINGLE-CENTRE EXPERIENCE OF EXCLUSIVE ENTERAL NUTRITION FOR INDUCTION OF REMISSION IN MORE THAN 100 CHILDREN WITH CROHN’S DISEASE

R. K. Russell, E. Buchanan, T. Cardigan, K. Hassan, D. Young, P. McGrogan. Paediatric Gastroenterology, Yorkhill Hospital, Glasgow, UK

Introduction: Enteral nutrition achieves remission rates between 20–84% in Crohn’s disease (CD) patients. We describe our experience of treating CD with a primary course of enteral nutrition and factors effecting treatment outcome.

Aims & Methods: Patients treated with enteral feeds over the past 3 years were studied. CD was diagnosed by standard criteria. Standard treatment was 8 weeks of exclusive enteral nutrition. Disease remission was assigned by a combination of clinical parameters (patient well being, weight gain, stool frequency and inflammatory markers). Disease phenotype was assigned using the Vienna classification and a paediatric classification that examined disease activity throughout the GI tract. Inflammatory markers were assessed at treatment start and end. Anthropometry (z scores) were calculated on starting treatment and at 2, 3 and 6 months later.

Results: 114 children were treated (4 were excluded from further analysis). There were 65 males (59%), median age at diagnosis 11.6 years (IQR 9.5–13.0) and median age at treatment of 12.1 (IQR 9.8–13.5). 57 (52%) took feeds orally, 48 (44%) via n-g tube and 4 (4%) via a PEG. Median length of treatment was 55.5 days (IQR 49.0–58.0) with 83 (75%) completing at least 7 weeks of treatment. Modulen IBD (Nestle) was used in 105 and EO28 (SHS) in 5. Median energy delivered was 107% of EAR (105–108) and protein 212% (IQR 204–220). Clinical remission was achieved in 88 (80%). In patients in clinical remission there was a significant reduction in both the median ESR and CRP at the start and end of treatment (38 vs 13 and 21 vs 6, respectively p<0.0001 for both). Patients in clinical remission gained a significant amount of weight (median weight gain 4 kg, p = 0.009 95% CI 1 to 7) reflected in a significant improvement in weight z score at the start and end of treatment (−0.93 vs 0.35, p<0.0001) but not in height z-score (p = 0.32). Unifactorial analysis examining sex, age, method of feeding, disease location and disease behaviour did not demonstrate any significant difference in patients achieving clinical remission. There were 19% of patients who had isolated colonic disease. The clinical remission rate in patients with isolated colonic disease was no different compared to other disease locations (79% vs 84%, p = 0.56). This was reflected in a significant drop in ESR in these patients at the start and end of treatment (46 vs 16 p = 0.001).

Conclusion: This large study demonstrates enteral nutrition is well tolerated and results in clinical remission, normalisation of inflammatory markers together with significant improvement in weight/BMI z-score in most patients. Importantly this study demonstrates clinical remission is not influenced by CD location suggesting enteral nutrition should be offered to all patients regardless of disease phenotype at diagnosis.

Gastroduodenal free papers

INFLUENCE OF ASPIRIN AND NON-STEROIDAL ANTI-INFLAMMATORY DRUG USE ON THE OUTCOMES OF BLEEDING PEPTIC ULCERS: 10-YEAR EXPERIENCE OF A SPECIALISED BLEEDING UNIT

U. Basavaraju, J. M. Thomson, D. Armour, N. A. G. Mowat, A. Fraser. Gastroenterology, Aberdeen Royal Infirmary, Aberdeen, UK

Introduction: The use of Aspirin, non-steroidal anti-inflammatory drug use (NSAID) and Warfarin has steadily increased over the past decade. Their use is amongst the main contributors to peptic ulcer bleeding requiring hospital admission. Since its inception in 1991, the Gastrointestinal Bleeding Unit (GIBU) has accepted all suspected bleeding admissions from a stable population of 485 600. We previously reported that while taking warfarin is associated with an increased incidence of having a significant bleed, an ulcer with stigmata of recent haemorrhage (SRH) and need for endotherapy, there is no statistically significant difference in the rebleeding, surgery or mortality rates. Here we analyse the effects of Aspirin and NSAID use on the outcomes of peptic ulcer bleeding.

Aims & Methods: From October 1991–2001, data were prospectively collected for all admissions with peptic ulcer bleeding to the GIBU. Patients were assigned to 4 groups: the “Aspirin” group were taking Aspirin alone; the “NSAID” group were taking NSAID alone; the “Combined” group were taking both Aspirin and NSAID; the “Control” group were taking neither drug. Statistical analysis with χ2 test was used and p value of <0.05 was considered significant.

Results: There were 1978 admissions with peptic ulcer bleeding, of whom 476 (24%) were on Aspirin alone, 336 (17%) were on NSAID alone, 137 (6.9%) on both Aspirin and NSAID and 1018 (51%) were taking neither drug. The use of these drugs has steadily increased over the 10 year period. Significant bleeding (Hb<10 g/dl, haemodynamic compromise) was more common in both the Aspirin (65%) and NSAID (69%) groups compared to 61% in the control group, but only statistically significant in the NSAID group (p<0.05). 73% in Aspirin and 74% in NSAID group required blood transfusion to maintain Hb>10 g/dl compared with 59% in Control group (p<0.001 for both). Use of these drugs made these patients more likely to have SRH at endoscopy (Aspirin 46% (p = 0.04), NSAID 50% (p = 0.002), Control 41%). Endotherapy rates were higher in the Aspirin (29%) (p<0.02) and NSAID groups (23.2%) (p = NS) compared to the Control group (22.9%).The rebleeding rate was significantly higher in Aspirin (19%) (p<0.02) and NSAID (20%) (p<0.01) groups than in the Control group (14%). Surgery rates were not significantly different among these groups. The 30-day mortality was 6.5% in the Aspirin group (p = NS), 8.9% (p<0.05) in the NSAID group and 13% (p<0.01) in the Combined group compared to 6% in the Control group.

Conclusion: Patients taking Aspirin or NSAID prior to admission with peptic ulcer bleeding were more likely to have a significant bleed, require blood transfusion and have SRH at endoscopy. Although these patients were no more likely to require surgery, they had an increased mortality suggesting that deaths were as a result of comorbid disease rather than bleeding.

THE EFFECT OF NATURAL POLYMORPHISMS UPSTREAM OF THE HELICOBACTER PYLORI VACUOLATING CYTOTOXIN GENE, VACA, ON VACA TRANSCRIPTION LEVELS

C. G. Masters, D. P. Letley, N. C. Doherty, J. C. Atherton. Wolfson Digestive Diseases Centre and Institute of Infections Immunity and Inflammation, University of Nottingham, Nottingham, UK

Introduction: The vacuolating cytotoxin is an important H pylori virulence factor, and different strains produce different amounts. This is due mainly to differences in vacA transcription. A vacA promoter and a transcriptional start point (TSP) have been identified between the upstream housekeeping gene, cysteinyl-tRNA synthase (cysS) and the vacuolating cytotoxin gene, vacA. Sequence analysis revealed that −35 (TTTATG) and −10 (TAAAAA) motifs are well conserved. Mutations made in this region by Forsyth et al (1999)1 resulted in a decrease in vacA transcription in vitro. In 2004 Ayala et al identified an inverted repeat upstream of the −35 motif present in strains transcribing high levels of vacA.2

Aims & Methods: Biopsies were taken from 30 patients attending for endoscopy in Nottingham. H pylori strains were isolated then grown in broth culture. RNA was extracted both from broth cultures and directly from biopsy specimens, vacA transcription levels were analysed by real-time PCR. The cysS-vacA intergenic region was sequenced for each strain, and natural polymorphisms correlated with vacA transcription in vitro and in vivo.

Results: Analysis of the cysS-vacA intergenic region revealed that differences between strains in the −35 and −10 regions were not significantly associated with levels of vacA transcription, and that regions within the promoter associated with differences in vacA production between strains differed in vitro and in vivo. Sequence alignment revealed that strains with a fully palindromic inverted repeat upstream of the −35 motif expressed more vacA than other strains in vitro (p = 0.01). Analysis of vacA transcription from the same strains in vivo revealed that strains with an A within the second stem of a different stem loop near the transcriptional start site expressed more vacA than those with a G at this position (p = 0.006).

Conclusion: Naturally occurring polymorphisms in two stem loops upstream of vacA are associated with differences in vacA transcription level. Mechanisms remain to be determined: effects on regulator binding or mRNA stability are possible. Different and opposing effects in vitro and in vivo imply in vivo regulation of vacA transcription. Differences in toxin production between H pylori strains could help explain why only some infections result in disease.

GHRELIN INCREASES NUCLEAR FACTOR-KAPPA B EXPRESSION IN A HUMAN LYMPHOCYTE CELL LINE

E. Z. H. Sung1, N. F. da Silva1, S. J. Goodyear1, R. Prajapati1, P. J. McTernan2, C. U. Nwokolo1. 1Gastroenterology, University Hospital Coventry, 2Unit for Diabetes and Metabolism, Warwick Medical School, Coventry, UK

Introduction: Ghrelin is a 28 amino acid orexigenic peptide produced predominantly by the stomach. Interest has increased in its other putative roles including immune-modulation. There is some evidence that ghrelin can affect the release of cytokines from primary lymphocytes and monocytes. The aim of this study was to assess the effects of octanoyl-ghrelin on the p65 subunit of nuclear factor-kappa B (NFκB) (a nuclear transcription factor involved in cellular inflammatory response) in a human B-lymphocyte cell line.

Aims & Methods: A B-lymphocyte cell line (WILCL) was cultured in RPMI 1640 media enriched with L-glutamine, fetal bovine serum and penicillin/streptomycin. Expression of ghrelin receptors (GHSR 1A and 1B) was confirmed by western blotting. The cells were incubated in an atmosphere of 5% CO2 at 37°C and cell density was maintained between 1–3×106 cells/ml. Cells were then transferred to smaller flasks maintaining a density of 1×106 cells/ml. There were 4 starting cell groups which were treated with either control solution or 3 concentrations of octanoyl ghrelin (1 nM, 10 nM and 100 nM). Cell aliquots from each group were cultured for 6, 24 and 48 hours with or without phyto-haemagglutinin (PHA). The cells were then washed in PBS, protein extracted and NFKB activity and concentration estimated using the Trans Am NFKB activity assay. Experiments were done in triplicate.

Results: After 6 hours of culture, for PHA treated (activated) cells mean (±SEM) NFKB activity for control cells was 0.31±0.083, and was increased to 0.35±0.102, 0.43±0.133, 0.32±0.062 for 1 nM, 10 nM and 100 nM ghrelin concentrations respectively. After 6 hours of culture, for non-PHA activated (resting) cells a similar trend was observed with mean (±SEM) NFKB activity for control cells of 0.3±0.044 and 0.3±0.055, 0.37±0.121, 0.32±0.063 for 1 nM, 10 nM and 100 nM ghrelin concentrations respectively. Octanoyl ghrelin was associated with a stepwise increase in NFKB activity at 1 nM and 10 nM concentration in activated cells but NFKB expression at 100 nM concentration of ghrelin was less than at 10 nM concentration. In all experiments PHA activated cells express more NFKB than resting cells.

Conclusion: Octanoyl-ghrelin seems to up regulate NFKB activity in the WILCL B lymphocyte cell line which expresses the ghrelin receptor. This should lead to the transcription of various cytokines involved in inflammatory responses. Our observation is contrary to the data from some recent studies which suggest an immune-inhibitory role for ghrelin. The reason for the tailing-off of the ghrelin effect at 100 nM concentration is uncertain.

GASTRIC CANCER CELL MIGRATION IS REDUCED BY NOVEL MIGRASTATIN ANALOGUES

G. Horgan1, D. Murray2, A. Madden2, G. Bennett2, G. Anquetin3, P. Murphy3, J. Lennon1, P. MacMathuna1, P. Doran2. 1Gastrointestinal Unit, Mater Misericordiae University Hospital, 2UCD School of Medicine and Medical Sciences, UCD/Mater Clinical Research Centre, Mater Misericordiae University Hospital, 3UCD Centre for Synthesis and Chemical Biology, Belfield, Dublin, Ireland

Introduction: Gastric cancer (GC) aetiology is unclear, its prognosis is poor and effective chemotherapy regimes have yet to be developed. We have previously reported NET1 to be increased in GC, whereby it facilitates cell migration and invasion via RhoA activation and subsequent cytoskeletal re-arrangements. Migrastatin analogues (MA) have been shown to reduce tumour cell migration and metastasis in a range of malignancies but have yet to be investigated in GC.

Aims & Methods: In this study we assessed the effect of seven novel MAs on: (A) GC cell proliferation and migration and (B) NET1 mRNA expression and cytoskeletal re-arrangements. GC cells were treated with varying doses of the seven novel MAs. Cell proliferation was assessed using an MTS assay. GC cell migration was investigated using a fluorescent trans-well assay and also using a scratch “wound-healing” assay. NET1 mRNA expression was quantitated using real-time PCR. F-actin assembly was assessed by immuno-fluorescently staining.

Results: 5 of the 7 compounds caused reduced tumour cell migration as assessed using the transwell assay over 24 h (p<0.001). The IC50 migration values for these compounds for were; Compound A; 17 uM, Compound B; 29 uM, Compound D; 32 nM, Compound F; 1 uM, Compound G; 2 uM. 5 of the compounds caused reduced tumour cell migration as assessed using the scratch “wound-healing” assay. (p<0.05). Treatment with Compound E resulted in 65% decreased NET1. No changes in cytoskeletal architecture were observed upon treatment with the compounds. No significant effect on GC cell proliferation was observed upon treatment with the compounds.

Conclusion: We have demonstrated for the first time that novel MAs significantly inhibit GC migration while exhibiting no cytotoxic effects. Furthermore, the reduced NET1 expression upon treatment with Compound E is promising and worthy of further study. Ongoing studies into the underlying mechanism of these compounds may very well establish them as ideal therapeutic agents in GC.

ASSESSMENT AND OUTCOMES IN GASTRIC CARCINOID

L. Marelli1, C. Toumpanakis1, R. Srirajaskanthan1, T. Eves1, J. Watkins2, M. Caplin1. 1Centre for Gastroenterology, 2Department of Histopatology, RFH, London, UK

Introduction: Gastric carcinoid tumours are rare, but increasing in number.1 There are three types: Type I associated with autoimmune atrophic gastritis and hypergastrinemia, which are indolent in nature; Type II associated with multiple endocrine neoplasia (MEN I) and sporadic Type III with poor prognosis.

Aims & Methods: To evaluate clinical, biochemical, endoscopic and histological features of patients with gastric carcinoid. 37 consecutive patients with gastric carcinoid (mean age: 58±13 years; M/F: 23/14) were referred to our Unit between 1998 and October 2007. 31 patients with type I (84%), 3 with type II (8%) and 3 with type III (8%). Data collected retrospectively in 31 patients (n/a in 6 patients with type I).

Results: Median follow-up was 52 months (1–200). 3 patients, all with type III, died for metastatic carcinoid. Symptoms at presentation were: anaemia (55%), dyspepsia (23%), abdominal pain (16%) and GI bleeding (6%). In type I: chromogranin A (CgA) was normal in 65%, increased <×2 normal in 26% and >×2 normal in 9%; all patients had high gastrin (>400 pg/ml in 61%); antiparietal cell antibody positive in 63%. At endoscopy, there was gastritis only in 13%, <3 polyps in 50% and multiple small polyps in 37%. Size of largest polyp (n = 15) was <1 cm in 11, 1–2 cm in 3 and >2 cm in 1. At histology, 20% were micro-carcinoid and 80% low grade carcinoid; MIB-1 proliferation index (n = 18) was <2% in 13, 3–5% in 4 and 5–10% in 1. 79% had atrophic gastritis. EUS (n = 11) showed no infiltration beyond the muscolaris mucosae. OctreoScan (n = 10) showed abnormal uptake only in 1 patient with liver metastases. 2 patients had partial gastrectomy (lymph node involvement in 1 with 25 mm polyp), 3 mucosectomy and 1 polypectomy. In type II: both CgA and gastrin were increased; 2 patients had low grade (MIB-1<2%) and 1 had high grade (40–70%) carcinoid. In type III: gastrin was normal, no atrophic gastritis, all high grade (MIB-1: 35–60%) carcinoid and all developed metastases. One patient died for post-embolisation complication, 1 had total gastrectomy, chemotherapy and died after 19 months and 1 had liver transplant but died of disease recurrence 3 years later.

Conclusion: Management of patient with gastric carcinoid is based on careful identification of tumour type—endoscopy with biopsy (MIB-1), CgA, gastrin, CT scan, OctreoScan, EUS. Type I have good prognosis (at 5 years all patients were alive and only 1 of 31 patients had metastases); polyp >2 cm may be more aggressive. Type II are similar to type I, management depends on the associated MEN1. Type III are more aggressive—in our series, mean survival time was 29 months despite treatment.

Pancreatic section meeting

AZATHIOPRINE IN THE MANAGEMENT OF RELAPSING AUTOIMMUNE PANCREATITIS/IGG4 SCLEROSING CHOLANGITIS

N. S. Sandanayake, G. J. Johnson, C. M. Smyth, N. I. Church, L. Ayaru, M. H. Chapman, A. R. Hatfield, S. P. Pereira, G. J. Webster. Gastroenterology, University College London Hospital, London, UK

Introduction: A rapid clinical, biochemical and radiological response to steroids is a diagnostic feature of autoimmune pancreatitis/IgG4 sclerosing cholangitis (AIP/SC). However, disease relapse may occur, and its clinical importance and management has not been well defined. The use of maintenance immunosuppression, including Azathioprine (AZA), has not been reported in a case series of AIP/SC.

Aims & Methods: Aim of the study was to report our experience of disease relapse and its management in patients diagnosed with AIP/SC between 2004–7, following an initial course of steroids. Disease activity was assessed prospectively on clinical, biochemical, serologic and radiological criteria. Patients with post-steroid relapse, or failed remission, were the focus of this study.

Results: Of 26 AIP/SC patients followed for a median 24 months (range 2–46) from start of initial steroid course, relapse/failed remission occurred in 11 (42%). All were male, median 55 years (range 27–81). Two patient groups were defined: (1) Post-steroid relapse: 7/11 (64%) patients relapsed after initial steroid course, a median 3 months (range 1–7) after stopping steroids, following achieved remission. Indicators of relapse included jaundice and new/worsening hilar +/− intrahepatic biliary stricturing in all patients, and renal failure in one patient with renal IgG4-positive plasma cell infiltrates. Serum IgG4 was elevated in 3/7 patients, with a mean level of 2.31 (normal 0–0.3). All 7 patients received a second tapering course of 30 mg prednisolone daily, while 5/7 patients commenced AZA 2 mg/kg/day. Objective evidence of disease control was seen in all patients, including resolution of jaundice and removal of biliary stents following improvement of strictures. Of the 5 patients maintained on AZA, 3 remained in remission 3, 14 and 19 months after stopping steroids respectively, and the rest were clinically improved on a tapering steroid regimen. 1 patient needed dose reduction of AZA due to persistent nausea. (2) Failed weaning: 4/11 patients responded to initial steroid course, but failed steroid weaning and cessation, at a median 10 months (range 3–22). 3/4 patients commenced AZA 2 mg/kg/day (one declined AZA) with subsequent cessation of steroids in 2/3, and slow weaning of steroids in the remaining patient on AZA.

Conclusion: More than 40% of patients with AIP/SC fail to achieve stable remission after initial short-course steroids, with progression of intrahepatic biliary stricturing commonly seen. Elevated serum IgG/IgG4 titres may predict relapse. Maintenance therapy with AZA, as used in autoimmune hepatitis, appears to be effective, and merits further study.

Colorectal/anorectal free papers

COULD SURVEILLANCE PROTOCOLS BE ALTERED DUE TO LOW YIELD OF SIGNIFICANT COLORECTAL NEOPLASIA IN A HIGH RISK COHORT? INITIAL FIVE YEARS EXPERIENCE OF A COLLABORATIVE HIGH RISK FAMILIAL CANCER SCREENING INITIATIVE IN IRELAND

G. Horgan1, G. Lawlor1, L. O’Brien1, J. Leyden1, N. Ullah1, V. Sibartie1, M. O’Riordain2, G. O’ Sullivan2, J. Lennon1, P. MacMathuna1. 1Gastrointestinal Unit, Mater Misericordiae University Hospital, Dublin, 2Department of Surgery, Mercy University Hospital, Cork, Ireland

Introduction: Identifying and removing significant polyps is the rationale underlying both an average and high risk colorectal cancer screening strategy. However significant variation has been observed in polyp yield which may question current surveillance intervals.

Aims & Methods: To critically evaluate advanced neoplasia yield from a cohort of high risk asymptomatic individuals identified at two dedicated family cancer clinics in Dublin and Cork. In 637 individuals undergoing screening colonoscopy, we analysed the polyp characteristics according to number and pathology subset in relation to age (<50, >50) and risk stratification (medium risk n = 487 and HNPCC n = 103). Colonoscopy results in 47 low (but above average) risk were also included. Fisher’s exact two-tailed t test was used for statistical analysis.

Abstract 062 Screened population colonoscopic findings (n = 637)

Results: Tubular adenoma rates were statistically significant between individuals aged <50 compared to those aged >50 in: (a) low risk cohort (5.6% vs 37.9%; p<0.02) and (b) all risk groups combined (16.4% vs 24%; p<0.02). Tubulovillous adenoma yields were statistically significant between individuals aged <50 compared to those aged >50 in: (a) medium risk cohort (1.6% vs 6.2% p<0.01); (b) HNPCC cohort (0% vs 9.5% p<0.03) and (c) all risk groups combined (1.2% vs 6.7% p<0.03).

Conclusion: Advanced neoplasia in all at risk groups less than 50 years of age is low. Our results raise the possibility to delay initiation of screening in low and medium risk groups until the age of 50. It is worth considering delaying initiation of screening of HNPCC until the 5th decade and widening the screening intervals below the age of 50 as currently recommended.

STROMAL CELL CYCLOOXYGENASE-2 PROMOTES COLORECTAL TUMORIGENESIS IN THE APCMIN/+ MOUSE MODEL OF FAMILIAL ADENOMATOUS POLYPOSIS

M. A. Hull1, D. J. Scott1, R. J. Cuthbert1, S. Ko1, S. L. Perry1, E. Cartwright1, C. Bonifer2, A. F. Markham1, L. Coletta1. 1Section of Molecular Gastroenterology, 2Section of Experimental Haematology, Leeds Institute of Molecular Medicine, Leeds, UK

Introduction: Cyclooxygenase-2 (Cox-2) plays an important role in the early stages of intestinal tumorigenesis. However, the mechanism(s) underlying the pro-tumorigenic activity of Cox-2 remains unclear. In human and ApcMin/+ mouse intestinal adenomas, Cox-2 is localised predominantly to stromal macrophages. We have previously demonstrated that paracrine macrophage Cox-2 activity drives tumorigenic behaviour of intestinal epithelial cells in vitro. Therefore, we tested the hypothesis that stromal cell Cox-2 promotes intestinal tumorigenesis in vivo using a transgenic mouse model of macrophage-specific Cox-2 over-expression.

Aims & Methods: A complete mouse genomic Cox-2 clone was expressed under the control of the chicken lysozyme (cLys) locus, which directs macrophage-specific transgene expression in mice. Transgenic C57Bl/6 (B6) x 129Sv cLys-Cox-2 founders were generated by pro-nuclear injection and one line (termed GC25) chosen for further experiments based on Southern blot analysis and transgene expression studies in bone marrow-derived macrophages. cLys-Cox-2 mice underwent a N6 backcross to the B6 background before crossing with B6 ApcMin/+ mice. Intestinal phenotype analysis was performed in a blinded manner at day 100.

Results: As expected, cLys-Cox-2 ApcMin/+ mice exhibited increased Cox-2 expression in the brain, spleen and intestine compared with non-transgenic ApcMin/+ mice. There was elevated macrophage-specific Cox-2 expression in adenomas and histologically normal (HN) intestinal mucosa, along with increased intestinal tissue prostaglandin (PG) E2 levels (9.1±1.0 ng/mg total protein vs 6.2±1.1 ng/mg for HN colon), in cLys-Cox-2 ApcMin/+ mice compared with non-transgenic ApcMin/+ animals. There was a significant increase in colorectal adenoma multiplicity in cLys-Cox-2 ApcMin/+ mice (n = 12; median 5, IQR 2–11) compared with ApcMin/+ controls (n = 11; 3, 1–6; p = 0.03, Mann-Whitney). There was also a significant increase in tumour size in cLys-Cox-2 ApcMin/+ animals (p = 0.03 for the difference in number of polyps of 4.1–6 mm diameter; Student’s t test). However, there was no significant difference in small intestinal tumour number between cLys-Cox-2 ApcMin/+ mice and non-transgenic controls.

Conclusion: This is the first demonstration that mucosal Cox-2 activity (as opposed to models of exogenous PGE2 administration or Cox-2 inhibition/deletion) drives ApcMin/+ mouse intestinal tumorigenesis. Macrophage Cox-2 activity increases adenoma multiplicity and size in the colon, but not the small intestine. The cLys-Cox-2 mouse model will be useful for future studies of paracrine mucosal Cox-2 signalling as a specific target for colorectal cancer chemoprevention.

ABERRANT CRYPT FOCI FORMATION IS INHIBITED BY PEROXISOME PROLIFERATOR-ACTIVATED RECEPTOR/ALPHA IN APCMIN+/- MICE

A. Shonde1, N. Mandir2, D. Bell3, R. Goodlad2, A. Bennett4, C. Hawkey1. 1WDDC, University Hospital Notingham, Nottingham, 2Histopathology Unit, London Research Institute, Cancer Research, London, 3Biology, 4Biological Sciences, University of Nottingham, Nottingham, UK

Introduction: We have reported that ligand activation of peroxisome proliferator activated receptor (PPAR)-α reduces tumour burden in APCmin+/- mice while deletion of PPAR-α increases formation of colonic tumours.1 We investigated whether aberrant crypt foci (ACF) which are thought to be precursors of sporadic colorectal cancer, played a role in this process.

Aims & Methods: Our initial aim was to investigate ACF formation in APCmin+/- and APCmin+/- PPAR-α-/- mice. Once weaned C57BL/6J APCmin+/- mice (n = 20) and C57BL/6J APCmin+/- PPAR-α-/- mice (n = 20) were fed standard chow (rat and mouse standard diet, BEEKAY, Humberside) throughout the study. These mice were sacrificed when they showed signs of anaemia or lost >20% of weight (term). To investigate ACF formation over time five mice from each genotype were sacrificed at weeks 12, 18, 24 and term. The colons were opened and laid out before being soaked in 0.2% methylene blue. ACF were then analysed using a dissecting microscope.

Results: Tumours developed sooner in APCmin+/- PPAR-α-/- mice (from 12 weeks) than in APCmin+/- mice (from 18 weeks). ACFs were seen in APCmin+/- PPAR-α-/- mice from 12 weeks but never in APCmin/+ mice. Among mice sacrificed because of anaemia or weight loss there was an increase in both ACFs and colonic tumours in APCmin+/- PPAR-α-/- mice. There were no ACF in APCmin+/- mice at term, while in APCmin+/- PPAR-α-/- the mean number of ACF was 1.95 (SEM 0.16) in the mice sacrificed at term. The mean number of large bowel polyps in APCmin+/- mice at term was 0.75 (0.25) and 2.43 (0.30) in APCmin+/- PPAR-α-/-. There were no ACF found in APCmin+/- mice sacrificed at week 12, 18 and 24. However, the mean number of ACF at week 12, 18 and 24 in APCmin+/- PPAR-α-/- was 0.80 (0.36), 0.80 (0.2), 1.80 (0.37) respectively.

Conclusion: Tumours in APCmin+/- mice may not develop from ACF but these appear to be the source of the additional polyps that develop as a result of PPAR-α deletion. PPAR-α may be a key target for chemo prevention of sporadic polyps and cancers.

WITHDRAWN

INCIDENCE OF PERFORATED COLONIC DIVERTICULAR DISEASE AND ITS ASSOCIATED MORTALITY: A UK GENERAL POPULATION BASED STUDY

D. J. Humes1, M. Solaymani-Dodaran2, K. Fleming2, J. Simpson1, R. C. Spiller1, J. West2. 1Wolfson Digestive Disease Centre, Nottingham University Hospital NHS Trust, 2Division of Epidemiology and Public Health, The University of Nottingham, Nottingham, UK

Introduction: Perforated colonic diverticular disease (PCDD) represents one of the most serious complications of diverticular disease and is thought to be associated with high levels of mortality. However, little is known regarding the occurrence and consequence of the disease. We aimed to determine the incidence and mortality associated with PCDD in the UK.

Aims & Methods: Computerised general practice records from the General Practice Research Database were analysed. The patients were registered in the database from 1987 to 2006. Each case was matched to 10 age, sex and practice matched controls. The outcomes measured were incidence and mortality rates by calendar period, age group and gender using Poisson and Cox regression modelling to provide adjusted estimates.

Results: We identified 894 incident cases between 1990 and 2005. The overall incidence of PCDD was 2.49 (95% CI 2.33 to 2.66) per 100 000 person years, and this was highest in older age groups (>85 years 12.27 per 100 000 person years 95% CI 9.88 to 15.23). The incidence rates increased 2.17-fold (95% CI 1.70 to 2.77) when corrected for age and gender between 1990 and 2005 (Graph 1). A total of 312 (34.90%) PCDD patients died compared to 1822 (20.41%) controls. The crude mortality rates for the PCDD and control cohorts were 90.29 (95% CI 80.53 to 101.25) and 46.06 (95% CI 44.00 to 48.23) per 1000 person years. Over the time period of the study PCDD patients were twice as likely to die compared to the general population when corrected for age and sex (HR 2.22 95% CI 1.96 to 2.51).

Conclusion: We found that the incidence of PCDD has increased twofold during the 15 years of our study. Those at older age are more at risk and the mortality associated with the condition is high. There is a high early mortality following PCDD how much of this relates to the perforation or to concomitant disease remains to be determined. With the ageing demographic in the UK we would expect this condition to be a large future burden on health services.

ONCOLOGICAL MEDIUM TERM OUTCOME OF LOCAL ENDOLUMINAL RESECTION OF LOW RISK PTIM/T1 RECTAL CANCER: CAN WE CHANGE THE MANAGEMENT PARADIGM?

D. P. Hurlstone1, S. R. Brown2, S. Amin2, I. Adam2, A. J. Shorthouse2, P. Skinner2, L. Hunt2, W. Baraza3, S. S. Cross3, N. Tiffin2. 1Endoscopy, Royal Hallamshire Hospital, 2Colorectal MDT Working Group, STH NHS Foundation Trust, 3Colorectal MDT Working Group, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Local excision of early rectal cancer is controversial, despite the non-inclusion of locally draining lymph nodes. Limited resection of low risk pTim/T1 disease can potentially achieve an adequate oncological outcome with lower morbidity and mortality compared with radical resection. We have previously shown R0 resection is achievable (both for Paris 0-II/Is morphology tumours) using EMR +/- combination endoscopic submucosal dissection (ESD) with full thickness excision achievable using the latter technique. Furthermore, overall 30-day mortality, procedure related complications and in-patient stay were superior to conventional transanal approaches in our previous series. Concern regarding local recurrence, nodal disease positivity and overall medium/long-term oncological outcome has limited practice of these techniques in the UK due to the lack of this mandatory evidence base.

Aims & Methods: Single centre, prospective cohort analysis of pTim/T1 rectal cancer resected using endoscopic local excision with curative intent. Outcome measures: post-dissection mortality/morbidity, local recurrence rates, tumour free and overall survival. 43 patients underwent local endoscopic excision using EMR+/-ESD. All lesions were staged pre-resection using triple modality imaging (magnification chromoscopy/EUS/pelvic MRI). Patients with T2 disease or node “+” were referred for surgery. Post-resection follow-up: 1. Chromoscopic colonoscopy + 12.5/20 MHz EUS at 3, 6, 12 months post-resection and annually thereafter. Higaki and Chak criteria were used to define recurrent disease at colonoscopy and EUS respectively.

Results: Data were available for analysis on 41 patients (2 lost to follow-up). 30-day re-admission/death rate = 0%. Early/delayed bleed rates = 3/41 (7%), haemastasis achieved in all using an endoscopic approach. Perforation rate = 1/40 (2.5%). Median in-patient stay = 8 hours (range 2–76). Resection status: R0 28/41(68%)/Rx 7/41(17%)/R1 6/41(15%). Of the 6 patients defaulting to R1 all underwent a R0 endoscopic dissection. Local recurrence rates = 2/41 (4.8%) – 1 patient Higaki criteria “+”/1 patient Higaki “−”(poorly differentiated mucinous compared to well-differentiated adenocarcinoma with TVA at index resection). Median follow-up 38 months (range 10–66): tumour free survival = 39/41 (95%)/overall survival 40/41 (97%).

Conclusion: This is the first UK study to show endoscopic resection of low-risk early rectal neoplasia can result in an oncological acceptable outcome in the medium term. Survival results are comparable to conventional transanal/transanal endoscopic microsurgical dissection.

MICRODISSECTING THE CARCINOGENESIS PATHWAYS IN SPORADIC AND COLITIS-ASSOCIATED COLON CANCER

S. J. Leedham1, T. Graham2, M. Rodriguez-Justo3, R. Harrison4, M. Novelli3, J. A. Jankowski5, N. A. Wright1. 1Histopathology Unit, 2Molecular and Population Genetics, Cancer Research UK, 3Pathology Department, University College Hospital, 4Pathology Department, Leicester General Hospital, London, 5Clinical Pharmacology, University of Oxford, Oxford, UK

Introduction: The adenoma-carcinoma sequence in sporadic colon cancer is the most established and best studied carcinogenesis pathway. It is thought that founder mutations in adenomatous polyposis coli (APC) are followed by K-ras lesions as polyps grow. p53 mutations occur late as an adenoma progresses to a cancer. The pathway to cancer in ulcerative colitis (UC) is less clearly understood.

Aims & Methods: We aimed to investigate these carcinogenesis pathways by assessing the mutation load of individual crypts at different stages of the dysplasia-carcinoma sequence in sporadic and colitis-associated cancer. Individual crypts were microdissected across 4 colitis-associated and 4 sporadic colectomy blocks containing carcinoma arising from dysplasia, as well as 15 sporadic polyps of varying sizes and 5 colitic dysplasias. APC, K-ras and p53 mutation load of normal, dysplastic and malignant crypts were then analysed. Individual lesions were also assessed for aneuploid change.

Results: Sporadic pathway: a small minority of sporadic polyps appeared to be polyclonal for APC. K-ras mutations, when present usually affected the entire adenoma, and p53 mutations were present in adjoining dysplastic as well as malignant crypts. Colitis-associated: p53 mutations were founder lesions in about 40% of cases, however one patient had a founder K-ras mutation. Dysplastic plaques are clonal lesions but synchronous tumours are usually made up of different clones. There was an increased frequency of aneuploidy in colitis-associated lesions.

Conclusion: Sporadic pathway: K-ras mutation may occur earlier or sweep through polyps more quickly than expected. p53 mutations do not always appear to be directly responsible for malignant change as they can be present in adenomatous crypts as well as their malignant counterparts. Colitis-associated: dysplastic plaques appear to be clonal as they shared identified founder mutations. Localised field cancerisation does appear to occur, probably as a consequence of increased crypt fission. No encompassing mutation sequence could be established and multiple carcinogenesis pathways are likely.

EXTRACELLULAR GALECTIN-3 INTERACTS WITH CELL SURFACE MUC1 TO PROMOTE CANCER CELL AGGREGATION, ADHESION AND TRANS-ENDOTHELIAL INVASION: AN IMPORTANT MECHANISM FOR METASTASIS

Q. Zhao1, M. Barclay1, J. Hilkens2, J. M. Rhodes1, L. G. Yu1. 1School of Clinical Science, University of Liverpool, Liverpool, UK, 2Divison of Cancer Biology, Netherlands Cancer Institute, Amsterdam, The Netherlands

Introduction: Cancer cell homotypic aggregation, heterotypic adhesion to and invasion through the endothelium are three key steps in the cancer metastasis cascade. Molecular mechanisms underlying these steps are not fully understood. We recently showed that the human lectin galectin-3, which can show up to fivefold increased serum concentration in cancer patients, interacts with the oncofetal Thomsen-Friedenreich (TF) (Galβ1,3GalNAc) carbohydrate antigen on MUC1 to promote cancer cell adhesion to endothelium by cross-linking cell surface MUC1 to reveal underlying adhesion molecules (Yu et al, J Biol Chem, 2007). We have now assessed the effects of this interaction on homotypic cancer cell aggregation and trans-endothelial invasion.

Aims & Methods: The aim of this study is to investigate the effects of MUC1-galectin-3 interaction on homotypic cancer cell aggregation and trans-endothelial invasion.

Results: Human colon cancer HT29-5F7 cells, a sub-population of HT29 cells that strongly express MUC1, in comparison with parent HT29 cells show less cell-cell aggregation (3.79±0.79%, mean ±SE vs 8.64±0.1%, p<0.01) assessed by FACS and less adhesion to human umbilical vein endothelial cells (HUVEC) either unstimulated: (8.6±0.77% vs 14.3±1.17%, p<0.01) or pre-stimulated with TNFα (15.5±2.75% vs 21.9±3.44%, p<0.05). MUC1-transfected human breast epithelial cancer cells (HCA1.7+) and human melanoma cells (ACA19+) also show much less cell-cell aggregation and adhesion to HUVEC than MUC1-negative transfectants (HCA1.7− and ACA19−). Recombinant galectin-3, at concentrations (0.1–1 μg/ml) similar to that found in the sera of cancer patients, caused cell surface clustering of MUC1 and increased cell aggregation and adhesion of high MUC1-expressing HT29-5F7 and HCA1.7+ cells, but not of HT29 and HCA1.7− cells. Recombinant galectin-3 increases invasion of ACA19+ (164±24%, p<0.05), but not ACA19−, through HUVEC monolayer. These effects of galectin-3 were largely prevented by the presence of lactose and also by anti-E-cadherin (in aggregation) or anti-CD44H or –E-selectin antibody (in adhesion).

Conclusion: These results suggest that circulating galectin-3 in patients with epithelial cancers including colon cancer may promote metastasis by interacting with MUC1/TF and revealing underlying adhesion molecules that allow homotypic aggregation and trans-endothelial invasion. Prevention of this interaction might plausibly prevent cancer metastasis.

INCIDENCE AND IMPLICATIONS OF PRE-POUCH ILEITIS IN PATIENTS FOLLOWING RESTORATIVE PROCTOCOLECTOMY

S. D. McLaughlin1, S. K. Clark2, A. J. Bell3, P. P. Tekkis1, P. J. Ciclitira4, R. Nicholls1. 1Department of Biosurgery and Surgical Technology, Imperial College, 2Department of Surgery, St Mark’s Hospital, London, 3Department of Gastroenterology, Weston-Super-Mare General Hospital, Weston-Supermare, 4Department of Gastroenterology, Nutritional Sciences Research Division, King’s College, London, UK

Introduction: Pouchitis following RPC for ulcerative colitis (UC) is common. Inflammation proximal to the pouch; pre-pouch ileitis (PI) has recently been described,1 its incidence and implications are unknown. Some authors have suggested it may indicate missed Crohn’s disease (CD).

Aims & Methods: The aim of this study was to identify the incidence of pre-pouch ileitis at pouchoscopy and correlate this with symptoms, diagnosis and outcome. It is routine practice at our institution to intubate the pre-pouch ileum at pouchoscopy. We searched the endoscopy database for the terms “pouchitis” and “ileitis” in all pouchoscopy procedures performed at our institution between April 2004 and September 2007, hospital records were then reviewed. Those with CD were excluded.

Results: 1448 pouchoscopies were performed on 754 patients (polyposis 179), (UC/indeterminate colitis (IC) 568), (CD 5). Pouchitis was diagnosed in 254 (34%) patients. 35 patients were identified with endoscopic signs of pre-pouch ileitis (UC/IC), (FAP 1). 1 patient took NSAIDs and was excluded from further analysis. All had concurrent pouchitis. 3 had ileal stricturing. PI length was recorded in 32 procedures (65%), mean PI length  = 13.3 cm. Of the 35 patients with PI 25% were asymptomatic. Histology of the terminal ileum at the time of colectomy was available in 17 patients, 6 (35.4%) had histological backwash ileitis. At follow-up (mean 13 months) n = 0 were reclassified as Crohn’s disease, and no patient required an ileostomy for poor function. Histology of the terminal ileum at time of colectomy was available in 19 patients, 6 (34%) had histological backwash ileitis.

Conclusion: 34 of 568 patients with an ileo-anal pouch for UC/IC and 1 of 179 patients with polyposis were found to have PI. All had associated pouch inflammation, however not all patients were symptomatic. The incidence of backwash ileitis in this group was twice that expected (18%) at colectomy for UC;2 this may suggest a similar disease process. Our results demonstrate that PI is more common than previously thought; it does not imply missed Crohn’s disease or predict an increased rate of pouch failure at least in the short term.

MOST BOWEL CANCER SCREENING COLONOSCOPIES ARE UNDERTAKEN FOR WEAK POSITIVE FAECAL OCCULT BLOOD RESULTS

C. J. Rees1, M. Ritchie1, I. Ward2, D. Nylander3, J. Painter3, J. Singh2, L. Hurst4, M. Rutter4. 1South of Tyne Bowel Cancer Screening Centre, South Tyneside General Hospital, South Shields, 2South of Tyne Bowel Cancer Screening Centre, Queen Elizabeth Hospital, Gateshead, 3South of Tyne Bowel Cancer Screening Centre, Sunderland Royal Infirmary, Sunderland, 4Tees Screening Centre, University Hospital of North Tees, Stockton, UK

Introduction: Individuals who participate in bowel cancer screening complete 6 faecal occult blood (FOB) testing windows on 3 separate stool samples (2 windows for each day). If 5 or 6 windows are positive the result is classed as abnormal and they are invited to attend for colonoscopy. If 1–4 windows are positive this is classed as an unclear result and individuals undergo repeat FOB testing. Two further sets of 6 normal windows are then required to classify a result as normal. If any windows on the two repeat kits are positive the individuals are invited to attend for colonoscopy. This is classed as a weak positive result.

Aims & Methods: Data from the South of Tyne and Tees bowel cancer screening centres were collected from March to early October 2007. These centres serve 1.5 million people. Analysis of data was undertaken to identify the % of screening kits returned with an abnormal FOB result; those with a weak positive result and those with an unclear result that was subsequently classed as normal. Positive results (abnormal and weak positive) were compared to the pathology diagnosed at colonoscopy.

Abstract 071 Colonoscopy results compared to faecal occult blood (FOB) result

Results: 1.6 % of FOBs returned in the screening population served were classed as positive (abnormal (0.16%) or weak positive (1.44%)) and individuals were subsequently invited for colonoscopy. 521 colonoscopies were undertaken in 2 centres by 5 colonoscopists. 469/521 (90%) of patients undergoing colonoscopy had weak positive FOB results and 52/521 (10%) had abnormal FOB results. The table demonstrates the pathology diagnosed. Patients diagnosed with cancer were more likely to have an abnormal FOB result when compared to patients with a normal colonoscopy (p<0.05). 2.1% of FOBs returned demonstrated an unclear result initially with 2 subsequent FOB kits normal. These patients were classed as normal and did not undergo further investigation.

Conclusion: The overwhelming majority of colonoscopies undertaken as part of bowel cancer screening are on individuals with weak positive FOB results. High levels of pathology are diagnosed. Cancer patients were more likely to have an abnormal FOB result when compared to patients with normal colonoscopies however most cancers were still diagnosed in patients with weak positive FOBs.

Research submitted on behalf of the Northern Region Endoscopy Group.

Endoscopy free papers

ENDOSCOPIC MUCOSAL RESECTION FOR HIGH-GRADE DYSPLASIA AND EARLY CANCERS IN THE UPPER GASTROINTESTINAL TRACT: PORTSMOUTH EXPERIENCE

M. Duku1, R. Mead1, D. Poller2, S. Somers3, S. Toh3, P. Goggin1, R. Ellis1, H. Duncan1, C. Archer4, P. Bhandari1. 1Gastroenterology, 2Pathology, 3Surgery, 4Oncology, Portsmouth Hospitals Trust, Cosham, UK

Introduction: Endoscopic mucosal resection (EMR) is as an alternative treatment option to radical surgery in selected patients with early neoplasia of the upper GI tract. The risk of invasive cancer and lymph node metastasis in these patients has been cited as reasons to advocate radical surgery for these lesions. In addition to its therapeutic role, EMR also serves as a staging tool.

Aims & Methods: To evaluate effect of EMR on the histopathological staging and the rate of recurrence and/or development of metachronous lesions following EMR. We analysed a prospective series of 45 EMR cases performed on 35 consecutive patients with early oesophago-gastric neoplasia (non-squamous) by a single endoscopist between 2004 and 2007 in Queen Alexandra Hospital Portsmouth. All patients had a staging CT scan and a selected group had EUS as well. A staging chromoendoscopy, with acetic acid was carried out on all patients before EMR. Different EMR techniques (cap and snare, duette, strip biopsy, and insulated tip knife) were used either alone or in combination depending on the size and location of the lesions. Patients were followed-up with chromoenoscopy and biopsies of the EMR site and any remaining Barrett’s segments every three months for one year then every six months.

Abstract 072 Histopathological staging pre and post EMR

Results: 28/35 patients underwent only one EMR while 7/35 had two or more EMR sessions to achieve complete clearance. 29/35 patients (82.9%) had curative EMR. 6/35 patients had submucosal invasion (SM-cancer), of whom 4 had oesophagectomy and 2 (unfit for surgery) were referred for palliative therapy. There was one delayed bleeding at 48 hours, which was controlled endoscopically, and one stricture, managed by endoscopic dilation. 8/30 (26.7%) patients with HGD pre-EMR were up-staged to cancer post-EMR. After a median follow-up of 11.5 months, recurrent/metachronous lesions were identified in 6 (17.1%) patients on surveillance (5 HGD & 1 IM-cancer). 4/6 have had repeat EMRs and are now in remission, 2/6 are awaiting EMR.

Conclusion: Our series show that: (1) EMR is an excellent therapeutic tool in selected patients. However, following curative EMR patients need regular surveillance to identify and treat recurrent/metachronous lesions. (2) EMR has no mortality and low morbidity. (3) It is the best available staging tool for early gastrointestinal cancers.

CONTRAST ENHANCED ENDOSCOPIC ULTRASOUND FOR DISCRIMINATION OF FOCAL CHRONIC PANCREATITIS AND PANCREATIC CANCER

M. Hocke1, T. Topalidis2, C. F. Dietrich3, A. Stallmach1. 1Internal Medicine II, Friedrich Schiller University Jena, Jena, 2Private practice, Cytological laboratory, Hannover, 3Internal Medicine II, Caritas Hospital, Bad Mergentheim, Germany

Introduction: Recognising malignant neoplasia in chronic inflammatory pancreatic lesions has always been difficult. However, the introduction of endoscopic ultrasound guided fine needle aspiration made this much easier. Nevertheless, a non-invasive method for detecting pancreatic malignancy would still be preferable.

Aims & Methods: In 192 patients with suspected chronic pancreatitis (age 62±12 years) pancreatic lesions were examined by conventional endoscopic B-mode, Power Doppler ultrasound and contrast enhanced power mode (Hitachi EUB 525 or Aloka-Olympus UCT 160, SonoVue, 2.4 ml, Bracco) using the following criteria for malignant lesions: no detectable vascularisation using conventional power Doppler scanning, irregular appearance of arterial vessels over a short distance using SonoVue contrast enhanced technique, and no detection of venous vessels inside the lesion. A malignant lesion was assumed if all criteria were fulfilled (gold standard EUS-guided fine needle aspiration cytology with 1-year follow-up or operation). The criteria for chronic pancreatitis without neoplasia were defined as no detectable vascularisation before injection, regular appearance of vessels over at least a distance of 20 mm after injection of SonoVue and detection of arterial and venous vessels.

Abstract 073 Results of non-contrast enhanced and contrast enhanced endoscopic ultrasound

Results: Using conventional EUS-criteria sensitivity was 82.2% and specificity of 87.8 % for pancreatic cancer. Using contrast enhanced EUS 108 of 118 patients with malignant pancreatic lesion displayed all criteria (increasing sensitivity to 91.5%). In 70 of 74 patients with chronic inflammatory pancreatic disease, the diagnosis was correctly recognised, giving a specificity of 94.6%.

Conclusion: Contrast enhanced endoscopic ultrasound improves the differentiation between chronic pancreatitis and pancreatic carcinoma.

IS ENDOSCOPIC ULTRASOUND WITH FINE NEEDLE ASPIRATION NOW ESSENTIAL FOR THE MANAGEMENT OF PANCREATICOBILIARY MALIGNANCY?

G. C. Bennett1, S. O’ Riordan1, R. Sugrue1, R. Bambury1, S. Sengupta1, E. Clarke1, N. Mulligan2, J. Lennon1, T. Kelleher1, G. McEntee3, P. MacMathuna1. 1Gastrointestinal Unit, 2Department of Pathology, 3Department of Surgery, Mater Misericordiae University Hospital, Dublin 7, Ireland

Introduction: Endoscopic ultrasound (EUS) with fine needle aspirate +/− biopsy (FNAB) is increasingly used in the loco-regional staging, and pathological diagnosis of pancreaticobiliary malignancy. Prior to its introduction, CT was the modality of choice for staging, and a tissue diagnosis was often difficult to obtain preoperatively.

Aims & Methods: To evaluate our experience of EUS-FNA in the management of suspected pancreaticobiliary malignancy. All procedures were conducted over a 5-year period on a national referral basis. In 107 patients, we examined demographics, imaging, pathology and clinical follow up. FNA was performed on pancreas, ampulla, abdominal nodes, ascites. We examined the indication for FNA, diagnostic yield and complication rate.

Results: An adequate sample for diagnosis was obtained in 94/107 cases (88%). Complete follow up was available for 76/94 (81%). Malignancy was confirmed in 57/76 (75%) (sensitivity 100%, specificity 100%). The diagnosis was benign in 19/76 (25%). In 7 (36%) cases, the diagnosis was subsequently shown to be incorrect following surgery. (sensitivity 63%, specificity 100%) One major complication (perforated duodenum) was encountered.

Conclusion: EUS with FNA provides a safe, effective “one stop shop” approach to loco-regional staging and tissue diagnosis in pancreaticobiliary malignancy, with a high PPV. Benign histology, in the context of a high clinical suspicion for malignancy, must be interpreted with caution, since sampling error is a generic phenomenon, regardless of modality.

“UPSIDE DOWN” COLORECTAL ENDOSCOPIC DISSECTION FOR RETROFOLD AND HEMI-CIRCUMFERENTIAL INTRAEPITHELIAL NEOPLASTIC DISEASE: EXPANDING THE ENDOSCOPIC TECHNICAL PARADIGM

D. P. Hurlstone1, W. Baraza2, S. R. Brown2, N. Tiffin3, S. S. Cross4. 1Endoscopy, Royal Hallamshire Hospital, 2Colorectal Surgery, 3Histopathology, STH NHS Foundation Trust, 4Histopathology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Retrofold disease, hemi-circumferential luminal involvement and consecutive haustral fold expansion have historically been contraindications to endoluminal resection therapy for Paris type I, 0-II, LST (G/F) and stage N0/Tim/T1 carcinoma. Intraluminal access using conventional colonoscopes is often impossible due to limitations in acute tip angulation. Retroflexion can be achieved at any anatomical location in the colorectum with retrofold access using a GIF-XQ240 gastroscope.

Aims & Methods: To prospectively assess the technical efficacy and medium term clinical outcome of lesions resected using retroflexion assisted dissection as salvage endoscopic therapy. Patients were recruited for endoscopic retroflexion assisted dissection following an index diagnostic colonoscopy that precluded conventional en face endoscopic resection. Exclusion criteria were pre-resection high-frequency 12.5 MHz EUS suggestive of T1/N1 disease, index histopathology suggestive of invasive neoplasia and non-correctable coagulopathy. Endoscopic mucosal resection (EMR) and endoscopic submucosal cap assisted dissection (ESCA) was performed in all eligible cases. Primary end point: Curative excision rates. Secondary end points: 1. Complications 2. Intraepithelial neoplasia recurrence 3. Procedure time analysis.

Results: 188 patients with 188 lesions fulfilled eligibility for salvage therapy. Caecal/neoterminal-ileal intubation using the GIF-XQ gastroscope was achieved in 184/188 (98%). Median post resection inpatient stay 24 hours (1–72), Median intubation time 16.5 min (range 8–39). Median resection times for Paris class Is, 0-II, LST-G and LST-NG were 89 min (range 20–240), 32 min (15–64), 148 min (22–242) and 64 (10–119) respectively. LST-G morphology was significantly associated with prolonged resection time (as compared with all other morphologies, p<0.04) and a high median submucosal dissection volume requirement (p<0.01). 159/188 patients completed the post dissection surveillance programme to 36 months. Primary end point: Higaki criteria for recurrence/remnant disease = 9/159 (5.6%), 2 undergoing R0 revision ESD, 7 surgical intervention. Overall “cure” rate at 36 months = 95%. 30-day mortality rate 0%, 30-day readmission rate 2/188 (1%), delayed/acute bleed rates 7/188 (4%), perforation rate 2/188 (1%)—both not requiring laparotomy.

Conclusion: This is the largest prospective study to address the technical feasibility and clinical outcomes using retroflexion assisted endoluminal resection. Retroflexion assisted resection using EEMR and ESCA for complex intraluminal adenoma and early stage cancer is a safe, efficacious technique at medium term follow-up and an alternative to surgical excision.

NARROW BAND IMAGING FOR LESION CHARACTERISATION IN THE COLON, OESOPHAGUS, DUODENAL AMPULLA AND LUNG: A META-ANALYSIS

J. E. East1, E. Tan2, J. J. Bergman3, B. P. Saunders1, P. P. Tekkis2. 1Wolfson Unit for Endoscopy, St Mark’s Hospital, 2Department of Biosurgery and Surgical Technology, Imperial College London, London, UK, 3Department of Gastroenterology, Academic Medical Center, Amsterdam, The Netherlands

Introduction: Narrow band imaging (NBI) is a new endoscopic technology which can highlight mucosal surface structures and microcapillaries. Abnormal surface structure and microvasculature are predictive of neoplastic change. We aimed to assess the diagnostic precision of NBI for diagnosis of epithelial neoplasia compared to conventional histology.

Aims & Methods: We performed a meta-analysis of studies which compared NBI based diagnosis of neoplasia with histopathology, using the search keywords “narrow band imaging” and “endoscopy” published before 8 August 2007. Sensitivity, specificity, diagnostic odds ratios (DOR) were calculated for each study. Study quality and heterogeneity was assessed. Summary receiver operator characteristics (ROC) and subgroup analyses were performed. Meta-regression was used to compare subgroups.

Abstract 076 Meta-analysis of diagnostic NBI studies

Results: 51 studies were identified, of which 11 had extractable data (colon 5, oesophagus 3, lung 2, duodenal ampulla 1). Results for overall analysis, by organ, by assessment method, and by study quality are presented (table). Significant heterogeneity was found overall, and for studies in the colon and oesophagus. Meta-regression showed no significant difference between, colon versus oesophagus, mucosal pattern (pit pattern) versus microvascular measures, or magnification versus non-magnification for diagnostic accuracy. Studies with more than 50 patients (>100 lesions) had a higher DOR compared to those with <50 patients, relative DOR 7.0 (95% CI 1.1 to 44.5), p = 0.041, and NBI was more accurate than chromoendoscopy, relative DOR 3.6 (95% CI 1.2 to 11.1), p = 0.024.

Conclusion: NBI, a simple, push-button technology, appears highly accurate, with narrow confidence intervals, for in vivo diagnosis of neoplasia across a range of organs, using simple microvessel based measures. It has the potential to increase the yield of targeted biopsies or could even potentially replace formal histopathology for assessment of low-risk lesions.

Endoscopy videos

DOUBLE BALLOON ENTEROSCOPY: REACHING THE DEPTHS OF THE SMALL BOWEL

E. Despott1, S. Preston1, E. Tripoli1, A. Polecina1, K. Konieczko2, C. Fraser1. 1Wolfson Unit for Endoscopy, St Mark’s Hospital, 2Department of Anaesthesia, St Mark’s and Northwick Park Hospitals, London, UK

Introduction: The advent of double balloon enteroscopy (DBE) has advanced the investigation and management of small bowel disease to a higher level. In addition to complementing the findings of capsule endoscopy (CE), it has the capability of being a useful therapeutic device in areas of the small bowel which were previously inaccessible outside the operating theatre.

Aims & Methods: Since the DBE service began at our unit in February 2005, all DBE procedures have been recorded on digital video. This short film taken from these recordings, shows amalgamated video footage that exemplifies the diagnostic and therapeutic capabilities of DBE.

Results: The viewers will be shown cases where DBE was instrumental in making a diagnosis that led to definitive patient management. This is illustrated by video clips showing a selection of small bowel tumours, mucosal ulceration and a large vascular lesion. The therapeutic ability of DBE will be highlighted by video footage of direct percutaneous endoscopic jejunal feeding tube insertion, coagulation of bleeding vessels and balloon dilatation of Crohn’s disease related strictures.

Conclusion: Double balloon enteroscopy is highly effective for the diagnosis and treatment of small bowel pathology.

ENDOSCOPIC SUBMUCOSAL DISSECTION OF EARLY GASTRIC CANCER IN UK SETTING

M. Duku1, R. Mead1, D. Poller2, P. Bhandari1. 1Gastroenterology, 2Pathology, Portsmouth Hospitals Trust, Cosham, UK

Introduction: Endoscopic submucosal dissection (ESD) is an established therapeutic modality for early gastric cancers in Japan. Although the incidence of antral gastric cancer is low in the Western world, the incidence of gastric cardia cancers is rising. Traditionally, all these patients have been treated by radical surgery. There is increasing enthusiasm about using EMR. En bloc endoscopic resection of early gastric cancer is essential for accurate histological staging and reducing the recurrence rate. Unfortunately, simple EMR techniques do not allow en bloc resections of lesions more than 2 cm. ESD is a new, technically challenging technique which is used for en bloc resection of early gastric cancers of more than 2 cm.

Aims & Methods: We have been performing ESD for early gastrointestinal cancers since 2005 and have developed a good understanding of its principles, problems and role in Western settings. We have found that gastric lesions smaller then 4 cm, situated in antrum, incisura and O-G junction are less challenging and are the type of lesions that Western endoscopists can attempt to resect with ESD after suitable training. We aim to show video clips to illustrate the basic principles of ESD: Role of chromoendoscopy in lesion identification; Pre-marking the margins; Submucosal injection; Circumferrential incision; Submucosal dissection; Specimen retrieval and fixation; Post-ESD plans; Risks and management of ESD associated complications.

Results: It is technically feasible to perform ESD in Western settings but it requires training, good understanding of the basic principles, risks and skills in managing complications.

Conclusion: En bloc resection is the main principle of onco-surgery and ESD enables us to meet this principle when dealing with lesions larger then 2 cm. We have demonstrated that ESD is feasible in the Western settings but its exact role in the West needs further clarification.

EN BLOC RESECTION OF EARLY GASTRIC NEOPLASIA USING THE ENDOSCOPIC SUBMUCOSAL DISSECTION TECHNIQUE: WEST MEETS EAST!

K. Ragunath1, R. Singh1, P. Kaye1, M. Ilyas1, K. Yao2. 1Wolfson Digestive Diseases Centre, Queens Medical Centre, Nottingham University Hospital, Nottingham, UK, 2Department of Gastroenterology, Fukuoka University Chikushi Hospital, Fukuoka, Japan

Introduction: Endoscopic mucosal resection (EMR) is now a well established technique to excise early gastrointestinal neoplasia. The efficacy of the procedure is directly related to the ability of the endoscopist to achieve complete resection. Until recently, single piece resection of lesions larger than 10 mm was technically difficult. This resulted in piecemeal resections being performed with high recurrence rates. However, yet another Japanese innovation in this field, endoscopic submucosal dissection (ESD), has resulted in overcoming this shortfall.

Aims & Methods: To report our experience in ESD for en bloc resection of early gastric neoplasia by (1) demonstrating the identification of early gastric neoplasia and the technique of ESD, (2) elucidating common complications and their management, (3) discussing learning curve and training issues. The first author (KR) undertook a period of sabbatical in Japan observing advanced endoscopic imaging in identifying early GI neoplasia and endoscopic resection techniques. The procedure was registered and performed as per Safety and Efficacy Register of New Interventional Procedures (SERNIP) guidelines. KR then performed all the procedures in Nottingham with guidance and hands-on technical advice by a senior Japanese endoscopist (KY). High definition zoom endoscopy with Autofluorescence Imaging and Narrow Band Imaging were done in addition to Chromoendoscopy to identify and delineate the lesion. The Insulated Tip Knife and Hook Knife from Olympus, Tokyo were used for the resections. High frequency EUS was performed prior to resections. All the procedures were digitally recorded and this presentation will include short edited clips.

Results: Four male patients (mean age: 72, range: 63–80) had undergone en bloc resection by ESD for early (Paris Type IIa or IIb) gastric neoplasia under intravenous sedation (pethidine 50–100 mg, diazemules 10–20 mg). The mean procedure time was 53 min (45–180). Three patients developed brisk bleeding during the procedure which was controlled with hot biopsy graspers. All patients recovered well without any immediate or delayed complications. Complete resection was achieved in all cases and the mean size of the resected specimen was 2.5 cm (1.7–3.5). Histopathology demonstrated high grade dysplasia, low grade dysplasia, intramucosal carcinoma and carcinoid tumour respectively. Follow-up endoscopy with detailed examination of the resected sites and biopsies were performed at 3 months, 6 months then yearly. To date after a mean follow-up of 16 months, there were no recurrences in this cohort of patients.

Conclusion: ESD is a safe technique for en bloc resection of early gastric neoplasia and can be adopted in the Western world provided it is done under optimal conditions. It is technically challenging and requires systematic hands-on training.

A 7 CM G-TYPE LATERALLY SPREADING TUMOUR OF THE RECTUM EXCISED EN BLOC USING ENDOSCOPIC SUBMUCOSAL DISSECTION WITH THE FLUSH KNIFE

N. Suzuki1, B. P. Saunders2. 1Wolfson Unit for Endoscopy, 2Endoscopy, St Mark’s Hospital, Harrow, UK

Introduction: Endoscopic submucosal dissection (ESD) is recognised in Japan as an effective therapeutic modality for the clearance of gastrointestinal neoplasms, providing an en bloc specimen to confirm lateral and deep resection margins and minimising the risk of recurrence. However, due to the technical complexity and risk of complications, few European centres have introduced this technique. This video shows how an ESD technique can be safely applied to resect a 7 cm laterally spreading tumour (LST) in the rectum.

Aims & Methods: The margins of the rectal LST were lifted by injection of sodium hyaluronate using a 23 gauge injection needle. The first incision was made through the mucosa adjacent to the lesion to gain access to the submucosal plane with a 1.5 mm flush knife (Fujinon), followed by cutting around the edge to enable a flap of mucosa to be lifted. The submucosal dissection was performed with the help of a distal attachment, which was fitted at the end of the scope to achieve optimum retraction and visualisation, helping to expose the submucosal layer by creating a counter traction effect. Any bleeding during the procedure was treated with an electrosurgical haemostatic forceps (Cograsper (Olympus)). The electrosurgical unit used was ICC200 (ERBE) using the following settings, “endocut 80W”, “soft coagulation 35W” and “forced coagulation 60W” for the mucosal incision, submucosal dissection and haemostasis, respectively.

Results: The patient received conscious sedation with midazolam (1.25 mg) and pethidine (25 mg). En bloc resection was achieved with a procedure time of 180 min. No early or late complications were encountered. The resected specimen was 65 by 70 mm in size. Histologically the lesion was a tubulovillous adenoma with high grade dysplasia with clear lateral and deep margins (R0 resection).

Conclusion: ESD was effective for the excision of this large rectal LST. Large sessile polyps and some T1 cancers are potentially best treated by this technique which provides definitive histology and very low risk of local recurrence. With the advent of bowel cancer screening and an increased detection of such lesions this new minimally-invasive approach is becoming more applicable and with careful training can be performed safely at specialist endoscopic centres in the UK.

SINGLE OPERATOR PERORAL CHOLEDOCHOSCOPY SYSTEM: A SINGLE CENTRE EXPERIENCE

M. A. Zubir, M. L. Wilkinson, R. J. Ede, T. Wong. Deparment of Gastroenterology, Guy’s And St Thomas Hospital NHS Foundation Trust, London, UK

Introduction: The SpyGlass (Boston Scientific) peroral choledochoscopy system provides 4-way deflected steering by a single operator which allows direct visual examination of bile ducts, visually guided intraductal biopsies, and therapeutic procedures.

Aims & Methods: To report our experience with using SpyGlass in 8 consecutive ERCP cases referred for choledochoscopy. ERCP with SpyGlass examinations with or without biopsy were performed in 8 patients. The patients selected were in those whom direct visualisation of intraductal pathology was indicated following previous investigations including ERCP which were inconclusive.

Results: Eight patients underwent ERCP with SpyGlass. The mean age of the patients was 68 years (range 34–84). The mean time taken for the whole procedure including ERCP was 43 minutes (range 28–56). Primary indications for SpyGlass were to biopsy indeterminate strictures (n = 5) and confirm intraductal abnormalities (n = 3). The primary SpyGlass visual diagnoses were strictures (n = 5), bile duct stone (n = 2) and normal findings (n = 1). Intraductal biopsies were taken in 4 patients. In patients in whom biopsies were not taken, one was due to technical difficulties and 3 others were not clinically indicated. No complications were identified post procedure.

Conclusion: Our early experience with SpyGlass suggest that it is safe, well tolerated, technically feasible and useful in the diagnosis and management of duct diseases. More experience of the procedure will be needed to the determine the accuracy of diagnosis with the Spyglass system.

ENDOSCOPIC MUCOSAL RESECTION OF HIGH GRADE GASTRIC DYSPLASIA USING THE DUETTE MUCOSECTOMY DEVICE

A. Dhar1, A. Allan1, E. Millington2. 1Department of Gastroenterology, Bishop Auckland General Hospital, County Durham, 2Endoscopy, Cook Medical (UK) Ltd, Hertfordshire, UK

Introduction: Endoscopic mucosal resection (EMR) is a relatively new technique for resection of early cancer in the oesophagus and stomach. It has been widely used in Japan for some years but has been incorporated into practice in the Western world for less than a decade. It can be done using conventional saline injection and snare resection using the Cap-EMR technique or by lift and cut technique using a dual channel scope. The Duette mucosectomy device is a new accessory developed by Cook Medical, using a combined multi-band ligator device and snare without the need for saline injection. In the UK, it has been used predominantly for resection of high grade dysplasia and intramucosal carcinoma of the oesophagus, in patients with Barrett’s oesophagus.

Aims & Methods: We present a patient with a 15 mm lesion in the stomach, histologically high grade gastric dysplasia, who was treated successfully with EMR using the Duette mucosectomy device.

Results: A 78-year-old gentleman with chronic obstructive airways disease, atrial fibrillation and heart failure underwent an upper GI endoscopy for new onset dyspepsia. At endoscopy, there was a 15 mm reddish, slightly elevated lesion in the distal body of the stomach on the lesser curvature which was better delineated with 1% indigo carmine dye spray. Multiple biopsies taken on 2 separate occasions, 3 months apart were reported to show high grade dysplasia, but no cancer. CT scan was normal. The patient underwent Endoscopic Mucosal Resection of the gastric lesion using the Duette mucosectomy device after marking and the entire lesion was removed in 4 separate pieces. Post-resection histology was reported to show dysplasia in only the superficial half of the mucosa with complete excision, and free margins. A 6 monthly endoscopic surveillance is planned for early 2008.

Conclusion: It is possible to safely resect early gastric cancer using the Duette mucosectomy device and the technique is easier to learn than the Cap-EMR resection and can be performed through a standard single channel therapeutic endoscope.

ENDOSCOPIC PAPILLECTOMY FOR AMPULLARY NEOPLASIA

R. J. Mead, M. D. Duku, P. Goggin, R. D. Ellis, H. Duncan, P. Bhandari. Gastroenterology, Portsmouth Hospitals Trust, Cosham, UK

Introduction: Management of benign ampullary tumours remains controversial due to the uncertainty in diagnosis and risks associated with various treatment options. We believe that endoscopic papillectomy (EMR) is technically feasible and safe. EMR can serve as a final diagnostic step or an ultimate therapeutic option for these patients.

Aims & Methods: We have been performing endoscopic resections of all benign ampullary tumours which has given us a good insight into the principles of endoscopic papillectomy. We aim to illustrate the following principles of endoscopic papillectomy: Role of chromoendoscopy; Role of submucosal injection; Snaring techniques; Billiary and pancreatic stents and their timings; Adjunctive treatment.

Results: We have recently performed 10 endoscopic papillectomies in last one year at our centre with very good outcome. We have had one patient with mild self-limiting pancreatitis and no cases of bleeding, perforation or mortality. These encouraging results have influenced the management of these patients in our region.

Conclusion: We have found that endoscopic papillectomy is technically feasible, safe and potentially curative. Our video clips illustrate some basic principles of the procedure including submucosal injections, snaring techniques, pancreatic and billiary stenting. We aim to discuss other controversies and principles of management of complications associated with this procedure.

Gastrointestinal pathology free papers

THE FIXATION OF MUTATIONS AND CLONAL EXPANSION IN INTESTINAL METAPLASIA: FIELD CANCERISATION IN THE HUMAN STOMACH

S. A. C. McDonald1, L. Gutierrez-Gonzalez1, M. Rodriguez-Justo2, S. J. Leedham1, M. Novelli2, I. Mitchell3, D. Stoker3, J. A. Jankowski4, N. A. Wright5. 1Histopathology Unit, London Research Insitute - Cancer Research UK, 2Histopathology, 3Department of Surgery, University College London, London, 4Department of Clinical Pharmacology, University of Oxford, Oxford, 5Histopathology, Barts and the London School of Medicine and Dentistry, London, UK

Introduction: The clonal origins and mechanisms of spread of intestinal metaplasia (IM) within the human stomach are poorly understood. Mihara et al1 suggested that IM is a polyconal disorder with distinct crypts having different methylation patterns. We have previously shown, using mitochondrial DNA (mtDNA) mutations as a marker of clonal expansion, that normal gastric glands are clonal: all cells within a gland contain the same mtDNA mutation. We have also shown that clonal patches occur with multiple glands each containing the same mutation.

Aims & Methods: Here we investigate the clonality of IM crypts in the human stomach and reveal a mechanism by which they spread. Double enzyme histochemistry was performed on frozen sections of human gastric resections with IM for reactivity of Cytochrome c oxidase (CCO) in along with the nuclear-encoded enzyme succinate dehydrogenase (SDH), the principle being that a mutation in the mtDNA that leads to inactivity of the CCO enzyme will result in only SDH activity being detected. Immunostaining for CD10, MUC2 and chromogranin was used as markers of differentiated epithelial cells.

Results: Entirely CCO −ve IM crypts were observed indicating that such crypts are clonal. Immunostaining revealed that all the major differentiated lineages (enterocyte, neuroendocrine and goblet cells) were present in such crypts. Partially mutated glands were also present suggesting that multiple stem cells reside in these crypts. We observed frequent patches of CCO −ve crypts which all contain the same mtDNA mutation suggesting that there is one founder crypt which has expanded by fission.

Conclusion: We propose that IM has its origins as a clonal disorder and that fission is a method by which it spreads in the stomach. Furthermore, the presence of partially and completely mutated crypts suggests that monoclonal conversion occurs—the process by which one stem cell takes over the entire stem cell compartment eventually leading to all their progeny becoming mutated. This is in conflict with previous studies which argue that IM is polyclonal and that “mixed” IM crypts are a separate classification of IM. We would suggest that mtDNA mutations are a more reliable method of showing clonal expansion as they take many years to accrue to detectable levels, whereas methylation is a far more rapid process and that patterns diverge quickly. We therefore conclude that in IM, crypt fission fixes mutations, enables the clonal expansion of mutated stem cells, and thus is responsible for field cancerisation in the human stomach.

IRON MEDIATED CELLULAR PROLIFERATION, MIGRATION AND COLONY FORMATION IS DEPENDENT ON ADENOMATOUS POLYPOSIS COLI STATUS IN COLORECTAL ADENOCARCINOMA

M. J. Brookes, K. Roberts, T. Iqbal, C. Tselepis. Cancer Studies, University of Birmingham, Birmingham, UK

Introduction: There is emerging evidence suggesting increased cellular iron is important in the development of colorectal cancer. Colorectal cancer is characterised by mutations in adenomatous polyposis coli (APC); a tumour suppressor gene crucial in Wnt signalling. We have previously demonstrated that iron can activate Wnt signalling in the background of an APC mutation. However, the effect of iron on cell phenotype to date has not been addressed. Thus the aims of this study were to determine if iron could enhance cellular proliferation, migration and colony formation and establish if these effects were APC dependent.

Aims & Methods: Human colorectal APC wild type (HEK-293 and RKO) and APC mutant containing lines (Caco-2, SW480) were challenged with either ferrous sulphate (FeSO4) (100 μM) or hemin (50 μM) in the presence or absence of the GSK-3β inhibitor lithium chloride (LiCl) (20 mM) for 24 h. Cellular iron loading was confirmed using the ferrozine assay. Cellular proliferation was determined by BrdU incorporation and FACS analysis. Cellular migration was assessed using a wound healing assay, and in vitro anchorage independent colony forming determined using a soft agar colony forming assay. To further test the dependency of APC SW480 cells transfected with or without full length wild type APC were also used in these studies.

Results: Challenging APC mutant containing lines with FeSO4 or hemin significantly increased cellular proliferation, wound healing and anchorage independent growth. Conversely, wild type APC containing cell lines only responded to FeSO4 or hemin when co-cultured with LiCl. The dependency for APC was exemplified by the inhibition of iron mediated cellular proliferation in SW480 cells transfected with human wild type APC.

Conclusion: This study demonstrates that iron can induce cellular proliferation, migration and anchorage independent growth and we believe this is a consequence of iron mediated Wnt signalling. Interestingly iron could only induce these cellular effects in the background of abrogated β-catenin degradation, such as a truncating mutation in APC or inhibition of GSK-β.

THE AETIOLOGY OF HYPERPLASTIC POLYPOSIS SYNDROME IN A LARGE FAMILY FROM THE WEST OF IRELAND

K. J. Monahan1, L. Carvajal-Carmona2, T. Guenther3, T. O’Gorman4, J. Cazier2, I. P. Tomlinson2, H. J. W. Thomas1. 1Family Cancer Clinic, St Mark’s Hospital, 2Molecular and Population Genetics Lab, Cancer Research UK, 3Academic Department of Pathology, St Mark’s Hospital, London, UK, 4Department of Medicine, University College Hospital, Galway, Ireland

Introduction: We have identified a family with a dominantly inherited predisposition to mixed histology multiple polyps and colorectal cancer. The family meet WHO criteria for hyperplastic polyposis. We have attempted to elucidate the gene which predisposes to this condition.

Aims & Methods: Mutations in APC, MYH, SMAD4 and genes which cause HNPCC were excluded by direct sequencing and other methods. Somatic mutations were screened in genes often mutated in colorectal cancers. Genome wide genotypes were obtained for over 10000 SNPs using Affymetrix 10k plus 2 arrays and linkage analysis was performed. Genome wide copy number variation analysis was also performed using the Goldengate SNP platform. Whole genome expression analysis profiles were obtained on cell line RNA using the Affymetrix U133 Plus 2.0 GeneChip oligonucleotide arrays.

Results: The polyps appear to follow a hyperplastic/serrated polyp to mixed serrated/adenomatous (see fig) to adenocarcinoma sequence. Linkage analysis shows a maximum parametric LOD score of 2.71 at 8p22–21.3. Genome wide copy number and loss of heterozygosity (LOH) studies reveal LOH at 8p and 17p in the cancers. A number of genes at that locus including BMP1 and MTUS1 have been screened, the latter gene having more than a fivefold up-regulation in expression.

Abstract 086

Conclusion: We have identified a locus on chromosome 8p which predisposes to hyperplastic polyps and colorectal cancer in a large family from the West of Ireland. This is the first time such an association has been identified and may be useful in screening families at risk of colorectal neoplasia.

INHIBITING EGFR AND VEGR REDUCES SMALL INTESTINAL AND COLONIC POLYP FORMATION IN A DSS ENHANCED APCMIN/+ MOUSE MODEL

D. Alferez1, R. W. Wilkinson2, A. J. Ryan2, R. A. Goodlad1. 1Histopathology, CRUK, London, 2Departments of Cancer and Infection Research, AstraZeneca Pharmaceuticals, Macclesfield, UK

Introduction: The ApcMin/+ mouse is a well-characterised model of intestinal tumorigenesis, and animals with this genetic mutation develop macroscopically detectable adenomas from approximately 6 weeks of age. Most of the polyps are found in the small intestine (SI) so that the resolution of events in the colon is limited. Inducing colitis with dextran sodium sulphate (DSS) can enhance the development of lesions in the colon.1 The contribution of VEGFR- and EGFR-dependent signalling pathways to the development of polyps in DSS treated or untreated ApcMin/+ mice was examined using a selective inhibitor of the tyrosine kinase activity of these receptors (ZD6474).

Aims & Methods: 8-week-old ApcMin/+ mice were given 1.8% DSS or water for 4 days followed by a 2 day pause, after which they were given ZD6474 or vehicle for 28 days. Intestines were opened2 and fixed in Carnoy’s fluid and then stored in 70% ethanol before polyp number and size were measured.3

Results: ZD6474 reduced the number of polyps in the SI by 76% in the water group and by 65% in the DSS treated (p<0.025). ZD6474 reduced the number of polyps in the colon by 60% (from 3.9 to 1.4 (p = 0.1)) of the non-DSS mice and in the DSS pre-treated by 78% (from 13.8 to 3.1, p<0.001). Multivariate analysis demonstrated that the effects of ZD6474 were similar in both DSS treated and untreated, as there were no interaction effects.

Conclusion: The DSS model thus does not alter the response of the colon; it only increases the resolution of the model. VEGF/VEGFR-2-dependent angiogenesis and EGFR-dependent tumour cell proliferation and survival are thus key mechanisms in polyp development. Targeting VEGFR- and EGFR-dependent signalling is a promising therapeutic strategy in intestinal tumours.

THE DELTA-SAMP MOUSE: A NOVEL EXPERIMENTAL MODEL FACILITATING ASSESSMENT OF THE ROLE OF C TERMINUS OF APC IN INTESTINAL TUMORIGENESIS

M. G. Deheragoda1, A. Lewis2, E. Nye3, N. Wright1, I. Tomlinson2. 1Histopathology Unit, 2Molecular and Population Genetics unit, 3Experimental Pathology Unit, Cancer Research UK London Research Institute, London, UK

Introduction: Mutation of the adenomatous polyposis coli (APC) gene is an initiating step in colonic carcinogenesis. APC is a multifunctional protein involved in the Wnt signalling pathway which controls proliferation and differentiation in the intestine. The C terminus of APC has been proposed to have significance in tumourigenesis through cytoskeletal interactions. We previously presented our data on a novel experimental model of intestinal tumourigenesis, the APC1322 mouse,1 which carries a mutated APC allele analogous to that most commonly found in patients with familial adenomatous polyposis (FAP), the APC1309 mutation. We used a modified version of the APC 1322T gene construct, to develop a novel model of intestinal tumorigenesis, the delta-SAMP mouse. The mutant APC protein in the delta-SAMP mouse has the same APC N terminus as the APC1322T (lacking SAMP repeats involved in beta catenin interactions) but retains the C terminus of APC.

Aims & Methods: To develop and characterise a novel experimental model, the delta-SAMP mouse, with which to assess the relative contribution of N terminal effects versus C terminal effects of mutant APC in intestinal tumorigenesis. A novel APC knockout model of intestinal tumorigenesis, The delta-SAMP mouse, was developed following homologous recombination in ES cells. The phenotype of the delta-SAMP mouse was characterised and mechanisms of tumour growth were assessed relative to the APC1322T model. Any differences in tumorigenesis were attributed to the presence of the C terminus of APC.

Results: Western blotting confirmed the presence of a mutant APC protein of the expected length in delta-SAMP mice. Intestinal tumour numbers, site of polyp development, cellular composition of polyps, parameters of tumour growth such as crypt fission, and levels of Wnt signalling, as measured with immunohistochemistry for nuclear beta catenin levels, were comparable with those seen in the APC1322T mouse, with no significant differences between the two models.

Conclusion: The C terminus of APC does not appear to have a significant effect on intestinal tumorigenesis. The action of the N terminus of mutant APC appears sufficient to account for the effects of mutant APC on tumorigenesis.

Inflammatory bowel disease free papers

SUSTAINED FISTULA HEALING AND RESPONSE WITH ADALIMUMAB FOR UP TO 2 YEARS IN PATIENTS WITH FISTULISING CROHN’S DISEASE: 12 MONTHS OF OPEN-LABEL MAINTENANCE WITH ADALIMUMAB FOLLOWING THE CHARM TRIAL

J. F. Colombel1, M. A. Kamm2, D. Schwartz3, W. J. Sandborn4, J. Li5, K. G. Lomax5, P. F. Pollack5. 1Hepatogastroenterology, CHU Lille, Lille, France, 2Gastroenterology and Medicine, St Mark’s Hospital, Harrow, UK, 3Inflammatory Bowel Disease Center, Vanderbilt University Medical Center, Nashville, TN, 4Gastroenterology, Mayo Clinic, Rochester, MN, 5Immunology Development, Abbott, Parsippany, NJ, USA

Introduction: Adalimumab (ADA), a fully human monoclonal antibody targeting tumour necrosis factor, is effective in the induction and maintenance of remission in patients with Crohn’s disease (CD). Fistulising disease complicates the course of CD in 20%–30% of patients and frequently leads to surgical resection. The efficacy of ADA in fistula closure and response was previously seen in the CHARM trial.1

Aims & Methods: We assessed the long-term efficacy of ADA in sustaining fistula healing and response in patients with fistulising CD. In the CHARM trial, patients were randomised to receive placebo, ADA 40 mg every other week (EOW), or 40 mg every week (EW). At or after Week 12, patients experiencing flare or non-response could receive open-label (OL) ADA. At the end of CHARM (56 weeks), patients were eligible to enrol in an OL extension during which they received ADA EOW OL or remained on ADA EW OL. In the extension trial, escalation from EOW to EW therapy was permitted for flares or non-response. Data from the two ADA dosage groups were pooled and evaluated using both non-responder imputation (data not shown) and last observation carried forward (LOCF) analyses. Endpoints were defined as complete healing or ⩾50% reduction of draining fistulas at time points since CHARM baseline.

Abstract 089 Long-term fistula healing and response through 24 months of adalimumab therapy

Results: The percentage (%) of patients with healed fistulas and the % with ⩾50% fistula response at 6 and 12 months in the blinded study and at 6 and 12 months of the OL extension (up to 24 total months of ADA therapy) are presented in the table (LOCF analysis). Months 18 and 24 in the table represent 6 and 12 months in the OL extension after up to 12 months of blinded therapy in the CHARM trial.

Conclusion: Adalimumab has sustained fistula healing and response through 2 years of therapy. Nearly two thirds of the patients in the OL extension who had fistulas at baseline of CHARM showed continued response with ADA maintenance treatment, and more than half of ADA-treated patients experienced sustained fistula closure.

A RETROSPECTIVE ANALYSIS OF SMALL BOWEL MRI IN EVALUATION OF DISEASE ACTIVITY IN ILEAL CROHN’S DISEASE

C. W. Lees1, C. Parisinos1, D. Subedi2, A. G. Shand1, I. D. R. Arnott1, S. Glancy2, J. Satsangi1. 1Gastrointestinal Unit, 2Radiology Department, Western General Hospital, Edinburgh, UK

Introduction: Crohn’s disease (CD) patients frequently have ileal involvement (70%). Barium follow-through has been the standard investigation of the small bowel (SB), but is limited by lack of discrimination between inflammatory and fibrotic strictures and high radiation exposure. Small bowel magnetic resonance imaging (SBMRI) has recently been introduced as a tool for assessment of disease activity, and may impact on the therapeutic decision pathway of CD patients.

Aims & Methods: We aimed to assess the reliability of SBMRI in detecting disease activity and its influence on subsequent management. 186 patients had SBMRI performed at WGH from 2004–7. Patients were asked to drink 1.5 l hyper-osmolar water solution pre-MRI. 103/186 had SBMRI to investigate possible ileal CD. 75/103 had either known CD (Lennard-Jones criteria) or proven ileal CD at SBMRI and were included in the main analysis. A total of 86 SBMRIs were performed in these 75 CD patients. There were significantly more females in this cohort (73.4%, p<0.001). Case-notes were reviewed in detail with documentation of serum and faecal inflammatory markers and histopathology pre- and post-SBMRI. Agreement between inflammatory markers and SBMRI findings was determined with Cohen’s kappa statistic.

Results: 32/86 (37.2%) of SBMRIs showed active ileal CD and 41/86 (47.6%) quiescent CD, while 11/86 (12.7%) were suboptimal (inadequate ileal distension secondary to limited contrast ingestion). There was substantial agreement between active ileal CD on SBMRI and CRP (κ = 0.61, p = 0.0004), faecal calprotectin (κ = 0.72, p = 0.047) and histopathology (κ = 0.66, p = 0.0006). SBMRI had 75% sensitivity and 93% specificity for active CD when compared with histopathology. Management was influenced by SBMRI reports following active (29/34, 75% treated medically) or quiescent (19/27, 85.3% managed conservatively) disease. Fibrotic strictures were predominantly treated surgically (5/6, 83.3%). 3/8 patients with inflammatory ileal strictures required surgery; only 1/3 did not have a trial of medical therapy. Active colonic CD was identified on 6/86 (7.0%) SBMRIs. In addition, a total of 26 extraintestinal findings were detected in the 103 SBMRIs performed for possible ileal CD.

Conclusion: SBMRI provides accurate information on ileal CD activity, with close correlation to inflammatory markers and histopathology. It represents a substantial advance in the staging of CD, with the potential to direct medical and surgical management, while avoiding harmful radiation exposure in young patients.

GENE SPECIFIC HYPERMETHYLATION IN COLORECTAL MUCOSA OF PATIENTS WITH ULCERATIVE COLITIS

R. P. Arasaradnam1, K. T. J. Khoo1, D. M. Commane1, J. Mathers1, M. Bradburn2. 1Human Nutrition Research Centre, School of Clinical Medical Sciences, Newcastle University, 2Department of Surgery, Northumbria Healthcare NHS Trust, Newcastle, UK

Introduction: We have previously shown evidence of global DNA hypermethylation in patients with ulcerative colitis (UC) and accompanying low folate status.1 Epigenetic modifications such as gene specific methylation are thought to occur early in the multi-step pathway of colorectal carcinogenesis. For example, p16INK4a a tumour suppressor gene (inactivated in most human cancers including colon cancer) has been shown to be silenced as a result of promoter methylation.2 Similarly, ESR-1 a putative tumour suppressor gene has been shown to be hypermethylated in both colon cancer as well as UC3 but few studies exist on gene specific (tumour suppressor) methylation in UC.

Aims & Methods: To assess gene specific methylation of ESR-1 (oestrogen receptor-1) and N-33 (tumour suppressor candidate-3) in the macroscopically normal mucosa of UC patients. 68 subjects were recruited (24 UC patients and 44 age- and sex-matched controls). Colorectal mucosal biopsies were obtained and DNA extracted. Gene specific methylation was quantified using the COBRA method described by Xiong et al.4

Abstract 091 Results

Results: Mean duration of UC was 9 years with on average 1 flare-up/year and mean SCCAI score was 2. Percentage methylation of both ESR-1 and N-33 were significantly (p = 0.015 and p<0.001 respectively) higher in UC subjects compared with age and sex matched controls.

Conclusion: In this study, we have confirmed the observation of ESR-1 hypermethylation but in addition, demonstrated a novel finding of hypermethylation of another putative tumour suppressor gene, N-33, in UC subjects. This preliminary finding suggests perhaps an alternative pathway for colorectal carcinogenesis in UC—that is, inactivation through methylation of putative tumour suppressor genes. Ongoing work is being undertaken to determine if lifestyle factors such as smoking and diet influence gene specific methylation.

This study is funded by the Food Standards Agency (N12015).

PROSPECTIVE MRI TO GUIDE PERINEAL CROHN’S FISTULA TREATMENT WITH INFLIXIMAB, ADALIMUMAB AND THALIDOMIDE

S. C. Ng1, S. Plamondon1, A. Swatton1, A. Gupta2, M. A. Kamm1. 1Gastroenterology, 2Radiology, St Mark’s Hospital, London, UK

Introduction: Anti-tumour necrosis factor (TNF) therapies are effective in treating perineal fistulising Crohn’s disease. We prospectively evaluated the efficacy of infliximab, adalimumab and thalidomide in treating perineal fistulising Crohn’s disease, and assessed the value of MRI to determine deep healing and influence duration of treatment.

Aims & Methods: Infliximab naïve patients received infliximab, while patients who had previously failed infliximab received adalimumab or thalidomide. Clinical fistula evaluation was performed at baseline and each 8 weeks. MRI was performed at baseline and weeks 34 and 52. Clinical healing, MRI assessment of fistula track healing, Crohn’s Disease Activity Index (CDAI), and Perianal Crohn’s Disease Activity Index (PCDAI) were prospectively assessed.

Results: 34 patients with perineal fistulas (29 anorectal and 5 anovaginal) were included. Median follow-up to date was 44 weeks (range 22–93). 19 patients received infliximab, 7 had adalimumab and 8 had thalidomide. 30 patients had concurrent luminal disease. Concurrent thiopurine was used in 17 of 19 infliximab, 1 of 7 adalimumab, and 3 of 8 thalidomide- treated patients. Adalimumab-treated patients had previously failed infliximab: 4 lost response, 2 anaphylaxis, 1 no response. At week 22, response (more than 50% reduction draining fistulas) and complete fistula closure, respectively, was seen in 16 (84%) and 10 of 19 (53%) infliximab-treated patients, 4 (57%) and 1 of 7 (14%) adalimumab-treated patients, and 4 (50%) and 1 of 8 (13%) thalidomide-treated patients. Three patients on thalidomide stopped treatment at week 2 due to side effects. At week 34, 19 patients have had a repeat MRI. Of 15 patients with clinical response, MRI showed complete healing in 4, improvement in 8 and unchanged in 3 patients. Of 4 patients without clinical response, MRI showed improvement in 1 and unchanged in 3. At week 52, of 13 patients who have had a repeat MRI, fistulas remained healed in the same 4 patients who had healed at week 34, further improvement in 5, and unchanged in 4 patients, compared with MRI at week 34. Mean PCDAI reduced from 9 at baseline to 4 at last follow-up. Mean CDAI reduced from 257 at baseline to 152 at last follow-up (p<0.05).

Conclusion: Anti-TNF therapy is highly effective in treating perineal Crohn’s fistulas. Limited preliminary data suggest that MRI may indicate when treatment can cease. In keeping with previous findings in luminal disease, in fistulising disease the use of a second antibody is proportionally less effective than the first. Fistula tracks are slower to resolve than clinical remission. Prolonged treatment is often required to achieve internal track resolution.

THE SAFETY PROFILE OF INFLIXIMAB FOR INFLAMMATORY BOWEL DISEASE IN EDINBURGH: FINAL ANALYSIS OF OVER 200 PATIENTS

C. W. Lees1, A. Ali1, A. I. Thompson1, R. O. Forsythe1, L. Marquez1, S. Aitken1, J. Fennell1, P. Rogers2, A. G. Shand1, I. D. Penman1, K. R. Palmer1, D. C. Wilson2, I. D. R. Arnott1, J. Satsangi1. 1Gastrointestinal Unit, Western General Hospital, 2Department of Paediatric Gastroenterology and Nutrition, Royal Hospital for Sick Children, Edinburgh, UK

Introduction: The efficacy of infliximab therapy is established for Crohn’s disease (CD) and to a lesser extent ulcerative colitis (UC). However, serious adverse events have been reported. We therefore set out to review the short and long-term safety of all patients treated with infliximab at the Western General Hospital (WGH) and Royal Hospital for Sick Children (RHSC), Edinburgh from 1999 to 2007.

Aims & Methods: Data were available on 202 patients (182 WGH, 20 RHSC) comprising 157 (77.7%) CD, 42 (20.8%) UC, and 3 (1.5%) coeliac disease, who received a total of 718 infliximab infusions. The median age at diagnosis was 23 years, with median duration diagnosis to first infusion of 4.9 years. Patients had a median of 3 infliximab infusions (range 1–35), with a median 2.5 years follow-up. 59.4% of patients received induction therapy only, 20.7% induction plus episodic and 19.8% induction plus maintenance therapy. The primary outcome was the number of adverse events occurring in the 90 days following infliximab infusion. Serious adverse events were defined as fatal or life-threatening, or leading to/prolonging hospitalisation.

Results: There were a total of 95 adverse events in 58/202 (28.7%) of patients in the 90 days following infliximab, with 30/202 (14.9%) having a serious adverse event. Of a total of 7 deaths in this cohort, 6 were >1 year post-infliximab and considered unrelated. 42/202 (20.8%) had an infectious event of which 22 were serious. 6/202 (3.0%) patients were diagnosed with malignancy after infliximab therapy. There were 3 cases of lung cancer in current or ex-smokers, all >65 years at diagnosis. 27/202 (13.4%) of patients had an infusion reaction. 20 acute infusion reactions were recorded in 19 patients; 4/20 were serious, including 1 anaphylactic reaction requiring adrenaline. 8 serum-sickness reactions occurred; 3/8 were serious, requiring hospitalisation for iv steroids. The median longest duration between previous infusions was 895 days. One patient (15 years old, CD) developed severe drug-induced lupus.

Conclusion: Important safety considerations of infliximab were identified in this retrospective cohort and the risk/benefit analysis needs to be carefully assessed and discussed with patients before commencement of therapy.

COMMON POLYMORPHISM IN THE ALDEHYDE OXIDASE GENE IS A MARKER OF NON-RESPONSE TO AZATHIOPRINE THERAPY IN INFLAMMATORY BOWEL DISEASE

M. A. Smith1, T. Marinaki2, A. Ansari1, J. Sanderson1. 1Gastroenterology, 2Purine Research Laboratory, Guy’s and St Thomas’ NHS Foundation Trust, London, UK

Introduction: Azathioprine (AZA) and 6-mercaptopurine (6MP) remain the immunosuppressants of choice for inflammatory bowel disease (IBD). Their metabolism has been extensively studied and both are thought to be pro-drugs activated by conversion to 6-thioguanine nucleotides (6TGNs). There are reports of aldehyde oxidase (AOX) acting on AZA, 6MP and the products of their metabolism, but the role played by the metabolites produced has not been established. There is only one reported coding SNP in the AOX gene: 3404A>G (encoding the amino acid substitution Asn1135Ser) it has a reported gene frequency of 0.16. We set out to establish whether the presence of this SNP is associated with clinical outcome in IBD patients treated with AZA.

Aims & Methods: A cohort of 182 prospectively recruited patients starting 2 mg/kg AZA for IBD was divided into responders, non-responders and those discontinuing treatment due to side effects. The whole cohort was genotyped for the presence of the coding SNP in AOX using real-time PCR Taqman drug metabolism genotyping assay. Contingency tables, chi-squared testing, Fisher’s exact test and unpaired Student’s t test were used for statistical analysis. DNA sequencing was used to confirm the accuracy of the SNP probe.

Results: 58% of patients recruited had Crohn’s disease. 74 (41%) withdrew due to side effects, 66 (36%) had a complete response. The SNP AOX 3404A>G had an allele frequency of 0.12 in our population over all. The allele frequency in responders was 0.08, but in non-responders the allele frequency was 0.18. The presence of SNP AOX 3404A>G strongly predicted a lack of response to AZA treatment as a dominant effect (p = 0.007). Sequencing data has confirmed the accuracy of the SNP probe. Failure to respond to therapy could not be attributed to differences in TGN level which did not differ significantly between those with and without the SNP (p = 0.56). Likewise, TPMT was not significantly different between those with and without the AOX SNP (p = 0.27). Importantly patients with both AOX 3404A>G and a TPMT greater than 35 had the lowest chance of responding (27%) compared to 55% for those with one of these predictors of non-response and an 86% chance of responding for those with neither predictor (p = 0.0002).

Conclusion: We have identified a novel pharmacogenetic marker of non-response to AZA. Our findings suggest that the thiopurine metabolites generated by AOX action may be clinically significant and warrant further study. Furthermore and additive effect between AOX 3404A>G and high TPMT predicts a group unlikely to respond to thiopurine therapy.

CORTICOSTEROIDS INCREASE INTERLEUKIN-10 AND INHIBIT INTERLEUKIN-12P40 PRODUCTION BY INTESTINAL DENDRITIC CELLS IN ACUTE ULCERATIVE COLITIS: NOVEL MECHANISMS OF THERAPY

S. C. Ng1, S. Plamondon1, H. O. Al-Hassi1, M. A. Kamm2, S. C. Knight1, A. J. Stagg1. 1Antigen Presentation Research Group, 2Gastroenterology, St Mark’s Hospital, London, UK

Introduction: Intestinal dendritic cells (DC) interact with the luminal flora and play a key role in maintaining immune homeostasis. Altered DC function and dysregulated cytokine production contribute to the inappropriate recognition of bacteria that drives inflammation in inflammatory bowel diseases.1 2 Corticosteroids, widely used in the treatment of acute ulcerative colitis (UC), work in part by suppression of T cell activation,3 4 but their influence on DC, key inflammatory initiators, is unknown.

Aims & Methods: We assessed cytokine production by ex vivo colonic DC from patients with acute UC, and determined the effects of in vivo corticosteroids. Rectal biopsies were obtained from inflamed tissue of patients with active UC, symptomatic for less than four weeks (n = 30; sigmoidoscopy score more than 2) and healthy controls (n = 16). Myeloid DC were identified by multicolour flow cytometry as CD11c+ HLA-DR+ lineage- cells (lin = anti-CD3, 14, 16, 19 and 34) in lamina propria mononuclear cells extracted from collagenase digested intestinal tissue and compared with CD11c- cells within the same gate. “Spontaneous” interleukin (IL)-10, IL-12p40, IL-6, and IL-13 production by lamina propria DC was measured by intracellular staining of permeabilised cells in the absence of exogenous stimulation. In seven UC subjects, cytokine production by colonic DC was assessed before and after 8 weeks of oral steroid treatment.

Results: In acute UC a significantly greater proportion of colonic myeloid DC produced IL-10 and IL-12p40 than did equivalent cells from control tissue (p<0.05). In contrast, production of IL-6 and IL-13 by myeloid DC did not differ between UC patients and controls. Compared with myeloid DC, fewer CD11c- HLA-DR+ lineage-cells in UC patients and controls produced the cytokines examined. Following high dose corticosteroids, there was an increase in IL-10 production by colonic myeloid DC (p<0.05), and a reduction in IL-12p40 production by both colonic myeloid DC and CD11c- HLA-DR+ lineage-cells in UC patients (p<0.05). There were no significant differences in IL-6 and IL-13 production by DC before and after corticosteroids.

Conclusion: Acute UC is associated with enhanced colonic DC production of the regulatory cytokine, IL-10 and the proinflammatory cytokine, IL-12p40. Oral steroids suppress production of IL-12p40 and potentiate secretion of IL-10 by colonic DC. These findings provide new insights into the potential mechanism of glucocorticoid gut immunosuppression and therapeutic efficacy in inflammatory bowel disease.

EFFECTIVE PROBIOTIC TREATMENT (VSL#3), BUT NOT PLACEBO, IN ACUTE ULCERATIVE COLITIS IS ASSOCIATED WITH DOWN-REGULATION OF INFLAMMATORY INTESTINAL DENDRITIC CELLS

S. C. Ng1, S. Plamondon1, A. L. Hart1, M. A. Kamm2, S. C. Knight1, A. J. Stagg1. 1Antigen Presentation Research Group, 2Gastroenterology, St Mark’s Hospital, London, UK

Introduction: In active ulcerative colitis (UC) gut bacteria drive inflammation, and preliminary data suggest that probiotic bacteria are therapeutically effective. Gut dendritic cells (DC) are pivotal in early bacterial recognition and T cell signalling and probiotic bacteria cause decreased pro-inflammatory DC interleukin (IL)-12p40 and increased regulatory IL-10 production in vitro.1 2

Aims & Methods: As part of a double-blind, placebo-controlled, therapeutic study investigating the effect of the probiotic mixture VSL#3 (8 bacterial strains: 4 lactobacillus, 3 bifidobacteria, and 1 streptococcus) in acute UC, we studied for the first time how probiotics modify DC function ex vivo, and how this correlates with therapeutic efficacy. In patients with acute UC (UC disease activity index 3–8), rectal biopsies were taken at baseline and after 8 weeks of treatment with VSL#3 (n = 10) or placebo (n = 9). Using multicolour flow cytometry, myeloid DC were identified as CD11c+ HLA-DR+ lineage-cells (lin = anti-CD3, 14, 16, 19, 34) in freshly isolated lamina propria mononuclear cells from collagenase digested colonic tissue.3 Surface expression (level and proportion of positive cells) of the activation markers (CD40, CD86), pattern recognition receptors (Toll-like receptors (TLR), TLR-2 and TLR-4), and spontaneous IL-10, IL-12p40, IL-6, and IL-13 production by DC, measured by intracellular staining of permeabilised cells in the absence of exogenous stimulation, were measured before and after VSL#3 or placebo.

Results: More patients on VSL#3 responded to treatment, compared with placebo. In patients treated with VSL#3, proportion positive cells and level of CD86 expression by colonic myeloid DC was significantly reduced (p<0.05), TLR-2 expression on DC was downregulated (p<0.05), IL-10 positive colonic DC were significantly increased and IL-12p40 producing DC were significantly reduced (p<0.005). The level of expression of IL-12p40 was also significantly decreased (p<0.05). There were no significant changes in TLR-4, IL-6 and IL-13 after VSL#3. In patients who received placebo, colonic DC expressed significantly more TLR-2 on repeat sampling (p<0.05), but no significant changes were detected for CD40, CD86, TLR-4 or intracellular cytokines.

Conclusion: In acute UC, probiotic treatment with VSL#3 differentially modulates the function of human intestinal DC studied ex vivo, inducing regulatory cytokine and down-regulating co-stimulatory molecules and pro-inflammatory cytokines. These changes correlate with clinical improvement. Probiotic effects on DC may be central to their therapeutic effect in UC.

DIETARY N-3 POLYUNSATURATED FATTY ACIDS PROTECT AGAINST THE DEVELOPMENT OF ULCERATIVE COLITIS: A UK PROSPECTIVE COHORT STUDY

S. John1, R. Luben2, H. J. Kennedy1, A. Welch3, K. Khaw4, S. A. Bingham5, A. R. Hart1. 1Gastroenterology, Norfolk and Norwich University Hospital, Norwich, 2Department of Public Health, University of Cambridge, Cambridge, 3Population Health, University of East Anglia, Norwich, 4Clinical Gerontology, 5Cancer Sciences, University of Cambridge, Cambridge, UK

Introduction: The aetiology of ulcerative colitis is unknown, although it is plausible that dietary n-3 polyunsaturated fatty acids (n3-PUFAs) may have a protective role. Metabolites derived from these nutrients, which are found in the colonic mucosa of patients, are less proinflammatory than those derived from n-6 PUFAs.

Aims & Methods: As no prospective cohort studies have investigated this association, the aim of this study was to report the effects of both total dietary n-3 PUFAs and the specific n-3 PUFAs eicosapentaenoic acid (EPA)and docosahexaenoic acid (DHA) on the risk of ulcerative colitis. A total of 25 630 participants without ulcerative colitis, living in Norfolk UK, completed detailed 7-day food diaries. These were interpreted using a specifically designed computer programme which converted food items into nutrients including n-3 PUFAs. The cohort was monitored for participants who developed ulcerative colitis and the diagnosis was confirmed by experienced physicians. Each case was matched for age and gender with four controls and analyses performed using unconditional logistic regression.

Results: In the cohort, 22 cases of incident ulcerative colitis were identified after a median follow-up time of 4.2 years (range 1.8–8.3 years). Statistically significant protective odds ratios for trends across quartiles were found for: total n-3 PUFAs (OR 0.49, 95% CI 0.25 to 0.99), EPA (OR 0.52, 95% CI 0.28 to 0.99) and DHA (OR 0.40, 95% CI 0.20 to 0.79) after adjusting for age, gender, total energy intake and cigarette smoking. These protective associations persisted after adjusting for n-6 PUFAs which can affect the metabolism of n-3 PUFAs.

Conclusion: Total dietary n-3 PUFAS, EPA and DHA are significantly associated with protection from the development of ulcerative colitis. This association needs to be investigated in other cohort studies to see if the effect is consistent. If the association is causal, then increasing the intake of n-3 PUFAs in the population should be helpful in reducing risk of ulcerative colitis.

MONITORING AZATHIOPRINE TREATMENT IN INFLAMMATORY BOWEL DISEASE: RITUALS RE-VISITED—BUT NOT REVISED!

M. Walsh, C. Royston, M. Srinivas, B. Hoeroldt, P. J. W. Willemse, P. Basumani, K. D. Bardhan. Gastroenterology, Rotherham General Hospital, Rotherham, UK

Introduction: Inflammatory bowel disease (IBD) patients on long-term AZA routinely have their haematology (± biochemistry) monitored regularly,1 a time consuming process often inconvenient to patients otherwise well.

Aims & Methods: Our aim was to assess if such monitoring is effective. All patients with IBD were followed-up, their data prospectively updated at each visit and later transferred to our electronic database. AZA starting dose was generally 50 mg (n = 205). Full blood count (FBC) and liver function tests (LFT) were routinely checked at baseline, 1, 2, 4, 8 and 12 weeks and 3 monthly thereafter if in remission and whenever an untoward event or unscheduled visit occurred. The process was repeated whenever dosage was increased. Thiopurine methyl transferase (TPMT) levels (introduced 1999) were checked whenever possible before starting AZA.

Abstract 098 Time to adverse events

Results: Numbers treated: Crohn’s disease 141/339 (42%), ulcerative colitis 103/393 (26%), indeterminate colitis 10/50 (20%), total 254/782 (32%). Adverse Events (AE) occurred in 122/254 (48%) of patients and usually led to the withdrawal of AZA. AE were independent of disease type hence pooled data are presented. Type of AE: 81/122 patients (66%) had constitutional disturbance (CONS) alone; 25 (20%) had silent FBC or LFT changes; 15 (12%) had both. CONS (total 96) included GI and/or influenza-like symptoms in 64. FBC changes (17) included falling total white cell or neutrophil count in 12; none had neutropenic crisis. LFT changes occurred in 24. 1 patient had FBC and LFT changes. Separately, 1 patient developed myeloma after 8 years. Dose vs time: 91 patients developed AE on the 50 mg dose and the majority of AE, irrespective of dose, occurred early (⩽3 m). Most late constitutional but only half of late FBC changes were associated with dose increase. TPMT: Tested in 157/254. Normal/high n = 134, intermediate/carrier n = 23. AE occurred in 56% and 57% respectively. None had severe deficiency.

Conclusion: (1) AZA associated AE were common, independent of starting dose, IBD type or TPMT levels (carrier or normal). (2) The majority of constitutional events and LFT changes occurred early in treatment. In contrast FBC changes were spread over time. (3) Most late AE followed dose increase. (4) The yield of clinically useful information from regular monitoring diminished markedly after a year of stable dose. The infrequency of late events gives us much reassurance. Nevertheless we are concerned with the unforeseen consequences of prolonged immunosuppression, therefore, our “ritual” of close monitoring remains unchanged.

Liver free papers

EUROPEAN LIVER FIBROSIS MARKERS CAN PREDICT CLINICAL OUTCOME IN PATIENTS WITH CHRONIC LIVER DISEASE

J. Parkes1, P. Roderick2, S. Harris2, C. Gough3, M. Wheatley2, G. Alexander4, J. Collier5, C. Day6, M. Lombard7, D. Mutimer8, J. Ramage9, G. Dusheiko10, A. Burt11, E. Sampson12, W. Rosenberg13. 1Public Health Sciences and Medical Statistics, University of Southampton, 2Public Health Sciences and Medical Statistics, University of Southampton, 3Wellcome Clinical Research Facility, Southampton General Hospital, Southampton, 4Department of Hepatology, Addenbrooke’s Hospital, Cambridge, 5Department of Gastroenterology, John Radcliffe Hospital, Oxford, 6Department of Hepatology, Freeman Hospital, Newcastle upon Tyne, 7Department of Gastroenterology, Royal Liverpool & Broadgreen Hospital, Liverpool, 8Department of Hepatology, Queen Elizabeth Hospital, Birmingham, 9Department of Gastroenterology, N Hants Hospital, Basingstoke, 10Department of Hepatology, Royal Free and University College School of Medicine, London, 11Department of Histopathology, Freeman Hospital, Newcastle upon Tyne, 12Siemens Diagnostics, Siemens, Tarrytown NY, United States, 13Institute of Hepatology, University College London, London, UK

Introduction: When attempting to predict clinical outcomes of liver disease clinicians use the surrogate of fibrosis in a liver biopsy as a reference standard. Performance of non-invasive tests as predictors of clinical outcomes has been evaluated against fibrosis on biopsy but use of clinical outcomes as the reference standard would be ideal. The ELF panel of markers (TIMP1, HA, PIIINP) was derived and validated in a cohort of patients recruited in 1998–9 and shown to have excellent ability to identify significant fibrosis on biopsy. To evaluate ELF performance in predicting clinical outcomes, the original ELF cohort was followed-up at 8–9 years.

Aims & Methods: Patients recruited to the ELF study at 7 English centres were followed up for liver morbidity and mortality by examination of clinical data. Those lost to follow-up were followed up for morbidity by questionnaires to the GP. All deaths were ascertained and death certificates were obtained. Primary outcome measure was liver related morbidity/death; secondary outcome was all-cause mortality.

Results: 498 patients were followed up after a median of 6.7 (0.3–9.1) years. The median age = 43 years (19–75), 66% male, 42% CHC. 41% were lost to follow-up/discharged. 85% of GP questionnaires were returned reporting no liver outcomes. There were 67 liver-related outcomes (39 liver-related deaths), 70 deaths overall. Kaplan Meier plots showed that tertiles of baseline ELF score can predict liver outcomes and all-cause mortality (log rank test p<0.001) Cox proportional hazards analyses show that ELF remains predictive when adjusted for age, gender and disease aetiology (p<0.0001); adjusted HR for the middle tertile ELF score versus lowest tertile ELF score = 6, and for highest tertile ELF score versus lowest tertile HR = 29. From logistic regression at 6 years follow-up the OR for liver related outcome was 2.14.

Conclusion: ELF score can predict clinical outcomes in patients with CLD and is likely to be a useful prognostic tool. A unit change in ELF score is associated with a doubling of the odds of getting a clinical outcome at 6 year follow-up.

HIGHLY EFFICIENT DIFFERENTIATION OF HUMAN EMBRYONIC STEM CELLS TO FUNCTIONAL HEPATOCYTES IN VITRO AND IN VIVO REQUIRES ACTIVIN A AND WNT3A SIGNALLING

D. Hay1, J. Fletcher1, C. Payne1, J. Terrace1, R. Gallagher1, J. Black1, A. Pryde1, C. Filippi1, S. Forbes1, K. Samuel1, J. Ross1, J. Iredale2, P. N. Newsome3. 1Centre for Regenerative Medicine, 2Centre for Inflammation Research, University of Edinburgh, Edinburgh, 3Institute of Biomedical Research, University of Birmingham, Birmingham, UK

Introduction: Human embryonic stem cells (hESC) are a potentially valuable source of human hepatocytes, although their homogeneous differentiation to specific germ layers/cell types has proven difficult to achieve. Wnt signalling pathways have been shown to play important roles in development and cellular differentiation in vivo.

Aims & Methods: Our aim was to define the role of Wnt signalling in hepatic development and its effect on the hepatocytic differentiation of hESC. Wnt3a expression was examined in first (T1) and second (T2) trimester liver sections by immunostaining. We used cytokine combinations of Wnt, Activin A and Sodium Butyrate to direct hESC towards the hepatocyte lineage. Hepatocyte phenotype was confirmed by immunofluorescent staining and metabolic assays. Hepatocyte-like cells (hESHs) were injected (1×106 cells) intrasplenically (IS) or intrahepatically into NOD-scid mice. Functionality was confirmed by histological analysis and measurement of human albumin in serum.

Results: T1 liver exhibited Wnt3a staining adjacent to developing portal structures, but was not detected in the parenchyma. However, in T2 livers the pattern of expression was both portal and parenchymal suggesting that Wnt3a parenchymal expression correlated with hepatocyte differentiation in the developing liver, and may play a role in early hepatocyte differentiation. After 10 days culture hESC expressed a wide range of hepatocytic genes (albumin, AFP, HNF4a, tyrosine aminotransferase, tryptophan oxygenase, apolipoprotein F and cytochrome p450 3A4) and proteins (Albumin, Glycogen storage). The final population of hESHs was 90% pure as assessed by albumin and HepPar1 expression. Metabolic function was confirmed by urea synthesis (0.9 μmol/h/g TP), Gluconeogenesis (1.5 μmol/h/g TP) and p450 metabolism. We have been able to direct the hepatocytic differentiation of hESC using standard culture methodology and also from clonal cell lines. After IS injection hESHs were seen in the murine liver and spleen, expressed albumin and cytokeratin 18/19 and were seen to be cycling (Ki67 expression). These cells produced albumin which was detected in murine serum (10 μg/ml).

Conclusion: (1) We demonstrate that Wnt3a is expressed at critical stages of human liver development in vivo, highlighting its important role in development. (2) Wnt3a directed differentiation of hESC results in a functional, highly homogenous population of human hepatocytes which function in vitro and in vivo studies.

HOSPITALISATION OF HEPATITIS C DIAGNOSED PERSONS IN SCOTLAND FOR DECOMPENSATED CIRRHOSIS: A POPULATION-BASED RECORD-LINKAGE STUDY

S. A. McDonald1, S. Hutchinson2, S. Bird3, P. Mills4, J. Dillon5, P. Hayes6, C. Robertson7, D. Goldberg2. 1BBV-STI Section, 2Health Protection Scotland, Glasgow, 3MRC Biostatistics Unit, Institute of Public Health, Cambridge, 4Gartnavel General Hospital, Glasgow, 5Ninewells Hospital & Medical School, Dundee, 6Edinburgh Royal Infirmary, Edinburgh, 7Department of Statistics and Modelling Science, University of Strathclyde, Glasgow, UK

Introduction: Few studies have examined progression to the decompensated stage of cirrhosis in hetereogenous populations of persons infected with the hepatitis C virus (HCV). We investigated hospitalisation for decompensated cirrhosis among all individuals who have been diagnosed with HCV in Scotland.

Aims & Methods: The population studied consisted of 20 969 individuals diagnosed with hepatitis C through laboratory testing in Scotland between 1991 and 30 June 2006. These data were linked to other national databases on (1) all in-patient hospital discharges to retrieve episodes with discharge diagnoses of cirrhotic complications or hepatic failure (ICD-10: R18, I850, I982, K721, K729, K704, K709), and (2) all HIV tests to determine HIV co-infection status.

Results: 1027 individuals were admitted to hospital for the first time with decompensated cirrhosis during the follow-up period. The number of first-time admissions increased over the period 1996–2005, with an average annual change of 10.3%. The relative risk of hospitalisation was significantly greater for males (hazard ratio 1.3, 95% CI 1.1 to 1.6), for those with a previous alcohol-related hospital admission (hazard ratio = 5.7, 95% CI 4.7 to 6.9), for HIV co-infected persons (hazard ratio 1.6, 95% CI 1.0 to 2.5), and for those aged over 30 years (30–39: hazard ratio 3.6, 95% CI 2.3 to 5.6; 40–49: hazard ratio 10.1, 95% CI 6.5 to 15.7; 50–59: hazard ratio 17.8, 95% CI 11.1 to 28.6; 60+: hazard ratio 29.5, 95% CI 17.7 to 49.1).

Conclusion: The burden to the Scottish healthcare system from HCV-infected persons developing cirrhotic complications is increasing, due to the advancing age of this Scottish HCV-diagnosed population. Of the epidemiological risk factors examined, a history of alcohol-related hospitalisation was most strongly associated with progression to decompensated cirrhosis. A greater proportion of HCV-infected individuals needs to be identified, treated and educated about the importance of minimal alcohol consumption to achieve a greater impact on the morbidity of this disease.

EARLY LIVER BIOPSY IS CRUCIAL FOR ACCURATE DIAGNOSIS OF ALCOHOLIC HEPATITIS AND PROVIDES INDEPENDENT PROGNOSTIC INFORMATION IN PATIENTS WITH DECOMPENSATED ALCOHOLIC LIVER DISEASE

R. P. Mookerjee1, C. Lackner2, V. Stadlbauer1, R. Williams1, R. Jalan1. 1Hepatology, Institute of Hepatology, UCL, London, UK, 2Institute of Pathology, University of Graz, Graz, Austria

Introduction: Severe alcoholic hepatitis (AH) has a high mortality and accurate diagnosis is essential to allow appropriate intervention. Selecting patients for anti-inflammatory therapy without increasing risk of sepsis has proved difficult through inaccurate diagnosis of AH.

Aims & Methods: This study questioned the importance of liver biopsy in diagnosing AH in decompensated alcoholic cirrhosis and applied a histological score to see if this would identify patients with the highest mortality. Methods: Patients were studied with an appropriate alcohol history presenting acutely with decompensated alcoholic liver disease, and excluded if they had known additional liver pathology or culture positive infection within 72 hours of admission. All patients had a liver biopsy within 3-days of admission, graded 1–5 for alcoholic steatohepatitis: (0: no AH; 1: ballooned hepatocyes comprising <5% of parenchyma and no inflammatory cells (mononuclear/neutrophils); ranging up to grade 5: ballooned hepatocytes comprising >50% of parenchyma and moderate to many inflammatory cells). Hepatocyte balloon degeneration was also verified immunohistochemically by loss of cytoplasmic keratin 8 and 18 staining.

Abstract 102

Results: 37 patients were graded as having no significant steatohepatitis (score 0–1) despite 18 patients having elevated systemic inflammatory response syndrome components (SIRS- possibly reflecting underlying infection), and 62 had features of established cirrhosis. 19 patients died, 6 with no significant AH. Cytokeratin 8/18 staining to qualify balloon degeneration in a subset of 21 AH+ve patients, showed a good correlation with steatohepatitis grade (r = 0.83, p<0.0001). The odds ratio (95% CI) for mortality in patients with AH was 4.3 (1.4 to 13.6) and for SIRS 7.6 (1.9 to 30). There was a significant difference in mortality in patients with and without SIRS and AH (p<0.01). In SIRS positive patients, discriminant function and Child score had no significant association with outcome.

Conclusion: Liver biopsy is crucial for the diagnosis of alcoholic hepatitis. Clinical assessment and the use of SIRS overestimates AH, whilst its diagnosis is missed in a proportion of decompensated patients without SIRS. Histological grading of AH has prognostic value independent of SIRS, which may help define treatment strategy.

CURRENT MEASUREMENTS OF RENAL FUNCTION DO NOT REFLECT TRUE GLOMERULAR FILTRATION: A COMPARISON OF STANDARD AND NOVEL MARKERS OF RENAL FUNCTION IN PATIENTS UNDERGOING LIVER TRANSPLANT ASSESSMENT

A. J. Portal1, M. McPhail1, I. Coltart1, M. Bruce1, S. Hughes1, N. Taylor1, R. Sherwood2, J. O’Grady1, N. Heaton1, M. A. Heneghan1. 1Institute of Liver Studies, 2Department of Biochemistry, Kings College Hospital, London, UK

Introduction: Renal dysfunction in cirrhotic patients is associated with significant mortality both pre and post transplantation. Creatinine clearances of <70 ml/min/1.73 m2 are associated with decreased survival following liver transplantation. Serum creatinine (SCr) is a poor reflection of glomerular filtration rate (GFR) and therefore novel methods of assessing GFR using Cystatin C (CyC) and estimated GFR (eGFR) have previously been investigated. We compared these methods against SCr and a new biomarker, neutrophil gelatinase associated lipocalin (NGAL), in 79 patients undergoing assessment for liver transplantation.

Aims & Methods: Data from 79 patients undergoing LT assessment were prospectively collected at the time of assessment. Renal function was measured by 24-hour urinary creatinine clearance (CrCl). NGAL and CyC were retrospectively analysed on stored samples. eGFR was calculated using the MDRD equation. Renal dysfunction was defined as a CrCl of <80 ml/min/1.73 m2. Results are expressed as medians and interquartile ranges.

Results: There were significant (p<0.001) correlations between CrCl and 1/NGAL (r = 0.6), CyC and eGFR (r = 0.58). The relationship between SCr and CrCl was less apparent. In patients with renal dysfunction, median SCr was within normal laboratory limits (99 mmol/l (84–130)) and 85 mmol/l (73–97, p = 0.001) in those without. Median values in those with and without renal dysfunction were: eGFR were 62 ml/min/1.73 m2 (49–82) vs 94 (81–104), p<0.001; NGAL 121 ng/ml (59–246) vs 73(48–104), p = 0.016; CyC 1.32 mg/l (0.9–1.67) vs 0.95 (0.74–1.07), p<0.001. In patients with Childs B or C cirrhosis, median SCr remained within normal limits even when CrCl was reduced to <45 ml/min/1.73 m2 (102 mmol/l (99–177)). Both CyC and NGAL were elevated outside of normal reference ranges. In patients with a CrCl of <80 ml/min/1.73 m2, eGFR of <74 ml/min/1.73 m2 had the highest receiver operator characteristic area under the curve (AUC 0.83, sensitivity 66%, specificity 93%, p<0.001). However, in those with lower GFRs, CyC and NGAL more accurately predicted low GFR than eGFR or SCr (CrCl <ml/min/1.73 m2: CyC >1.45 mg/l AUC0.85, sensitivity 80%, specificity 82%, p<0.001; NGAL >101 ng/ml AUC 0.8, sensitivity 91%, specificity 69%, p<0.001; eGFR and SCr AUC 0.78).

Conclusion: Normal SCr values do not accurately represent glomerular filtration as measured by urinary creatinine clearance. Both CyC and NGAL have good predictive value especially in patients with more severely impaired renal function. They also have the advantage of being elevated prior to a rise in SCr, making them useful screening tests in both the pre- and post-transplant period.

SILENCING TIMPS WITH SIRNA: DEFINING A ROLE FOR TIMP-1 IN HEPATIC STELLATE CELL PROLIFERATION

A. J. Fowell1, W. M. C. Rosenberg2, R. C. Benyon1. 1Liver and Pancreas Research Group, Division of Infection, Inflammation and Repair, University of Southampton, Southampton, 2Institute of Hepatology, University College London, London, UK

Introduction: Tissue inhibitors of metalloproteinase (TIMP)-1 and -2 may play a key role in hepatic fibrogenesis both by protecting extracellular matrix from matrix metalloproteinase (MMP) mediated degradation and by supporting survival of fibrogenic myofibroblastic hepatic stellate cells (HSC) which are pivotal to liver fibrosis. HSC are the major source of TIMPs during chronic liver injury, and both TIMPs and HSC represent important potential therapeutic targets.

Aims & Methods: Our aim was to investigate the phenotypic effect of silencing TIMP-1 and -2 in HSC using small inhibitory RNA (siRNA). Culture-activated rat HSC were transfected with TIMP-1 and -2 siRNA or a negative control siRNA by electroporation. Cell lysates and culture supernatants were collected to quantitate TIMP-1 and -2 mRNA and protein by real-time PCR and ELISA, respectively. Potential toxicity of siRNA electroporation was assessed by lactate dehydrogenase activity assay. Conditioned media from siRNA treated HSC were tested for their ability to inhibit the activity of recombinant active matrix metalloproteinase-2 (MMP-2) using a collagenase activity assay. HSC proliferation following siRNA mediated TIMP silencing was determined by [3H]-thymidine incorporation in serum-containing media.

Results: TIMP-1 and -2 siRNA electroporation was an effective means of silencing HSC TIMP expression. At 24, 48, 72 and 192 hours after electroporation, TIMP-1 siRNA induced 91%, 92%, 86% and 45% knockdown of TIMP-1 protein expression respectively, compared with negative control siRNA (p<0.05 at 24–72 h). TIMP-2 siRNA induced 24%, 41%, 63% and 57% knockdown of TIMP-2 protein at the same time points (p<0.05 at 48–72 h). Similar silencing was seen at the mRNA level. There was no excess toxicity in TIMP-1 siRNA treated cells compared with electroporation alone and no effect on TIMP-2 protein. Conditioned media from TIMP-1 and TIMP-2 siRNA treated cells showed 40% and 25% less inhibition respectively of MMP-2 collagenase activity than media from negative control siRNA treated cells, verifying that TIMP silencing removed a functionally important MMP-suppressive effect in HSC cultures. TIMP-2 knockdown had no effect on proliferation, whereas silencing of TIMP-1 produced a significant (p<0.05) 40% reduction in HSC proliferation.

Conclusion: These data suggest that siRNA are an effective means of silencing functional TIMP-1 and -2 activities in HSC. A novel autocrine mitogenic effect of TIMP-1 on HSC has been defined, and this effect might promote liver fibrosis in addition to TIMP-1’s previously described anti-apoptotic effect on HSC. Furthermore, siRNA targeting TIMP-1 may offer potential as a hepatic anti-fibrotic therapy.

ADRENAL INSUFFICIENCY IN PATIENTS WITH AOCLF ADMITTED TO INTENSIVE CARE. A SINGLE UNIT EXPERIENCE

M. J. Austin, A. J. Portal, I. Coltart, L. Sizer, G. Auzinger, W. Bernal, J. Wendon. Liver Intensive Care Unit, Kings College Hospital, London, UK

Introduction: Adrenal insufficiency (AI) is common in critically ill patients with septicaemia, having significant impact on prognosis.1 Patients with acute on-chronic liver failure (AoCLF) have a marked systemic inflammatory response (SIRS), and increased susceptibility to infection.24 This study was designed to determine prevalence of adrenal insufficiency, vasopressor requirements, severity of illness, and survival in a large cohort of patients with AoCLF admitted to intensive care.

Aims & Methods: All patients with AoCLF admitted to liver ITU (LITU) between November 2000 and June 2007 were eligible for the study. Cirrhosis was confirmed by histological or clinical, biochemical and radiological evaluation. Routine admission biochemical and physiological parameters were recorded allowing calculation of APACHEII, SOFA, MELD and CP score. AI was determined by short synacthen test (SST). This involved measurement of serum cortisol at time 0 (T0), and at time 30 (T30) and 60 (T60) min following injection of 250µgs of ACTH (synacthen). Requirements for inotropic support, severity of illness and survival were noted. Results expressed as n (%) or median (IQR).

Results: 259 patients underwent SST on admission to LITU; 160 (62%) male and 119 (46%) admitted with alcohol-related chronic liver disease. Median age 49 years (40–58), 50 (19%) admitted with encephalopathy, 62 (24%) with sepsis and 66 (25%) with variceal bleeding (VB). Mortality 73%. Median T0 cortisol 370 nmol/l (250–541), T30 cortisol 531 nmol/l (382–702). 82 (32%) required inotropes. ApacheII score 24 (18–29), SOFA score 10 (8–13), Child Pugh score 12 (11–13) and MELD score 28 (20–34). Those with AI: 64 (25%) were more likely to be admitted with multi-organ failure (MOF) SIRS (p = 0.05), treated with steroids (p<0.001) and require inotropic support p = 0.014. Serum albumin and lipid profile were no different between those with or without AI. LITU and 3-month survival was not related to presence of AI.

Conclusion: Adrenal dysfunction is common in patients with AoCLF. Those with AI are more likely to present with MOF or SIRS and require inotropic support. Survival and severity of illness were no different between those with or without AI.

IDENTIFICATION OF BIOMARKERS THAT MEASURE DISEASE ACTIVITY IN PATIENTS WITH NON-ALCOHOLIC FATTY LIVER DISEASE

R. Malik1, M. Lai1, K. Bhaskar1, I. Nasser2, M. Curry1, D. Schuppan1, N. Afdhal1. 1Liver Center, Department of Medicine, 2Histopathology Department, Beth Israel Deaconess Medical Center Harvard Medical School, Boston, USA

Introduction: Non-alcoholic fatty liver disease (NAFLD) is a heterogenous group of disorders ranging from simple steatosis to non-alcoholic steatohepatitis (NASH).

Aims & Methods: We characterised a group of NAFLD patients to determine the predictive value of serum markers to identify histological NASH. We assessed demographic, serological, radiological and histological variables on 95 consecutive patients with NAFLD. Diagnostic criteria for NAFLD included an alcohol intake less than 175 g/week, negative hepatitis serology and liver biopsy with greater than 10% lipid content. The serum markers assessed were CK 18 a marker of apoptosis and markers of liver fibrogenesis hyaluronic acid, TIMP 1 and YKL 40.The two tailed t test was used for parametric variables and the χ2 test for categorical variables with differences reported if p<0.05. The clinical performance of each significant variable was assessed using the area under the ROC curve.

Results: The NAFLD cohort of patients was initially stratified according to the degree of necroinflammation into simple steatosis (NAS 1–3) vs NASH (NAS score 4–8). Severe necroinflammation/NASH was associated with type 2 diabetes and high serum ALT levels. CK 18 levels were significantly higher in the NASH group compared to the simple steatosis group (394+53 u/l cf 194+26 u/l; p<0.05). There were no differences between groups in the levels of markers of liver fibrogenesis. CK 18 performed the best at identifying NASH patients yielding an AUC of 0.8, with a cut-off value of 350 u/l giving a PPV of 84% and NPV of 81%. Subsequently, we stratified the NAFLD cohort according to the degree of liver fibrosis into Metavir F0 vs F1–F4. Fibrosis was associated with advanced age, high BMI, type 2 diabetes and a high NAS score (NAS 4–8). There were significantly higher levels of CK 18 (372+49 u/l cf 206+31 u/l; p<0.05), hyaluronic acid (64+11 ng/ml cf 20+2.5 ng/ml; p<0.05) and YKL 40 (125+13 ng cf 79+10 ng/ml; p<0.05) in the fibrosis group compared to the control group. A high NAS score of 4–8 performed the best at predicting fibrosis yielding an AUC of 0.9.

Conclusion: These results indicate that CK 18 levels reflect both severity of inflammation and subsequent fibrosis in NASH, where as HA and YKL 40 exclusively mirror fibrosis/fibrogenesis. CK 18 can be used as a surrogate biomarker to assess disease severity in patients with non-alcoholic fatty liver disease.

PREVALENCE AND DETERMINANTS OF PRIMARY SCLEROSING CHOLANGITIS IN A COHORT OF PATIENTS WITH INFLAMMATORY BOWEL DISEASE AND NORMAL LIVER FUNCTION TESTS

H. K. Bungay1, O. C. Buchel2, F. Cummings2, S. P. L. Travis2, R. W. G. Chapman2. 1Radiology Department, 2Gastroenterology Department, John Radcliffe Hospital, Oxford, UK

Introduction: Primary sclerosing cholangitis (PSC) is the most frequent hepatobiliary disease associated with ulcerative colitis (UC), and is diagnosed in 2.1–4% of cases. Alkaline phosphatase is usually elevated, but may be normal in 8.5% of patients in large case series. Magnetic resonance cholangiopancreatography (MRCP) is the established method for the diagnosis of PSC. The prevalence of PSC in patients with IBD and normal serum liver enzymes is unknown.

Aims & Methods: The aim of this study was to investigate the prevalence and possible determinants of PSC in a cohort of patients with IBD and normal serum levels of liver enzymes. 58 patients with colitis and normal serum liver enzymes, recruited from the IBD clinic consented to having an MRCP. Of these, 51 had an MRCP. Demographic, lifestyle, medical and laboratory data were collected retrospectively from existing case notes. 9 volunteers without IBD and with normal serum liver enzyme levels, had MRCP examinations, and served as a control group. Variables analysed included age at MRCP investigation, duration of IBD at time of MRCP, gender difference, smoking status, Azathioprine use and presence or absence of extra-intestinal manifestations. Continuous data (age and disease duration) were compared using the student t test. Other variables were analysed using the χ2 test.

Results: All MRCP examinations in the control group were normal. The study group consisted of 32 men (63%) and 19 women (37%), aged 18 to 84 years (median age 48 years). 45 patients had pan UC, 2 Crohn’s colitis, 2 right sided colitis, 1 distal colitis and 1 pan colitis yet to be classified. Of the 51 patients, 9 (18%) had biliary abnormalities consistent with PSC on MRCP examination. There was no difference between the mean age of patients with and without PSC respectively (53.4 vs 48.4, p = 0.435). Similarly duration of IBD did not differ significantly between the two groups (17.7 vs 12.0 years, p = 0.390). The prevalence of PSC in females was 6/19 (32%) and 3/32 (9%) in males (p = 0.053, OR for males 0.224, 95% CI 0.048 to 1.038). 2/4 (50%) of active smokers had PSC as opposed to 6/41 (16%) of those not currently smoking (NS, p = 0.624). Only 2/19 patients (11%) who had ever been treated with Azathioprine had PSC, with those never receiving the drug having PSC in 7/28 (25%) of cases (NS, p = 0.197). 5/19 (26%) patients with extra-intestinal manifestations had PSC as compared to 3/26 (12%) without extra-intestinal manifestations (NS, p = 0.188). No patient has yet had a liver biopsy.

Conclusion: 18% patients with normal liver enzymes and colitis had appearances consistent with PSC on MRCP. There was a trend for more females to have the appearances of PSC, but this did not reach significance. This study suggests that PSC may be more prevalent than appreciated in patients with IBD.

STANDARDISED MORTALITY RATIO IN AUTOIMMUNE HEPATITIS

B. S. Höeroldt1, E. McFarlane2, M. Campell3, P. Basumani1, M. Karajeh2, D. Gleeson2. 1Gastroenterology, Rotherham General Hospital, Rotherham, 2Hepatology, Royal Hallamshire Hospital, 3Medical Statistics Group, School for Health and Related Research, Sheffield, UK

Introduction: The long-term outlook of autoimmune hepatitis (AIH) treated with immunosuppression is considered to be good, although follow-up data beyond 10 years are limited. We have previously reported substantial 20-year all-cause (45%) and liver-related (24%) mortality rates in AIH, despite >90% initial remission rate; however, age at presentation (median 57 years) was older than in other series. 222 patients with definite (n = 147) or probable (n = 75) AIH by International AIH Group criteria presenting to Liver Unit over 1971–2005. Capture of cases presenting since 1987 (n = 177) is thought to be near complete. 182 patients (82%) were referred directly from primary care.

Aims & Methods: To assess whether mortality in this cohort is higher than that expected for the general population. Standardised mortality ratios (SMRs), using age- (10 year blocks) and gender-specific life expectancy tables for the population of Yorkshire and Humberside were calculated.

Results: During follow-up for median (range) 9.2 (0.2–34) years, 9 patients had a liver transplant, 2 of whom subsequently died. 23 other patients died from liver disease without transplant and 1 from probable immunosuppressive drug toxicity. Reasons for not transplanting were: advanced age+cormorbidity (n = 9, age 72–79 years), major comorbidity (n = 8), advanced hepatoma (n = 2), cause unclear (n = 4). 35 other patients died, from non-liver related causes. 145 of the remaining 154 patients were alive without transplant on 31/12/05 and 9 were lost to follow-up. Overall SMR (n = 222), considering all-cause deaths and liver transplants was 1.88 (95% CI 1.42 to 2.33). SMR was similar: 1.87 (1.31 to 2.44) in patients presenting since 1987 (n = 177, presumed consecutive series). Overall SMR (n = 222) was not much reduced: 1.68 (1.22 to 2.13) when only actual deaths were considered (transplant not considered as death). SMR for non-liver-related death alone was close to unity: 0.97 (0.51 to 1.43); thus, most of the excess mortality is due to liver disease. The excess of all-cause deaths (including transplants) (A): persisted after the first decade of follow-up: SMR 1.92 (1.32 to 2.53) and 1.75 (1.02 to 2.49) for first and second decade respectively (B): was higher in patients presenting at a younger age: SMR 9.79 (3.9 to 15.7) if <45 years, 2.00 (1.24 to 2.75) if 45–65 years and 1.38 (0.87 to 1.89) if >65 years at presentation and (C): whilst higher in patients (n = 76) with cirrhosis at presentation (SMR 2.51 (1.62 to 3.40), was also present in patients (n = 145) without (SMR 1.53 (1.02 to 2.04).

Conclusion: Despite a good initial response to immunosuppression and despite the potential availability of liver transplantation, mortality in patients with AIH is higher than in the general population. This excess is due to liver disease and is relatively greater in those presenting at a younger age.

Interactive case presentations

AN UNUSUAL CAUSE OF RECURRENT CHEST PAIN

J. R. A. Skipworth1, A. Humphries2, B. Patel1, F. Hughes1. 1Department of Upper GI Surgery, Royal London Hospital, 2Research Fellow, Cancer Research UK, London, UK

Introduction: A 55-year-old man presented to the casualty department, with chest pain and vomiting. 8 months after this episode, he represented with similar symptoms. On both occasions, he was critically ill with clinical signs of septic shock. Further investigation revealed raised inflammatory markers, haemoconcentration, raised serum urea and positive chest radiograph findings (pneumothorax, pleural effusion, pneumomediastinum). We describe the strategies utilised in the multidisciplinary management of this patient’s recurrent condition.

Aims & Methods: Prospective assessment and collection of data, during the patient’s in-hospital stay, was undertaken, and combined with retrospective analysis of the patient’s notes, laboratory results and radiological images. A summary of the case is presented, together with a discussion of the literature.

Results: A combination of clinical, radiological, endoscopic and operative techniques were required to diagnose and manage recurrent, spontaneous oesophageal rupture and septic shock in this patient. During both admissions, the patient was treated operatively with direct closure of the oesophageal defect and conservation of the oesophagus, together with the insertion of mediastinal drains, drainage gastrostomy and feeding jejunostomy. Prolonged intensive care admission for cardiorespiratory organ support was crucial to ensure a successful outcome.

Conclusion: A multidisciplinary approach combining intensive care, radiological, endoscopic and surgical strategies is required in patients with recurrent Boerhaave’s syndrome, and patients with severe sepsis following late presentation. Literature searches reveal few detailed series of the diagnosis and management of Boerhaave’s syndrome, and only 3 descriptions of recurrence. Boerhaave described a transmural tear of the oesophageal wall, secondary to excessive vomiting, and occurring in association with excessive food and alcohol. However, clinical presentation can be misleading, and a high index of suspicion, combined with accurate radiological evaluation, is necessary to achieve a successful outcome. Swift diagnosis is crucial as the time elapsed between onset of symptoms and initiation of treatment correlates well with prognosis and hospital stay. Complications include mediastinitis, septic shock and death in up to 70%; and therefore patients commonly require prolonged intensive care admission and organ support. Management approaches include conservative, endoscopic and operative strategies, depending upon the patient’s condition, timing of presentation, and the site and size of the tear. Literature reviews of operative techniques that conserve the native oesophagus are usually associated with reduced morbidity and mortality, as compared to oesophageal replacement or resection approaches.

PRIMARY PANCREATIC LYMPHOMA: A DIAGNOSIS TO REMEMBER

K. E. Evans1, A. S. Mee2. 1Gastroenterolgy, Wycombe Hospital, High Wycombe, 2Royal Berkshire Hospital, Reading, UK

Introduction: This is a series of three cases of primary pancreatic lymphoma (PPL). PPL shares similar clinical and radiological features with the more common pancreatic adenocarcinoma. This series highlights the need to distinguish between the two because PPL is amenable to non-surgical treatment and carries a much better prognosis.

Aims & Methods: Case 1: A 51-year-old man presented with abdominal pain, fevers and obstructive jaundice. Ultrasound of the abdomen revealed pancreatic head mass. ERCP demonstrated a 10 mm stenosis in the pancreatic head. CT abdomen showed multiple cystic lesions within the pancreas, and multiple, large lymphnodes. Biopsy confirmed diffuse large B cell non-Hodgkin’s lymphoma. Case 2: A 74-year-old man presented with epigastric pain, weight loss, obstructive jaundice and a right upper quadrant mass. CT demonstrated a large pancreatic mass with large volume lymphadenopathy. ERCP demonstrated a distorted duodenum and a tight stricture in the common bile duct. The tumour and lymphnodes were PET negative. He then developed melaena and biopsy of the responsible gastric ulcer demonstrated a high grade non-Hodgkin’s lymphoma. Case 3: A 62-year-old man presented with symptoms of pancreatic insufficiency. This was confirmed by a pancreolauryl test. ERCP demonstrated irregular ducts in the pancreatic body, and common bile duct narrowing. CT abdomen showed a large soft tissue mass in the mesentery. A CT guided biopsy of the mesenteric mass confirmed a follicular B cell non-Hodgkin’s lymphoma.

Results: Two patients are in remission following chemotherapy. The third died of multiple complications within three months of diagnosis.

Conclusion: Primary pancreatic lymphoma is a rare extra-nodal lymphoma accounting for less than 0.5% of pancreatic tumours.1 As demonstrated here PPL often clinically and radiologically mimics the more common pancreatic adenocarcinoma. Nearly all PPLs are B-cell tumours.2 Most are treated with variants of CHOP (cyclophosphamide, doxorubicin, vincristine, prednisolone) chemotherapy. Tumour burden and LDH levels are the most important prognostic factors affecting outcome in primary gastrointestinal NHL.3 Multimodality treatment can give cure rates of up to 30% of PPL. This contrasts with the 5% 5-year survival for pancreatic adenocarcinoma.

BE AWARE OF THIS RISK WITH FEEDING

J. Mannath, I. Mohammed. Gastroenterology, Sandwell General Hospital, West Bromwich, UK

Introduction: Case report. A 57-year-old gentleman was recently admitted to gastroenterology ward with persistent vomiting and possibility of aspiration. He had an extradural haematoma in 2002 following trauma and underwent craniotomy. He was admitted under the gastroenterology team in 2003 with poor oral intake and was assessed by the SALT team. His swallow was poor and he was NG fed for 8 weeks before a PEG insertion was done. He was transferred to a nursing home after establishing PEG feed. In 2006 he was re-admitted with a PEG site infection and blocked PEG. He had a buried bumper for which he underwent laparotomy and removal. During laparotomy a gastric feeding tube was inserted. However he had persistent vomiting post gastrostomy insertion and a contrast examination through the PEG was done.

Aims & Methods: There were no signs of obstruction or reflux in the pegogram to account for persistent vomiting. However efforts to sustain his feed were futile due to vomiting. A jejunal extension tube (PEG-J) was thought to be the way forward and this was performed in July 2006 (Freka, ch-9). He settled well after this and his feed was gradually increased to 75 ml/h and he was discharged to nursing home. He tolerated the feeds well over the next 9 months and re-presented in April 2007. He had persistent vomiting and symptoms of aspiration. In view of his persistent vomiting a further pegogram and jejogram was done

Results: This showed that the jejunal extension tube has migrated to the oesophagus. The tube was probably displaced as a result of vomiting, causing further regurgitation and aspiration during feeding. The jejunal extension tube was removed and he was started on PEG feeds. He tolerated this well and was discharged with a plan to observe him and if he continues vomiting, a direct percutaneous endoscopic jejunostomy (DPEJ) or surgical jejunostomy will be considered.

Conclusion: Technical success of initial feeding tube insertion into the jejunum (PEG-J) can be achieved, but the function of these systems is often unsatisfactory due to frequent retrograde J-tube migration and dysfunction due to kinking and clogging. J-tube dysfunction is reported in 27–84% of cases and the average longevity of the jejunal extension tube is 3–6 months. DPEJs may be more dependable than PEG-Js and are being performed with increasing frequency, however some studies have shown a failure rate of around 30% due to lack of transillumination.

LATERAL THINKING AND GASTROINTESTINAL BLEEDING

E. Wesley1, R. Ali1, R. Ramnarace1, M. A. Puckett2, J. R. Lowes1. 1Gastroenterology, 2Radiology, Torbay Hospital, Torquay, UK

Introduction: Gastrointestinal bleeding is a common problem seen by gastroenterologists. We present a rare cause of this and discuss the possible causes and controversial treatment options available.

Aims & Methods: An 80-year-old man presented to our Accident and Emergency department with a short history of fresh haematemesis. He had not passed melaena and had no abdominal pain. Past medical history included atrial fibrillation, angina and deep vein thrombosis for which he was taking warfarin. There was no history of use of non-steroidal anti-inflammatory drugs, no history of excess alcohol consumption and no previous history of upper GI bleeding. On initial presentation he was shocked with a blood pressure of 90/70 mmHg and a pulse rate of 120. He had no stigmata of chronic liver disease and physical examination was otherwise unremarkable. His initial Rockall Score was 4. Initial blood tests showed a haemoglobin of 11.8 g/dl, platelet count of 302×109/l and INR of 2.9. His urea was normal at 6.4 mmol/l. He was resuscitated with intravenous fluids including blood and his coagulopathy was reversed with vitamin K and fresh frozen plasma. A gastroscopy was arranged and took place within six hours of admission. His stomach contained a large residue of altered and fresh blood and in the antrum an extensive adherent clot was noted. Adrenaline was injected around the edge of this and the heater probe used. Initially, this stopped the bleeding but the clot could not be removed to inspect underneath. A repeat gastroscopy and CT scan were arranged for the following day. Prior to this he received a pro-kinetic agent to improve mucosal views.

Results: The endoscopy and CT scan both showed a choledochogastric fistula.

Conclusion: This diagnosis is extremely rare. More common is fistulation between the gallbladder or bile ducts and other sites in the gastrointestinal tract. The main types of fistulae are cholecystoduodenal, cholecystocolonic and choledochoduodenal. Complications and therefore methods of presentation include gastrointestinal haemorrhage as well as gallstone ileus, ascending cholangitis and severe diarrhoea. Treatment of enterobiliary fistulae is controversial and often complex. Due to the extremely rare nature of the diagnosis, the area is lacking any firm evidence from randomised controlled trials. Some have advocated surgery in all patients because of the risk of carcinoma of the gallbladder which can be missed without surgery. However the mortality from such procedures is high. Our patient was treated conservatively. His bleeding stopped with the initial endoscopic therapy and correction of his coagulopathy. Given his age and comorbidity it was felt that surgical options to correct the fistula did not warrant the high risk of complications. He has so far not been readmitted and was asymptomatic at follow-up.

PYREXIA OF UNKNOWN ORIGIN: A DIAGNOSTIC CHALLENGE

A. Anand, R. Parmeshwar, A. K. Bohra. General Surgery, Royal Albert Edward Infirmary, Wigan, UK

Introduction: Pyrexia of unknown origin is an important clinical entity. It remains a diagnostic challenge because of the diversity and complexity of the clinical conditions it can represent.

Aims & Methods: We report a case of pyrexia of unknown origin where the lack of specific symptoms and signs and the absence of positive radiological findings presented a significant diagnostic challenge.

Results: A 45-year-old builder was admitted with a 3-week history of decreased appetite, fever and weight loss. The only positive examination findings were tachycardia and persistent pyrexia. Blood tests were unremarkable except for a low haemoglobin and albumin. Chest x ray and ECG were normal. An extensive infection screen and an autoimmune screen were negative. However, the patient developed diarrhoea and his general condition continued to deteriorate. A CT scan of the thorax, abdomen and pelvis was unremarkable. A bone marrow biopsy indicated depleted iron stores hinting at a gastrointestinal cause. However, before a gastroscopy and colonoscopy could be performed, he developed severe abdominal pain. On examination, he was shocked and his abdomen was peritonitic. He underwent an urgent laparotomy, which showed extensive small bowel thickening and patchy necrosis leading to perforation. He also had a partially obstructing lesion in the distal transverse colon. A resection-anastomosis of the worst affected segment of the small bowel was performed with fashioning of a loop colostomy. Although our patient went on to receive the recommended treatment for his disease, he developed neurological complications and succumbed to the disease barely five months after the onset of symptoms.

Conclusion: Our case illustrates the difficulties in the diagnosis and management of this clinical entity. We conclude that a high index of suspicion and appropriate early investigations are essential to improve the prognosis, which in general continues to be poor.

THE CONFUSING ALCOHOLIC

D. G. Craig1, I. Curzon2, K. Prasad3, N. Quinn4, J. Hancock1. 1Gastroenterology, 2Radiology, 3Histopathology, 4Endocrinology, Friarage Hospital, Northallerton, UK

Introduction: We report a rapidly fatal case of acute pancreatitis associated with severe hypertriglyceridaemia in an alcoholic admitted with newly diagnosed diabetic ketoacidosis. Serum amylase may be unhelpful in such cases and alternative diagnostic tests are required. The role of plasma exchange in hyperlipidaemic pancreatitis is discussed.

Aims & Methods: A 32-year-old Caucasian alcoholic was admitted with a 2-day history of progressive polydipsia and confusion. A diagnosis of diabetic ketoacidosis (DKA) was made after finding massive glycosuria and ketonuria along with a profound metabolic acidosis. Serum amylase was mildly elevated at 185 U/l. Liver function tests showed preserved synthetic function and serum lactate was normal. He was noted to have a milky, lipaemic serum with a cholesterol of 22.5 mmol/l and a hypertriglyceridaemia of 71.21 mmol/l. CT of chest and abdomen demonstrated marked peripheral infiltration of both lung bases but shortly afterwards the patient had a cardiorespiratory arrest culminating in his death 12 hours after admission. Postmortem examination revealed acute pancreatitis (AP) with fat necrosis.

Results: Coexistent diagnoses of DKA and AP present the physician with a complex diagnostic problem. The presence of abdominal pain is non-specific1 and the coexistence of AP with DKA is associated with more severe metabolic acidosis and hyperglycaemia, and a worse prognosis, than cases of isolated DKA. Hypertriglyceridaemia is a recognised consequence of DKA due to lipolysis of adipose tissue and severe hypertriglyceridaemia (>55 mmol/l) can cause chemical irritation around pancreatic acinar cells. Non-specific hyperamylasaemia is frequent in DKA while patients with hyperlipidaemic pancreatitis may have low initial levels of serum amylase2 and this provides a further diagnostic hurdle. Urinary trypsinogen-2 levels have high negative predictive values and may provide a more sensitive marker in hyperlipidaemic pancreatitis than serum amylase.3 Plasma exchange has been used successfully in hyperlipidaemic pancreatitis but its effectiveness has recently been called into question.4

Conclusion: This case demonstrates the continued diagnostic difficulties presented by patients with concurrent DKA and AP. Early recognition of the possibility of AP in patients with lactescent serum may allow earlier institution of intensive care input and radiological diagnosis.

SHORTNESS OF BREATH IN CROHN’S DISEASE

P. Ellul1, J. Mann2, E. T. Swarbrick1, D. C. Rowlands3, I. Morgan4. 1Gastroenterology, 2Respiratory, 3Histopathology, 4Cardiothoracic Unit, New Cross Hospital, Wolverhampton, UK

Introduction: A case of shortness of breath in Crohn’s disease.

Aims & Methods: To discuss the differential diagnosis and management of such pathologies.

Results: Our presentation is about an interesting case of a young man with Crohn’s disease, diagnosed 9 years ago who presented with weight loss, anorexia, nasal blockage, chest tightness and shortness of breath on exertion. He was having balsalazide and azathioprine as to keep his disease stable, although he had chronic perianal disease. In our case we first go through the investigations, differential diagnosis and treatment of his current medical pathologies. We demonstrate some interesting images of his chest x ray, computed tomography scan of his thorax and of histology samples. Then, we discuss the respiratory syndromes associated with IBD as well as the drugs used in the management of IBD. Finally we discuss the management of such challenging coexistant pathologies.

Conclusion: We discuss in detail, with the aid of images, the above differential diagnosis and the challenges in management of such pathologies.

AN UNUSUAL CAUSE OF GASTROINTESTINAL HAEMORRHAGE

C. E. Kober1, M. C. Gallagher1, S. de Sanctis2, C. Topham3, C. J. Tibbs1. 1Department of Gastroenterology, 2Department of Histopathology, 3Department of Haematology, Royal Surrey County Hospital NHS Trust, Guildford, UK

Introduction: A 59-year-old man presented with a 4-day history of malaena. His medical history revealed that he had undergone an anterior resection for colorectal carcinoma 36 months earlier. Postoperatively he received 12 cycles of adjuvant chemotherapy with Oxaliplatin and 5 Fluorouracil.

Aims & Methods: He had no prior history of gastrointestinal haemorrhage or bleeding diathesis, no known risk factors for chronic liver disease, and a modest alcohol intake (16 units per week).

Results: Laboratory tests revealed anaemia and thrombocytopenia (Hb 6.7, platelets 91), with normal renal function and liver biochemistry. Following resuscitation an upper gastrointestinal endoscopy was arranged. Multiple oesophageal varices with stigmata of recent haemorrhage were identified, and endoscopic band ligation performed. Ultrasound and CT examinations revealed splenomegaly, a structurally normal appearing liver and patent portal vessels. There was no evidence of recurrent colorectal carcinoma. Laboratory investigations for causes of chronic liver disease were negative (normal immunological profile and iron studies, and negative viral serology). Liver histology revealed features of hepatic venous outflow obstruction (veno-occlusive disease). Haematological studies for prothrombotic states were negative, and a bone marrow showed no evidence of a myeloproliferative disorder.

Conclusion: Although liver fibrosis and venocclusive changes have been reported in patients treated with oxaliplatin, this is the first report of a case of variceal haemorrhage secondary to oxaliplatin-induced liver injury.

A RARE COMPLICATION OF LIVER BIOPSY

M. Wireko1, S. Bridger2, C. Hovell2, R. Ward3. 1Gastroenterology, Poole Hospital NHS Trust, Poole, 2Gastroenterology, 3Radiology, Dorset County Hospital Foundation Trust, Dorchester, UK

Introduction: Case report. A 66-year-old man was admitted under the surgeons with abdominal pain and haematochezia. 7 days previously he had been diagnosed with autoimmune hepatitis with a pre-treatment score of 14.

Aims & Methods: His liver function tests (LFTs) were as follows: ALT 2442 IU/l,Alkaline Phosphatase 284 IU/l, GGT 195 IU/l, Bilirubin 383 μmol/l, Albumin 36 g/l and raised IgG of 24.4. His ultrasound guided liver biopsy was consistent with this diagnosis. He had been started on prednisolone 40 mg daily with rapid improvement in his LFT.

Results: On this admission, he was noted to be hypotensive, dropped his haemoglobin by 2 g/dl. Flexible sigmoidoscopy was negative. Gastroscopy revealed blood in the second part of the duodenum and a CT scan revealed an interesting abnormality. He responded successfully to a specific treatment and has now made a full recovery.

Conclusion: The symptomatic triad of gastrointestinal bleeding, biliary pain, and jaundice is characteristic of haemobilia and tends to occur about 5 days after liver biopsy. This complication is very rare occurring in only 4 out of 68 276 (0.006%) percutaneous liver biopsies.1 Most cases of haemobilia settles spontaneously, however significant cases causing symptoms will need embolisation. If the latter is unsuccessful, then surgical intervention would be required. This case highlights the need for vigilance and close follow-up of patients for rare but potentially serious complications post liver biopsy.

Plenary posters

MUCOSALLY-ADHERENT E COLI ISOLATES FROM HUMAN COLON CANCER INDUCE NUCLEAR TRANSLOCATION OF BETA-CATENIN AND CYCLIN D1 EXPRESSION IN HUMAN COLON CANCER EPITHELIAL CELLS

P. D. Collins, L. Yu, J. M. Rhodes. School of Clinical Science, University of Liverpool, Liverpool, UK

Introduction: Dysregulation of Wntβ-catenin signaling, nuclear translocation of β-catenin and subsequent T-cell factor/lymphoid-enhancing factor-mediated transcription are early events in colon cancer development.1 The Wnt target gene, Cyclin D1, is a major cell-cycle regulator. Non-virulent bacteria can influence β-catenin signaling in human epithelia.2 Mucosally-adherent E coli are found in increased numbers in colonic mucosa in patients with colon cancer.3 Bacterial involvement in cancer progression could explain the much higher rate of cancer in the colon than the small intestine.

Aims & Methods: DLD-1 cells, a colon cancer cell line with biallelic inactivation of APC gene, were cultured with E coli mucosal isolates from patients with colon cancer, isolates from control patients, heat-killed bacteria or bacterial supernatant. Cells were incubated with Prostaglandin-E2 (0.1 μM) as a positive control. Cellular localisation of β-catenin was assessed by immunofluorescence microscopy. Nuclear localisation of β-catenin was quantified and expressed as a ratio of the intensity of nuclear to whole cell fluorescence for over 100 cells from each blind-labelled sample. Expression of β-catenin and Cyclin D1 in whole cell lysates was analysed by western blotting.

Results: Colon cancer E coli isolate HM545 increased nuclear localisation of β-catenin (mean ratio of nuclear to whole cell fluorescence (N:WC)  = 1.064; SD 0.406) compared to the negative control (no bacteria N:WC = 0.982; SD 0.087) (p = 0.017). Neither HM545 supernatant (N:WC = 0.983; SD 0.085) (p = 0.90) nor heat-killed HM545 (N:WC = 0.997; SD 0.117) (p = 0.25) increased nuclear localisation. Two E coli isolates from patients with normal colons did not induce nuclear localisation of β-catenin (HM452; N:WC = 0.983; SD 0.115 (p = 0.95) and HM455; N:WC = 0.973; SD 0.083 (p = 0.42)). Six of nine mucosal E coli isolates from colon cancer induced nuclear translocation. Western blotting showed that Cyclin D1 expression is increased in response to three of five E coli isolates (including HM545), but not in response to HM545 supernatant or heat-killed bacteria. The stimulatory effect of E coli HM545 on Cyclin D1 was inhibited when cells were pre-incubated with the cox-2 inhibitor NS398 (10 μM) or soluble plantain fibre (5 mg/ml), an inhibitor of E coli epithelial cell adhesion. Neither bacterial haemagglutination status nor phylotype corresponded with Cyclin D1 expression or β-catenin nuclear localisation. Bacterial isolates had no effect on total levels of β-catenin in whole cell lysates.

Conclusion: Mucosal E coli isolates from colon cancer induce nuclear localisation of β-catenin and activation of Cyclin D1 in human colon cancer cells. This effect requires whole, live bacteria. This is a plausible mechanism by which mucosa-associated bacteria may promote the development of colon cancer.

THE MISUSE OF THE FAECAL OCCULT BLOOD TEST UNDER THE LOWER GASTROINTESTINAL TWO-WEEK WAIT RULE

A. Shaw1, C. Longman2, J. N. Lund3, A. F. Goddard4. 1Gastrointestinal Surgery, University of Nottingham, 2Colorectal Surgery, Derby City General Hospital, 3Gastrointestinal Surgery, University of Nottingham, 4Gastroenterology, Derby City General Hospital, Derby, UK

Introduction: In July 2000 the two-week wait (2WW) referral system was introduced to reduce the delay in referral, diagnosis and subsequent treatment of patients with colorectal cancer. General practitioners are required to complete a proforma and state the necessary referral criteria for each patient. It is not part of the criteria for patients to have a FOB performed before referral. The NHS Bowel Cancer Screening Programme was introduced in England in July 2006 using the faecal occult blood (FOB) test. Screening demands a total of 6 samples be tested with 5 or 6 positive samples classified as a positive result.

Aims & Methods: To find the proportion of patients with an FOB performed as part of the referral for the lower gastrointestinal 2WW referral system, and whether this correlates with the cancer status. The last two thousand patients (from 01/08/05–31/07/07) referred as a 2WW were identified using the hospital’s cancer audit database. Performance of FOB and cancer status were recorded for each patient.

Abstract 119 Table comparing faecal occult blood to cancer status

Results: A total of 2000 patients (F:M 1097:2000; median age 58; range 18–98) were referred by general practitioners. The FOB test was only performed on three samples in all cases. 1 positive (or more) FOB was classified as a positive result. In total, 169/2000 patients (8.5%) had an FOB performed prior to their referral, with 55/169 patients (32.5%) as part of the referral for 2WW. 16/169 patients (9.5%) had an FOB performed in the presence of overt rectal bleeding. In only 2/169 patients (0.01%) did the FOB correlate with a colorectal cancer (one of these had overt rectal bleeding). Unnecessary testing for FOB costs has cost £4001.92 in 2 years.

Conclusion: There is significant misuse of FOB testing. The detection rate, even in this symptomatic group of patients, is very low. FOBs are being performed in the community on three samples and not six, and those with one or two positive samples are being referred rather than being recalled for further testing, contrary to protocol. This, combined with the high false positive rate, leads to patients not only experiencing psycho-social consequences but could potentially lead to significant risks from unnecessary invasive investigation as well as the added financial burden of the test itself.

DOUBLE BALLOON ENTEROSCOPY IS HIGHLY EFFECTIVE: THE FIRST 100 CASES AT ST MARK’S HOSPITAL

E. Despott1, E. Tripoli1, A. Polecina1, K. Konieczko2, C. Fraser1. 1Wolfson Unit for Endoscopy, St Mark’s Hospital, 2Department of Anaesthesia, Northwick Park and St Mark’s Hospitals, London, UK

Introduction: The investigation and management of suspected small bowel pathology has been revolutionised by the introduction of double balloon enteroscopy (DBE). DBE is unique in that it provides the facility to biopsy lesions and to apply therapeutic procedures such as the coagulation of bleeding vessels and balloon dilatation of strictures, potentially averting the need for surgery.

Aims & Methods: We report our experience of DBE since its introduction to our unit in 2005. A comprehensive DBE database was collated and analysed for indications, concordance of capsule endoscopy (CE) and DBE findings, therapy applied at DBE, sedation practices and complications.

Abstract 120

Results: 100 cases (56 males) were performed. Mean age was 58.8 years (16–92 years). 81 cases were performed via the oral route, 14 via the rectal route 4 cases via both the oral and rectal routes and 1 case through an ileostomy; 49 cases were done under general anaesthesia (mean age 49.1 years); 51 cases had conscious sedation (mean age 67.5 years). The indications and therapeutic interventions are shown (table). DBE was preceded by CE in 82 cases, the concordance of diagnoses at CE and DBE was 71%. Where there was divergence in the CE and DBE findings, abnormalities seen at CE were not confirmed at DBE in 15 cases. Conversely, DBE showed lesions that were not picked up by CE in 7 cases. Lesions missed at CE included small bowel ulceration, a GIST and a case of strongyloidiasis. The findings at DBE led to successful surgical management in 9 cases (small bowel tumours and a large vascular lesion). In cases referred for stricture dilatation, surgery was averted in 6 of 10 cases. In our cohort of patients the overall diagnostic and therapeutic yields (on an intention to treat basis) of DBE were 76% and 77% respectively. Limiting factors included bowel fixation by adhesions, scope looping, obesity and patient discomfort despite sedation. There were no clinically important complications.

Conclusion: DBE is a feasible, safe and reliable technique for the diagnosis and treatment of suspected small bowel pathology. Importantly, we confirm that in selected cases it offers an alternative to surgery. Our experience compares favourably to other results published in the literature to date.

DEFINING IN VIVO MUCOSAL INFLAMMATORY ACTIVITY AND DISEASE EXTENT IN ULCERATIVE COLITIS: DEVELOPING A CLINICALLY APPLICABLE MODEL USING LASER SCANNING CONFOCAL ENDOMICROSCOPY

D. P. Hurlstone1, N. Tiffin2, S. S. Cross3. 1Endoscopy, Royal Hallamshire Hospital, 2Histopathology, STH NHS Foundation Trust, 3Histopathology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Colonoscopy with mucosal biopsy is currently considered the “gold standard” for the assessment of disease activity and extent assessment in ulcerative colitis (UC). High-magnification Saitoh criteria are superior to Baron indices for all histopathological stages of disease but is complicated by complex classifications and significant inter/intra-observer variability. Using confocal laser scanning endomicroscopy we propose a novel, simple in vivo optical “biopsy” classification using combined surface cellular resolution and deep vascular architecture crypt imaging.

Aims & Methods: To evaluate the efficacy of live cellular endomicroscopic imaging for the in vivo diagnosis of disease extent and activity in UC. Consecutive patients with UC were recruited to ascertain current mucosal activity parameters and extent. Disease activity and extent was documented using conventional endoscopy (Baron indices), Sheffield endomicroscopy criteria with mucosal biopsy as the assumed “gold standard” from all 5 colonic segments and the rectum. Endomicroscopy was performed using topical acroflavin for surface mucosal nuclear imaging in addition to deep z-axis vascular imaging following 5 ml of iv. 10% sodium fluorescene. Endomicroscopy was performed at 10 mm quadrantic segmental intervals. Confocal greyscale macro data were collected at 1.6 frames/sec/optical z-axis sectioning 0.7 μm—field view 500 μm×500 μm, minimum 12 macros per segment. The in vivo endomicroscopic diagnosis (according to Sheffield criteria) was then compared to targeted biopsies equating to the raster scanned mucosal segment.

Results: 48 patients completed protocol. A total of 28-targeted biopsies were taken from each patient (in a segmental distribution) exacting to a 12 “slice” z-axis segmental endomicroscopy-scanning plane. A total of 1344 biopsies were subsequently compared to 18 816 endomicroscopy images graded according to Sheffield criteria. The K coefficient of agreement between the endomicroscopy indices (grade 0–2) and Matts grade 1/2, 3a–4 and 5 were 0.94, 0.82 and 0.79 respectively. Histopathological grade of disease was represented more accurately using endomicroscopy criteria as compared to conventional Baron scores for all clinical parameters (r = 0.91; p<0.01). Endomicroscopy was significantly better than conventional colonoscopy for predicting in vivo disease extent (p<0.001).

Conclusion: Endomicroscopy permits live cellular, sub-cellular and deep vascular imaging in UC which when interpreted using Sheffield criteria provide a high accuracy in vivo “virtual biopsy” and hence provide an instant biomarker for disease relapse. High-accuracy virtual biopsy can limit the number of biopsies required for the procedure with significant cost savings to pathology services.

“SALVAGE” ENDOSCOPIC SUBMUCOSAL DISSECTION AS AN ALTERNATIVE TO SURGICAL RESECTION FOR RESIDUAL OR LOCAL RECURRENT INTRAEPITHELIAL NEOPLASIA IN THE COLORECTUM: A PROSPECTIVE ANALYSIS

D. P. Hurlstone1, S. R. Brown2, W. Baraza2, N. Tiffin3, S. S. Cross4. 1Endoscopy, Royal Hallamshire Hospital, 2Colorectal Surgery, 3Histopathology, STH NHS Foundation Trust, 4Histopathology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Curative intent for residual or locally recurrent colorectal neoplasia following EMR is often not technically feasible due to submucosal fibrosis. Such cases often demonstrate false positive submucosal invasion status at high-frequency 20/12.5 MHz EUS and are associated with a conventional non-lifting sign at submucosal catheterisation. Primary ablative therapies preclude therapeutic efficacy confirmation histologically in this group. Endoscopic submucosal dissection may permit curative intent “salvage” therapy without the requirement for formal surgical resection as the technique permits endoluminal transection of fibrotic tissue.

Aims & Methods: A prospective technical feasibility study of cap assisted ESD for “curative intent” in patients with residual or local neoplastic recurrence following EMR. Primary end points were second stage R0 resection rate, safety and recurrence. Salvage endoscopic submucosal dissection was performed using the Olympus GIF-XQ240 gastroscope and KD-630L insulation tipped knife. Submucosal injection solute comprised sodium hyaluronic acid (10 ml) mixed with 1 ml 1/10 000 adrenaline and 1 ml of 0.5% indigo carmine solution for enhanced delineation of the muscularis at endoscopic dissection. 30-day mortality, re-admission rates, complications and histological resection status were collected prospectively up to 9 months following index resection.

Results: 30 patients met eligibility criteria. 29/30 (96%) demonstrated non-lifting or asymmetric lifting at index submucosal catheterisation. Index R0 resection was achieved in 25/30 (83%) lesions. 1 patient underwent surgical excision with a second receiving a curative second stage dissection. 96% (29/30) patients were discharged within 24 hours of the procedure with a 0% 30-day mortality and re-admission rate. Bleeding occurred in 5/30 (16%) treated successfully with endoluminal haemostasis. There were no perforations. Overall “cure” rates at short-term follow-up (median 6/12 (range; 3–18)) was 96%.

Conclusion: This novel application of endoscopic submucosal dissection for first line “salvage” therapy in treating residual or locally recurrent neoplastic disease may be a safe, minimally invasive and cost effective alternative to direct surgical resection in a select patient cohort.

PILLCAM COLON CAPSULE ENDOSCOPY COMPARED TO COLONOSCOPY IN DETECTION OF COLON POLYPS AND CANCERS. INTERIM ANALYSIS OF A PROSPECTIVE MULTICENTRE TRIAL

A. J. Postgate1, C. Fraser1, A. Fitzpatrick1, M. Munoz-Navas2, I. Fernandez-Urien2, M. Lapalus3, T. Ponchon3, G. Gay4, M. Delvaux4, H. Neuhaus5, M. Philliper5, G. Costamagna6, M. Riccioni6, C. Spada6, F. Haegenmuller7, M. Keuchel7, N. Schoofs8, A. Van Gossum8, J. Deviere8. 1Wolfson Unit for Endoscopy, St Mark’s Hospital, London, UK, 2Digestive Diseases Division, University of Navarra, Pamplona, Spain, 3Department of Digestive Diseases, Hôpital Edouard Herriot, Lyon, 4Department of Internal Medicine and Digestive Diseases, Hôpitaux de Brabois, Vandoeuvre les Nancy, France, 5Department of Gastroenterology, Evangelisches Krankenhaus, Düsseldorf, Germany, 6Digestive Endoscopy Unit, Catholic University, Rome, Italy, 7Department of Gastroenterology, Asklepios Klinik Altona, Hamburg, Germany, 8Department of Gastroenterology, Erasme University Hospital, Brussels, Belgium

Introduction: In pilot studies, PillCam colon capsule endoscopy (PCE) has been shown both feasible and able to demonstrate colonic polyps and cancers. An 8-centre prospective study with a target of 329 patients is ongoing in Europe. The primary objective is to assess the yield of PCE in detecting polyps and other pathologies in comparison to traditional colonoscopy. This study reports the first planned interim analysis.

Aims & Methods: Patients included in this study were known to have (age >18 years) or suspected of having (age >50 years) colonic disease and were referred for colonoscopy. Patients went through a traditional PEG colon prep and ingested the capsule in the morning. The procedure included prokinetic agents and additional small doses of laxatives. Independent physicians performed capsule imaging analysis and traditional colonoscopy (gold standard) after excretion. Significant findings were defined as at least one polyp ⩾6 mm or ⩾3 polyps of any size.

Abstract 123 Accuracy of Pillcam Colon capsule endoscopy for polyp detection

Results: One hundred and sixty six patients, mean age 60 (23–84) years, were included in this interim report. Capsules were expelled within 10 hours post-ingestion in 96% of patients, and remained in the stomach for the entire procedure in 2 (1.2%) patients who were excluded from the analysis. Quality of colon preparation in PCE was evaluated as excellent to good (68%), fair (24%) and poor (8%). A polyp of any size was found in 111 patients (68%) and 75 (46%) patients had significant findings detected by colonoscopy. The yield of PCE compared to colonoscopy is shown in the table. Fifteen polyps were detected only by PCE. No severe adverse events related to PCE were reported.

Conclusion: This interim analysis shows encouraging sensitivity and negative predictive values for the detection of significant findings compared to colonoscopy. If this data is further validated at the conclusion of this study, this new non-invasive technology might challenge colonoscopy for CRC screening and polyp detection.

ACCURATE QUANTIFICATION OF H PYLORI DENSITY IN HUMAN GASTRIC BIOPSIES USING REAL-TIME PCR

S. R. Patel, K. Robinson, D. P. Letley, J. C. Atherton. Gastroenterology, Wolfson Digestive Diseases Centre, Nottingham, UK

Introduction: The density of H pylori on gastric mucosa may be influenced by a number of factors, including the host immune response and virulence determinants expressed by the colonising strain. Current methods for assessing colonisation densities, however, are based on quantitative culture or histopathology scoring techniques, which have limited sensitivity and accuracy.

Aims & Methods: We aimed to develop a quantitative real-time PCR assay for measuring the colonisation density of H pylori in gastric biopsies. The H pylori 16S rRNA gene was selected as a target. Primers were designed and tested to ensure that they were specific for amplification of a product from diverse strains of H pylori. Standard curves were prepared using purified H pylori J99 genomic DNA. For greater accuracy, bacterial copy numbers were normalised against the human GAPDH housekeeping gene. The assay was validated by adding a known number of bacteria to a known number of AGS cells (a human gastric epithelial cell line), and also by spiking uninfected biopsy specimens with a known number of bacteria. The data were normalised against the copy number of human GAPDH, presenting the results as bacterial genome copy per human genome copy.

Results: Primers for the 16S rRNA gene were specific and could be used to amplify a product from 30 diverse H. pylori strains. Both bacterial and human standard curves demonstrated a high degree of fit (R2 = 0.99691 and 0.99582 respectively), and good reproducibility in the range of 20 ng/μl–10 fg/μl. Validation of the quantitative assay using known numbers of bacterial and AGS cells resulted in good accuracy with the ratio of number of bacteria/human cell detected almost exactly equal to the number added. We have examined antral and corpus biopsies from 10 patients and more are in progress. Mean ratios of bacteria to eukaryotic cells in the antrum were 0.3 (range 0.02–0.92), and in the corpus were 0.55 (range 0.009–3.78). There was no obvious association between bacterial numbers and neutrophil or lymphocyte scores. In preliminary results, all 3 patients with DU had an antral predominant colonisation with a mean antral: corpus bacterial ratio of 7.0. In contrast, 10 patients without ulcers had widely varying antral: corpus bacterial ratios.

Conclusion: The real-time PCR assay developed was highly sensitive and accurate for the quantification of H pylori. Preliminary results support antral predominant colonisation predisposing to DU. Further work is in progress.

UPREGULATION OF UROKINASE-TYPE PLASMINOGEN ACTIVATOR RECEPTOR IS ASSOCIATED WITH HELICOBACTER PYLORI IN PREMALIGNANT GASTRIC TISSUE

J. M. Mikhail1, P. Griffiths2, J. Manson3, L. Thomas4, L. Williams5, J. N. Baxter3, J. M. Parry1, G. J. S. Jenkins1. 1School of Medicine, Swansea University, 2Pathology, 3Surgery, 4Medicine, Swansea NHS Trust, Swansea, 5Medicine, Royal Glamorgan Hospital, Llantrisant, UK

Introduction: Helicobacter pylori (H pylori) associated gastritis, induces the production of reactive oxygen species (ROS) and has been associated with disease progression. ROS induced expression of IL-8, uPAR and PAI-1 has been described: NF-κB may be involved. It is unclear whether expression of these genes plays a role in gastric cancer progression.

Aims & Methods: We have investigated the expression of IL-8, uPAR and PAI-1 in gastric tumours and +/−H pylori infection in premalignant tissue to determine the stage in disease progression that changes occur. Tumours and adjacent normal mucosal tissue were obtained from surgical gastric resections. Gastric biopsies were isolated from consenting patients attending Endoscopy clinic. All samples were stored in RNA later, histology was confirmed and RNA extraction performed (Qiagen). Analysis of the gastric tumours using microarrays (Superarray) was followed by qPCR (Biorad). Premalignant biopsies were analysed using qPCR.

Results: In gastric tumours, IL-8 was upregulated >twofold in 17/19 (89.5%), PAI-1 was upregulated in 14/19 (74%) and urokinase-type plasminogen activator receptor (UPAR) was upregulated in 7/19 (37%), compared to the corresponding normal mucosal tissue. In the premalignant biopsy samples, IL-8 showed upregulation in chronic gastritis and gastritis associated with H pylori infection. Upregulation of uPAR was significant in samples with coexisting or previous H pylori infection. PAI-1 was not differentially expressed.

Conclusion: This study has shown upregulation of two ROS and NF-κB regulated genes, IL-8 and uPAR, in both gastric tumours and premalignant tissue. Further investigation, may show these genes to be markers of disease progression and NF-κB a potential therapeutic target.

THE COST-EFFECTIVENESS OF BIOLOGICAL THERAPIES FOR CROHN’S DISEASE: MARKOV COHORT ANALYSES INCORPORATING UK PATIENT-LEVEL COST DATA

K. Bodger1, T. Kikuchi2, D. A. Hughes2. 1Department of Medicine, Clinical Science Centre, University Hospital Aintree, Liverpool, 2Centre for Economics and Policy in Health, University of Bangor, Bangor, UK

Introduction: Anti-TNF-α agents for Crohn’s disease (CD) have good clinical efficacy but high acquisition cost compared to rival drugs. Previous modelling estimates of cost-effectiveness suggest an unacceptably high incremental cost per QALY compared to standard care. However, existing models were limited by a lack of primary data for costs of care and health state utility. The aim of the present study was to undertake an independent analysis incorporating data from recent trials and observational studies from the perspective of the UK National Health Service. The present analysis focuses on infliximab and adalimumab.

Aims & Methods: Lifetime Markov analyses constructed to simulate outcomes and costs. CD was represented by 5 disease states: Full Response (CDAI<150, remission); Partial Response (CDAI 150–220, mild-to-moderate activity); Non-Response (CDAI >220, moderate-to-severe activity); Surgery and Death. The course of CD under standard care was modelled from a transition matrix derived from a large population cohort.1 Systematic review identified ACCENT I (infliximab) and CHARM (adalimumab) as sources for efficacy data. We modelled an intention-to-treat strategy for biologics by including a non-responder cohort (representing patients excluded from the trials after failed induction). Surgical rates were based on observational data, cost estimates were taken from our UK dataset2 and utilities were derived for each state from an algorithm converting CDAI to EQ5D utilities.3

Abstract 126 Lifetime cost-effectiveness of biological treatments for Crohn’s disease

Results: In the base-case analysis (lifetime horizon; 1–2 years continuous therapy; discount rate 3.5%) both agents achieved acceptable incremental cost effectiveness ratios (ICERs) compared to standard care (table). Lifetime therapy was dominated by standard care.

Conclusion: We have taken advantage of recent studies to develop a model for estimating the cost-effectiveness of rival therapies for CD. Unlike earlier attempts, our model suggests acceptable ICERs for biological agents for periods of up to 2 years continuous therapy. Apparent differences between rival biological agents must be interpreted cautiously as head-to-head trial data are not available.

NICOTINE EXERTS ITS ANTI-INFLAMMATORY EFFECTS ON CELLS OF COLONIC PHENOTYPE VIA INHIBITION OF TRANSCRIPTION FACTOR, NF-κB

K. M. Khatab, W. Dhaliwal, D. S. Rampton. Centre for Gastroenterology, Barts and The London Queen Mary’s school of Medicine and Dentistry, London, UK

Introduction: Smoking is beneficial for ulcerative colitis, while the opposite is true for Crohn’s disease; the reason for this paradox is unknown. We have shown in previous work (Khatab, Gut 2005;54:A97) that nicotine significantly inhibits TNF-α-induced IL8 production by epithelial cells of colonic phenotype (HT29) while it has no effect on IL8 production by epithelial cells of small intestinal phenotype (Caco2). We have also shown that nicotine exerts its anti-inflammatory effects via α7-nicotinic acetyl choline receptors (nAChR) which are expressed by colonic epithelial cells (HT29) and myofibroblasts but not by those with small intestinal phenotype (Caco2) (Khatab, Gut 56(Suppl 2):A113–A114, April 2007). The exact intracellular signalling pathways involved in the anti-inflammatory potential of nicotine are not well established. Studies in different cell types including monocytes and macrophages have indicated that the anti-inflammatory potential of the alpha7-nAChR is mediated by the inhibition of the transcription factor, NF-κB.

Aims & Methods: We studied the hypothesis that the anti-inflammatory potential of nicotine by colonic intestinal cells is mediated by the inhibition of the transcription factor, NF-κB. Confluent Caco2, HT29 and myofibroblast cells were stimulated with TNF-α (50 ng/ml) or IL1 (10 ng/ml) in the absence or presence of nicotine (10−5 M). The nuclear extracts of these cells were collected after 24 hours. NF-κB (p65) levels were assayed using p65 NF-κB ELISA kit which detects the active form of NF-κB (p65).

Results: In all cell types used in these experiments; both TNF-α and IL1 stimulated p65 NF-κB. Nicotine had no effect on p65 NF-κB activation by Caco2 (fig 1) while it significantly inhibited p65 NF-κB activation by both HT29 (fig 2) and colonic myofibrolasts (not shown).

Abstract 127

Conclusion: TNF-α and IL1 up-regulated the activation of NF-κB in gut cell lines. Nicotine inhibits activation of NF-κB in cell lines of cells of colonic phenotype that express α7-nAChR phenotypes (HT29 and myofibroblasts) while it had no effect on cells of small intestinal phenotype (Caco2) which do not express this receptor.

SAFETY OF ADALIMUMAB MAINTENANCE THERAPY IN CROHN’S DISEASE PATIENTS: OPEN-LABEL EXTENSION OF CHARM AND GAIN TRIALS

J. F. Colombel1, P. Rutgeerts2, W. J. Sandborn3, R. Panaccione4, W. Lau5, K. G. Lomax5, P. F. Pollack5. 1Hepatogastroenterology, CHU Lille, Lille, France, 2Gastroenterology, University Hospital of Gasthuisberg, Leuven, Belgium, 3Gastroenterology, Mayo Clinic, Rochester, MN, USA, 4Medicine, Health Science Centre, Calgary, AB, Canada, 5Immunology Development, Abbott, Parsippany, NJ, USA

Introduction: Approved in the EU for the induction and maintenance of remission in adults with severe Crohn’s disease (CD), adalimumab (ADA) has demonstrated efficacy and safety in patients with and without prior anti-TNF treatment.13

Aims & Methods: We assessed the safety of ADA maintenance therapy in patients who completed either of two Phase III, double-blind, placebo-controlled, ADA safety and efficacy trials, GAIN (4-week induction in infliximab failures) and CHARM (1-year maintenance). Patients had active CD (CDAI 220–450) and were naïve to prior anti-TNF therapy (approximately 50% of 854 patients in CHARM) or had lost response to/were intolerant of prior infliximab therapy (GAIN). Safety was routinely assessed and data were collected throughout CHARM, GAIN, and an open-label extension (OLE) trial following patients through 2 years since the start of CHARM and 1 year since the start of GAIN. Patients randomised to placebo in GAIN who discontinued prior to Week 4 with no exposure to ADA were not included in the safety analysis. Adverse events (AEs) were tabulated in events/100-patient-years (E/100-PY).

Abstract 128 Adverse events of interest: GAIN/CHARM and OLE Combined (n = 1169; 1299 patient-years)

Results: In all, 1169 patients who received ADA in CHARM (n = 854) or in GAIN (n = 315) participated in the OLE. AEs of interest are presented in the table. Of the 28 opportunistic infectious AEs, 27 were non-systemic candidiasis, with 1 case of coccidiodomycosis. Of the 21 malignant AEs, there were 11 skin cancers and 1 lymphoma; the remainder varied in type.

Conclusion: Adalimumab was well-tolerated in this follow-up analysis of patients from CHARM and GAIN, with a safety profile consistent with those observed in previous ADA studies in CD and other indications.

AUDIT OF INFLAMMATORY BOWEL DISEASE CLINICAL RESEARCH ACTIVITY AND RESOURCES IN THE UNITED KINGDOM

B. A. Theis, S. L. Bloom. Department of Gastroenterology, University College London Hospitals NHS Foundation Trust, London, UK

Introduction: Results from the 2006 UK IBD Audit1 suggest that less than 1% of inpatients admitted with flares of IBD participated in clinical trials of treatments for their disease.

Aims & Methods: To try to determine the reasons behind such poor clinical trials recruitment an audit of current research activity, available research resources and perceived barriers to the successful conduct of clinical trials in patients with IBD in Gastroenterology units across the UK was undertaken. A questionnaire was sent via email to all members of the BSG IBD Section in September 2007. After two weeks a reminder was sent to non-responders.

Results: A total of 103 questionnaires from 99 units were returned: 85 (85.9%) were from England and 38 (38.8%) belonged to a teaching hospital. A median of 3 gastroenterologists with an “interest” in IBD (range 1–6) worked in the units. Only 7 (7.1%) units knew the size of the patient population they were treating, this despite 40 (40.4%) having a “regularly maintained database” of patients. The estimated median patient population for all respondents was 600 (range 50–3500). A specialist IBD Clinic was conducted in 66 (66.7%) units. Forty eight (48.5%) units were undertaking one or more clinical trials (median 2, range 1–5): in comparison units not currently doing so (over 60% of which reported past clinical trials experience) had a smaller estimated patient population (median 450 vs 800) and were less likely to conduct a specialist IBD Clinic (53.1% vs 81.3%), employ a clinical nurse specialist (57.1% vs 79.2%), research nurse (6.1% vs 33.3%) or research fellow (2.0% vs 44.7%), or have access to a clinical trials coordinator (14.3% vs 29.2%). While 9 respondents considered conducting clinical trials “easy”, 61 (61.6%) found it to be “difficult”. Forty eight (48.5%) respondents reported clinical workload to be a major barrier to clinical trials participation, with inadequate personnel (41.4%) and paperwork/bureaucracy, particularly in relation to LREC and R&D applications being cited as problematic. Measures that units felt would help support their clinical trials work included new/additional research nurse appointments (44.9%), clinical nurse specialists (15.3%) or a clinical trials coordinator (24.5%), along with extra funding (15.3%) and time for research being built into job plans (10.2%).

Conclusion: This audit has demonstrated the difference in availability of resources between research active and inactive units, as well as highlighting those factors identified by respondents as hampering the conduct of clinical trials and those deemed likely to be of most use towards developing clinical trials capability.

THE EFFECTS OF RIBAVIRIN OF DENDRITIC CELL MATURATION AND CYTOKINE PRODUCTION

S. J. Barker1, A. E. Semper2, W. M. Rosenberg2. 1Liver Group, Southampton University, 2iQur, Southampton General Hospital, Southampton, UK

Introduction: Hepatitis C is a major cause of liver related morbidity and mortality. Ribavirin in combination with interferon alpha (IFN\α) is currently the mainstay of treatment, leading to sustained viral eradication in 40–80% of infected patients. Ribavirin is a purine nucleoside analogue whose mechanisms of action are poorly understood although thought to include anti viral and immune modulatory effects. Dendritic cells (DCs) are potent antigen presenting cells and play a pivotal role linking innate and adaptive immunity and plasmacytoid DCs (pDCs) are the main producers of IFN\α in response to viral infection.

Aims & Methods: We tested the effect of ribavirin at different concentrations on expression of DC maturation markers (HLA-DR, CD83, CD86, CD40) and cytokine secretion from normal healthy donors and chronic hepatitis C patients (CHC) using ex vivo-isolated plasmacytoid DCs (pDCs) induced by CD40L, TLR9 or TLR7 agonists and cultured, monocyte-derived DCs (MoDCs) induced by CD40L.

Results: In pDCs, ribavirin upregulated the expression of CD86 (p<0.05) but had no effect on HLA class I expression in NHDs. CD40 ligation induced IFN-α production from NHDs and CHC patients was significantly reduced by ribavirin at all concentrations (p<0.05). In NHDs, ribavirin also caused a significant reduction in IL8 (p<0.05) and TNF-α (p<0.05). When a TLR9 agonist and a TLR7 agonist were used as alternative stimuli, the reduction in IFN-α production from pDCs by ribavirin was again seen in NHDs (p<0.005 and <0.05 respectively). The suppressive effects of ribavirin on cytokine production were not due to induction of pDC apoptosis or cell death, as assessed by cytometric staining for annexin V and PI. In MoDCs, ribavirin had no effect on maturation but was found to significantly suppress the production of TNF-α (p = 0.05).

Conclusion: The suppression of TNF-α seen from both MoDCs and pDCs treated with ribavirin may explain the reduction in hepatic inflammation seen with ribavirin monotherapy. The reduction of IFN-α from ribavirin treated pDCs may give some explanation as to why ribavirin alone has no beneficial effects on patients viral loads and does not lead to a sustained viral response and it underlines the importance of IFN in the treatment of HCV. The exact mechanisms causing the reduction in IFN-α production needs to be explored further.

RETINAL VASCULAR CHANGES IN PATIENTS WITH CIRRHOSIS

A. Mukhopadhya1, T. MacGillivray2, N. McAvoy1, P. C. Hayes1. 1Department of Gastroenterology, 2Clinical Research Facility, Royal Infirmary of Edinburgh, Edinburgh, UK

Introduction: The retinal microvasculature is unique in that it is the only part of the human circulation that can be directly visualised non-invasively in vivo and easily photographed. The subsequent application of digital image analysis to measure vascular topography in an automated and objective fashion is increasingly more common in modern ophthalmology and clinical research. Circulatory changes are common in cirrhosis and it is well known that vasodilatation worsens with worsening liver dysfunction. Study of retinal circulation may reflect severity of liver disease.

Aims & Methods: A digital fundus camera was used to take colour images of the human retina from patients with documented cirrhosis and compared with images from healthy controls. Post-processing, in the MATLAB environment, creates an enhanced grey-scale image for quantitative analysis. Vessel tracking describes the process of automatically searching along a vessel segment to detect centre positions and widths at regular intervals. Automatic segmentation of blood vessels enables the measurement of vascular tortuosity. We aimed to compare retinal vascular tortuosity in patients with cirrhosis and in healthy controls.

Results: The Lilliefors test confirmed that the tortuosity data of the arterioles and venules of cirrhotic patients did not come from a normal distribution. Median of veins tortuosity was 1.4 E-4 and the median of arteries tortuosity was 5.1 E-4. The tortuosity of veins was reduced and that of the arteries increased as compared to the normal controls. There was good correlation of indices from left and right eyes (correlation coefficient 0.7). The tortuosity of the arteries was significantly more in patients with Child’s C cirrhosis (11.3 E-4) as opposed to that with Child’s A (3.2 E-4) and Child’s B patients (4.0 E-4).

Conclusion: We show that liver disease increases tortuosity of arteries and lowers tortuosity of veins. We show that there is a relatively strong correlation between arterial tortuosity in left and right eyes in the presence of liver disease. We also demonstrate a strong correlation between arterial tortuosity and the stage of liver disease.

BILIARY EPITHELIAL TO MESENCHYMAL TRANSITION CONTRIBUTES TO PORTAL TRACT FIBROGENESIS IN PRIMARY SCLEROSING CHOLANGITIS

H. L. Marshall1, H. Robertson1, A. D. Burt1, D. E. J. Jones1, J. A. Kirby1, M. Hudson2. 1Institute of Cellular Medicine, University of Newcastle, 2The Liver Unit, Freeman Hospital, Newcastle upon Tyne, UK

Introduction: Damage and loss of intra-hepatic bile ducts are characteristic features of a range of chronic inflammatory liver diseases, including primary sclerosing cholangitis (PSC). Progressive cholestasis may predispose to portal fibrosis and biliary cirrhosis and is generally accompanied by an increase in bile ductules at the interface with the parenchyma, in turn thought to be a driver for periportal fibrosis. Intrahepatic biliary epithelial cells (IBEC) are obvious targets for immune-mediated injury since they bind leukocytes. Indeed, recent studies have shown immune cells induce IBEC to undergo phenotypic transition to myofibroblasts. Proteomic features of epithelial mesenchymal transition (EMT) include loss of epithelial markers and acquisition of mesenchymal markers such as S100A4. Importantly, induction of S100A4 occurs sufficiently early to visualise epithelial cells in situ during the initial phases of EMT.

Aims & Methods: The study was designed to quantify EMT at different stages of PSC. Tissue sections from all histopathological stages of PSC were labelled by immunohistochemistry to detect antigens associated with bile ducts (cytokeratin (CK)-19 and E-cadherin), EMT (S100A4, vimentin and matrix metalloproteinase-2 (MMP-2)) and myofibroblasts (α-smooth muscle actin: αSMA). S100A4/CK-19 dual-labelled cells were counted in biopsy sections and expressed as the number of S100A4+ IBEC per mm2 of portal tract. Biopsies from donor livers on the day of transplant provided normal controls.

Results: Normal bile ducts did not express S100A4, vimentin, MMP-2 or αSMA. IBEC in PSC expressed S100A4, vimentin, and MMP-2 but not αSMA; colocalisation of both S100A4/CK 19 and S100A4/E-cadherin was observed in bile ducts and the ductular reaction. There was loss of CK-19 and cell-surface E-cadherin in some S100A4-expressing IBEC and a tendency for these cells to migrate from the ductular structures. αSMA-expressing myofibroblasts were only observed in areas of established fibrosis. There was a positive relationship between number of S100A4+ IBEC per mm2 portal tract and PSC stages 1–3 (including normal, stage 0 p<0.02).

Conclusion: The association between EMT in both bile ducts and the ductular reaction and stage of PSC provides good evidence of a pathogenetic role for EMT in the development of portal tract fibrosis in this disease.

CORTICOTROPHIN RELEASING HORMONE INCREASES POSTPRANDIAL BLOATING BUT NOT DISTENSION IN PATIENTS WITH IRRITABLE BOWEL SYNDROME

A. Agrawal, R. Lea, P. J. Whorwell, L. A. Houghton. Neurogastroenterology Unit, University Hospital of South Manchester, Manchester, UK

Introduction: Corticotrophin releasing hormone (CRH) mediates the stress response causing increases in both visceral sensitivity and motility.1 2 The effect of stress on the symptom of bloating and abdominal distension (that is, change in girth) in patients with IBS, which are also modulated by changes in visceral sensation3 and gastrointestinal (GI) transit4 remain unknown.

Aims & Methods: To assess the effect of CRH on postprandial bloating and distension in patients with IBS compared with healthy volunteers. A randomised, double-blind, placebo controlled, cross-over study in which the effect of CRH on meal induced bloating and distension were compared between 17 female IBS (Rome II) patients with a history of bloating (aged 21–42 years) and 10 healthy female volunteers (21–32 years). Abdominal girth was measured using the validated technique of Ambulatory Inductance Plethysmography5 and bloating using a Likert Scale of 0–5 (5 = very severe) for one hour before and two hours after a standard meal (1114 kcal). CRH 1 μg/kg or placebo was intravenously administered immediately after the meal and patients were studied during the luteal phase of the menstrual cycle or while taking the contraceptive pill.

Results: In IBS patients, CRH significantly increased postprandial bloating (change fasting to fed: CRH, +1.35 (1.04 to 1.66) (mean (95% CI)) vs placebo, +0.53 (0.23 to 0.83); p = 0.001) but not distension (+2.02 cm (1.37 to 2.67) cm vs +1.28 cm (0.74 to 1.82) cm) compared with placebo control. No such effects were seen in healthy volunteers (bloating: +0.50 (0.1 to 0.9) vs +0.27 (0.09 to 0.45,), distension: +1.41 cm (0.30 to 2.52) cm vs +1.38 cm (0.85 to 1.91) cm). Thus, the postprandial bloating response to CRH compared with placebo was greater in IBS patients compared with healthy volunteers (p = 0.02).

Conclusion: IBS patients exhibit an enhanced postprandial bloating response to CRH compared with healthy volunteers. This may be due to the sensation of bloating in IBS patients being in part related to visceral sensitivity3 which is increased by CRH.1 The lack of effect on distension, which has previously been shown to be directly related to transit time in IBS4 is probably explained by the fact that CRH increases motility and/or transit.

MEASUREMENT OF GLIADIN ALONE CANNOT PREDICT TOTAL GLUTEN CONTENT OF FOODS FOR COELIAC DISEASE SUFFERERS

H. J. Ellis1, M. C. Bermudo-Redondo2, M. L. Butero2, T. Suligoj1, C. K. O’Sullivan2, P. J. Ciclitira1. 1Nutritional Sciences King’s College London, Rayne Institute St Thomas’ Hospital, London, UK, 2Chemical Engineering, Universtidad Rovira i Virgili, Tarragona, Spain

Introduction: The gliadin fraction of wheat gluten has long been known to be coeliac disease (CD) activating. The other major wheat protein group, the polymeric glutenins, comprises low molecular weight and high molecular weight glutenin subunits (LMW-GS and HMW-GS). The latter was recently shown to exacerbate the CD. Less is known of the LMW-GS, but they are likely to be CD activating. Purified wheat starch is a standard constituent of many gluten-free foods for individuals with CD. Methods for gluten measurement have focussed on gliadins, whose content is multiplied ×2 to obtain total gluten content. However, this calculation is invalid, since high pressure liquid chromatography showed the ratio glutenin:gliadin varies from 0.2× to 5.8× in wheat starches. Draft legislation suggests the maximum permissible level of gluten in foods labelled as gluten-free should be 200 ppm gluten for rendered and 20 ppm for naturally gluten-free foods.

Aims & Methods: We previously developed monoclonal antibodies to HMW-GS and to the immunodominant gliadin epitope, α-2-57-75. Both antibodies recognised gluten from a wide variety of wheat cultivars. We used these antibodies in competitive ELISAs to measure HMW-GS and gliadin in a wide variety of gluten-free food types.

Results: In 5 wheat starches the ratio of HMW-GS to gliadin varied from 0.45× to 4.2×. The derived gluten content from gliadin ×2 in the last example would be 4.9 ppm, whereas gliadin plus HMW-GS was 12.4 ppm. In other foods the ratios of HMW-GS to gliadin were as low as 0.07×. Gluten-free pasta had 75 ppm gliadin, multiplied ×2 this would imply the product contained 150 ppm gluten, whereas gliadin plus HMW-GS comprised 88 ppm.

Conclusion: Within wheat varieties gliadin comprises 58–77% of total gluten and glutenins form 26–38%. Within glutenins LMW-GS comprise 19–25% of total gluten and HMW-GS 7-13%. However, owing to differing solubility characteristics of gluten protein fractions, huge variation in their final content may occur after processing. Thus the ratio of HMW-GS to gliadin in wheat starches varied 10-fold, and in other food samples variations of up to 68× were seen. We did not measure LMW-GS, however it is most unlikely that the proportion of these proteins varied inversely with HMW-GS sufficient to account for the wild fluctuations in the HMW-GS to gliadin ratio. Measurement of gliadin ×2 will lead to potential gross under and overestimations of total gluten content. If new guidelines are to be introduced with limits as strict as 20 ppm it will be essential to measure all of the toxic fractions independently rather than relying on the calculation gliadin ×2  = gluten.

A COMPARISON OF OUTCOME PREDICTIVE SCORES FOR ACUTE HAEMORRHAGE FROM OESOPHAGEAL VARICES

P. D. Dunckley1, N. Rajoriya2, K. Tweed2, J. Booth2. 1Gastroenterology Department, Gloucestershire Royal Hospital, Gloucester, 2Gastroenterology Department, Royal Berkshire Hospital, Reading, UK

Introduction: Several validated scoring systems have been developed to predict mortality from an acute upper gastrointestinal haemorrhage including the Rockall (RS) and Blatchford (BS) scores. Oesophageal variceal bleeding represented only 4% of the sample studied for the development of the RS and 6% for the BS. Whether these scores can be accurately used to predict outcome in suspected oesophageal variceal haemorrhage is less clear. The MELD and Child-Pugh (CPS) scores have also been shown to predict mortality in patients with acute variceal haemorrhage.

Aims & Methods: The aim of this study was to compare these four scores and their ability to predict inpatient mortality, 30-day re-bleed rates and transfusion requirements in patients with acute oesophageal variceal haemorrhage. Case notes of patients admitted to a single centre (RBH) were reviewed for the period between January 2000 and December 2006. Demographic, clinical, biochemical and haematological parameters were collected in order for the MELD, PS, CPS and BS scores to be calculated. Main outcome measures of in-patient mortality, 30-day re-bleed rates and transfusion requirement were documented.

Results: During the period of study, there were 86 admissions in 63 patients with proven acute oesophageal haemorrhage. The overall inpatient mortality was 17.4%; the 30-day re-bleed rate was 15.1% and the mean number of units of blood transfused was 5.1 units (SD-4.5). Receiver operating characteristic (ROC) curves were calculated to assess the performance of the scores. All four scores significantly predicted in-patient mortality (RS – area 0.71, 95% CI 0.57 to 0.84, p = 0.01; BS – area 0.76, 95% CI 0.64 to 0.87, p<0.01; CPS – area 0.74, 95% CI 0.61 to 0.87, p<0.01; MELD – area 0.73, 95% CI 0.58 to 0.88, p<0.01). Likewise all scores correlated with the total units of blood prescribed (RS – p = 0.001 (one-way ANOVA), CPS – p = 0.02 (one way ANOVA); MELD – Pearsons R –0.28, p<0.05; BS – Pearson’s R – 0.25, p<0.05). However, only the RS predicted 30-day re-bleed (RS – area 0.70, 95% CI 0.56 to 0.84, p = 0.03; BS – area 0.59, 95% CI 0.42 to 0.77, p = 0.29; CPS – area 0.55, 95% CI 0.37 to 0.73, p = 0.60; MELD – area 0.54, 95% CI 0.35 to 0.72, p = 0.69).

Conclusion: All four scores (RS, BS, CPS and MELD) predict in-patient mortality and correlate with the number of units of blood received by the patients suffering from acute oesophageal haemorrhage. The RS is the only score to predict 30-day re-bleeds. We propose therefore that the RS is the score of choice for predicting outcome in acute oesophageal haemorrhage, especially in light of its widespread use in upper gastrointestinal bleed integrated care pathways and its simplicity of use.

VIP ENDOSCOPY: THE RED CARPET TREATMENT?

C. W. Pritchard, P. M. Neville, M. J. Webberley. Department of Gastroenterology, Nevill Hall Hospital, Abergavenny, UK

Introduction: The VIP (virtual in-patient) system was introduced into our hospital in 2006. VIPs are patients who present to the acute hospital services or who are currently in-patients, who need urgent investigation, but do not necessarily require in-patient care. They are ‘admitted’ onto a virtual ward on the Clinical Work Station (CWS), but discharged from the hospital. Once their investigations are complete and their management plan decided, they are discharged from the virtual ward. The system has been used by both physicians and surgeons. It has allowed early discharge of patients from hospital but minimises their risk of being lost to follow-up.

Aims & Methods: VIP has successfully included many investigative modalities but here we have evaluated the patients referred for endoscopy.

Abstract 136

Results: 103 procedures were performed over a 4-month period. Of this, 74 (72%) were OGDs, 10 (9.5%) were colonoscopies and 19 (18.5%) were sigmoidoscopies. 70% were booked within 48 hours of presenting and 30% were in-patients. The median time to procedure from booking, was 5 days, with 90% having been performed within 10 days. This amounted to 441 bed days saved over the study period (3.6/day). Of the gastroscopies, dyspepsia and abdo pain accounted for 50% of the indications, whereas upper GI bleed was only 22% (Rockall 1 or less), and a normal examination was found in more than 50% of patients. Of the colonoscopies, anaemia was the commonest indication (50%) and in 8 (70%) the procedure was normal. PR bleeding accounted for 60% of the indications for sigmoidoscopy and in 48% the procedure was normal. There were no adverse events recorded whilst patients were on the virtual ward and all patients were safely discharged to home following their procedure and no-one was lost to follow-up. 33 patients went on to have further investigations as out-patients.

Conclusion: We believe VIP to be a safe, popular and bed-efficient system for managing GI patients who would otherwise be admitted for investigation. It is easy and cheap to set up and requires little administration.

IS THERE A CASE FOR ROUTINE DUODENAL BIOPSY AT GASTROSCOPY OR IS A TARGETED APPROACH ADEQUATE? A RETROSPECTIVE AND PROSPECTIVE EVALUATION

A. D. Hopper, T. S. Chew, D. S. Sanders. Gastroenterology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Current UK recommendations advocate duodenal biopsy (DB) at the time of gastroscopy if there is evidence of anaemia or mal-absorption (weight loss or diarrhoea). However in “real” clinical practice DB rates have been reported as varying considerably (31–74%). For this reason routine DB has been suggested in order to avoid missing treatable disease—for example, coeliac disease or giardia.

Aims & Methods: We aimed to compare the diagnostic yield of routine DB against a policy of targeted DB using “high risk” indications. Initially, we retrospectively examined the endoscopy and histopathology database in our hospital from January 2003 to January 2004 to obtain data on our current practice. We obtained the number of gastroscopies and DB performed. Subsequently, consecutive adult patients referred for gastroscopy were prospectively recruited by a single endoscopist over a 26 month period (January 2004 to April 2006). All these patients had routine DB performed. In this prospective group patients were identified if they had target symptoms that should prompt an endoscopist to obtain DB. Patients were excluded if they had a coagulopathy, active gastrointestinal bleed or a suspected carcinoma observed during the examination (n = 220). We noted the number of new diagnoses of coeliac disease and giardia made from the DB in both groups.

Abstract 137

Results: In the retrospective group 5979 gastroscopies were performed and analysed, 1464 with DB (DB rate 24.5%). In the prospective group 2220 gastroscopies were performed with routine DB taken in 2000 (DB rate 90.1%), 739 of these patients had target biopsy indication symptoms: a prevalence of 37.0%. The biopsy rates and diagnostic yields of the two groups are shown in the table. By performing routine DB, the DB rate increased by 3.7 (90.1/24.5)-fold. This was reflected by a similar and significant 4.17 (0.4/0.084)-fold increase in giardia prevalence and 3.9 (3.9/1.0)-fold increase in coeliac disease prevalence at gastroscopy (p<0.01). When comparing the prevalence of findings of targeted versus a routine DB policy there was a small non significant (p = 0.68) decrease in the prevalence of coeliac disease (3.9 to 3.6%) and also similar findings were observed in the giardia prevalence (0.4 to 0.1%) (p = 0.11). However there was a 2.7 (2000/739)-fold decrease in duodenal biopsy rates (p<0.0001) by adopting the targeted approach.

Conclusion: The diagnostic yield for coeliac disease and giardia is significantly increased by performing routine duodenal biopsies. However with accurate targeting of patients with anaemia or malabsorption, the number of DB can be significantly reduced without a significant reduction in the detection rate of small bowel disease.

Endoscopy posters

ENDOSCOPIC ULTRASOUND PREDICTS PATIENTS WITH OESOPHAGEAL CANCER WHO BENEFIT FROM NEOADJUVANT THERAPY

N. S. Balaji1, R. Melhado1, A. Fowler1, N. Wariach1, A. Goddard2, P. C. Leeder1, A. Cole2, S. Iftikhar1. 1Upper GI Surgery, 2Gastroenterology, Derby Royal Infirmary, Derby, UK

Introduction: Endoscopic ultrasound (EUS) remains the most accurate tool for staging of loco regional disease in oesophageal cancer. Neoadjuvant chemotherapy is proven beneficial in operable locally advanced T stages (>T2) and Nodal (N1) disease based on recent UK trials. The aim of this study was to assess the role of EUS in identifying patients with advanced locoregional disease who are likely to benefit from neoadjuvant treatment.

Aims & Methods: The study group included patients who underwent resection for carcinoma of the oesophagus following preoperative EUS over a 10-year period. Those who received neoadjuvant therapy were excluded. EUS stage was compared with pathologic staging. The accuracy of EUS in identifying patients who would benefit from neoadjuvant therapy (T2 N1, T3N0 or greater) was determined.

Results: 118 patients had EUS staging prior to resection. 31 patients underwent preoperative chemotherapy and were excluded. The mean age was 63 (SD 8.9) with an M: F distribution 4:1. Oesophagectomy was performed by an Ivor Lewis approach in 41 patients, left thoracotomy in 45 and a 3-stage approach in 1 patient. R0 resection was achieved in 66% of patients. The mean lymph nodal yield was 13 (SD 6). EUS predicted T stage was accurate in 55% of T1, 45% of T2, 66% in T1+T2 combined and 77% of T3 disease. T4 disease was excluded (unresectable). EUS predicted N stage was accurate in 56% of N0 and 60% of N1 disease. Locally advanced disease (T2N1, T3NO or greater) was predicted accurately in 89% of patients. The sensitivity of EUS for predicting the above group of patients who would benefit from neoadjuvant chemotherapy was 90% with a positive predictive value of 84%.

Conclusion: Although the sensitivity of EUS varies with T and N stage, the combined stage prediction of locally advanced disease is accurate. EUS remains a valuable tool in identifying this subset of patients in whom neoadjuvant chemotherapy offers a survival benefit following surgery.

OUTCOMES AND PERCEPTIONS OF A HIGH-STAKES COLONOSCOPY SKILLS ASSESSMENT: THE JOINT ADVISORY GROUP/BOWEL CANCER SCREENING PROGRAMME SCREENING COLONOSCOPY ASSESSMENT

R. Barton. For the Joint Advisory Group on Gastrointestinal Endoscopy and NHS Bowel Cancer Screening Programme, Newcastle University and Northumbria Healthcare NHS Trust, Newcastle upon Tyne, UK

Introduction: The introduction of an assessment of candidate colonoscopists wishing to provide the NHS Bowel Cancer Screening Programme created strong feelings in the endoscopy community, with concerns about the appropriateness and robustness of the process.

Aims & Methods: The aims of this study were to evaluate the performance and outcomes of candidates in the assessment, and their perceptions of the process. Data were collected as part of the accreditation process, and comprised quantitative data on performance in MCQ and DOPS assessments, and qualitative data on candidates’ perceptions of the assessment.

Results: Seventy six assessments were undertaken, by 67 candidates. Candidates had performed a mean (SD) of 2490 (1506) colonoscopies, range 500–7500, and detected polyps in a mean 29% (8%) of cases, range 18–53%. Mean MCQ score was 80% (6.3%), range 59–98%. Eight assessments were repeated, three for breach of process protocol with the DOPS assessment on the 1st or 2nd occasion, and 5 as a true re-sit assessment. Of 73 secure assessments, 54 (74%) candidates met the criteria. There were comments about the ambiguity of some MCQ questions “content fine but considerable ambiguity”, the clarity of the photographs “pictures unclear to make a measured judgement”, the relevance of some questions, and the inclusion of some controversial topics “There is some disagreement between experts on the management of depressed polyps”. Suggestions included using video clips rather than static photographs. There was confusion around whether or not the MCQ was negatively marked. With respect to the DOPS, candidates commented on the stress and anxieties provoked, “The process is hugely stressful. The examiners put me at ease but preassessment stress was very high” and “nerve racking but enjoyable”. They noted the luck involved in the cases encountered, “Very friendly and relaxed. Only problem it relies on individual patients having appropriate colons” and the lack of familiarity with the assessing unit, “Better in base hospital, could have been made less threatening.” They largely felt welcomed, “very relaxed non intimidating atmosphere” and that the assessment was fair “I think it is a fair assessment and very little improvement required” and “The process as written entirely fair, difficult two cases”. There have been five appeals against the results referred to the Accreditation Panel, with one upheld as a process breach.

Conclusion: Despite significant initial concerns, the overall process seems to have a reasonable level of acceptance. The MCQ clearly needs work and no-one has yet been failed on this basis. The DOPS, while nerve-racking in an unfamiliar environment, is generally well accepted. Further work on refining the process continues.

INITIAL EXPERIENCE OF THE NEW CYTOBRUSH IN THE ENDOSCOPIC ULTRASOUND MANAGEMENT OF PANCREATIC CYSTIC LESIONS- PROCEDURAL CHARACTERISTICS, CYTOLOGICAL YIELD AND SAFETY PROFILE

H. Karageorgiou, J. R. Bebb, K. Ragunath, G. Aithal. Gastroenterology, University Hospital, Nottingham, UK

Introduction: Endoscopic ultrasound (EUS) guided FNA of pancreatic cyst fluid is one of the most accurate methods of diagnosis of the nature of pancreatic cystic lesions but is still of limited sensitivity. A new through-the needle cytologic brush system has recently been approved for use during evaluation of cystic lesions of the pancreas. An initial study of 10 patients1 suggested use of brush was superior to standard FNA.

Aims & Methods: In a tertiary referral centre we have reviewed the clinical applicability of the cytobrush in the EUS evaluation of pancreatic cystic lesions in 38 consecutive EUS examinations.

Results: 38 EUS examinations were carried out on 32 patients with pancreatic cystic lesions over 30 months (median age 60, range 24–83, 53% male). 15/32 patients had a history of pancreatitis or EUS criteria for chronic pancreatitis on the day. The echobrush was used in 23 cases- on 15 occasions through the stomach. We were unable to use the echobrush in all cases for the following reasons: technically difficult or unable to use 19G needle, unable to puncture cyst wall, clinically inappropriate, technical malfunction. Cellularity yield was adequate in 19/23 (83%) cases. There were no major complications after any of the procedures. The final diagnoses (after minimum 6 months follow-up, including surgery) were pseudocyst (15), benign cyst (8), premalignant lesion (4) and malignant lesion (4). The cytobrush was used in all 4 malignant cases (adequate cells in 3 cases) with malignant cells (C5) seen in 2 of the patients.

Conclusion: This is the largest reported series of use of the cytobrush in routine EUS assessment of pancreatic cystic lesions. Our initial experience is that it is safe and technically feasible to use in most cases with good cellular yield. We feel that the cytobrush probably has a complementary role alongside other modalities in the management of cystic pancreatic lesions.

ENDOSCOPIC ULTRASOUND FINE NEEDLE ASPIRATION: A VALUABLE TOOL FOR EVALUATION OF MEDIASTINAL PATHOLOGY

G. C. Bennett1, R. Bambury1, S. O’Riordan1, R. Sugrue1, S. Sengupta1, E. Clarke1, J. Lennon1, N. Mulligan2, T. Kelleher1, P. MacMathuna1. 1Gastrointestinal Unit, 2Department of Pathology, Mater Misericordiae University Hospital, Dublin 7, Ireland

Introduction: Endoscopic ultrasound (EUS) with fine needle aspiration +/− biopsy (FNAB) is established in the diagnosis and staging of upper GI tract malignancy. Mediastinoscopy has been the usual method for sampling mediastinal nodes and masses. EUS FNAB provides a promising alternative in select cases.

Aims & Methods: We report our experience with EUS FNAB for mediastinal disease. In 35 patients referred for FNAB, we reviewed demographic information, radiology, surgical records, follow-up information, indication, diagnostic yields and complication rates.

Results: The indications were: mediastinal mass (n = 10), mediastinal nodes (n = 24) and oesophageal submucosal lesion (n = 1). A specimen adequate for diagnosis was obtained in 32/35 (91%). Follow-up data were available in 30 patients. Malignancy was identified in 15/30 (50%) (NPV 100%, PPV 100%). In 12/26 (46%), benign disease was found. No false negatives were identified. Post-procedure chest pain occurred in 2 patients.

Conclusion: EUS FNAB is a safe and reliable method for obtaining tissue from the mediastinum with high diagnostic accuracy. It provides a relatively non-invasive, safe and effective alternative to mediastinoscopy/CT guided biopsy in the management of mediastinal disease.

OUTCOME OF PATIENTS WITH UPPER GASTROINTESTINAL HAEMORRHAGE UNDERGOING DELAYED OUT-PATIENT ENDOSCOPY

S. Bhat, M. McLoughlin, T. C. K. Tham. Division of Gastroenterology, Ulster Hospital, Belfast, UK

Introduction: Ideally, all patients who present with an upper GI bleed should have an in-patient endoscopy. However, studies have shown that patients felt to be at low risk, according to a verified scoring system, could have the procedure as an out-patient without suffering any increase in morbidity or mortality. However, these studies were performed in healthcare systems where out-patient endoscopy can be performed within 48 hours. No studies have evaluated the outcome of such patients awaiting delayed outpatient endoscopy in a healthcare system with limited resources such as the NHS.

Aims & Methods: To determine if patients who were discharged for delayed out-patient endoscopy suffered adverse events while awaiting their procedure. We identified consecutive patients with upper GI haemorrhage who were discharged for out-patient endoscopy. A clinical Rockall score was generated retrospectively for each case. Those with a score of 0 were considered low risk for adverse events. Follow-up was obtained from the patients’ notes.

Results: 44 patients (27 males, 17 females), with a median age of 45 (range 17–91), were discharged for an out-patient endoscopy. The average delay from discharge to endoscopy was 85.5 days (range 1–318). 18 patients had a clinical Rockall score of 0, 12 had a score of 1, 5 had a score of 2 and 9 had a score of more than 2. The average length of stay for all patients was 3.6 days (range 1–21). In general, those with a Rockall score over 2 had longer hospital admissions (average 5.8 days). There were no deaths but 2 patients, with initial scores of 1 and 0, were re-admitted at 10 and 60 days respectively while awaiting OGD. The first required a blood transfusion and was diagnosed with a Mallory-Weiss tear. The second discharged himself after readmission without waiting for an inpatient endoscopy. Of the 33 patients who proceeded to out-patient endoscopy 14 had a normal OGD, 11 had either oesophagitis, gastritis or duodenitis, 6 had a hiatus hernia, 1 had Barrett’s oesophagus and 1 had a Mallory Weiss tear. 11 patients did not have an OGD. Of these, 8 did not attend, 1 patient was still awaiting endoscopy, intubation was not possible in 1 patient and in 1 elderly patient the family cancelled the procedure.

Conclusion: Despite a delay in out-patient endoscopy for those with upper GI haemorrhage, only 2 of 44 (4%) were re-admitted with an upper GI haemorrhage while awaiting their endoscopy. It is doubtful if an in-patient endoscopy in these cases would have changed their outcome. The decision to discharge patients before endoscopy on the basis of a low clinical Rockall score, or a length of stay of more than 3 days when there has been no rebleeding, appears to be associated with a satisfactory outcome in the majority of cases despite a delay in endoscopy.

MISSED DIAGNOSES IN PATIENTS WITH UPPER GASTROINTESTINAL CANCERS

D. H. Chowdhury, P. Bhandari, P. Goggin, R. Ellis, Q. Amanda, T. Trebble, S. Toh, S. Sommers, H. Duncan. Gastroenterology, Queen Alexandra Hospital, Portsmouth, UK

Introduction: Upper GI cancers are common and in spite of all the effects and improvements in the treatment modalities the overall survival is not improving. This is due to the cancer being diagnosed at a relatively advanced stage. There is evidence to suggest that most patients presenting with cancer have dyspeptic symptoms for many years predating these presentations with the increasing use of endoscopy and imaging techniques a lot of these patients would have had gastroscopies in the past. We believe that oesophageal and gastric cancers in their early stages are difficult to diagnose and could be missed at endoscopy.

Aims & Methods: The aim of our study was to: identify what percentage of patients have had imagings in the 3-year period prior to their presentation with cancer; to correlate the endoscopic findings with their current presentation; to identify the “missed early cancer” rate. Missed cancer was defined as a cancer patient who had a gastroscopy within 3 years prior to the index endoscopy (at which cancer diagnosis was made) and was found to have either normal or benign findings. We retrospectively analysed the patient records of 124 consecutive patients diagnosed with oesophageal and gastric cancers at our centre. We recorded timing of previous endoscopy, endoscopic findings, and surveillance or non-surveillance type of endoscopy, stage of the cancer and histological type of cancer. The reasons for failure or missed diagnoses were documented.

Results: Of the 124 patients, 14 (12%) had undergone a minimum of one endoscopy within the previous 3 years, 5 (36%) of these within the previous 1 year. 80% (11/14) of these patients had significant symptoms at the time of previous endoscopy. 12/14 of these patients with missed cancers had oesophageal cancers while 2/14 had gastric cancers. The endoscopy findings at the time of previous endoscopy (missed cancer) included normal (4/14), oesophagitis (8/14) and gastritis (2/14).

Conclusion: In our series endoscopic cancer miss rate is 12% for oesophago-gastric cancers. The exact reasons for this miss rate is not clear from our work but we believe that majority of these patients must have had very subtle endoscopic changes, the understanding of which lacks in western endoscopists. Advances in endoscopic imaging and increasing awareness of endoscopic appearances of early cancers should help reduce this miss rate.

LOCALISATION OF COLORECTAL TUMOUR AT COLONOSCOPY: DO WE KNOW WHERE WE ARE?

B. J. Colleypriest, P. F. Marden, A. Nicol, E. Griffiths, J. D. Linehan. Gastroenterology, Royal United Hospital, Bath, UK

Introduction: Quality and safety indicators for colonoscopy from the JAG have recently been revised and await ratification. It is suggested that suspected malignant polyps and that tumours if small, or the position unclear to the endoscopist are tattooed.1 Quality standards specific to colonoscopy in the National Bowel Cancer Screening Programme set as an objective in the planning of surgery the identification of tumour position in the correct segment of the colon in >95% of cases and tattooing of suspected malignant polyps in 100%. A previous American study has quoted a figure as high as 21% for endoscopist’s inaccuracy in correctly identifying tumour position at colonoscopy.2

Aims & Methods: This study aims to assess the ability of UK colonoscopists to identify the correct anatomical position of a colorectal tumour during the procedure. The endoscopy reporting system was interrogated for the findings of tumour or carcinoma on reports from June 2004 to August 2006. The endoscopically reported location of the tumour was recorded and compared with the pathology report of all patients proceeding to surgical resection. The colon was divided into segments as specified on the endoscopy reporting program (see table). The endoscopist was deemed inaccurate if there was discrepancy of more than one segment.

Abstract 144

Results: 251 patients with a colorectal tumour found at colonoscopy were identified, 142 were male and 109 female. 200 of these had histologically confirmed cancers. 159 patients (87 male and 72 female) proceeded to surgical resection. 4 histology reports could not be found. Complete agreement between colonoscopy report and histology report was found in 94 (61%) cases. The endoscopic localisation was found to be inaccurate in 26 (17%) cases.

Conclusion: This study shows a significant inaccuracy in colonoscopy reporting of colorectal tumour location similar to figures published from North America.2 In 17% of cases there was a discrepancy of more than 1 segment and only 61% were accurate. Current BSG Quality and safety guidelines suggest tattooing colorectal tumours if small or the position is unclear and is an auditable outcome. Our data suggest that endoscopic localisation of tumours is unreliable and tattooing of all tumours should be mandatory.

COLORECTAL CANCER FOUND DURING COLONOSCOPY: HOW MANY BIOPSIES ARE NECESSARY?

B. J. Colleypriest, P. F. Marden, L. Ayres, B. Doshi, A. Nicol, E. Griffiths, J. D. Linehan. Gastroenterology, Royal United Hospital, Bath, UK

Introduction: Identification and diagnosis of colorectal tumours usually result from biopsies taken at colonoscopy. On occasions when the diagnosis is not clear from the first set of biopsies further sampling may necessitate a repeat colonoscopy with its inherent risks. Currently there are no guidelines recommending the number of biopsies that should be taken on the index colonoscopy. Studies asking the same question in oesophageal1 and combined upper GI cancer2 suggest 6 and 7 biopsies respectively. BSG guidelines for the management of oesophageal and gastric cancer reference a paper suggesting 6 biopsies.3

Aims & Methods: To suggest the optimal number of biopsies necessary. Data were collected on all colonoscopies with a diagnosis of suspected cancer over a 26-month period. Histological data were collected for initial and repeat biopsies and subsequent surgical specimens. The patients were divided into 2 groups, carcinoma correctly diagnosed at initial biopsy and those needing a second colonoscopy or surgery for diagnosis. Severe dysplasia was counted as a correct diagnosis unless this diagnosis was unsure and a repeat colonoscopy resulted.

Abstract 145 Sensitivity (S) of cancer diagnosis according to number of biopsies taken

Results: 217 patients were identified of whom 198 had a histological diagnosis of cancer. Of these 198 patients 182 were correctly diagnosed at index colonoscopy and 16 were diagnosed either on repeat colonoscopy (10) or at surgery (6). Nine patients with suspected cancer not confirmed on initial biopsies neither progressed to surgery nor were re examined endoscopically. 10 patients with suspected cancer were re-examined and diagnosed with non-malignant disease. The number of biopsies in the 16 patients with cancer “missed” at index colonoscopy were statistically lower when compared with the correct group (5.729±0.1392 vs 4.250±0.4031, p = 0.0025).

Conclusion: This study demonstrates that the sensitivity of diagnosis of a suspected cancer increases with the number of biopsies. A high degree of statistical significance was noted in the lower number of biopsies taken in the group of patients with an initial missed diagnosis. The combined sensitivity of taking 5 or fewer biopsies (85%) was significantly lower compared to more than 5 (98%) or 6 (99%). We suggest that at least 6 biopsies should be mandatory as a national auditable standard in the context of suspected cancer.

DOUBLE BALLOON ENTEROSCOPY: A LEAP FORWARD IN THE MANAGEMENT OF SMALL BOWEL CROHN’S STRICTURES

E. Despott1, E. Tripoli1, A. Polecina1, K. Konieczko2, C. Fraser1. 1Wolfson Unit for Endoscopy, St Mark’s Hospital, 2Department of Anaesthesia, St Mark’s and Northwick Park Hospitals, London, UK

Introduction: Small bowel strictures are a common, debilitating complication of Crohn’s disease (CD) that frequently result in the need for surgical treatment. Apart from the inherent risks of the morbidity and mortality associated with major surgery, afflicted patients also face the threat of resultant short bowel syndrome. Double balloon enteroscopy (DBE) with its unique ability to provide endotherapy to previously inaccessible parts of the small bowel, may make it possible to treat CD related strictures safely and at low risk to patients. We report our preliminary experience with DBE and CD stricture dilatation since the introduction of the service in February 2005.

Aims & Methods: A prospective proof of concept analysis of stricture dilatation by DBE in CD, assessing the feasibility of the procedure, symptom improvement and surgical requirements post procedure. Patients referred for stricture dilatation at DBE had the procedure done under general anaesthesia. Stricture dilatation was performed using controlled radial expansion (CRE) balloon dilators with fluoroscopic guidance. These patients remain under follow-up for recurrence of their symptoms and the need for surgical intervention.

Results: Eight patients (mean age 44.8, range 30–52 years) with CD related strictures were referred for DBE stricture dilatation due to obstructive symptoms. The strictures were diagnosed radiologically, were shown to be <5 cm in length and were refractory to medical therapy. Balloon stricture dilatation was performed successfully in 6/8 patients with a mean dilatation diameter of 15.4 mm (range 12–20 mm). In 2/8 patients, fixation of the small bowel due to adhesions, made reaching the strictures at DBE impossible. These 2 patients required surgical stricturoplasty. In the 6 patients where stricture dilatation was done at DBE, symptoms improved dramatically and to date there was no need for surgical intervention. The mean follow-up period was 8.8 months (range 2–28 months). There were no clinically important complications.

Conclusion: Our preliminary proof of concept experience regarding the use of DBE in CD related stricture dilatation shows promise. The procedure was feasible and safe and to date has had satisfying results. The potential for DBE to revolutionise the current management of CD small bowel strictures and avert surgical intervention, merits further evaluation in larger studies with more prolonged follow-up. The impact of this approach on patient quality of life, avoidance of surgery associated risks and small bowel saving potential could be significant.

A PROFILE OF COLONIC POLYPS IN THE NORTHEAST OF ENGLAND AND ITS IMPACT ON COLONOSCOPIC PRACTICE: THE SOUTH DURHAM POLYP PILOT STUDY

T. Wood1, C. Westwood1, K. Gunning1, A. Dhar2. 1Colorectal Surgery, Darlington Memorial Hospital, 2Department of Gastroenterology, Bishop Auckland General Hospital, County Durham, UK

Introduction: The clinico-epidemiological profile of colorectal polyps in the Northeast of England is not well described. The performance of colonoscopy and polypectomy in the region needs to be analysed particularly in preparation for bowel cancer screening, due to start in 2007 in the County.

Aims & Methods: To study the clinical characteristics of colonic polyps in the South of Durham, using a pilot cohort of approx 200 patients, and also analyse the performance of colonoscopic polypectomy and adherence to BSG guidelines. The histopathology database (SNOMED) was interrogated to identify a random set of 200 patients with colonic polyps between July 2005 and June 2006. Case notes and endoscopy reports were manually reviewed to include patient demographics, polyp characteristics, dysplasia grade, completeness of polypectomy and adherence to surveillance guidelines.

Results: 160 patients (M: F = 1.46, 55% over age 65 years) were included in this pilot study, with an adjusted caecal intubation rate of 93.8%. 213 polyps were found in these patients (mean 1.33 polyps/patient), with 78% having only one polyp and 22% having >2 polyps. 63% of all patients with polyps had adenomas. Of all polyps, 45% were metaplastic, 26% tubular adenomas and 21% tubulovillous, 4% villous and 1% had intramucosal carcinoma. Polyp sizes were as follows: <5 mm–68%, 6 to 10 mm–18%, 11 to 20 mm–6% and >20 mm–7%. Interestingly, 75% of the polyps were sessile or flat and only 20% were pedunculated. The base of the polyp was examined in only 16% patients after polypectomy. Histologically, the polyp removal was reported to be complete in 20%, incomplete in 14%, indeterminate in 34% and not commented in 32%. BSG guidelines for surveillance were adhered to in 85% of cases.

Conclusion: This pilot study shows that majority of the patients in this region have a single polyp, which is less than 10 mm in size but most of the polyps are sessile. This is likely to mean advanced saline lift polypectomy as the norm for most patients, but also a careful examination of the base to ensure complete excision. There is a need to improve adherence to BSG guidelines and improve our colonoscopy training by examining the base post polypectomy. The bowel cancer screening programme will give us more information as to whether the asymptomatic population has a different profile from the symptomatic population.

SAFE SEDATION IN ENDOSCOPIC PROCEDURES: IMPACT OF PETHIDINE ON 30-DAY POST-PROCEDURE MORTALITY

B. Doshi1, P. Marden1, B. Colleypriest1, G. Taylor2, D. Robertson1. 1Gastroenterology, Royal United Hospital, 2School for Health, University of Bath, Bath, UK

Introduction: Current endoscopy sedation protocols are based on the British Society of Gastroenterology guidelines issued in 2003. It recommends that minimum amounts of drug are used and that certain procedures may require a combination of benzodiazepine and opioid for adequate pain relief. The NCEPOD “scoping our practice” 2004 findings suggested that increased sedation may have contributed to 30-day post-procedure mortality (PPM). However, some studies have shown that reduced doses of midazolam did not result in a significant reduction in 30-day PPM.1

Aims & Methods: We retrospectively investigated the impact of a change in sedation practice on 30-day PPM in our unit. A total of 7122 endoscopic procedures performed over 12 months were identified on our electronic reporting system. Patients who died within 30 days of an endoscopic procedure were identified using the hospital IT system. The procedures were then assessed in two consecutive 6-month periods P1 (1/07/2006 to 31/12/2006) and P2 (1/01/2007 to 30/06/2007). Primary outcomes assessed were average dose (SD) of midazolam administered, frequency of pethidine use and 30-day PPM. Changes in sedation practice was assessed in different operator groups such as Consultants (Con), Non-Training Grade (NTG) such as associate specialist and general practitioners, Endoscopy Nurse Specialist (ENS) and Trainees.

Results: Total cases performed in P1 and P2 were 3509 (49.3%) and 3613 (50.7%) respectively. There was a significant increase in the percentage of cases who received pethidine in P1 compared to P2 (10% (223 cases) vs 19.1% (440 cases) (p<0.05)). This was accompanied by a reduction in average midazolam dose from P1 to P2 (3.5(1.2) mg vs 3.2 (1.2) mg (p<0.05)). The 30-day PPM remained similar in P1 and P2 (2% vs 2.5% (p = 0.23)). The Trainee group had the most significant increase in frequency of pethidine use (2.2% in P1 vs 14.3% in P2 (p<0.05)) and this was accompanied by a significant reduction in average midazolam dose (3.6 (1.3) mg in P1 vs. 2.9 (1.0) mg in P2 (p<0.05)). Their 30-day PPM did not alter significantly (2.6% in P1 vs 2.8% in P2 (p = 0.887)).The NTG group was the only other group to have had a statistically significant increase in frequency of pethidine use (5.9% in P1 vs 12.2% in P2 (p<0.05)). But the NTG group did not have a significant reduction in average midazolam dose (3.2 (0.9) mg in P1 vs 3.1 (0.9) mg in P2 (p = 0.46)). Interestingly, the 30-day PPM in the NTG group had a trend towards significance at 1.2% in P1 and 2.4% in P2 (p = 0.09).

Conclusion: Any interpretation of sedation related mortality should include analysis of both benzodiazepine and opioid administration. Despite a reduction in midazolam dose, a lack of improvement in 30-day PPM could be due to increased pethidine administration.

ENDOSCOPIC MANAGEMENT OF NON-VARICEAL UPPER GASTROINTESTINAL HAEMORRHAGE: RESULTS OF A QUESTIONNAIRE EVALUATING CURRENT PRACTICE

J. M. Dunn, A. McNair, V. Saxena. Gastroenterology, The Queen Elizabeth Hospital, London, UK

Introduction: Non-variceal upper gastrointestinal (UGI) haemorrhage is associated with a significant morbidity and mortality, which has remained unchanged for 50 years. There is good evidence that higher doses of 1:10 000 adrenaline (13–20 ml) and use of a second modality can improve outcome.1 Consultants should make trainees aware of this and ensure that they acquire these important endoscopic skills.

Aims & Methods: To evaluate current practice a questionnaire was sent to 38 specialist registrars in 16 acute hospitals across the South East Thames region. The questionnaire comprised of a clinical vignette describing a patient presenting with haematemesis, Rockall score = 8. A photograph showed a 1.5 cm duodenal ulcer with visible vessel. Questions on endoscopic management followed. A multinomial model of confidence in ability using 4 endoscopic therapies was evaluated. A 5-point quiz objectively assessed knowledge of UGI haemorrhage management. Training facilities were evaluated at each centre.

Abstract 149 Results of multinomial model

Results: Completed forms were returned from 15 hospitals (a 94% response rate) and from 29 trainees (an 80% response rate). The trainees were divided into two groups according to seniority; group A 3rd year or above; group B 1st and 2nd year. Results of multinomial model are shown (table). There was no significant difference in quiz scores (78% vs 68%, p = 0.12). 100% would use adrenaline as a first line. 69% of trainees would routinely use less than 10 ml of adrenaline, 71% in group A and 67% in group B (p = NS). There was considerable geographic variation in access to a second modality: heater probe (HP) or endoclip was available in 100% of hospitals inside M25, 78% outside. 50% of hospitals in the region had the choice of both. 73% of hospitals provided argon plasma coagulation (APC).

Conclusion: The majority of SpRs in South East Thames are not using high enough doses of adrenaline for treatment of bleeding ulcers. There was no significant difference in the proportion of senior trainees using higher doses of adrenaline, despite greater confidence and experience. As expected, senior trainees were significantly more confident in their ability to use a second modality. However 24% would use APC when not indicated and 76% did not know the energy setting for HP. Assuming the participants were representative of gastroenterology trainees in the UK, the quality and provision of training in the management of non-variceal UGI haemorrhage is currently inadequate.

GASTROINTESTINAL BLEEDING AND WARFARIN

P. Eadala, R. Crimmins, S. Dolwani, G. L. Swift, J. T. Green. Department of Gastroenterology, Llandough Hospital, Cardiff, UK

Introduction: Acute upper gastrointestinal bleeding (UGIB) is a common medical emergency with an incidence ranging from 50 to 150 per 100 000 of the population each year.1 Warfarin is commonly used to treat and reduce thrombo-embolic events. The GI tract is the most common site of bleeding in patients receiving anticoagulant therapy.2 Immediate endoscopy has been shown to have a lower yield in this group of patients and the group with negative findings have a favorable outcome.3

Abstract 150

Aims & Methods: All patients presenting with acute UGIB between 1 June 2000 and 31 May 2005 to a single UK centre were identified. At the time of referral for endoscopy an initial proforma was completed as part of an ongoing audit and case notes were later retrieved to capture other data. Patients who were on warfarin were identified together with its therapeutic indications. They were subdivided on the basis of their target INR (International Normalised Ratio). Significant endoscopic findings were defined as: oesophageal ulcer, gastric ulcer, duodenal ulcer, carcinoma, angiodysplasia and varices. The presence of significant endoscopic findings and mortality was compared between the groups. Stastical analysis was performed using Fisher’s exact test.

Results: During this period 808 patients presented with UGIB, 80 (10%) were on warfarin. 47 (59%) had INR above their therapeutic range when compared to 33 (41%) whose INR was therapeutic/subtherapeutic at presentation. There is a lower yield of significant endoscopic findings (p<0.037) but no significant difference in mortality in those with INR above their therapeutic range when compared to those in the therapeutic/subtherapeutic range at presentation. Despite the above there was stastically demonstrable difference in mortality (p<0.049) but not in significant endoscopic findings in those on warfarin compared to those who are not.

Conclusion: Most of those who present with UGIB while on warfarin have an INR above their therapeutic range. Endoscopy in these patients has low yield but mortality is higher probably due to the fact that patients on warfarin tend to be older with significant comorbidities. Adequate resuscitation and correction of coagulopathy prior to endoscopy still form the mainstay of treatment.

A PROSPECTIVE STUDY TO DETERMINE FACTORS ASSOCIATED WITH INCREASED DISCOMFORT DURING COLONOSCOPY

D. A. Elphick, M. T. Donnelly, K. Smith, S. A. Riley. Department of Gastroenterology, Northern General Hospital, Sheffield, UK

Introduction: Colonoscopy can be an uncomfortable procedure. To increase safety, there is a trend towards lower levels of sedation and analgesia use during colonoscopy in the UK. In many endoscopy units, including our own, unsedated colonoscopy with or without use of inhaled nitrous oxide (entonox) is common practice and opiates are rarely used.

Aims & Methods: This study aimed to determine patient factors predictive of higher discomfort levels during colonoscopy. Patients undergoing outpatient colonoscopy were recruited. Patient age, sex, height, weight, surgical history, indication for colonoscopy, previous colonoscopy details were recorded. Patients were then asked to grade how uncomfortable they anticipated the test to be on a Numeric Rating Scale (NRS: zero (no discomfort) to 10 (maximal pain)). Scores were then recorded on the NRS for cannulation (22G), rectal examination, colonoscope insertion, every 2 minutes during the procedure, retroflexion in the rectum and extubation. An overall score for the procedure on the NRS was recorded on completion. The endoscopist graded procedure difficulty (0–10) and a hospital anxiety and depression score was filled by the patient. NRS scores of peak and overall discomfort were recorded between 2 and 3 months post procedure.

Results: 109 patients (44 male, 65 female; median age 61.5 (21–80) years) were recruited. 99 procedures were completed to caecal pole or terminal ileum, 44 received midazolam (median 2 (2–5) mg) and 88 used entonox. Mean overall NRS score was 4.72 (men 3.93; women 5.25; p<0.01) and mean peak score 6.97. Discomfort scores were higher in those with: irritable bowel syndrome (p = 0.03); diverticular disease (p<0.01); midazolam (p = 0.03), buscopan (p<0.001) or entonox (p<0.001) use; endoscope tracker use (p<0.01); incomplete procedures (p<0.001) or preceding gastroscopy (p = 0.02). Discomfort scores correlated with: anticipated discomfort rate (p<0.001), peak discomfort rate (p<0.001), endoscopist difficulty grade (p<0.001), patient anxiety (p<0.001) and depression (p = 0.02) scores. Discomfort during the procedure did not correlate to cannulation or rectal examination NRS score. Recollected overall discomfort scores 2–3 months following procedure were significantly lower than the originals (mean reduction 0.85) (p<0.001). However, this reduction was not influenced by the use of midazolam.

Conclusion: Colonoscopy can be an uncomfortable procedure. Factors indicative of difficult colonoscopy are associated with higher levels of discomfort, as are the presence of irritable bowel, high anxiety, depression and the anticipation of discomfort. Preceding gastroscopy worsens colonoscopy discomfort. Low dose midazolam neither relieves discomfort nor makes patients forget it. Selected patients may benefit from increased analgesia use.

ERCP TRAINING: WHAT HAS CHANGED IN THE LAST 5 YEARS?

A. D. Farmer, M. J. Brookes, P. G. Wilson. Department of Gastroenterology, City Hospital, Birmingham, UK

Introduction: Endoscopic retrograde choleangiopancreatography (ERCP) is perhaps the most difficult and skilled endoscopic procedure to learn. Despite less invasive diagnostic modalities being developed, demand for ERCP has remained static at a rate of 0.75–1 procedures per 1000 population.1 Training, and competency, in ERCP has long been a game of numbers and targets. In 1988, completion of 35 procedures deemed a trainee competent although this figure has risen to 300–450 procedures in 2007, with selective duct cannulation rates (SDCR) being a surrogate marker of the quality of training.2 The British Society of Gastroenterology ERCP survey in 20061 identified that only one third of trainees needed to become competent in ERCP to maintain the service.

Aims & Methods: The purpose of this survey was to assess trainees’ attitudes and exposure to ERCP training in the West Midlands and how this has changed over the last 5 years. All gastroenterology trainees in the West Midlands were approached to complete the survey. Trainees were surveyed in 2002 and 2007.

Abstract 152 Results

Results: The response rates were 23 out of 28 (82.3%) in 2002 and 36 out of 38 (94.7%) in 2007. The distribution of trainees through the training programme was similar. The table summarises the results.

Conclusion: It is apparent from this regional study that there has been a dramatic change in trainees’ attitudes and experience of ERCP training over the last five years. Because demand for training has decreased, with changing attitudes towards ERCP, smaller numbers of trainees are getting higher volume exposure to ERCP training. Training appears to have improved as reflected by the increase in SDCR, the reduction in major complications encountered and the total number of procedures completed and this is probably related to the numbers of ERCPs individual trainees are undertaking.

ENDOSCOPIC ULTRASOUND TRAINING: THE TRAINEES’ PERSPECTIVE

A. D. Farmer1, M. J. Brookes1, J. I. W. Jones2, P. G. Wilson1. 1Department of Gastroenterology, City Hospital, Birmingham, 2Department of Gastroenterology, Princess Royal Hospital, Telford, UK

Introduction: Endoscopic ultrasound (EUS) training within the UK is haphazard with piecemeal exposure. In addition, the lack of definitive guidance regarding curriculum and competence assessment from the Joint Advisory Group on Gastrointestinal Endoscopy serve only to compound this problem. These problems were highlighted as early as 2004 by the UK EUS Users group.1

Aims & Methods: The purpose of this survey was to assess trainees’ attitudes, and exposure, to EUS training in the West Midlands. Currently, there are 4 centres performing EUS in the region. All specialist registrars (SpRs) in gastroenterology in the region (n = 38) were approached either in person, by email or by telephone to complete the survey.

Results: 36 out of 38 SpRs responded to the survey (94.7%). 26 respondents (72%) were interested in pursuing training in EUS. 12 trainees (33.3%) intended to achieve accreditation in EUS prior to completion of training, although of these only 10 (27.7%) had done any EUS training. 5 respondents (13.8%) felt that obtaining accreditation in EUS was an essential part of training as a gastroenterologist. Indeed, 7 respondents (19.4%) felt that it would improve their job prospects if they were accredited in EUS. Experience of diagnostic EUS was limited with only 3 trainees having completed more than 40 procedures. Only 2 trainees had had any experience of interventional EUS. Of those intending to achieve EUS accreditation, they felt that training was hindered by the concomitant training of consultant staff (50%), other trainees (20%), lack of trainer availability (15%) and lack of time (15%). Only 1 respondent (2.7%) felt the current level of EUS training was adequate. No trainee had attended a national or international centre of excellence for specific EUS training. Trainees stated that the most important factor in improving EUS training should be the development of a dedicated training module (100%).

Conclusion: The experience of EUS among trainees in the West Midlands is low, although interest and enthusiasm is high. The reasons for this are several. These include the small numbers of centres performing the procedure, the relative low volume of procedures performed in the region and the concomitant training of local consultant staff. Training opportunities are likely to remain limited while capital costs of setting up EUS services remain high and the number of trainers small. If EUS is to grow as a diagnostic and therapeutic modality it is vital to have highly organised and coordinated training across the region. Clearly, provision needs to be made in terms of training courses and access for trainees to acquire EUS skills. We conclude that the establishment of a dedicated training module in EUS with defined goal orientated objectives and assessment, leading to accreditation, will improve the quality of EUS training and service within the region.

MORTALITY FOLLOWING PEG INSERTION IN A SCOTTISH DISTRICT GENERAL HOSPITAL

S. Ghosh, A. K. Ghosh, A. Roychoudhury, J. Rose. General Medicine, The Ayr Hospital, Ayr, UK

Introduction: Percutaneous endoscopic gastrostomy (PEG) is a widely used method for long term enteral nutrition whose appropriateness and safety have been critically examined in a recent report (Scoping our Practice. NCEPOD, 2004).1

Aims & Methods: To establish the safety of this technique as practiced in our hospital, we performed a retrospective case note analysis of the indications, risk factors, complications and mortality in patients with PEG insertion for the 10 years from 1996–2005. All patients were assessed by medical, dietetic and endoscopic unit staff and received parenteral antibiotics routinely before PEG insertion.

Results: Of 125 patients treated, adequate records were available for 109 (86%). Of these 36 were males and 73 females with a mean age of 65.4 years. The procedure was used in patients with CVA (60), dysphagia (41), aspiration (39), malnutrition (26) and MND or dementia (22) and was performed an average of 8.5 days after referral. Over 65% (71 patients) were alive at 28 days (cf: NCEPOD 75%) and a total of 58 (53%) were able to be discharged or transferred to long-term care. The common complications were chest infection (25), wound infection (11), peritonitis and sepsis (9) and pain (7). 35% (38 patients) died within 28 days, nearly three quarters (28 patients) of these within 14 days of the procedure. In a retrospective study of patients with multiple diseases it was not possible to ascertain the exact final cause of death. Although patients with aspiration may not benefit from PEG insertion, there was no excess of deaths of patients in this subgroup.

Conclusion: The majority of patients treated with a PEG tube in our hospital survived both the first month and to be discharged or transferred. Compared to the NCEPOD data early mortality (especially within the first two weeks) was higher in our audit (carried out on patient prior to NCEPOD) could probably be improved by better pre-assessment and stricter adherence to selection criteria. A prospective audit is essential to identify potential pitfalls in selection of patients and avoidance of complications related to PEG insertion.

USE OF ENDOSCOPIC ULTRASONOGRAPHY VERSUS OESOPHAGOGASTRODUODENOSCOPY IN ASSESSING INTESTINAL FAILURE ASSOCIATED LIVER DISEASE IN CHILDREN ASSESSED FOR INTESTINAL TRANSPLANTATION

P. McKiernan, K. Sharif, G. Gupte. Liver Unit, Birmingham Children’s Hospital, Birmingham, UK

Introduction: The severity of intestinal failure associated liver disease (IFALD) along with the presence/absence of portal hypertension determines the type of transplant recommended as per the international consensus criteria. The severity of IFALD is usually assessed histologically on liver biopsy and the presence of portal hypertension at esophagogastroduodenoscopy (OGD). Liver biopsy carries a risk of significant morbidity in this vulnerable group. Endoscopic ultrasonography (EUS) is a newer technique for detection of oesphagogastric varices (OGV) which can also detect extraluminal varices not seen by OGD. Using EUS in children being assessed for intestinal transplant might reduce the need for liver biopsy in this group.

Aims & Methods: To evaluate if EUS is more sensitive than OGD in detecting OGV in progressive IFALD. Children referred for liver and small bowel transplantation were assessed prospectively. 16 children (9M:7F, median age 13 months (range 7–88)) with progressive IFALD and splenomegaly were assessed by EUS and OGD to detect presence of GOV. EUS was performed with a high resolution endoluminal ultrasound miniprobe (UM-2R-3 20 MHz, Olympus UK) during conventional OGD. Results of EUS findings were compared with OGD examination.

Results: In 9 patients results of EUS and OGD were concordant, i.e., both positive (n = 2) or both negative (n = 7) for GOV. In 7 patients GOV were identified by EUS, but were not seen on OGD.Liver biopsy was avoided in 5 children by the use of EUS.

Conclusion: EUS can detect GOV not seen on OGD. EUS has changed our assessment strategy and is now used as the initial investigation method in evaluating the degree of IFALD.

MID-POINT RESULTS OF A RANDOMISED BLINDED CONTROLLED STUDY OF HANDS-ON THERAPEUTIC ENDOSCOPY TRAINING

A. V. Haycock, S. Thomas-Gibson. Wolfson Unit for Endoscopy, Imperial College London, London, UK

Introduction: Despite recent investment, there is limited evidence for the efficacy of intensive hands-on training courses in endoscopy. The St Mark’s Basic Therapeutic Endoscopy Course (BTEC) involves lectures and hands-on training in four therapeutic modalities: non-variceal upper GI bleeding, polypectomy and hot biopsy, oesophageal stricture dilatation and PEG insertion. Feedback from previous courses has shown it is considered to be an excellent training opportunity (mean rating 9.1 out of 10), but there has been no higher level evaluation of the benefits of the course.

Aims & Methods: The primary aim of the study is to evaluate the efficacy of the BTEC in delivering training in a range of basic therapeutic endoscopic techniques. Recruitment aims to enrol 32 delegates overall. Applicants to the course are randomised into subjects and controls. All delegates are initially assessed by blinded experts on procedural skills using procedure checklists and a global rating scale. They also complete a multiple choice questionnaire (MCQ) assessing knowledge of therapeutic endoscopy. Delegates are randomised to assessment before or after attending the lectures. The subjects then receive a half-day practical hands-on training as per the previous BTEC course, while the controls receive no hands-on training. All delegates are then re-tested on their procedural skills on each therapeutic modality.

Abstract 156 Difference between initial and final assessment scores

Results: We report the mid-point results of the study (n = 17). There were no significant demographic differences between subjects and controls. There were no significant differences in initial practical assessment scores or MCQ score between delegates who had the lectures before or after testing. There was a significant improvement in the checklist score and overall score for subjects compared to controls for polypectomy and control of upper GI bleeding. There were differences for oesophageal dilatation and PEG insertion nearing significance.

Conclusion: There was no detectable benefit on practical skill or knowledge acquisition from the lectures. There were significant benefits from hands-on teaching on both skill at polypectomy and control of upper GI bleeding, with strong improvements in skill at oesophageal dilatation and PEG insertion. To confirm these results enrolment is continuing.

THE UK NATIONAL ENDOSCOPY TRAINING PROGRAMME HAS IMPROVED COLONOSCOPY TRAINING OVER A 5 YEAR PERIOD: CLOSING THE AUDIT LOOP

A. V. Haycock, J. H. Patel, S. Thomas-Gibson. Wolfson Unit for Endoscopy, Imperial College London, London, UK

Introduction: An audit in 20021 showed that colonoscopy training in the North West Thames deanery was poorly structured, with low levels of supervision and high complication rates. Since then, the UK National Endoscopy Training Programme introduced centrally funded, JAG accredited courses and new assessment tools to standardise training and raise the quality of practitioners. The primary aim of this study was to determine if this has had an impact on the standard of colonoscopy training in this region.

Aims & Methods: This was a cross-sectional survey. Questionnaires used in the previous study were updated and emailed to all gastroenterology and GI surgery trainees in the deanery and those who participated in the previous study. Trainees could complete the forms electronically and return them using an anonymous web-based secure upload portal.

Results: 37 out of 113 trainees responded (33%). 92% were aware of the JAG guidelines for training. Significantly more trainees said that they had been formally taught the principles of colonoscopy (89% vs 65%; p = 0.02), polypectomy (76% vs 53%; p = 0.04) and extubation (82% vs 55%; p = 0.02) than in 2002. Trainers are significantly more likely to be present in the room for a trainee’s first 100 colonoscopies (77% vs 49%; p = 0.04) and attempt explaining a solution to a problem before taking over (55% vs 23%; p = 0.03). Course attendance significantly increased (76% vs 45%; p = 0.006), and although most had to wait between 4–6 months to attend, all trainees were “quite” or “very” satisfied with the course. There was an increase in trainees who record their complication rate (51% vs 32%) although this was not significant (p = 0.11). Reported complication rates were lower, with 3 of 18 (16%) having had a perforation compared to 8 of 13 (61%) in 2002. 78% of trainees think that their training has been adequate or better than adequate, compared to 25% in 2002. The average caecal intubation rate, a surrogate marker of competence, did not change (90% vs 93%; p = 0.09).

Conclusion: There has been a significant improvement in colonoscopy training both at base hospitals and in access to specialist courses compared to 2002, although 83% of trainees believe that training can still be improved further. New technologies such as simulators and the 3D imager were considered by the majority to have contributed towards the improvement, but most trainees still feel they need greater regular access to dedicated training lists and courses. The central training programme and funding has made a significantly positive impact in this region. The loss of such investment may have a detrimental effect on future colonoscopy training.

GASTROINTESTINAL CANCER PREVALENCE IN IRON DEFICIENCY AND THE PREDICTIVE VALUE OF IRON INDICES

B. Hayee, S. Patel, J. O’Donohue, D. Reffitt. Gastroenterology, University Hospital Lewisham, London, UK

Introduction: In attempting to provide an efficient and cost-effective endoscopy service, “cut-off” criteria for investigating patients are useful tools to reduce the number of potentially inappropriate investigations. An audit of endoscopic investigation of patients referred for asymptomatic anaemia was carried out, changing local practice in terms of creating a new referral form with a mandatory requirement to enter iron indices to indicate the presence of true iron-deficiency anaemia (IDA). Concerns were raised regarding the validity of the use of such criteria to direct investigation. Results were therefore analysed further.

Aims & Methods: To determine the sensitivity, specificity and positive and negative predictive values of iron indices in the diagnosis of gastrointestinal (GI) cancer, data from an audit of endoscopic investigation of IDA were analysed over two three-month periods (2006–7). Iron deficiency (ID) was defined as a either mean cell volume (MCV) <76 fl or ferritin <15 μg/l (or <50 if ESR raised). With haemoglobin (Hb) level of <12.5 g/dl for men and <11.5 g/dl for women, these patients were termed “IDA”. Those with low Hb not fulfilling either of the two additional criteria were termed “non-IDA”.

Results: 221 patients (mean age 68.5 ± 16.9 years; 149 female) were analysed. 14 GI cancers were identified in the IDA group (n = 112) and 4 in the non-IDA group (n = 109): an overall prevalence of 8.1%. However, there were fewer cancers in the non-IDA group (3.5% vs 12.5%; two-tailed p = 0.025). Examination of the case-notes of this group revealed the presence of alarm symptoms in two of the four patients with GI cancer. Including these two patients, our results demonstrate a sensitivity (Sn) and specificity of iron-deficiency for detecting GI cancer of 77.8% and 51.7% respectively. Positive and negative predictive (NP) values were 12.5% and 96.3%. Raising the “cut-off” level for ferritin to 50 μg/l for all patients would increase the Sn and NP value to 94.4% and 99.1% respectively. There was no advantage comparing MCV to ferritin as the sole determinant of ID and insistence on both criteria being fulfilled would lead to a drop in Sn and NP values (55.6% and 93.3%).

Conclusion: The sensitivity and negative predictive value of the above criteria for GI cancer are high. Patients without true iron-deficiency are unlikely to have a sinister cause for their anaemia. While the possibility of GI cancer in patients without iron-deficiency should not be overlooked, a thorough evaluation and the presence of alarm symptoms should reduce the potential for missed diagnoses in this group. Defined criteria for IDA may be useful to determine whether endoscopic examination is warranted.

VIRTUAL CHROMOENDOSCOPY BY MEANS OF REAL-TIME ADAPTIVE HISTOGRAM EQUALISATION OF HD VIDEO WITH THE SONY CELL RSX GRAPHICS PROCESSING UNIT

S. Hebbar1, R. Beresford2, D. Lennon2, M. Williams2, J. Anderson3, N. Shepherd4, J. Brown1. 1Gastroenterology, Gloucester Royal Hospital, Gloucester, 2Broadcast & Professional Research Labs, Sony, Basingstoke, 3Gastroenterology, Cheltenham General Hospital, Cheltenham, 4Histopathology, Gloucester Royal Hospital, Gloucester, UK

Introduction: Accurate and early endoscopic diagnosis demands a precise assessment of mucosal detail. Chromoendoscopy, Narrow Band Endoscopy (Olympus) and Virtual Chromoendoscopy (Fujinon Intelligent Chromo Endoscopy) have been developed to supplement the information derived from white light illumination. The choice of techniques available to the clinician is often related to the availability of endoscopic equipment since many of the new technologies are device-specific. Our objective has been to devise a system that will generate real time, contrast-enhanced images to provide diagnostic information in a device and platform independent way.

Aims & Methods: We have undertaken Adaptive Histogram Equalisation on RGB images in each colour domain using matrix methods (MATLAB). This enhances contrast by stretching the pixel distribution in the original image across the entire dynamic range of the video monitor. We have also transformed RGB images to a novel endoscopy colour space and applied adaptive histogram equalisation to the luminance channel. A library comprising white light, adaptive histogram contrast enhanced, NBI and chromoendoscopy images has been accumulated for comparison.

Results: We have shown that it is possible to enhance endoscopic images using numerical methods which provide additional diagnostic information in a similar way to chromoendoscopy and NBI. Furthermore, real-time image processing can be accomplished by rewriting the MATLAB algorithms in C++ for compilation on the Sony Cell RSX graphics processing unit.

Conclusion: Numerical methods may provide enhanced diagnostic information for endoscopy and we have developed several candidate algorithms for real-time image processing. Further studies are planned to compare our numerical methods with the more established endoscopic techniques.

INTRAEPITHELIAL NEOPLASIA AND INTRAMUCOSAL CARCINOMA COMPLICATING BARRETT’S OESOPHAGUS: ENDOLUMINAL RESECTION IS POSSIBLE IN THE UK

D. P. Hurlstone1, N. Tiffin2, S. S. Cross3. 1Endoscopy, Royal Hallamshire Hospital, 2Histopathology, STH NHS Foundation Trust, 3Histopathology, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Endoscopic mucosal resection (EMR) and now endoscopic submucosal dissection (ESD) using the insulation tipped needle knife for curative intent in intraepithelial neoplasia and Barretts associated intramucosal carcinoma could assume a “gold standard” therapy in comparison to primary ablative techniques. Definitive histopathology can be retrieved and if circumferential excision is achieved, buried gland phenomena abolished. No prospective studies from the UK have addressed the technical feasibility, safety or medium term efficacy of this technique.

Aims & Methods: To evaluate the efficacy of a staged pan-endoluminal resection for the treatment of Barretts associated intraepithelial neoplasia (IN) +/−intramucosal carcinoma in the UK. Patients were referred for consideration of curative intent endoluminal resection of Barretts oesophagus with biopsy proven complicating IN or mucosal carcinoma that had declined formal surgical excision or primary ablative techniques. Exclusion criteria were; pre-resection high-frequency 20 MHz EUS suggestive of T1/N1 disease or beyond, non-correctable coagulopathy or previous ablative PDT or APC endotherapy. Circumscribed lesions in addition to native Barretts epithelium were resected using extended EMR or ESD using sodium hyaluronic acid and an insulation tipped needle knife – 60W pure cut current. Circumferential lesions were resected in 2 or 3 sequential sessions. Primary end-point: histopathologically confirmed resection status; secondary end-points: (1) complication rates (2) intra-epithelial neoplasia recurrence; (3) stricture formation; (4) medium-term clinical efficacy.

Results: 24 patients met inclusion criteria (18;75% male). Pre-resection histopathology: IN–low grade:n = 16/high-grade n = 6/intramucosal carcinoma n = 2. Median number resection session 2.5 (range 1–4)/median procedure duration 68 min (range 32–146). Resection modality: ESD n = 15 (63%), EEMR n = 9 (37%). Primary outcome: post-resection histopathological IN resection status: R0 = 19/24(79%)/Rx: = 3/24(13%)/R1 = 2/24(8%). Secondary outcomes: post-resection stricture formation n = 1/24(4%)/perforation rate 0/24/bleed risk (immediate/delayed) 6/24 (25%)/successful endoluminal haemostasis n = 6 (100%)/IN recurrence n = 1 (4%)/endoscopic follow-up: median 39 months (range 6–68).

Conclusion: EMR and ESD with circumferential extension is a safe and efficacious therapeutic modality for the treatment of intraepithelial neoplasia complicating Barrett’s oesophagus with low complication rates and good medium term clinical efficacy. Definitive resection status can be objectively assessed histopathologically and the risk of dysplastic buried gland phenomena negated. Long-term efficacy and learning curve characteristics for this technique are now required.

THE ROLE OF STRAIGHT-TO-TEST ENDOSCOPY FOR TWO-WEEK WAIT REFERRALS SUSPECTED OF UPPER GASTROINTESTINAL MALIGNANCY

G. R. Irving1, F. Chikwanda1, P. Morowski2, K. Kapur3, M. Shiwani1. 1Upper Gastrointestinal Surgery, 2Clinical Audit, 3Gastroenterology, Barnsley District General Hospital, Barnsley, UK

Introduction: The two-week wait (2WW) system was introduced from July 2000. The impact of this with respect to Upper Gastrointestinal services and the management of malignancy has been reported. Originally, all 2WW referrals were provided with a clinical appointment following which a second appointment for endoscopy would be arranged. This has lead to the evolution of straight-to-test (STT) referrals for endoscopy with an aim to improve cancer waiting times. Referral criteria have been developed in an attempt to improve the service.

Aims & Methods: Data were obtained prospectively on patients referred via 2WW pathway to the Upper gastrointestinal specialists during the last financial year. Patients referred with dysphagia were automatically given STT appointment. Patients without dysphagia were given a clinical appointment. During the STT appointment clinical history, examination and endoscopy were conducted and further follow-up arranged as appropriate. Data were cross-referenced with the Open Exeter National cancer waiting time database and the 2WW referral office. Additional data were obtained retrospectively from all cases of upper gastrointestinal cancer. Origin of referral, diagnostic route, clinical outcome and the chronological data were recorded.

Results: 295 patients were referred via the 2WW system and underwent endoscopy of which 144 referred with dysphagia went STT. There were 53 gastro-oesophageal malignancies diagnosed (23 gastric and 30 oesophageal) in one year. 34 were diagnosed via 2WW pathway of which 10 were diagnosed at STT appointment. Median waiting time for patients referred for STT = 9 days (1–17) with 1 patient breeching the 14-day target before consultation. The median time before 2WW clinic appointment was 9 days (1–24) with no patients breeching the target. Patients referred from clinic to endoscopy had a median wait of 14 days (1–49). 1 cancer patient from the STT group exceeded 31 days until decision to treat date and 2 patients from this group exceeded the 62 day target for treatment. In 2WW patients not referred straight to test 4 exceeded 31 days until diagnosis and 5 exceeded 62 days until treatment.

Conclusion: Straight-to-test endoscopy service does improve cancer waiting times. It does not increase the incidence of diagnosing malignancy and proportionately more cancers were diagnosed in the patients not referred STT. Overall incidence was 11.5% in all 2WW patients but only 2% in STT group. Our criteria for immediate endoscopy do not provide good sensitivity, which may be a reflection of the referral practices and the interpretation of symptoms by primary care practitioners. However, fast-tracking endoscopies will still remain important with waiting times for non-malignant disease also becoming implemented in the near future.

INDICATIONS, SUCCESS AND COMPLICATIONS OF NEEDLE KNIFE SPHINCTEROTOMY IN A TERTIARY REFERRAL CENTRE

V. Jairath, S. P. Pereira, A. R. W. Hatfield, G. J. M. Webster. Department of Gastroenterology, University College Hospital London, London, UK

Introduction: Needle knife sphincterotomy (NKS) has been controversial since its introduction in the 1980s. There is a large variation in the published literature regarding the incidence of its use, ranging from 1% to 30%. Similarly, published complication rates vary from 5–30% in large series. NKS emerges as a multivariate risk factor for post-ERCP pancreatitis and overall complications of ERCP.

Aims & Methods: The aim of this study was to examine the frequency, indications, success and complication rates of NKS in our centre. We analysed data from 100 NKSs (38 prospective, 63 retrospective). Success was defined as deep biliary cannulation sufficient to allow further diagnostic/therapeutic measures. We recorded major complications as defined by Cotton et al (pancreatitis, bleeding, perforation, infection graded as mild/moderate/severe). Thirty-day post-procedure complication data were collected.

Abstract 162 Final diagnoses in NKS patients

Results: We used NKS in 9.1% of ERCPs. Clinical scenarios in patients requiring NKS included: jaundice and dilated biliary tree (55%); abnormal LFTs and dilated biliary tree/bile duct stones (36%); postoperative bile leak (9%). Successful deep biliary cannulation was achieved in 80 patients (80%) requiring NKS, at first attempt at ERCP. Of the 20 initial failures, success was achieved on further ERCP in 6 cases within 7 days (that is, final endoscopic success in 86% of NKS cases), 5 had successful percutaneous transhepatic drains (all in the setting of malignancy), and 9 were managed expectantly. 30-day follow-up data were obtained for 87/100 patients with an overall complication rate of 11.5% (n = 10). Complications included pancreatitis in 4 cases (3 mild, 1 severe), bleeding in 5 cases (4 moderate, 1 severe), and 1 severe case of cholangitis leading to death in a patient with cholangiocarcinoma. There were no perforations.

Conclusion: In our analysis of 100 consecutive NKS the procedure was used in <10% of total ERCPs, with endoscopic success in 80% at first attempt. Overall 30-day complication rate was similar to data from other large series, with a pancreatitis rate lower than expected. Our data from a large volume centre suggest that NKS is an effective technique where biliary cannulation is difficult, with an acceptably low complication rate.

ASSESSMENT OF RAPID ON-SITE EVALUATION BY A BIOMEDICAL SCIENTIST IN ENDOSCOPIC ULTRASOUND GUIDED FINE NEEDLE ASPIRATIONS

S. Chaudhry1, V. Y. Kaushik2. 1Cytopathology, Tameside General Hospital, Ashton-under-Lyne, 2Gastroenterology, East Lancashire NHS Trust, Blackburn, UK

Introduction: Endoscopic ultrasound guided fine needle aspiration (EUS-FNA) is a technique that has achieved high sensitivities and specificities in the diagnosis of a range of malignancies many of which would otherwise require a more invasive and riskier procedure to allow confirmation. To aid the endosonographer in achieving a higher success rate, the laboratory can provide rapid on-site evaluation (ROSE) of the samples achieved. Other researchers have reported that the procedure time can be affected considerably by the presence of a cytopathologist, and therefore improving the overall benefits.1

Aims & Methods: The effectiveness of ROSE by a biomedical scientist (BMS) is evaluated in this study. During the study period of 6 months, a total of 24 patients had EUS-FNA performed. Data were gathered on the procedure times, the number of passes performed and the pass at which the biomedical scientist confirmed adequacy. The BMS staff had in-house training from a senior member of the technical staff, who accompanied them for the first 9 procedures.

Results: The data analysis showed a strong association between increasing numbers of procedures and a decrease in the overall procedure time. The average time for the pancreatic procedures decreased from 57 to 40 minutes, when comparing the first 7 procedures to the last 7 procedures performed (Pearson value = −0.911211). However, there was no correlation between procedures performed and the number of passes required for each patient, this remained at a median average of 2.5 passes with a range of 2–4. The overall inadequate rate achieved was 20%, while the FNA of pancreatic masses achieved an inadequate rate of 15%.

Conclusion: An improvement in the specific technical skills required by both the endosonographer and the liaising BMS leads to a decrease in the time required for each procedure. The absence of correlation between this and the number of passes can be explained by other underlying factors that were not accounted for in this study, for example the location and size of the mass and other clinical and pathological features of the patients. One of the other reasons was the cautious endosonographer’s desire to obtain a further sample after being informed on the adequacy of the FNA. The results achieved in this study prove that the use of ROSE in clinics by BMS is an effective method of reducing patient discomfort, and increasing diagnostic yield but there is a need to analyse the steps required to achieve proficiency in this technique.

BISPECTRAL INDEX MONITORING DURING ERCP

I. Khan1, M. Hegde1, A. Windsor2, J. Bateman1, M. Smith1. 1Gastroenterology, 2Anaesthetics, The Shrewsbury and Telford Hospital NHS Trust, Shrewsbury, UK

Introduction: In the UK most ERCP procedures are carried out using parenteral sedation, which carries risk of complications (for example, under/over sedation, aspiration). BIS gives an objective (EEG based) measure of the degree of sedation. Here data are shown from one unit where sedation was titrated using BIS during ERCP.

Aims & Methods: Data were collected prospectively during ERCP (at 5-min intervals): BIS scores (when used), heart rate, pulse oximetry, visible distress and sedative dose (midazolam and fentanyl). At the end of the procedure, the Observer’s Assessment of Alertness/Sedation (OAA/S) score (1, unresponsive to 5, responds to normal voice) was noted. When fully alert patient satisfaction with the procedure was assessed using visual analogue scales (VAS).

Abstract 164

Results: 111 patients were studied in total: BIS (n = 77, 35 male, mean age 71.6 years, range 22–96 years); no BIS (n = 34, 15 male, mean age 65.1 years, range 27–90 years). Mean data shown in table. There were no serious sedation related complications and VAS analysis from both groups showed that most patients were satisfied with the procedure irrespective of whether BIS was used.

Conclusion: In our experience BIS provides a useful method for objective monitoring of parenteral sedation during ERCP, but the procedure can be safely and comfortably conducted without it. However, further case controlled data is required to fully evaluate this technique.

EARLY EXPERIENCE OF DIAGNOSTIC AND THERAPEUTIC SINGLE-BALLOON ENTEROSCOPY

S. C. Kong, A. J. Morris. Department of Gastroenterology, Glasgow Royal Infirmary, Glasgow, UK

Introduction: Capsule endoscopy has resulted in an increase in the pathology seen in the small bowel that requires further assessment or intervention. Push enteroscopy is limited to examination of the proximal jejunum. Recently encouraging results of the yield of double-balloon enteroscopes have been reported. A new single-balloon enteroscope (Olympus XSIF-260SB) has recently become available. We describe our initial experience with the new enteroscope.

Aims & Methods: The single balloon enteroscope was assessed by endoscopists trained in the use of push enteroscopes. The Olympus single-balloon enteroscope system was used. This comprised an EVIS LUCERA small intestinal video scope (XSIF-260SB), a balloon control unit (MAJ-1440) and a single use splitting tube (ST-SB1). The splitting tube is a silicone rubber overtube with an inflatable and deflatable balloon at one end. The inner surface of the tube has been treated with a hydrophilic lubricant coating which is activated by passing water over the inner surface of the tube. The technique of enteroscope advancement comprised of sequential push and pull manoeuvres combined with inflation and deflation of the balloon. Real-time radiological imaging confirmed the progress of the single-balloon enteroscope and the straightness of the shaft of the enteroscope.

Results: Over a three month period, 7 patients (5 females) were referred to our tertiary centre for single-balloon enteroscopy. All patients had been extensively investigated prior to referral. Mean age was 66 years old (range 42–80 years). 5 patients were referred due to recurrent symptomatic iron deficiency anaemia and 2 were referred for suspected complicated coeliac disease. The mean length of small bowel intubation was 182 cm from the incisors (range: 150–200 cm). 4 out of the 5 patients with recurrent iron deficiency anaemia had arteriovenous malformations (AVM) identified and ablated. In one of the cases, the lesion was actively bleeding on initial viewing and haemostasis was achieved at end of procedure. In one of the complicated coeliac cases, an ulcer was identified and biopsied at 150 cm. There was no complication in any of our patients.

Conclusion: Early experience with single-balloon enteroscope shows it is an effective instrument in diagnostic and therapeutic small bowel intervention. Initial experience suggests that the single-balloon enteroscope may be useful in treating lesions or obtaining biopsy from pathology identified by capsule endoscopy.

PATIENT INTOLERANCE DURING CONSCIOUS SEDATION IS A MAJOR CAUSE OF PROCEDURE FAILURE: A RETROSPECTIVE AUDIT OF ERCP IN A LONDON TEACHING HOSPITAL

N. Kumar, P. Kennedy, J. Mawdsley, M. Thursz, J. Teare, J. Hoare. Gastroenterology and Hepatology, St Mary’s Hospital, London, UK

Introduction: The BSG audit1 has provided a target for gastroenterologists performing ERCP. The society has also published sedation guidelines, recommending a maximum dose of midazolam of 5 mg.2 However, sufficiency of such low dose conscious sedation for successful ERCP is debatable, with deep sedation using propofol preferred for ERCP in the US.3 We undertook a retrospective analysis of ERCP practice in a London teaching hospital, assessing cannulation rates, technical success rates and complication rates and compared them to the BSG audit. We also examined adherence to sedation guidelines and recommendations.

Aims & Methods: Data were retrospectively collated from medical records of all ERCP procedures over a one-year period. ERCP indication, desired duct cannulation, diagnosis, therapeutic intervention and procedure failures were recorded. In addition, 30-day complication and mortality rates were calculated and sedation use recorded.

Results: 122 patients (52 male, 70 female, median age 63 years) underwent 171 procedures over 1 year. Completion of intended treatment was achieved in 99/122 (81%) undergoing their first recorded ERCP, compared with the national average (70%) in the BSG audit. Of the 122 first recorded procedures, 10 (8%) procedures were terminated prematurely due to failed sedation; 10 (8%) procedures failed duct cannulation, and only 3 (3%) procedures failed intervention. Hence with 102/122 (84%) successful duct cannulations at first recorded ERCP with 8/102 (8%) requiring pre-cuts, our overall cannulation rate of 84% is comparable to the BSG audit. There were 3 cases of ERCP related pancreatitis (3/171, 1.8%), 1 case of cholangitis (1/171, 0.6%) and one ERCP-related death of a patient with biliary sepsis, jaundice and advanced malignant disease (1/171, 0.6%). Midazolam was administered in 163/171 procedures. A median dose of 7 mg was used. In 85/163(52%) cases the recommended 5 mg limit was exceeded, compared with 33% of cases in the BSG audit. An inverse relationship between age and midazolam dose highlighted in the BSG audit was supported by our data.

Conclusion: Our cannulation rates, technical success rates, complication and mortality rates are comparable to those observed in the BSG audit. The commonest reason for an unsuccessful procedure was sedation failure. This was despite higher than the recommended dose of midazolam being used in more than 50% of cases. In response to this finding we propose more procedures to be performed under propofol sedation with anaesthetic support, with a view to improving patient experience and reducing failed procedures.

THE EFFICACY, ACCEPTABILITY AND SAFETY OF THE NEW 2 L MOVIPREP VS THE 4 L KLEANPREP IN PATIENTS UNDERGOING ELECTIVE COLONOSCOPIES

B. C. Lee, D. A. Moyes, J. C. McLoughlin, P. L. Lim. Gastroenterology, Mater Infirmorium Hospital, Belfast, UK

Introduction: Colonoscopy is an established diagnostic modality for colonic diseases and it has been increasingly utilised in colorectal cancer surveillance programme. A successful colonoscopy is very much dependent on the quality of bowel preparation. Although existing bowel preparation such as Kleanprep (Norgine Ltd, Harefield, UK) is widely used in most endoscopy units, its large volume remains a disadvantage. Therefore, there is a need for a bowel preparation with equivalent bowel cleansing properties but better patient tolerability.

Aims & Methods: To compare the efficacy, patient acceptability and safety of a new 2 l polyethylene glycol + electrolytes + ascorbic acid (Moviprep, Norgine Ltd, Harefield, UK) to the standard 4 l polyethylene glycol + electrolytes (Kleanprep, Norgine Ltd, Harefield, UK) in patients undergoing elective colonoscopies. A single-blinded randomised study was carried out at our unit over a 4-month period (May to August 2007). Patients referred for elective colonoscopy were invited to participate. Consenting patients were randomised to either Moviprep or Kleanprep for bowel cleansing. Endoscopists were blinded to the type of bowel preparation the patients had taken. The quality of bowel preparation was scored by the endoscopists at the time of colonoscopy. The acceptability and side effects of either bowel cleansing agent were recorded in a patient-filled questionnaire. Patient compliance was also reported.

Results: A total of 56 patients (28 males, 28 females) participated in the study, with a mean age of 57.9 years (age range 28–81 years). Twenty two patients were randomised to take Kleanprep while 34 patients to Moviprep. Gender distribution was comparable in both groups. The overall success rate in bowel cleansing was 82% with Moviprep and 77% with Kleanprep (p = NS). Moviprep has better taste acceptability compared to Kleanprep (70% vs 50%; p = NS). The compliance rate was significantly higher in the Moviprep group (91% vs 50%; p = 0.001). The proportion of patients taking either bowel preparation in the future, if required, is similar (85% Moviprep vs 75% Kleanprep; p = NS). Side effects were not significantly different between the two groups.

Conclusion: Moviprep has comparable bowel cleansing efficacy and safety with Kleanprep. With its higher patient compliance, due to its low volume, and better acceptability, Moviprep may be the new kid on the block in bowel preparation for elective colonoscopies.

IN CASES OF OBSCURE GASTROINTESTINAL BLEEDING, NEGATIVE CAPSULE ENDOSCOPY PREDICTS GOOD LONG-TERM OUTCOME

J. C. Macdonald, V. Porter, D. A. McNamara. Department of Gastroenterology, Aberdeen Royal Infirmary, Aberdeen, UK

Introduction: Capsule endoscopy (CE) is an established method for investigating obscure gastrointestinal bleeding (OGIB) with positive diagnostic yield reported at between 40–90%. There is little data available regarding the clinical implication and long-term outcomes for patients with a negative study in these circumstances.

Aims & Methods: Between April 2005–October 2006, 47 patients with OGIB underwent CE. All patients had previously undergone appropriate endoscopic examination of the upper and lower GI tract. 42 patients (89%) have been followed up over a mean period of 17.3±6.4 months (mean age 59.8±14 years; 45% female). 24/42 studies (57.1%) were performed in view of overt OGIB, 18/42 for occult OGIB. Case records were analysed for evidence of recurrent GI bleeding (re-investigation, blood transfusion, unexplained acute drop in haemoglobin >2 g/dl).

Abstract 168 Re-bleeding episodes during follow-up

Results: Overall diagnostic yield was 57.1% (24 examinations), the remaining 18 studies (42.9%) being normal. Patients with a positive study tended to be older (65.7±14 years positive vs 53.8±14 years negative; p<0.01) but the nature of OGIB (overt vs occult) was not a predictor of study outcome. In the group with a positive study 19/24 (79%) were found to have angiodysplasia, and 7/19 (36.8%) of these patients underwent argon plasma coagulation. There was a re-bleeding rate of only 11.1% in the negative study group, while re-bleeding occurred in 11/24 (45.8%) following a positive study (p<0.01). Both individuals who re-bled following negative CE were anticoagulated with warfarin at the time of re-bleed. During the follow-up period there was one death amongst patients with a negative study (bacterial endocarditis), compared with 3 deaths in those whose CE had positive findings (1 renal failure, 1 gastrointestinal malignancy, 1 ischaemic heart disease). There was no statistical difference in period of follow-up for either group (16.2±4.6 months positive vs 18.6±5.9 months negative).

Conclusion: Patients with OGIB and a normal CE examination appear to be at low risk of subsequent re-bleeding. These results would indicate that this group of patients do not require further routine follow-up or additional gastroenterological investigations.

RATES OF CAECAL INTUBATION AS REPORTED IN THE ABSTRACTS OF THE BSG ANNUAL MEETING

M. R. Welfare1, S. J. Mills2, L. A. Macdougall3. 1Gastroenterology, North Tyneside General Hospital, 2Surgery, Wansbeck General Hospital, 3Foundation Year 2, North Tyneside General Hospital, Tyne and Wear, UK

Introduction: Individual colonoscopists frequently achieve the target of 90% crude completion rate but published data for whole units rarely do. We review the published completion rates in abstracts at the annual meeting of the BSG between 2000 and 2007.

Aims & Methods: We hand-searched for relevant abstracts, excluding all reports with <300 cases and <5 operators. The abstracts presented data in various formats and not all gave the absolute figures so we report only the percentage success. We ascribed the data to the year in which the colonoscopies were carried out, when possible, or to the last year of the audit if year by year breakdown was not given. As far as possible we reported crude completion rates, giving the range and the mode. We also looked for audit techniques used to improve performance.

Abstract 169

Results: Before 2005 only one study reported completion rates of >90% and that was from St Mark’s (table). 50% of reports in 2005/6 achieved 90% or greater completion. The most effective audit techniques appear to be feedback on performance, although several units reported sequential improvement with no specific intervention. Improving bowel preparation also contributed. One report found that intensive education by outside experts and use of a magnetic imager did not improve completion rates.

Conclusion: There is clear evidence that general performance in colonoscopy is improving in the UK if these audits are representative of overall performance. Many units improved over time and several factors could be contributing to this including the standards promoted by JAG, the drive towards bowel screening and the threat of revalidation.

IS THE COLONOSCOPIC MEDICAL SURGICAL DIVIDE A REALITY?

L. A. Macdougall1, M. R. Welfare2, S. J. Mills3. 1Foundation Year 2, 2Gastroenterology, North Tyneside General Hospital, 3Surgery, Wansbeck General Hospital, Tyne and Wear, UK

Introduction: The Joint Advisory Group on Endoscopy (JAG) has set standards for training in colonoscopy since 1998. All trainees, regardless of discipline, should be trained to the same standards which include a 90% crude completion rate, a target which trained colonoscopists should also aim for. Previous studies have suggested a systematic difference in completion rates between physicians and surgeons. This study examines the completion rates in our institution over an 8-year period since the publication of standards.

Aims & Methods: Data were collected prospectively on endoscopy reporting systems (Endosoft and ERS) in 2 teaching District General Hospitals in Northumbria Healthcare NHS Trust from 2000–7. The discipline and grade of the endoscopist were recorded as were crude completion rates. Completion was defined as endoscopist reported caecal or terminal ileum intubation based on intention to complete with no adjustment for inadequate bowel preparation or impassable strictures.

Results: A total of 15 869 colonoscopies were recorded, the majority performed by consultants, but 1103 by clinical assistants, 318 by nurse endoscopists and 2567 by training grade doctors. The number of colonoscopies performed increased from 946 in financial year 2000/1 to 2850 in 2006/7 with an overall 8-year crude completion rate of 85.4%, improving from 78.5% in 2000/1 to 90.5% in 2007. The remainder of the analysis focuses on the 14 448 examinations performed by consultants and training grade doctors, excluding 52 cases for whom extent of examination was not recorded. Overall, consultant gastroenterologists intubated the caecum more regularly than consultant surgeons with completion rates of 4869/5550 (88.1%) and 5206/6237 (83.5%) respectively. However, the gastroenterologists started from a lower baseline, improving from 73.9 to 92.0% compared with the surgeons 80.1 to 88.1% over the same 8-year period. Completion rates of trainees improved from 79.1 to 91.6% between 2000 and 2007 and there was no overall difference between gastroenterology (1119/1409, 84.5%) and surgical trainees (983/1155, 85.1%).

Conclusion: The standard of colonoscopy in respect of completion rate has risen consistently over the last 8 years. Although the gastroenterologists performed better against this standard overall, the gap has narrowed in the most recent 2 years and there is no difference in performance amongst the training grade doctors. It may be that guidance on training from the JAG, with clearly defined objectives, joint training courses, and heightened awareness is raising and unifying standards. Examining the remaining shortfall in performance and enabling mentoring may improve colonoscopy performance as a whole.

USE AND SAFETY OF NEEDLE KNIFE SPHINCTEROTOMY AT ERCP: AN AUDIT OF 8 YEARS’ EXPERIENCE AT A DISTRICT GENERAL HOSPITAL

P. G. Mayhead, R. H. Loke, G. Greenhall, Y. Umebuani. Gastroenterology, Kent and Sussex Hospital, Tunbridge Wells, UK

Introduction: Needle knife sphincterotomy (NKS) is a method for achieving biliary access. It is used when standard cannulation is unsuccessful. Complications include hemorrhage, perforation and pancreatitis. NKS is a controversial technique due to the complications, but studies have shown it to be safe,effective and comparable to standard sphincterotomy (SS).1 2

Aims & Methods: We analysed the data of one doctor’s use of sphincterotomy techniques to detect differences in indications and complications between NKS and SS. Data were collected from 1999–2006 from reports. Patients who had a NKS were recorded and 1 in 3 patients who had SS were analysed. Complications were detected by using the hospital computer system to identify anyone who had ERCP with sphincterotomy who stayed more than or equal to two days following the procedure and the notes were examined for complications. Pancreatitis was defined using recent BSG Guidelines.3 Indications and findings were recorded from the documentation of each ERCP report.

Results: From January 1999 to December 2006, 1081 ERCPs were performed averaging 135 cases per year. 148 NKS and 546 SS were performed. Notes were reviewed for 80% of the NKS patients and 82% patients selected in the SS group. Repeat ERCP was required following NKS in 10 cases and was successful in 9. There was no difference in indication for ERCP between each group. The overall complication rate was 12.7% for NKS (5% pancreatitis, 2.5% bleeding) and 2.2% for SS (0.55% pancreatitis). There were 2 deaths within 10 days in the NKS group (1/540) from bleeding, both in their 80s.

Conclusion: Our complication rates are similar to previously published reports. In a similar study in 2004, complication rates from NKS were 7% for haemorrhage and 4% for pancreatitis.4 Our complication rate of 12.7% for NKS is comparable to other centres performing high volumes of ERCP (12% complication rate, 5.5% bleeding and 0.5% pancreatitis).5 There were very few complications in the SS group. Overall, the indications for ERCP in which each modality of sphincterotomy was used was similar implying that NKS is a safe alternative to SS and is a highly effective method of gaining biliary access in patients in whom deep cannulation proves difficult or impossible and biliary access is considered essential.

Gastrointestinal pathology posters

THE REGULATION OF CELLULAR IRON IS DEFECTIVE IN APC MUTANT COLORECTAL CARCINOGENESIS

M. J. Brookes, K. Roberts, T. Iqbal, C. Tselepis. Cancer Studies, University of Birmingham, Birmingham, UK

Introduction: The tight regulation of cellular iron levels in normal epithelial cells is mediated through the interaction of iron regulatory proteins (IRPs) with iron response elements found in many genes implicated in iron metabolism. In instances of excessive iron levels IRE/IRP signalling causes a repression in iron import proteins including TfR1 and DMT1. However, in colorectal cancers we have previously demonstrated high iron import protein expression in the face of high cellular iron levels suggesting an aberration in IRE/IRP signalling.

Aims & Methods: The aim of this study is to determine if iron mediated Wnt signalling is a crucial in the aberration of iron sensing and in particular address the impact of c-myc. Colorectal lines RKO and SW480 were cultured with or without iron; the expression of iron transport proteins and Wnt-targets examined by Q-RT-PCR and Western blotting. C-myc and IRP2 over expression studies were performed in RKO cells followed by ferrozine assays for iron determination.

Results: Our results demonstrate that in RKO cells (APC wild-type) iron does not induce Wnt signalling and cells behave in an IRE/IRP dependent manner. However in SW480 cells (APC mutant) iron loading results in Wnt signalling and inductions in iron import proteins (TFR1 and DMT1) which were associated with increased c-myc expression. This observation was verified by c-myc over expression studies where expression of the iron import proteins TfR1 and DMT-1 were both induced as a consequence of c-myc.

Conclusion: Wnt signalling and in particular c-myc is crucial in interfering with normal IRE/IRP signalling.

CELL SURFACE BINDING AND INTERNALISATION OF CLOSTRIDIUM DIFFICILE TOXIN A BY HUMAN INTESTINAL EPITHELIAL CELLS

K. Solomon, A. Robins, Y. R. Mahida. Institute of Infection, Immunity and Inflammation, University of Nottingham, Nottingham, UK

Introduction: C difficile causes colitis via secreted toxins A and B. Epithelial cells are likely to be the first host cells that bind the luminal toxins but the nature of these interactions are currently undefined.

Aims & Methods: To investigated binding and internalisation of toxin A. Purified toxin A was fluorescently labelled and unincorporated dye removed by size-exclusion chromatography. The labelled toxin retained biological activity (cell rounding) in Vero and Caco-2 cells. Specificity of fluorescence following application of labelled toxin was confirmed by its loss when pre-incubated with beads coated with toxin A-specific monoclonal antibody PCG-4. To control for batch-to-batch and experiment-to-experiment variation, beads with different (but known/defined) PCG-4 binding capacity were used in each experiment to create a standard curve of fluorescence, after application of labelled toxin A to PCG-4-coated beads. From the standard curve, fluorescence of cells after application of labelled toxin A was used to calculate toxin A fluorescence units (TAFU). Labelled toxin A binding to the surface of Caco-2 cells was studied on ice after 1 min exposure. Internalisation of labelled toxin A was studied by flow cytometry and confocal microscopy, after 2 h and 5 h continuous exposure at 37°C. Data are expressed as mean (±SEM).

Abstract 173

Results: Caco-2 cell surface fluorescence was rapid (within 20 s) following application of labelled toxin A and was constant over the subsequent 5 min. When exposed to unfixed Caco-2 cells at 37°C, internalised toxin A was seen within the cell cytoplasm by confocal microscopy. Analysis by flow cytometry (table) showed a concentration-dependent increase in cell surface-associated TAFU (p<0.05). Internalised TAFU increased in a concentration and time-dependent manner. There was an inverse relationship between the concentration of labelled toxin A applied and the calculated proportion of cell surface-associated TAFU that was internalised at 2 h (2 µg/ml–60.7%, 5 µg/ml–33.9%, 10 µg/ml–24.2%) and 5 h (2 µg/ml–121.6%, 5 µg/ml–75.4%, 10 µg/ml–43.9%). Surface-associated TAFU was significantly greater than internalised TAFU after 5 h incubation with 5 and 10 µg/ml labelled toxin A (p<0.05).

Conclusion: (1) C difficile toxin A rapidly associates with the surface of human intestinal epithelial cells in a concentration-dependent manner. (2) There is an inverse relationship between the concentration of toxin A applied and the proportion of cell-associated toxin that is internalised. (3) The above could be explained by a concentration-dependent increase in non-specific binding of toxin A to the epithelial cell surface.

MALE PREDOMINANCE OF UPPER GASTROINTESTINAL CANCER IS RELATED TO INTESTINAL HISTOLOGICAL SUBTYPE NOT TUMOUR LOCATION

M. H. Derakhshan1, S. Liptrot2, J. Paul2, I. L. Brown2, K. E. L. McColl1. 1Medical Sciences, 2Division of Cancer Science and Molecular Pathology, University of Glasgow, Glasgow, UK

Introduction: Upper gastrointestinal adenocarcinomas show an unexplained male predominance which is more apparent in the oesophagus than stomach. We have conducted a population-based study to determine whether the gender phenomenon is primarily related to the anatomical site or the histological subtype of the tumour.

Aims & Methods: The study was based upon 3270 gastric and oesophageal cancers recorded in West of Scotland cancer Registry between 1998 and 2002. 812 of these cases were randomly selected for detailed analysis. Their records were reviewed for histological diagnosis and anatomical site of tumour using ICD-10 and ICD O-2. The Lauren histological subtype of adenocarcinoma was determined by reviewing 1204 original reports and 3241 biopsy and surgical slides. Logistic regression models were used to estimate relationship between male predominance and histological subtype, tumour location and age.

Abstract 174

Results: 405 (51.5%) cancers originated from the non-cardia region of the stomach, 173 (22.0%) from the gastric cardia and 209 (26.6%) from the oesophagus. Regardless of anatomical subsite, the upper GI cancers were more common in males (502, 63.8%) than females (285, 36.2%). Regardless of anatomical site, the crude incidence rate of intestinal type upper GI adenocarcinoma was higher in males (23.86/100 000/year) versus females (9.00/100 000/year), giving M/F of 2.65. M/F ratio of intestinal type cancer was 3.41 at age <50, reached a peak of 7.86 at age 50–59, and then showed a progressive decrease with a minimum of 2.29 at age group ⩾80 years. In contrast, the crude incidence rate of diffuse-type adenocarcinoma was similar in males and females (5.58 vs 5.20/100 000/year) yielding M/F of 1.07. Male predominance was greatest in the oesophagus with M/F of 3.50, less in the cardia (M/F = 2.00) and least in the non-cardia cancer (M/F = 1.65). The relationship between tumour location and male predominance lost its significance in the multivariate analysis when histological subtype was added (OR 1.37, 95% CI 0.88 to 2.12). Multivariate analyses including histological subtype, tumour location and age indicated that the male predominance was related to the histological type rather than anatomical location. Intestinal type adenocarcinoma showed similar male predominance of incidence irrespective of its anatomical location (OR 2.6, 95% CI 1.78 to 3.9).

Conclusion: Male predominance in upper GI adenocarcinomas is not a function of tumour location (table) but rather related to histological subtype. The mechanism of the strong gender association with intestinal subtype requires elucidation.

MALE PREDOMINANCE OF UPPER GASTROINTESTINAL CANCER IS DUE TO A 17.2-YEAR DELAY OF INTESTINAL HISTOLOGICAL SUBTYPE IN FEMALES

M. H. Derakhshan1, S. Liptrot2, D. Morrison3, I. L. Brown2, K. E. L. McColl1. 1Medical Sciences, 2Division of Cancer Science and Molecular Pathology, 3Scottish Universities Environmental Research Centre, University of Glasgow, Glasgow, UK

Introduction: We have recently demonstrated that the male predominance of upper GI adenocarcinoma is associated with the intestinal histological subtype and is independent of anatomical location—that is, oesophageal versus gastric. We have undertaken kinetic analysis of age specific incidence trends to determine the effect of the gender phenomenon.

Aims & Methods: The study was based on 812 cases of upper gastrointestinal adenocarcinoma randomly selected from the West of Scotland Cancer Registry between 1998 and 2002. The Lauren histological subtype of adenocarcinoma was determined by reviewing 1204 original reports and 3241 biopsy and surgical slides. A regression curve modelling was used to calculate age-specific curve equations of both diffuse and intestinal subtypes of cancer.

Results: The analysis compromised 405 non-cardia cancer, 173 cardia cancer and 209 oesophageal adenocarcinoma. The crude incidence rate of intestinal type upper GI adenocarcinoma was higher in males (23.86/100 000/year) versus females (9.00/100 000/year), giving M/F of 2.65. M/F ratio of intestinal type cancer was 3.41 at age less than 50, reached a peak of 7.86 at age 50–59, and then showed a progressive decrease with a minimum of 2.29 at age group 80 years and over. In contrast, the crude incidence rate of diffuse-type adenocarcinoma was similar in males and females (5.58 vs 5.20/100 000/year) yielding M/F of 1.07. Curve modelling of the age-specific rates for diffuse subtypes showed similar equations, y = 0.016×mean age 2.007+32.98, R2 = 0.999 and y = 0.016×mean age 1.954+35.97, R2 = 0.989, for male and females, respectively. For the intestinal histological type, the age-specific incidence rates were different for males and females. Curve modelling indicated the same kinetic equation for rise in incidence with age, in males (y = 0.016×mean age 2.315+28.75, R2 = 0.990) and in females (y = 0.016×mean age 2.316+46.07, R2 = 0.998). However, the female curve was displaced to the right indicating a delay in the cancer onset. The estimated cancer onset delay for females was 17.2 years (that is, +46.07 minus +28.75 = 17.22).

Conclusion: Male predominance in upper GI adenocarcinomas is a reflection of approximately 2 decades delay of development of intestinal subtype adenocarcinoma in females. The mechanism of the marked delay in development of intestinal subtype requires elucidation.

SHOULD WE BIOPSY ANTRAL GASTRITIS TO DETECT INTESTINAL METAPLASIA?

S. S. Samji1, B. Nedjat-Shokouhi1, N. Alband2, A. D. Millar1. 1Gastroenterology, North Middlesex University Hospital, 2Gastroenterology, Queen Mary School of Medicine, London, UK

Introduction: Antral gastritis is a common finding at endoscopy often caused by H pylori.1 In NHS practice biopsies from mild-moderate antral gastritis are usually taken for CLO test but not histology, potentially missing intestinal metaplasia; which has proven carcinogenic potential.2 We reviewed results of 272 endoscopies comparing macroscopic appearances of the antrum, CLO test results and antral histology to establish the rate of intestinal metaplasia and associated gastric pathology.

Aims & Methods: We reviewed endoscopy reports from 1999–2007 in patients who had undergone endoscopy and were >18 with no previous gastric surgery, and if antral histology, a CLO test and an image of the antrum were available for analysis. Demographics, symptoms, diagnosis, antral appearance, histology and CLO-test results were recorded. The patients were drawn from a single endoscopist’s practice in two independent hospitals. In all patients at least two random antral biopsies were taken for CLO test and for histology.

Abstract 176

Results: The age range was 17–93, median 53 years, with a male/female ratio of 1:1.2. 35 patients (13%) had a positive CLO test and 43 patients (16%) had gastritis on histology, of which only 13 had H pylori detected by CLO test or on histology; 3 had intestinal metaplasia, only 1 of which was positive for H pylori. As expected a positive CLO was strongly linked with GU or DU (p = 0.005) and dyspepsia (p = 0.04).

Conclusion: This study confirms that the common finding of gastritis is strongly linked with H pylori and GU or DU3 though ulcers were not commonly found probably because primary care practitioners follow national “test and treat” guidelines. CLO positive inflammation was therefore also uncommon in the study group. Three biopsies showed intestinal metaplasia which would not have been detected under common practice and many of these may be H pylori negative. Further studies are needed to establish the indication for antral biopsy to detect intestinal metaplasia and the effectiveness of surveillance of these patients.

ENHANCED BIOMEDICAL SCIENTIST CUT-UP ROLE IN COLONIC CARCINOMA: PRELIMINARY PERFORMANCE DATA AND COMPARISON WITH DEPARTMENTAL PERFORMANCE

E. J. V. Simmons1, D. S. A. Sanders1, A. Smith1, P. Murphy2, M. Osbourne2, K. Busby2, M. Stellakis2, J. Francombe2. 1Department of Histopathology, 2Department of Surgery, Warwick Hospital, Warwick, UK

Introduction: The extended Biomedical Scientist (BMS) role in surgical cut-up is increasing in importance and is likely to continue to do so. It allows more efficient use of Consultant time and resources and provides additional career progression for a highly skilled workforce. As the remit of BMS cut-up expands to include major cancer resection specimens, it becomes essential, on grounds of quality assurance, to apply the same performance criteria applied to specimens dealt with by Consultant and trainee pathologists. These are included in the recently published second edition of the RCPath colorectal cancer dataset.3

Aims & Methods: To compare performance data between colorectal cancer (CRC) cases trimmed by a pathologist, by our BMS in his extended practice role, RCPath guidelines and other published series. After preliminary review, CRC specimens of appropriate complexity are allocated to the BMS. The Consultant pathologist and the BMS then examine the specimen together, discussing block selection and any other areas of concern specific to the case. The Consultant is available for further advice and discussion at any point during trimming. The BMS then reviews the slides of the case plus their performance data in conjunction with the Consultant. Departmental figures are derived from 400 colorectal cases trimmed predominantly by a single Consultant Pathologist, with smaller numbers trimmed by trainee patholgists. The cases trimmed by our BMS are included in this series. At the time of abstract submission our BMS had trimmed 21 selected cases (colonic only).

Abstract 177 Comparison of performance data

Results: BMS cut-up performance meets RCPath guidelines and is comparable to overall departmental performance. Of note, our figures fall short of the exceptional results published by Peterson et al.2

Conclusion: The performance data from the cases handled by our BMS support the continuation of the BMS extended role within our department, with analysis of further data as it becomes available. The data collected, namely mean lymph node harvest and frequency of extramural vascular invasion and serosal involvement, have been demonstrated to be important in determining patient prognosis and appropriate adjuvant therapy.1 2 4

EGFR, HER-2, HER-3 AND HER-4 RECEPTOR EXPRESSION IN NEUROENDOCRINE TUMOURS

R. Srirajaskanthan1, T. Shah1, J. Watkins2, L. Marelli1, A. Quaglia2, K. Khan2, R. Sim2, M. Caplin2, D. Hochhauser3. 1Gastroenterology, 2Histopathology, Royal Free Hospital, 3Oncology, University College Hospital, London, UK

Introduction: The HER family is comprised of four distinct receptors: EFGR, HER-2, HER-3 and HER-4. These are transmembrane receptors composed of an extracellular ligand-binding domain and a cytoplasmic region with enzymatic activity. Over-expression of HER family receptors is associated with reduced survival in patients with breast cancer, colon and ovarian cancer.

Aims & Methods: In this study we sought to determine the pattern of expression and the prognostic significance of EGFR, HER-2, HER-3 and HER-4 in neuroendocrine tumours. Consecutive samples of formalin-fixed paraffin-embedded tumour tissue were available from 81 patients with a histologically confirmed diagnosis of neuroendocrine tumour (NET). The study population included all major NET subtypes including: foregut, mid-gut, hindgut, bronchial, paraganglionoma and NETs of unknown primary. Tumours were classified according to their site of origin, level of differentiation and their initial mitotic index. Scoring was based on intensity of staining whereby 0 = negative, 1 = weakly positive, 2 = moderate, 3 = strongly positive. Tumour staining was also scored, whereby, positive staining was marked as: 1 = <25%, 2 = 25–75% and 3 = >75%. The product of the density of staining and the percentage of tumour cells staining positive was used as the histological score, giving final values of 0, 1, 2, 3, 4, 6, 9. Score of ⩽2 were classed as negative.

Results: EGFR expression was identified in 69 of 81 (85.1%) samples, HER-2 was not expressed in any tissue, HER-3 was expressed in 7 samples (3 pancreatic, 3 paraganglionoma and 1 mid-gut), HER-4 was expressed 74 (91%). EGFR negative tumours were associated with a significantly worse prognosis than EGFR positive tumours (p = 0.014). There was no significant association between expression of HER-4 or HER-3 and survival. No significant correlation between EGFR and HER-4 expression.

Conclusion: EGFR and HER-4 are significantly expressed in NETs. EGFR and HER-4 share many the same ligands, this in part may explain the expression of these two receptors in NETs. The expression of EGFR does not appear to be a negative prognostic indicator.

THE ROLE OF ALPHA FETOPROTEIN AND HUMAN CHORIONIC GONADOTROPHIN-β; AS PROGNOSTIC TUMOUR MARKERS IN PATIENTS WITH NEUROENDOCRINE TUMOURS

R. Srirajaskanthan1, T. Shah1, M. Bhogal1, C. Toumpanakis1, T. Meyer2, A. Noonan1, T. Amin1, C. Witney- Smith1, S. Niroshan1, W. Ben1, D. Hochhauser3. 1Gastroenterology, 2Oncology, Royal Free Hospital, 3Oncology, University College Hospital, London, UK

Introduction: Serum chromogranin A is the best general tumour marker available for neuroendocrine tumour (NET) patients. The role for other tumour markers is less clear.

Aims & Methods: To determine the diagnostic and prognostic value of serum alpha-fetoprotein (AFP) and human chorionic gonadotrophin β (hCG-β) in NETs. A database containing biochemical, histological including Ki67 proliferation index, and survival data on 360 NET patients was constructed using an excel spreadsheet. These data were statistically assessed, using SPSS statistics package, to determine the utility of commonly measured tumour markers with particular emphasis on AFP and hCG-β.

Results: AFP and hCG-β were raised in 9.5% and12.3% of patients respectively. AFP levels associated strongly and positively with tumour grade, serum CgA, hCG-β levels and worse survival (mean 37.6 vs 69 months for AFP-high and control groups respectively; p = 0.001). hCG-β levels also associated strongly and positively with serum CgA, AFP levels and worsening survival (mean 48 vs 57.3 months for hCG-β high and control groups respectively; p = 0.037).

Conclusion: Both AFP and hCG-β are elevated in high-grade neuroendocrine tumours with a rapidly progressive course and poorer survival. They also correlate with chromogranin-A, which is known to be a marker of tumour burden and to have prognostic value. Thus AFP and hCG-β are clinically important in NETs and should be routinely measured.

TUMOUR GROWTH ASSESSMENT USING MINIMUM SPANNING TREE ANALYSIS IN PANCREATIC, AMPULLARY AND DISTAL BILE DUCT CANCER

C. S. Verbeke1, K. V. Menon2, J. Knapp3. 1Histopathology, 2Surgery, St James’s University Hospital, 3School of Physics and Astronomy, University of Leeds, Leeds, UK

Introduction: The definition of microscopic resection margin involvement (R1) in large bowel cancer (LBC) has been established as tumour cells <1 mm to the margin. This R1 definition is also used for the reporting of pancreatic head cancer specimens, but has never been validated. The Minimum Spanning Tree algorithm (MSTA) is a computer based method that analyses the topology of cancer growth patterns by determining the minimum distance between tumour cells.

Aims & Methods: To determine, using the MSTA, whether a similar growth pattern in LBC and cancer arising in the pancreatic head justifies the current R1 definition for reporting of the latter. In 10 cases of each LBC, pancreatic (PC), ampullary (AC) and distal bile duct cancer (DBC) a 200× picture was taken in the centre and periphery (at the invasive front) of the tumour as well as half way between both. For each picture, the distances between cancer cells were determined using the MSTA. The average minimum distance was compared between the 3 zones in the 4 cancer groups, as was the tumour cell density.

Results: In all cancer groups, the distance between cancer cells was comparable in the central zone, and larger in the periphery than in the centre of the tumour (p⩽0.02). However, this difference between centre and periphery was smaller in LBC than in PC, DBC and AC (p⩽0.03). Results were similar for PC and DBC, but differed from those of AC (p⩽0.03). On average, the tumour cell density dropped in the periphery of PC to 30% of that in the tumour centre, and only to 83% in LBC (p<0.0001), 62% in AC and 40% in DBC (p⩽0.01).

Conclusion: Tumour cell growth in PC, DBC and—to a lesser extent—AC is less dense than in LBC. Particularly in the periphery, tumour growth is more dispersed in PC, DBC and AD, and tumour cell density drops to as little as 30% at the invasive tumour front. This difference in tumour growth demands revision of the R1 definition for pancreatic head cancer. The MST algorithm is a useful tool to assess the topology of cancer growth patterns.

Gastrointestinal physiology posters

OVERWEIGHT AND WAIST CIRCUMFERENCE ARE RISK FACTORS FOR GASTRO-OESOPHAGEAL REFLUX DISEASE

R. Anggiansah1, R. Sweis2, T. Wong2, M. Fox1, A. Anggiansah1. 1Oesophageal Laboratory, 2Gastroenterology, St Thomas’ Hospital, London, UK

Introduction: The increasing prevalence of obesity and gastro-oesophageal reflux disease (GORD) strongly suggests an association between these two conditions. Previous studies indicate that waist circumference (WC) but not body mass index (BMI; kg/m2) is independently associated with oesophageal acid exposure.

Aims & Methods: This study aimed to examine the relationship between increasing WC and specific reflux parameters on 24-h pH monitoring to provide insight into the mechanism by which this occurs. A prospective study of 676 consecutive patients with suspected GORD referred for oesophageal investigation. Height, weight and waist circumference (WC) were measured. 24-h ambulatory pH studies were performed. The associations between demographic variables, reflux parameters (oesophageal acid exposure (%time-pH<4), number of reflux episodes (NR), longest reflux episodes (LR)) and the DeMeester score) was assessed in four time periods (total, upright, supine and postprandial) by univariate and multivariate analysis. In addition, symptom severity was assessed by a validated questionnaire.

Results: Complete 24-h pH data for 583 (86%) patients with M:F 44.%:56% were available. The prevalence of obesity was similar in both sexes. %time-pH<4 and LR were associated with increasing age in all time periods (all with p<0.001); however no such effect was seen for reflux symptoms (p>0.1). WC (adjusted for age and sex) but not BMI was independently associated with a significant increase in all reflux parameters during all four time periods (all p<0.001). There was a progressive increase in reflux parameters (%time-pH<4, NR and LR) from the low (WC <89 cm) through the medium (89–99 cm) to the high tertile group (WC >99 cm) which was significant in all time periods (all p<0.001). Similarly, the odds ratios for pathological reflux (DeMeester >14.7) adjusted for age and sex compared to patients with a low tertile WC was elevated in the medium tertile (OR 1.58, CI 1.02 to 2.44; p<0.001) and the high tertile groups (OR 3.36, CI 2.14 to 5.28; p<0.001). WC was not associated with the severity of reflux symptoms independent of oesophageal acid exposure (p = 0.1).

Conclusion: In patients referred for investigation of reflux symptoms WC is associated with increasing the number and duration of reflux episodes and overall oesophageal acid exposure, in all study periods. This indicates that obesity increases the severity of reflux disease both by increasing the frequency of reflux episodes in both the upright position likely by increasing the risk of reflux during transient lower oesophageal sphincter relaxation, but also in the supine position likely due to weak and incompetent reflux barrier and impaired clearance.

INTER-INDIVIDUAL VARIATIONS IN THE GUT-TO-BRAIN SIGNALLING RESPONSES TO HEXOSE SUGARS

A. Gopinath, T. Little, J. T. McLaughlin, D. Lassman, S. Rhodes, D. G. Thompson. GI Sciences, Hope Hospital, Manchester, UK

Introduction: The incidence of obesity is increasing rapidly. The increased consumption of fructose has been implicated in the pathogenesis of obesity. Therefore, understanding the mechanisms by which different sugars influence the gastrointestinal signaling to the brain, and thus the central regulation of appetite, is of paramount importance. Sugars are proposed to slow gastric emptying (GE), a proxy measure of brainstem activation, by stimulating osmoreceptors in the small intestine; however, glucose and fructose appear to empty differently.1 Furthermore, large inter-individual differences in the GE response to sugars were apparent.1 Previous studies have been unable to explore gastric emptying responses in detail within an individual due to limitations in methodology, therefore, it is unclear whether the reported differences are a true representation of an individuals responsiveness.

Aims & Methods: Primary aims of this study were, therefore, to determine: (1) the effects of hexose sugars, glucose, fructose and galactose on GE and (2) whether inter-individual differences in GE were consistent within individuals. 7 healthy subjects were studied after an overnight fast on 10 separate occasions to evaluate the effects of 500 ml solutions of glucose, galactose and fructose at 250, 500 and 1000 mOsmol, or water on GE (assessed using a 13C-acetate breath test). Subsequently, 3 subjects underwent 32 studies to evaluate intra-individual reproducibility in the GE of 250 mOsml hexose sugars.

Results: As the osmolality of the hexoses increased, GE was more potently slowed (p<0.001). At a low osmolality (250 mOsmol), a sugar-specific effect was apparent; the effects of fructose were more potent than those of glucose and galactose (p<0.05). After observing substantial inter-individual differences in the response to sugar ingestion at 250 mOsmol, we investigated whether the GE response within an individual was consistent. In Subject A, but not subjects B and C, fructose consistently slowed GE (p<0.05).

Conclusion: Hexose sugars slow gastric emptying in both an osmolality-dependent and sugar-specific fashion. At a low osmolality (250 mOsmol), fructose emptied from the stomach more slowly than both glucose and galactose; as the osmolality increased, the difference between the sugars decreased. Furthermore, while the effects of glucose and galactose on gastric emptying were relatively consistent between individuals, there was substantial inter-individual variation in fructose at 250 mOsmol; some individuals did not slow gastric emptying in response to fructose. Understanding the mechanisms underlying the inter-individual variations in the gastric emptying response to fructose may help to explain why some individuals are more susceptible to obesity in an obesogenic environment.

The below abstract is now an oral presentation in the Gastrointestinal Physiology Associates Group on Wednesday 12 March

DO PATIENT’S REPORTED FOOD TRIGGERS REALLY PROVOKE GASTRO-OESOPHAGEAL REFLUX AND HOW WELL ARE WE ADVISING THEM?

D. Vasant1, M. Deakin2, C. Cheruvu2, S. Evans1, F. C. Leslie1. 1Gastroenterology, 2Upper GI Surgery, University Hospital of North Staffordshire, Stoke-on-Trent, UK

Introduction: Different food intolerances and lifestyle practices are thought to trigger reflux. Foods act both as direct irritants or indirectly either pharmacologically or hormonally on the lower oesophageal sphincter—for example, fats. Lifestyle advice is important, but we hypothesised that there would be a wide variation in how effectively this was given.

Aims & Methods: A questionnaire was developed for gastro-oesophageal reflux disease (GORD) patients awaiting GI physiology appointments, asking about advice patients received, which foods precipitate their symptoms and nominated one food as a “test meal”. The post-prandial effect of the “test meal” on the oesophageal pH was analysed at 24-h pH/manometry.

Results: Data were obtained from 45 of 66 patients and 48 trigger foods identified. The top 4 trigger foods were fatty foods 76%, fried foods 71%, spicy foods 71% and pastry 69%. 75% (34) reported frequent heartburn (daily/several times/week), where the mean no of “trigger” foods was 14.0/patient, compared to those with infrequent heartburn (n = 11), mean 7.2. Despite referral for pH studies, only 71% recalled lifestyle advice. 63% to raise head of bed, 53% to eat regular small portions, 50% weight loss, 44% to avoid trigger foods, 25% to avoid late meals and 16% to reduce alcohol. In pH studies (n = 18), test meals chosen by patients included pastry (6), spicy food (4), chocolate, spearmint and muesli. Two patients had normal pH (Demeester <14.72). 88% (14/16) who had physiological reflux had increase in %time pH<4 2 hours post test meal, compared to total during 24-h study. Significant reflux was demonstrated in 5/6 patients who ingested pastry and 4/4 after spicy food test meals. Muesli and spearmint did not induce physiological reflux.

Conclusion: Reflux sufferers reported a large variety of “trigger” foods. Patients with more frequent reflux symptoms report double the amount of food triggers compared to those with infrequent heartburn. 88% of patients who ingested a food they believed to trigger their symptoms during 24-h pH testing, actually had a refluxogenic episode post-prandially. Better provision of consistent information about lifestyle and dietary measures could improve patient outcome.

WHAT DO GASTROENTEROLOGY TRAINEES KNOW ABOUT OESOPHAGEAL MANOMETRY AND PH MONITORING?

J. D. Shutt1, F. Robbie1, P. Boger2, A. Quine1. 1Gastroenterology, Queen Alexandra Hospital, Portsmouth, 2Gastroenterology, Southampton General Hospital, Southampton, UK

Introduction: Oesophageal manometry and pH monitoring is widely available for the investigation of oesophageal symptoms. However, trainees are frequently not exposed to these techniques. In November 2006 the BSG issued updated guidelines for oesophageal manometry and pH monitoring, acknowledging the advent of PPIs and their value in therapeutic trials.

Aims & Methods: We aimed to evaluate the knowledge of oesophageal manometry and pH monitoring against key recommendations made in the BSG guidelines, among gastroenterology SpRs in the Wessex region. All received a 25-point questionnaire that was completed electronically.

Results: 18/30 (60%) responded. One trainee worked in a hospital without oesophageal manometry and pH monitoring onsite. Knowledge of oesophageal manometry was good, although 22% considered it a first line investigation in patients with suspected oesophageal symptoms. Regarding ambulatory oesophageal pH monitoring: 44% wrongly considered it to have a very low false negative rate; 39% considered prior endoscopy necessary for accurate electrode placement; and while all knew that it allows correlation between reflux events and symptoms, only 41% understood the ideal time. Trainees overestimated normal oesophageal acid exposure, with mean total, upright, and supine times for oesophageal pH<4 for less than 11%, 7.6% and 13.4% respectively (5, 8 and 3% in the guidelines). Only 1 trainee was aware that the degree of supine reflux should be less than that upright, other responses ranged from 5–50%. The majority appreciated that pH monitoring should be the investigation of choice in patients with symptoms suggestive of gasto-oesophageal reflux disease (GORD) (89%) and in cases with atypical symptoms possibly due to GORD, a prolonged therapeutic trial of a PPI is considered more relevant (83%). The most confusion concerned having pH monitoring on or off a PPI in GORD: 44% correct in testing on a PPI in a failed therapeutic trial; 39% correct in testing off a PPI in endoscopy negative oesophagitis with a good response to PPI prior to anti-reflux surgery. In a patient with symptoms potentially due to GORD with a failed therapeutic trial, no trainee was correct in advising a pH study on a PPI prior to anti-reflux surgery. Knowledge of procedural issues were satisfactory, although only 33% and 11% appreciated the restrictions on diet and activity respectively, and only 44% were aware of the need for antibiotic prophylaxis if high risk of bacterial endocarditis.

Conclusion: The working knowledge of oesophageal manometry and pH monitoring among trainees was adequate at best, with limited in-depth knowledge and confusion regarding performing pH monitoring on or off PPIs. To ensure trainees refer patients appropriately, further training and exposure to these techniques is required.

Liver posters

ASSESSMENT OF “NUCLEATION TIME” AS A GOOD PREDICTOR OF CHOLELITHIASIS

V. Abeysuriya1, K. I. Deen1, S. K. Kumarage1, N. M. M. Navaratne2. 1Surgery, Faculty of Medicine, Ragama, University of Kelaniya, Sri Lanka, 2Medicine, National Hospital of Sri Lanka, Colombo, Sri Lanka

Introduction: In the formation of gallstones, crystal nucleation is the key step, which is followed by precipitation and gradual growth of cholesterol crystals.

Aims & Methods: A case control study was performed among 60 subjects (30 patients, 14 males: 16 females, median age 36 years, range 33–71 years, BMI = 25.1 +/−0.33 kg/m2, who underwent laparoscopic cholecystectomy; 30 controls, 15 males: 15 females, median age 38 years, range 33–70 years, BMI = 24.5+/−0.23 kg/m2, who underwent laparotomy and who had normal ultrasound scans of the gallbladder and no demonstrable stones). Bile aspirated from the common bile duct was ultrafiltered and anaerobically incubated at 37°C. Incubated bile was examined daily by polarised light microscopy (PLM), for appearance of cholesterol crystals. Nucleation time (NT) of bile was assessed as the time taken for the first crystals to appear under PLM.

Results: Controls were not different to cases studied. The overall mean NT was significantly shorter in patients versus controls (mean NT, patients: 1.76+/−0.2 SEM days; vs controls: 12.74+/−0.4 SEM days, p = 0.001). Of controls, females demonstrated a shorter NT compared with males (mean NT, females, 11.4 +/−0.36 SEM days; vs males, 14.1 +/−0.46 SEM days, p = 0.006). By contrast, there was no gender difference in NT in patients (mean NT, females: 1.7+/−0.24 SEM days; vs males: 1.8+/−0.2 SEM days, p = 0.7).

Conclusion: Overall, females tend to have shorter nucleation time compared with males. However, in those with demonstrable gallstones, there appears to be no significant difference in bile nucleation time.

EXPERIENCE OF A NURSE LED HEPATOMA SCREENING CLINIC

L. Bailey1, P. Dundas1, C. Clark2, A. Fraser1. 1Department of Gastroenterology, 2Department of Radiology, Aberdeen Royal Infirmary, Aberdeen, UK

Introduction: Hepatocellular cancer (HCC) is the fourth most common cancer in the world with the majority of cases having cirrhosis and/or chronic hepatitis B (HBV) infection. Once symptomatic, cure is seldom possible and prognosis is poor. Surveillance is recommended in high risk patients and aims to detect early tumours where curative therapy may be offered. Alphafetoprotein (AFP) is inadequate as a screening tool alone as the sensitivity is 60% and therefore ultrasonography with a sensitivity of between 65% and 80% may be used in combination with AFP. The ideal surveillance interval is unknown, an interval of 6–12 months has been proposed based on tumour doubling times.1 The clinic was established in September 2005. Patients with cirrhosis or at a high risk of HCC (with HBV infection) were referred to the clinic primarily from one consultant. The clinic operates twice monthly with 10 ultrasound appointments and nurse review/phlebotomy on the same day. A 6-month interval between appointments was selected. Although the clinic started in September 2005, there was an initial 6-month period before activity was monitored.

Aims & Methods: Details of patients with a diagnosis of cirrhosis were recorded prospectively into a database. An audit of all patients on the database and clinic attendance outcomes from March 2006–September 2007 were evaluated.

Results: As of September 07 there were 245 patients with cirrhosis or a diagnosis of chronic HBV infection on the database with 99 patients (40%) attending the clinic. Reasons for not offering appointments were; age >70 (depending on fitness) (30%), geographic distance from centre (23%), recurrent non-attender (20%), consultant discretion (primarily due to ongoing chaotic substance misuse) (19%) and lost to follow-up (8%). Of those being screened, the mean age is 54 years (range 21–76) and 68 (68%) are male. Since March 2006, there have been 270 appointments offered with 227 (85%) attending, 20 (6.5%) appointments were lost due to patients who could not attend giving prior notification and only 23 (8.5%) when the patient did not attend without prior notification. Since the clinic was established, there have been 3 new diagnoses of hepatocellular carcinoma. Patient A tumour size 1.5×6 cm required chemoembolisation awaiting transplant; Patient B tumour size 1 cm required surgery and chemoembolisation; Patient C tumour size 3.5×5.1 cm required surgery. All tumours have been diagnosed at the initial clinic visit with no tumours diagnosed through 6-monthly follow-up.

Conclusion: The clinic was well attended but this may be due to our selection criteria. The clinic has only operated for 2 years and the ability to identify tumours at an early stage requires further evaluation.

PHYSICAL ACTIVITY AND REDUCED RISK OF SYMPTOMATIC GALLSTONES: A PROSPECTIVE COHORT STUDY

P. J. Banim1, S. Bingham2, R. Luben2, N. Wareham2, A. Welch3, K. Khaw4, A. Hart3. 1Gastroenterology, Norfolk and Norwich University Hospital, Norwich, 2Institute of Public Health, University of Cambridge, Cambridge, 3School of Medicine, University of East Anglia, Norwich, 4Clinical Geronotology Unit, University of Cambridge, Cambridge, UK Kingdom

Introduction: Physical activity may protect against gallstone formation by reducing bile stasis and plasma triglycerides and by raising HDL levels. No prospective studies have previously investigated this hypothesis in both genders using a physical activity questionnaire validated against physiological measurements. The aim of this study was to investigate this relationship using a questionnaire to provide a simple global index of physical activity and to determine if physical activity predicts disease independently of known risk factors for gallstones.

Aims & Methods: A total of 25 639 men and women, aged 45–74, years were recruited into EPIC-Norfolk (European Prospective Investigation into Cancer) and completed a questionnaire which recorded both occupational and recreational physical activity. Participants were then ranked into four groups of physical activity, ranging from active to inactive. This questionnaire had been validated against both energy expenditure, as assessed by 4-day heart rate monitoring, and cardiorespiratory fitness, assessed by submaximal oxygen uptake. The cohort was monitored for the development of symptomatic gallstones and each case matched with four controls for age and sex. Odds ratios were calculated using unconditional logistic regression, adjusted for body mass index, alcohol intake, hormone replacement therapy and parity.

Results: In the cohort, 219 participants (67.6% women) developed symptomatic gallstones at a mean age of 61.3 (SD 9.0) years. Increased physical activity decreased the risk of symptomatic gallstones stones in a linear manner across the four categories of physical activity (odds ratio for trend 0.85, 95% CI 0.72 to 1.0, p<0.05). An active vs inactive level of physical activity gave an odds ratio of 0.48 (95% CI 0.26 to 0.89 p = 0.02). Physical activity was independently associated with a reduced risk of gallstones after multivariate adjustment for known risk factors (p<0.05).

Conclusion: Increased physical activity was associated with a decreased risk of symptomatic gallstones, a consistent finding in epidemiological studies, and now reported for the first time using a physical activity questionnaire validated against physiological measurements. If the level of physical activity in the population could be raised, then 33% of symptomatic gallstones could be prevented.

MYCOPHENOLATE MOFETIL IN PATIENTS WITH REFRACTORY AUTOIMMUNE HEPATITIS

A. T. Barnardo, U. Warshow, J. Mitchell, M. Cramp. Hepatology, Derriford Hospital, Plymouth, UK

Introduction: Autoimmune hepatitis has a prevalence of 16.9/100 000 in the European population and 65% respond to standard therapy of prednisolone and azathioprine by 18 months. However 15% are either intolerant of standard therapies or suffer long-term treatment failures.

Aims & Methods: A retrospective case study was performed for patients treated with mycophenolate mofetil (MMF) for autoimmune hepatitis (AIH) between 1999–2007. Eleven females and one male were identified all had definite autoimmune hepatitis on the International hepatitis scoring system with pre treatment scores greater than 15 (median 17, range 15–21).

Results: Patients had received MMF because of toxicity and intolerance to either azathioprine (n = 3) or 6-mercaptopurine (6MP) (n = 1); treatment failure with prednisolone, azathioprine and tacrolimus (n = 2); or intolerance and treatment failure (n = 5). One patient was commenced on MMF as the primary steroid sparing agent. While on MMF tranaminases normalised in 6/12 patients and prednisolone dosages were reduced in 10/12 and discontinued in a further case but 4 are now cirrhotic. Toxicity with Azathioprine caused Stevens-Johnson reaction and neutropaenia (normal TPMT) and 6MP toxicity caused haemolysis and presented with 4 weeks in all cases. Two further patients were intolerant with severe headaches and nausea. All 5 are in biochemical remission on lower doses of prednisolone and MMF, and one is off all treatment. One patient has features of NAFLD rather than AIH on their last biopsy. One patient had treatment failure on prednisolone and azathioprine and one was also refractory to the addition of with Tacrolimus. Both have persistent biochemical evidence of disease activity and prednisolone doses have not been reduced, and their biopsies reveal cirrhosis but no interface activity. One has now decompensated with jaundice and ascites. Five patients had treatment failure (prednisolone, azathioprine n = 4, prednisolone, azathiprine and tacrolimus n = 1) and severe intolerance. One has progressed to biopsy proven cirrhosis and one has stage 3 fibrosis and has had a variceal bleed. Three have entered biochemical remission on lower dose prednisolone but no histology is available. The patient who progressed to cirrhosis has had compliance difficulties. The patient treated with MMF as the primary steroid sparing agent because of pre-existing thrombocytopenia progressed to cirrhosis.

Conclusion: Patients in whom there was a drug intolerance but biochemical response were maintained in remission on MMF. Patients who did not respond to prednisolone and azathioprine/6MP were not controlled by MMF and progressed to cirrhosis. Response in patients with both intolerance and treatment failure had a variable outcome. Alternative immunosuppressive regimes should be considered for those patients refractory to prednisolone and azathioprine/6MP.

AUDIT OF MANAGEMENT OF ALCOHOL WITHDRAWAL AT DERRIFORD HOSPITAL

A. T. Barnardo, L. Walters, M. Cramp, J. Mitchell. Hepatology, Derriford Hospital, Plymouth, UK

Introduction: Alcohol-related morbidity and mortality is an increasing problem in the UK. Approximately 5% of males and 3% of females are drinking at hazardous levels. Hospital admissions for inpatient alcohol detoxification has dramatically increased from 49 in 2000 to 139 in 2006. Trust guidelines advise the use of Clinical Institute Withdrawal Assessment for Alcohol (CIWA) score for assessment of severity and only oral benzodiazepines for treatment.

Aims & Methods: Notes were reviewed for patients admitted in the last 12 months with alcohol withdrawal. Demographics and outcomes were recorded and their clinical management was compared to the trust guidelines.

Results: Fifty three men and 21 women, were identified. Chronic liver disease was evident in 19% men and 8% women and a previous recognised psychiatric diagnosis was present in 24% men and 12% women. CIWA score was recorded in the medical notes at admission in 77% patients and a repeat CIWA was recorded in 53% and 73% cases in the medical and nursing notes respectively. Diazepam was used in 54%, lorazepam 7% and a combination in 28%. Intravenously sedation was only administered in ITU. Pabrinex and oral vitamin supplements were given to 95%. Generalised seizures occurred in 38% of which 3% had a history of head trauma. Security was called for 14% patients and 7% were admitted to ITU for intravenous sedation and 3% required ventilation. Patients were managed on 17 wards and patients managed on surgical wards only had CIWA scores recorded in 3/12 cases. The median length of stay was 5.8 days (range 1–28) and 23% took their own discharge. Three patients with concomitant chronic liver disease died during their admission from PE, septicaemia and multi-organ failure respectively. A further 3 have died since discharge.

Conclusion: The assessment of severity of alcohol withdrawal was poor with documentation at admission in only 77% and in 27% no assessment was documented throughout the admission despite the dispensing of benzodiazepines. An unexpected high percentage of patients admitted with alcohol withdrawal have severe symptoms. Four of the admissions to ITU were unavoidable as oral sedation was not sufficient. Requests for security could have been obviated if the severity of withdrawal had been realised earlier and acted upon. Features of chronic liver were found only in a minority of patients with alcohol withdrawal but were present in all patients admitted to ITU. Diazepam was given inappropriately to patients with chronic liver disease and this led directly to one admission to ITU because of respiratory depression. The use of intravenous benzodiazepines was not identified outside of ITU and the nutritional aspects of the patients care were performed well. More education is required on assessment and treatment of alcohol withdrawal.

CHANGING EPIDEMIOLOGY AND TREATMENT DECISIONS IN CHRONIC HEPATITIS B

I. S. Bhattacharya, A. J. Bathgate. Liver Unit, Royal Infirmary of Edinburgh, Edinburgh, UK

Introduction: In recent years there has been a change in the numbers and demographics of chronic HBV patients attending the Liver Unit, Edinburgh. Treating chronic HBV infection can prevent cirrhosis, end stage liver failure and hepatocellular carcinoma (HCC), however one of the difficulties in chronic HBV management is deciding if and when it is appropriate to treat. High HBV DNA levels are associated with the development of cirrhosis. The normal levels for ALT and the ALT level which determines treatment remains controversial.

Aims & Methods: The aims of this study were to determine the clinical and demographic characteristics of patients attending the Liver Unit, to identify the predictors of treatment using both our current ALT Upper Limit of Normal (ULN) (45 IU) and recently proposed ALT ULN (30 and 19 IU for men and women respectively (rev ULN)) and to identify the predictors of cirrhosis. This was a retrospective case note study, which included patients aged ⩾18 years. Patient names were obtained from an existing HBV database, the Chinese HBV Clinic, and from Liver Unit out-patient clinics.

Abstract 190

Results: 204 patients with chronic HBV were identified; 8 patients were co-infected with HCV, 3 co-infected with HIV and 1 with HCV and HIV. 67% patients were male, and 33% female. 45% patients were Chinese, 30% were Caucasian and 9% Pakistani. The most common “likely mode of transmission” was perinatal (54%). Of the 204 patients, 31% were HBeAg +ve, 30% were currently receiving treatment, 3% had HCC, 21% were cirrhotic and 2% had received liver transplant. Following exclusion of our 12 co-infected patients the following were found; 31% were HBeAg +ve and 27% were receiving treatment. Predictors of treatment were the presence of HBeAg (p<0.0005) and an HBV DNA level ⩾500 000 copies (p<0.0005). Predictors of cirrhosis were HBV DNA ⩾500 000 copies/ml (p<0.0005) and HCV co-infection (p<0.005). The table shows the relationship of present and revised ALT levels.

Conclusion: HBV demographics in the Liver Unit are changing with an increasing number of Chinese patients attending. An increasing HBV DNA level is a strong predictor of cirrhosis which emphasises the importance of regular monitoring and suppression of HBV DNA levels. ALT was not a predictor of treatment using either the present or proposed values. This study shows that our decisions regarding treatment have not been rigorously based on ALT levels.

CONCOMITANT PRIMARY SCLEROSING CHOLANGITIS DOES NOT INCREASE THE RISK FOR COLORECTAL DYSPLASIA IN PATIENTS WITH COLONIC CROHN’S DISEASE

B. Braden1, S. Aryasingha1, B. Warren2, T. Kitiyakara1, R. Chapman1. 1Gastroenterology, 2Pathology, John Radcliffe Hospital, Oxford, UK

Introduction: It is established that patients with ulcerative colitis and concomitant primary sclerosing cholangitis have an increased risk of developing colorectal dysplasia or cancer compared with those patients with ulcerative colitis alone. Therefore, annual surveillance colonoscopies are recommended.

Aims & Methods: We investigated whether primary sclerosing cholangitis is a risk factor for developing colorectal dysplasia/cancer in patients with Crohn’s disease of the colon. A retrospective review of 168 patients with primary sclerosing cholangitis was performed. 118 had concomitant ulcerative colitis and 50 Crohn’s disease. 109 patients with colonic involvement of Crohn’s disease served as controls. The primary outcomes were colorectal cancer and dysplasia. The secondary outcome was overall mortality.

Abstract 191

Results: None of the patients with colonic Crohn’s disease developed dysplasia during follow-up of 9±6 years, neither the patients with concomitant PSC, nor with Crohn’s disease alone. In contrast, 2 cancers and 7 colorectal dysplasia were diagnosed in patients with ulcerative colitis and PSC during follow-up of 11±7 years giving a crude annual incidence rate of dysplasia or colorectal cancer of 1 in 144 patients. The general (p = 0.12) or PSC related (p = 0.20) mortality did not differ between PSC patients with ulcerative colitis or colonic Crohn’s disease.

Conclusion: The presence of primary sclerosing cholangitis in patients with colonic Crohn’s disease does not seem to increase the risk of devoloping dysplasia of the colon as it does in patients with ulcerative colitis.

PERIHEPATIC LYMPHADENOPATHY INDICATES AETIOLOGY IN ACUTE AND CHRONIC LIVER DISEASE

C. F. Dietrich1, D. Faust2, A. Ignee1, D. Schreiber1, A. Planner3, T. Hirche2, B. Braden4. 1Gastroenterology, Caritas Hospital, Bad Mergenteim, 2Gastroenterology, University Hospital, Frankfurt/Main, Germany, 3Radiology, 4Gastroenterology, John Radcliffe Hospital, Oxford, UK

Introduction: In chronic virus hepatitis C the total perihepatic lymph node volume reflects the underlying liver histology, viral load, and the host’s immunological response. Assessment of the perihepatic lymph node volume may represent an important diagnostic tool, and may help streamline the patient’s further management.

Aims & Methods: The purpose of this study was to prospectively assess whether perihepatic lymphadenopathy is associated with the aetiology of acute (and chronic) hepatitis. In 40 consecutive patients with transaminases >500 U/l without known liver disease the total perihepatic lymph node volume was assessed and compared to the ultrasound findings in 263 patients with known chronic liver disease and also 49 healthy controls.

Results: 31/40 patients were diagnosed with an acute viral hepatitis, while 9/40 patients were diagnosed with a toxic cause, resulting in acute liver damage. In all sonographically evaluated patients with acute viral hepatitis (29/31, 94%) perihepatic lymphadenopathy was found, whereas none of the patients with a toxic cause demonstrated lymphadenopathy. In chronic liver disease, perihepatic lymphadenopathy was present in 86% of viral, in 90% of autoimmune hepatitis, in 100% of primary sclerosing cholangitis, in 97% of primary biliary cirrhosis, but only in 6% of hemochromatosis, in 1% of fatty liver disease, and in 4% of cholecystolithiasis.

Conclusion: Perihepatic lymphadenopathy is found in infectious and autoimmune liver diseases, but not in metabolic or toxic liver damage. The absence of perihepatic lymph nodes in acute liver failure should lead to intensive search for a toxic or metabolic cause.

MISSED OPPORTUNITIES TO INTERVENE IN PROGRESSION TO ALCOHOLIC LIVER DISEASE

S. Benfield1, T. R. Card2. 1Gastroenterology, Kings Mill Hospital, Sutton in Ashfield, 2Department of Epidemiology and Public Health, University of Nottingham, Nottingham, UK

Introduction: Although the UK government’s alcohol harm reduction strategy of 2004 promised to “strengthen the emphasis on the importance of early identification of alcohol problems” it is far from clear that screening for alcohol problems has been widely adopted. If our patients are to receive benefit from intervention in their drinking behaviour, problem drinkers must first be identified.

Aims & Methods: To establish the efficacy of current practice (opportunistic case finding) in our trust we have audited the notes of those eventually referred to our drug and alcohol team. We have searched hospital records as well as their history provided to our drug and alcohol service to establish whether they have previously received any intervention for their drinking, and whether there had been an opportunity for their drinking problem to be detected at an earlier point.

Abstract 193

Results: We requested records of the last 120 patients referred to the drug and alcohol service of our trust. Reasons for referral are given in the table. Of these we were able to obtain notes on 119 from which we abstracted information. 53 (45%) of these had previously been referred to alcohol services, but in 47 (40%) it was possible to identify clear evidence of problem drinking on a visit to the hospital prior to their first referral to alcohol services. On average failure to refer at the first opportunity resulted in a 1.6 year delay. Among the 23 finally referred with alcoholic liver disease (21 decompensated), 14 had had an opportunity for referral missed at an earlier stage, and in 11 the delay in referral exceeded one year. Of these 11 with a prolonged delay in referral only 1 had clear evidence of end organ damage at the first presentation (decompensated alcoholic liver disease). As can be seen in the table this high rate of delayed referral (61%) is greater in alcoholic disease patients than in any of the other groups we examined.

Conclusion: Although evidence of the benefit of strategies to intervene in problem drinking is not strong, we clearly cannot develop successful interventions until we identify those requiring them. At present, despite a well-funded drug and alcohol service in our trust, we are failing to do even this. We therefore plan to move from a system of case finding to one of screening.

ACUTE ALCOHOL WITHDRAWAL. IS IT TIME FOR NATIONAL GUIDELINES?

B. J. Colleypriest, P. F. Marden, A. Fard, J. Maltby. Gastroenterology, Royal United Hospital, Bath, UK

Introduction: Alcohol withdrawal is commonly encountered in all in-patient medical disciplines and carries a significant mortality. The psychiatric, psychological, forensic and medical problems of alcohol abuse/dependence are all well documented. Alcohol consumption, per person, has doubled since 1980 and continues to rise. Currently there are no national guidelines for the pharmacological management of acute alcohol withdrawal, although many hospitals have local policies.

Aims & Methods: This study aimed to answer the questions were local guidelines followed and was appropriate medication prescribed. Data for a 6-month period were retrospectively collected from the case notes of patients admitted through the medical admission unit at the Royal United Hospital. All patients treated for alcohol withdrawal were identified and pharmacological intervention recorded and compared to local guidelines.

Abstract 194 Regular chlordiazepoxide (CDX) prescriptions

Results: 51 patients were identified of whom 34 were male. The number of units of alcohol being drunk by each patient was recorded in 71% of cases. In addition to a regular decreasing dose, chlordiazepoxide was prescribed on a PRN basis for 63% of patients. Oral thiamine was prescribed in 79% patients but only according to guidelines in 16% of cases. Oral Vitamin B was prescribed in 59% patients of which only 14% followed guidelines. 37% of patients received multivitamins. Wernike’s encephalopathy (WE) was identified in 4 patients all of whom were prescribed parenteral B vitamins, but none according to guidelines. 47 patients, on review of notes, were deemed to be at risk of WE of whom 37 received parenteral B vitamins but only 4% followed guidelines.

Conclusion: This study demonstrates considerable variability and inadequacies in the treatment of alcohol withdrawal. Oral thiamine, vitamin B and PRN chlordiaxepoxide were only prescribed in 79%, 59% and 63% of patients respectively. Local guidelines were not followed in the vast majority of patients for thiamine, vitamin B, parenteral B vitamins or chlordiazepoxide prescription. Considering the rapid changeover of junior staff adherence to guidelines may be easier if national standardised guidelines were available.

LENGTH AND ADEQUACY OF HEPATITIS C VIRUS STAGING LIVER BIOPSY IMPROVE AS REGISTRARS SPEND MORE TIME IN POST

M. L. Cowan1, S. J. Thomson1, S. Musa1, M. Gess1, C. Finlayson2, S. Krishna3, S. Clark1, D. M. Forton1, T. M. Rahman1. 1Department of Gastroenterology and Hepatology, 2Department of Histopathology, St George’s Hospital, 3Division of Cellular and Molecular Medicine, St George’s, University of London, London, UK

Introduction: Liver biopsy is widely accepted as the gold standard for the staging of chronic liver disease secondary to infection with hepatitis C virus (HCV). Accuracy is reduced in small samples. A gold standard biopsy should be longer than 20 mm or contain 12 portal tracts. In our hospital, liver biopsy is performed by gastroenterology registrars who rotate in October.

Aims & Methods: We sought to determine the temporal trends in use of liver biopsy in patients with HCV and to assess the adequacy of tissue obtained for liver biopsy in our hospital. Patients with HCV undergoing liver biopsy between January 2001 and December 2006 were identified by a search of the histopathology database. Patient characteristics and biopsy findings were obtained by a search of the hospital electronic patient record.

Abstract 195

Results: 258 percutaneous parenchymal liver biopsies were performed in patients with HCV. Eleven biopsies contained no liver tissue or were so small that the reporting pathologist described them as uninterpretable. The average patient was 46 years (range 10–78 years), 64% were male and 51% were infected with HCV genotype 1. 45% had minimal hepatic fibrosis (Ishak stage 0/1), 39% moderate fibrosis (stage 2–4) and 16% cirrhosis (stage 5/6). The average biopsy was 21.5 mm (range 3–57 mm) and contained 13.2 portal tracts (3–35). Sixty four per cent fulfilled the above criteria for biopsy adequacy. There were no differences in numbers of biopsies performed, viral genotype, fibrosis stage, biopsy size or adequacy when analysed by year between 2001 and 2006. There were significant differences in biopsy size and adequacy (p<0.05) depending upon the month that the biopsy was performed (but not in patient age, genotype or fibrosis).

Conclusion: Staging liver biopsy for assessment of HCV constitutes a significant workload for gastroenterologists and histopathologists. There has been no significant change in the numbers and findings of biopsies performed between 2001 and 2006. The adequacy of biopsies improves during the year that gastroenterology registrars spend in post, presumably reflecting increasing experience.

THE IMPACT OF CO-INFECTION OF HEPATITIS DELTA AND HUMAN IMMUNODEFICIENCY VIRUS ON THE NATURAL HISTORY OF CHRONIC HEPATITIS B

T. J. S. Cross, R. Dancer, M. Horner, P. M. Harrison, K. Agarwal. Institute of Liver Studies, King’s College Hospital, London, UK

Introduction: Background: Individuals with chronic HBV infection are at increased risk of developing cirrhosis, hepatic decompensation and hepatocellular carcinoma (HCC).

Aims & Methods: To assess the impact of co-infection with either hepatitis delta (HDV) or human immunodeficiency virus (HIV) on the natural history of chronic HBV infection. We performed a retrospective analysis of all chronic HBV infected patients attending our institution between 2000 and 2005. Patients were identified who were seropositive for anti-HDV antibody (HDV Ab), and anti-HIV Ab positive. Case records were reviewed for the presence of cirrhosis or HCC. Comparisons were made between HBV mono-infected patients against those co-infected with HDV or HIV. Statistical analysis was performed using χ2 tests and risk was calculated including 95% confidence intervals.

Results: 1026 patients were identified in whom HDV and HIV status was known in 941 patients. The median age for the entire cohort was 35 years (IQR 29–43). The median ages for the HBV mono-infected, vs HDV/HBV co-infected vs HIV co-infected, were 35 (29–43), 35 (28–46), and 42 (38–46) respectively (p = NS). 527 patients (56%) of patients were male. 841 patients had HBV mono-infection, 79 patients had HDV/HBV, and 22 patients had HIV co-infection. 1 patient had triple infection. HDV increased the risk of cirrhosis (22/78 (28%) vs 109/841 (13%), OR 2.64, 95% CI 1.55 to 4.49, p<0.0001), but did not increase the risk of HCC (8/78 (10%) vs 66/841 (8%), OR 1.34 95% CI 0.62 to 2.91, p = 0.46). The presence of HIV increased the risk of cirrhosis (13/21 (62%) vs 109/841 (13%), OR 10.91 95% CI 4.42 to 26.93, p<0.0001), and also increased the risk of HCC (7/21 (33%) vs 66/841 (8%), OR 5.87 95% CI 2.29 to 15.05, p<0.0001).

Conclusion: In our cohort, the presence of co-infection with HDV or HIV increased the risk of cirrhosis in chronic HBV patients. Interestingly, HIV but not HDV co-infection significantly increased the risk of HCC in our group. This may be due to lower HBV viral load often seen in HDV/HBV co-infection.

THE IMPACT OF STEATOSIS AND STEATOHEPATITIS IN TREATMENT RESPONSES IN PATIENTS WITH CHRONIC HEPATITIS C INFECTION

T. J. S. Cross, A. Quaglia, J. Nolan, I. Fletcher, K. Agarwal, P. M. Harrison. Institute of Liver Studies, King’s College Hospital, London, UK

Introduction: There is controversy regarding the impact that steatosis has on the rate of sustained virological response (SVR) amongst patients treated with pegylated interferon and ribavirin for chronic hepatitis C (CHC) infection.

Aims & Methods: To study the effect of pre-treatment liver biopsy steatosis and steatohepatitis on the likelihood of SVR with combination therapy. We studied liver biopsies and treatment responses in 204 consecutive patients treated for CHC between January 2001 to January 2005. Liver biopsies were graded according to the method described by Ishak. Steatosis and steatohepatitis were graded using the modified Brunt score. Patients were classified according to the presence or absence of steatosis and steatohepatitis. Patients were studied together and by their genotype. Patients achieving SVR were labelled as responders and those failing to achieve SVR as non-responders.

Results: 179 patients were included in the final analysis. 72% % were male, median age 46 years (interquartile range 40–52). 127 were Ishak Fibrosis grade 3 or less (71%), and 31 patients (17.3%) were cirrhotic (Ishak grade 5–6). The overall SVR rate was 59%, the SVR rates by genotype were: genotype 1 (35%), genotype 2 (82%), genotype 3 (79%), and genotype 4 (57%). 106 patients were responders (59%). Responders did not differ from non-responders in respect to gender, age, steatosis, necroinflammatory score or pre-treatment ALT. On univariate analysis reduced SVR was associated with non-genotypes 2 and 3 (p<0.0001), cirrhosis (p<0.0001), and steatohepatitis (p = 0.009). On multivariate analysis only non-genotype 2 and 3 (p<0.001), pretreatment weight (p = 0.002) were significantly associated with SVR. 98 patients had steatosis (51.9%). The Χ2 test for trend demonstrated a difference between the groups and the rates of steatosis (p = 0.14). Steatosis was lowest in genotype 2, 12/37 (32%) vs non-genotype 2, 81/142 (57%), p = 0.008. Genotype 3 had steatosis 34/54 (63%) vs non-genotype 3 59/125 (47%), p = 0.053. On exclusion of genotype 4 patients this reached statistical significance p = 0.028. Steatosis had no effect on the likelihood of being a treatment responder (p = 0.176). The presence of no or mild steatosis (grade 0–1) also had no effect on the SVR (p = 0.46). Steatohepatitis was present in 34 patients (19%). Patients with Steatohepatitis were older (49 years vs 44 years, p = 0.03), and were more likely to have more severe steatosis (grade 2–3 vs 0–1, p<0.0001), and cirrhosis (p<0.0001).

Conclusion: Steatosis has no effect on SVR. Steatohepatitis may predict poor outcome. SVR is mainly determined by genotype, patient weight, and the presence of cirrhosis. Therapeutic strategies aimed at treating steatohepatitis may improve SVR in at risk patients. Patients at risk should have liver biopsy to assess for steatohepatitis.

ALCOHOL INDUCED LIVER INJURY MOBILISES HAEMATOPOIETIC STEM CELLS THAT CONTRIBUTE TO HEPATIC FIBROSIS

E. Dalakas1, P. N. Newsome1, S. Boyle2, A. Pryde1, W. A. Bickmore2, P. C. Hayes1, D. J. Harrison3, J. N. Plevris1. 1Hepatology Department, Univesity of Edinburgh, 2MRC Human Genetics Unit, Western General Hospital, 3Pathology Department, Univesity of Edinburgh, Edinburgh, UK

Introduction: Haematopoietic stem cells (HSCs) can contribute to multiple epithelial lineages including myofibroblasts in the human liver. HSC mobilisation and migration to injured liver are essential components of this pathway.

Aims & Methods: Our aim was to study the effect of alcohol liver injury on HSC mobilisation, hepatic recruitment and differentiation into hepatic epithelial lineages. Peripheral blood samples were collected from patients with alcoholic hepatitis (AH, n = 15), paracetamol hepato-toxicity (n = 7) chronic liver injury (n = 17) and normal controls (NC, n = 12). Circulating CD HSC levels were quantified. Colony forming unit (CFU) assays were performed on CD34+ cells. Cross-sex (x-sex) liver biopsies with AH were analysed for HSC content (CD34+) and for markers of myofibroblasts (α SMA), hepatocytes and biliary epithelium using immunohistochemical and fluorescent in situ hybridisation techniques.

Results: (1) The AH group had higher circulating blood CD34+ cell levels (0.195±0.063%, p<0.05) and higher CFU counts (154±38, p<0.05) when compared with NC (0.058±0.008% and 44±28 CFUs). (2) There were increased numbers of recipient CD34+ cells in x-sex donor liver grafts with AH as compared with controls (1.834±0.605% vs 0.299±0.208%, p<0.05). 3) Increased numbers of recipient α SMA+ cells (7.9%–26.8%) were identified in the AH x-sex liver grafts. (4) There were increased numbers of co-staining α SMA+/CD34+ cells in x-sex donor liver grafts with AH as compared with controls (5.025±0.621% vs 0.690±0.290%, p<0.05). (5) No recipient hepatocytes or biliary cells were identified in the x-sex donor liver grafts with AH.

Conclusion: (1) In AH, CD34+ stem cells are mobilised into the circulation and display true stem cell potential. (2) Alcohol liver injury recruits CD34+ cells into the liver and contributes significantly to the hepatic myofibroblast population but not to parenchymal lineages.

PIG MEAT CONSUMPTION AND MORTALITY FROM CHRONIC LIVER DISEASE

H. R. Dalton1, C. Pritchard2, R. Bendall3. 1Cornwall Gastrointestinal Unit, Peninsula College of Medicine and Dentistry, and Royal Cornwall Hospital, 2Research and Development Research Unit, 3Clinical Microbiology, Peninsula College of Medicine and Dentistry, and Royal Cornwall Hopsital, Truro, UK

Introduction: In 1985 Nanji and French reported a correlation between pork consumption and mortality from chronic liver disease (CLD),1 based on data from the 1960s and 70s. This finding has never been confirmed nor adequately explained. This observation is currently of interest because of the emergence of locally acquired hepatitis E infection (HEV) in the developed world. Locally acquired HEV in developed countries is thought to be a zoonosis from pigs. It is much more common than previously realised, has a predilection for middle aged/elderly males, is caused by HEV genotype 3, and carries an adverse prognosis in CLD.2 Viable HEV has been isolated from retail pig meat in the USA,3 and HEV can survive cooking temperatures up to 60°C.4

Aims & Methods: To correlate the mortality from CLD with pork, beef and alcohol consumption and the seroprevalence of HBV and HCV in developed countries. Data from 1990–2000 in 18 developed countries for mortality from CLD, alcohol consumption, HBV and HCV seroprevalence were obtained from WHO databases, and pig meat and beef consumption from United Nations databases. Mortality from CLD was compared with the other variables using regression analysis.

Results: Bivariate regression showed that alcohol (t = 3.792, p = 0.002) and pig meat (t = 4.255, p = 0.001) consumption were associated with mortality from CLD, but beef consumption, HBV and HCV seroprevalence were not. Multivariate regression showed that alcohol (t = 3.403, p = 0.005) and pig meat (t = 4.691, p<0.001) consumption and HBV seroprevalence (t = 2.317, p = 0.037) were independently associated with mortality from CLD, but HCV seroprevalence was not.

Conclusion: Pig meat consumption is independently associated with mortality from chronic liver disease in developed countries. The reason for this observation is uncertain: it could be an epiphenomenon; a factor in pig meat (for example, pork fat) might cause cirrhosis; a factor in pig meat (for example, HEV) may cause increased mortality in patients with pre-existing chronic liver disease. The latter possibility is not as improbable as might at first be thought, as we have previously shown that 16% of blood donors in SW England are HEV IgG positive and up to 21% of US blood donors are seropositive. This suggests that unrecognised infection with HEV is common.

EVALUATION OF ABNORMAL LIVER FUNCTION TESTS IN PRIMARY CARE

T. Dickel, A. Sinha, M. Lancaster-Smith, H. Curtis. Gastroenterology, Queen Mary’s Hospital, Sidcup, UK

Introduction: Abnormal liver function tests (LFT) are a common finding in primary care. When significant underlying disease of the liver is suspected an ultrasound scan (USS) should be arranged in the clinical context together with additional blood tests, lifestyle and medication changes and referral to a specialist.1 2

Aims & Methods: 172 patients referred for open access USS to Queen Mary’s Hospital between February 2005 and January 2006 by their General Practitioner (GP) for investigation of abnormal LFTs had been identified. For 124 patients results were available from the pathology department. Data from GP notes were gathered for 78 patients by visiting the surgeries. The objective of this study is to assess the investigations arranged by GPs leading to and following open access USS for abnormal LFTs.

Results: USS for 93/124 (75%) patients showed fatty infiltration of the liver. Hepatitis A, B and C serology had been requested in 20/124 (16.1%), 43/124 (34.7%) and 39/124 (31.5%) patients respectively. Metabolic diseases had been screened for in 2/124 (1.6%) for Wilson’s disease, in 13/124 (10.5%) for α1-antitrypsin deficiency and in 33/124 (26.6%) patients for haemochromatosis. Antinuclear antibodies (ANA) were available in 20/124 (22.6%) patients. Smooth muscle antibodies (SMA) and antimitochondrial antibodies (AMA) were requested in 7/124 (5.6%) patients each. 17/124 (13.7%) and 11/124 (8.9%) patients had been referred to a gastroenterologist or surgeon. A liver screen including three groups of underlying liver diseases (viral, metabolic and autoimmune) was requested in 15/124 (12.1%), a screen including at least 1 or 2 groups in 46/124 (37.1%) and no further tests in 63/124 (50.8%) patients. 20/124 (16.1%) patients had LFTs higher than twice the upper limit of normal (ULN). Of this subgroup only 2/20 (10%) had further investigations and only 3/20 (15%) were referred to a specialist. Repeat LFTs were requested in 75/124 (60.5%) patients. 13/75 (17.3%) still had LFTs >2 ULN. For 78/124(62.9%) patients data from their GP notes was available. In 32/78 (41%) patients a potential cause was identified. Lifestyle advice was given in 16/78 (20.5%) cases. 7/78 (8.9%) and 9/78 (11.5%) patients were referred to a surgeon or gastroenterologist for further evaluation. In 38/78 (48.7%) cases no further investigations were arranged to follow up the results of the ultrasound.

Conclusion: Abnormal LFTs are often first diagnosed by the GP. Initial investigations and referral practice vary and are not always in line with current recommendations. The request for USS is often not accompanied by additional tests (screening for viral hepatitis, haemochromatosis and autoimmune disease). Implementation of guidelines for the evaluation of abnormal LFTs in primary care should be considered.

PROTEIN OXIDATION AND XANTHINE OXIDASE ACTIVITY IN HEPATITIS C VIRUS-RELATED HEPATOCELLULAR CARCINOMA: ROLE IN TUMOUR APOPTOSIS AND CYTOPROLIFERATION

H. A. El Aggan1, S. A. Mahmoud2, L. K. Younis3. 1Department of Medicine, Hepatobiliary Unit, 2Department of Medical Biochemistry, 3Department of Pathology, Faculty of Medicine, University of Alexandria, Alexandria, Egypt

Introduction: Hepatitis C virus (HCV) infection causes a state of chronic oxidative stress, which may contribute to hepatocarcinogenesis. The formation of reactive oxygen species (ROS) in the liver can promote oxidative damage of intracellular proteins, which tend to form high-molecular-weight aggregates and thereby influence a diverse array of cellular processes. Xanthine oxidase (XO) is a primary source of ROS and a key enzyme in protein oxidation.

Aims & Methods: The present work was designed to study the changes in protein oxidation and XO activity in patients with HCV-related hepatocellular carcinoma (HCC) in relation to the rates of tumour apoptosis and cell proliferation. To achieve this goal, the plasma levels of carbonyl proteins and advanced oxidation protein products (AOPP), as markers of protein oxidation, and plasma XO activity were determined in 50 patients with HCV-related cirrhosis (30 with histologically-proven HCC of different grades and stages and 20 without HCC) and in 20 healthy subjects. Core liver biopsies of hepatic tumours were examined for the rate of apoptosis using the terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labelling (TUNEL) method and for mutant p53 expression by immunohistochemistry. The proliferative activity of HCC was quantitatively assessed using proliferating cell nuclear antigen (PCNA) labelling.

Results: The plasma levels of carbonyl proteins and AOPP and plasma XO activity were significantly higher in cirrhotic patients with and without HCC than in healthy subjects and in patients with than in those without HCC (p<0.0001). The apoptotic rate was always less than the proliferative activity in HCC tissues and their ratio tended to decrease with higher tumour grade, size and stage and in p53-positive tumours (p<0.001). The markers of protein oxidation (but not XO activity) in HCC patients showed positive correlations with the proliferative activity and inverse correlations with the ratio between the apoptotic rate and the proliferative activity (p<0.05).

Conclusion: These findings suggest that enhanced protein oxidation and XO activity seems to play a role in the development and progression of HCV-related HCC with imbalance between rates of apoptosis and cytoproliferation. Better understanding of these processes might lead to the development of improved therapeutic approaches for the prevention and treatment of HCC.

WHICH SCORE FOR ALCOHOLIC HEPATITIS BEST PREDICTS MORTALITY?

P. Ellul1, K. M. Ashfaq2, R. Sekhar2, J. Keating2. 1Gastroenterology, New Cross Hospital, Wolverhampton, 2Gastroenterology, Furness General Hospital, Barrow-In-Furness, UK

Introduction: Alcoholic hepatitis (AH) is associated with a high mortality. With an alcohol-related mortality of 21.2 patients (UK −12.9) per year per 100 000 population, Barrow-in-Furness has one of the highest alcohol related mortality in the North West of England. There are various scores, which are used to risk stratify the patients. However, which score best predicts mortality?

Aims & Methods: Patients who died due to AH between January 2000 and December 2006 were identified. The modified Discriminant function (mDF), MELD score and the Glasgow AH score were retrospectively analysed. Data were obtained from the patients’ case notes. All patients had alcoholic liver disease and no other coexistent liver pathologies. The quoted 28-day mortality for patients with an mDF score >than 32 is 65%; MELD score >11 is 55%; Glasgow AH score >9 is 54%. Our aim was to assess which score best predicted mortality.

Results: We identified 62 patients and 51 were analysed. The others could not be analysed due to inadequate documentation. There were 37 male patients and 14 female patients. Mean age was 54 years (27–76 years). We calculated the scores by using the data on day 1 of admission. The scores were: 1. mDF >32; MELD>11; Glasgow score >9–25 patients; 2. mDF <32; MELD>11; Glasgow score <9–10 patients; 3. mDF >32; MELD>11; Glasgow score <9–14 patients; 4. mDF <32; MELD<11; Glasgow score <9–2 patients. The two patients with an mDF <32; MELD<11; Glasgow score <9 were admitted with alcoholic hepatitis but developed a gastrointestinal bleed (GI) and died from multiorgan failure after the GI bleed. From our results an mDF >32 was present in 39 patients who died (76.5%). The best predictor was the MELD score, as 49 patients (96%) with a score >11 died within 28 days. Only 25 patients (49%) with a Glasgow score >9 died within 28 days.

Conclusion: From our analysis the MELD score was the best predictor of 28-day mortality. An advantage of using the Glasgow AH score is that it is easy to calculate. However as information technology is easily accessible, it should be easy to calculate the MELD and mDF scores. Our limitation was that we could not calculate 3 month–1 year mortality. The obvious dilemma for the general clinician is which score to use for this common pathology with a high mortality. As if the above scores were not enough, recently the Lille prognostic model has been described. Its aim is the identification of patients at high risk of death at 6 months. Its accuracy should be higher than that of any of the other scores and is designed to predict survival in subjects with AH treated with corticosteroids. Above the ideal cut-off, the Lille model is able to predict 76% of the observed 6-month deaths. It suggests that it may be futile to treat patients for >7 days if the score is above the cut-off. What is definitely needed is more research on alternative treatments for patients with AH as this group of patients has a high mortality and a consensus for the best score is needed.

ALCOHOL-RELATED MORTALITY, SOCIOECONOMIC DEPRIVATION AND URBANICITY IN ENGLAND AND WALES

S. E. Erskine1, D. Gleeson2, R. Maheswaran1. 1School of Health and Related Research, University of Sheffield, 2Liver Unit, Sheffield Teaching Hospitals, Sheffield, UK

Introduction: Alcohol-related mortality is a serious problem in England and Wales. Alcohol-related mortality rates are increasing faster than in many other European countries. Such deaths are potentially preventable through education and regulation. It is important to identify groups of individuals who are most at risk of dying from alcohol-related disease to ensure that health interventions are effectively targeted.

Aims & Methods: Geographical ward level mortality data based on electoral wards for England and Wales from 1999 to 2003 were used to explore the association between alcohol-related mortality and socioeconomic deprivation by age and sex. The Carstairs index was used as an area based measure of deprivation. It is a standardised combination of four 2001 Census variables at ward level (male unemployment, overcrowding, residents in Social Class IV or V, no car access). Differences between urban and rural areas were also investigated. The Office for National Statistics definition of alcohol-related mortality was used. Deaths from liver cirrhosis accounted for 85% of the deaths. Other causes of death included in the definition were alcoholic gastritis, alcohol-induced chronic pancreatitis, alcohol poisoning, alcoholic neuropathies and alcoholic cardiomyopathy.

Results: People living in wards experiencing greater levels of deprivation had higher risks of dying from an alcohol-related cause. The rate ratios of mortality rates adjusted for age and urban-rural status for the most compared with the least deprived quintile of the population were 3.39 (95% CI 3.10 to 3.71) for males and 2.36 (2.16 to 2.57) for females. Rate ratios by age are as follows: 15–24 years: males: 1.45 (0.65 to 3.27) females: too few deaths to calculate; 25–44 years: males: 4.73 (4.00 to 5.59) females: 4.24 (3.50 to 5.13); 45–64 years: males: 4.18 (3.76 to 4.65) females: 2.67 (2.40 to 2.96); 65–79 years: males: 2.01 (1.75 to 2.32) females: 1.49 (1.29 to 1.71); 80+ years: males: 0.97 (0.75 to 1.26) females: 1.13 (0.86 to 1.48). The association peaked in the 25–44 years age category. People living in more urban areas experienced higher alcohol-related mortality rates. The rate ratios of mortality rates adjusted for age for urban relative to rural wards were 2.37 (2.05 to 2.75) for males and 1.68 (1.49 to 1.88) for females. After additional adjustment for deprivation, differences in mortality were attenuated but remained, with rate ratios of 1.35 (1.20 to 1.52) and 1.12 (1.01 to 1.25) for males and females respectively.

Conclusion: Alcohol-related mortality rates were found to be higher in more deprived areas and urban areas. The differences between the most and least deprived quintiles of the population were greatest between ages 25 and 44 years. Deprivation and urbanicity should be considered when designing alcohol-related health policies.

USING THE NEW AASLD-RECOMMENDED ALT RANGE INCREASES THE WORKLOAD FOR CHRONIC HEPATITIS B PATIENTS IN A CITY WITH A LARGE ETHNIC MINORITY POPULATION IN THE UK

N. Bholah, A. C. Ford, S. M. Moreea. Digestive Disease Centre, Bradford Royal Infirmary, Bradford, UK

Introduction: Bradford is a city with approximately 87 000 people from South Asia, in whom the prevalence of hepatitis B (HBV) is estimated at between 3 and 4%. In the latest AASLD guidelines1 the upper limits of normal for ALT at which to consider liver biopsy and treatment have been decreased from 40 IU/ml to 30 IU/ml for men, and from 40 IU/ml to 19 IU/ml for women. Treatment of HBV is based on eAg status, an ALT >1–2× the upper limit of normal (ULN), HBV viral load, and the degree of inflammation and/or fibrosis on liver biopsy.

Aims & Methods: We have prospectively collected data on all cases of HBV in Bradford between 2005 and 2007. Our database was analysed to obtain information concerning patient demographics, fluctuations in ALT and HBV viral load. The newly recommended ALT range and most recent viral load measurements were used to estimate any potential increase in the number of patients who would require a liver biopsy and possible treatment for their chronic HBV if the new AASLD guidelines were applied in our population. Patients were grouped according to HBe antigen (Ag) status and viral load.

Abstract 204 Possible increase in workload in chronic HBV patients

Results: After excluding those individuals currently receiving anti-HBV therapy, and those with incomplete data, there were 147 patients available for analysis. There were 55 (37%) males with a mean age of 39 years (range 21 to 77.5), 35 (64%) of whom were Asian, and 92 (63%) females with a mean age of 34 years (range 17 to 83), 73 (79%) of whom were Asian. The results of our analyses are shown in the table.

Conclusion: Despite the fact that our current database does not reflect the expected prevalence of HBV in the large South Asian population in Bradford, adhering to the AASLD guidelines would significantly increase our workload, in terms of obtaining and interpreting liver histology, and treating patients with chronic hepatitis B infection. The majority of this increase in workload would be due to the alteration in the ULN of ALT in women. This must be taken into account for service development and provision.

WHEN CORTICOSTEROIDS DON’T WORK: RESCUE TREATMENT WITH PENTOXIFYLLINE FOR ALCOHOLIC HEPATITIS

E. Watson, H. Lafferty, E. H. Forrest. Gastroenterology, Glasgow Royal Infirmary, Glasgow, UK

Introduction: Corticosteroid (CS) treatment for alcoholic hepatitis (AH) remains controversial. Recent data have indicated that CS treatment can improve survival amongst patients with a Glasgow Alcoholic Hepatitis Score (GAHS) greater than or equal to 9 and a Discriminant Function (DF) greater than or equal to 32. However some patients do not appear to respond to CS, and these patients have a particularly poor outcome. Pentoxifylline (PTX) has also been suggested as a treatment of AH.

Aims & Methods: The aim of this study was to assess the effect of “rescue” PTX therapy in patients in whom CS was ineffective or had to be stopped. We retrospectively identified those patients with AH treated with CS who did not respond to such treatment, or in whom CS treatment (prednisolone 40 mg) was discontinued on account of contra-indications. Non-response was defined as a failure of serum bilirubin to fall by 25% from baseline after 7 days of CS treatment. Survival to 84 days was recorded relative to the use of rescue treatment with PTX (400 mg tds for 4 weeks).

Results: Thirty eight patients were identified: 34 were CS non-responders, 4 had CS therapy discontinued after developing sepsis. All patients had a GAHS greater than or equal to 9 and a DF greater than or equal to 32 at the start of treatment. Twenty one patients received PTX as treatment; 3 in addition to continued CS treatment, the remainder as alternative treatment. Six of the PTX-untreated patients discontinued CS treatment, the remainder continued for 4 weeks. The 28-day survival for PTX treated and PTX-untreated patients was 52% and 76% respectively (p = 0.24). The 84-day survival for PTX and PTX-untreated patients was 38% and 44% respectively (p = 0.97). There was no survival difference on Kaplan-Meier analysis (HR 0.655; 95% CI 0.28 to 1.52; p = 0.32). Multiple logistic regression analysis failed to identify PTX therapy, the discontinuation of CS treatment, or the 7-day change in bilirubin as independent predictors of outcome.

Conclusion: Rescue treatment with PTX had no survival advantage for patients with AH who did not respond to CS, or in whom such treatment had to be discontinued. The management of these patients remains difficult and new therapeutic approaches need to be explored.

HEPATITIS B: A SINGLE CENTRE STUDY OF PRACTICE

M. R. Foxton1, W. Tong2, S. Palmer2, J. Oben1, M. L. Wilkinson1, T. Wong1. 1Gastroenterology, 2Virology, St Thomas’ Hospital, London, UK

Introduction: The management of chronic hepatitis B (HBV) infection has undergone major changes recently, particularly with the greater accessibility to HBV DNA testing and an increasing number of therapies. Moreover, the Health Protection Agency (HPA) has issued standards regarding the follow-up of HBV infection. We wished to assess the practice of the management of HBV infection within our trust, which covers a population of 400 000 of whom 33% are from ethnic minorities, prior to the release of these guidelines.

Aims & Methods: All patients with a positive HBsAg test in the calendar year 2005 were identified along with the referring source and whether the patients were under follow-up by a team with expertise in the management of chronic HBV infection. All patients under the care of the gastroenterology department from June 2002 to June 2006 were identified from outpatient clinic letters. Data were collected for each patient including demographic, biochemical, virological, radiological, histological and therapeutic parameters.

Results: There were 21 683 hepatitis B surface antigen tests performed in 2005. Of these 938 (4.33%) were positive in 671 patients. In 167/671 (24.9%)patients, the test was ordered by gastroenterology or HIV/GU medicine teams. 356 tests (53%) were ordered from primary care. 65.4% of patients with a positive HBsAg test were not under the care of gastroenterology or GU/HIV medicine. 243 patients were seen in the out-patient gastroenterology clinic for further management with a median follow up of 464 days (IQR 154–922 days).The median number of appointments per patient was 5 (IQR 3–8). The median age was 35 years and the majority of patients were of Sub-Saharan Africa (44%) or Oriental (19%) origin. Fifty four patients were lost to follow-up and 11 patients died at a median 172 days after being first seen. There were 27 biopsies performed with 11 having significant fibrosis. Fifty five patients had abnormal ultrasound scans, 20 of which generated 33 CT scans. There were 6 cases of hepatocellular carcinoma. 27/35 eAg positive patients had abnormal ALT and 20/35 had HBV >5log IU/ml. 70/151 eAg negative patients had elevated ALT with 13 patients having an HBV DNA >5log. 42 patients were on anti-viral therapy. 38 patients were initiated on monotherapy. 4 patients were on dual therapy after monotherapy at a range of 15–26 months.

Conclusion: HBsAg testing is commonly performed but a significant proportion of positive tests are not referred to secondary care. Patients with chronic HBV require ongoing follow up, since a proportion will have advanced fibrosis. However, if all patients with positive HBsAg were referred to secondary care, as per HPA guidelines, there would be a considerable increase in resource utilisation in terms of biochemical, virological and radiological investigations.

INSULIN RESISTANCE PREDICTS RAPID VIROLOGICAL RESPONSE IN PATIENTS WITH GENOTYPE 1 CHRONIC HEPATITIS C TREATED WITH PEGINTERFERON AND RIBAVIRIN

A. Grasso1, F. Malfatti1, H. Martines1, P. De leo2, F. Toscanini2, A. De Ferrari3, E. Venturino4, M. Anselmo2, G. Menardo5. 1Internal Medicine and Gastroenterology Unit, 2Infectious Disease Unit, 3Immunohematology and Transfusion Medicine Unit, 4Pathology Unit, 5Internal Medicine and Gastroenterology Unit, San Paolo Hospital, Savona, Italy

Introduction: Several baseline factors have been identified as predictors of response to treatment in patients with chronic hepatitis C. Of these, genotype and on-treatment viral dynamics at week 4 and 12 are the most important. Aim of the study was to evaluate both the predictive capacity and the influence grade of pre-treatment variables on viral dynamics in patients with genotype 1 chronic hepatitis C (G1CHC) treated with peginterferon and ribavirin.

Aims & Methods: Sixty two patients (33 male, median age 46 years, range 30–67) with G1CHC consecutively treated with peginterferon alpha and ribavirin in our Department, were included. Determination of viral load was made (Cobas Amplicor 2.0, Roche) at baseline, week 4, week 12 and six months after therapy had finished. We defined rapid virological response (RVR) as undetectable plasma viral load at week 4, “complete” early virological response (cEVR) as undetectable plasma viral load at week 12 and “partial” early virological response (pEVR) as at least 2 logs reduction of viral load from baseline at week 12. In 52 patients viral load was also determined at week 1. Insulin resistance index (HOMA = insulin (mUI/l) × glucose (mmol/l)/22.5), body mass index (BMI), steatosis at histology (<5%/>5%), fibrosis (F0-F4; Scheuer) were also assessed. Univariate and multivariate analysis were carried out by means of a statistical software. Statistical significance was defined as an alpha error <0.05.

Results: Twenty nine patients (47%) experienced sustained viral response (SVR). All 21 RVR (100%), 6 of 11 cEVR (55%) and 2 of 6 (33%) pEVR had SVR. At multivariate analysis RVR was independently associated with HOMA (OR 0.22; 95% CI 0.06 to 0.81; p = 0.022) whereas cEVR was independently associated with steatosis (OR 0.045; 95% CI 0.002 to 0.92; p = 0.044). HOMA (<2 vs >2), steatosis (<5% vs >5%) and fibrosis (<2 vs >2) but not BMI (<25 vs >25) were able to influence viral dynamic assessed as decline of viral load (logs) from baseline to week 1 (HOMA p = 0.018; steatosis p<0.000; fibrosis p<0.000; BMI p = 0.74), decline of log HCV-RNA from baseline to week 4 (HOMA p = 0.008; steatosis p<0.000; fibrosis p = 0.001; BMI p = 0.75) and decline of log HCV-RNA from baseline to week 12 (HOMA p = 0.011; steatosis p<0.000; fibrosis p = 0.004; BMI p = 0.59).

Conclusion: HOMA is the only pre treatment predictor of RVR in patients with genotype 1 chronic hepatitis C candidates to antiviral treatment. The presence of insulin resistance, steatosis and fibrosis negatively influence early on-treatment viral dynamics during antiviral treatment.

CANCER IN AUTOIMMUNE HEPATITIS

B. S. Höeroldt1, N. Farquharson2, K. Shepherd2, E. McFarlane2, P. Silcocks3, D. Gleeson2. 1Gastroenterology, Rotherham General Hospital, Rotherham, 2Hepatology, Royal Hallamshire Hospital, 3Medical Advisor, Trent Cancer Registry, Sheffield, UK

Introduction: Autoimmune hepatitis (AIH) is an autoimmune liver disease, which may lead to cirrhosis and is often treated with long-term Azathioprine. Autoimmune diseases have been associated with lymphoproliferative cancers. In rheumatoid arthritis and following renal transplantation Azathioprine treatment has been associated with in increased risk of cancer. AIH has been associated with a low risk of hepatocellular carcinoma but a possibly increased rate of extrahepatic cancer; however published data are inconclusive.

Aims & Methods: To assess if cancer rates in a large cohort of patients with AIH are higher than those expected for the general population. Patients: 223 patients (186 female, mean age at diagnosis 57 years, mean follow-up 9.75 years) with definite (n = 147) or probable (n = 75) AIH by International AIH Group criteria presenting between 1971–2005. Patients with HCV-AB or HBsAg were excluded. Data were extracted from clinical records regarding site and time of diagnosis of first cancer after diagnosis of AIH. Cancer rates in AIH patients were compared with age- and gender-adjusted rates for the general population, using the Trent Region Cancer Registry. Follow-up was censored at diagnosis of cancer (n = 36), death (24 liver related, 19 non-liver, non-cancer), liver transplantation (9 patients), cessation of follow-up (9 patients) or at 31.12.2005. Cancers in 10 patients prior to the diagnosis of AIH were not included.

Abstract 208 Cancer in AIH and in the general population

Results: 36 patients with AIH developed cancer: 5 HCC (all cirrhotic and female, mean age 73.3 yrs) and 31 other cancers including 7 GI tract, 6 breast, 4 lung, 3 GU tract and 3 maxillo-facial. Whether Azathioprine exposure was classified as (a) ⩽28/7 and >28/7 or (b) as no exposure and any Azathioprine exposure the relative risk was consistent to with up to a double risk of all malignancies (except non-melanoma skin cancer) but equally well with a halving respective relative risk; (a) RR 0.70 (95% CI 0.305 to 1.609) and (b) RR 1.00 (95% CI 0.458 to 2.148). By Cox multivariate analysis development of cancer in AIH was significantly associated with age at AIH diagnosis (p = 0.013) but not with duration of prior Azathioprine therapy.

Conclusion: The overall risk of cancer in AIH is slightly increased, but that for extrahepatic cancer not significantly so, relative to the local reference population. Cancer risk is unrelated to duration of Azathioprine treatment. These data do not suggest a substantially increased cancer risk in patients with AIH treated with Azathioprine. The risk of hepatocellular cancer is higher than in some previous reports.

PRIMARY SCLEROSING CHOLANGITIS IN PATIENTS WITH A DIAGNOSIS OF AUTOIMMUNE HEPATITIS

M. Hunter1, M. Loughrey2, M. Gray2, P. Ellis3, N. McDougall1, M. Callander1. 1Liver Unit, 2Pathology Department, 3Radiology Department, Royal Victoria Hospital, Belfast, UK

Introduction: Overlap syndromes between the autoimmune liver diseases (autoimmune hepatitis (AIH), primary sclerosing cholangitis (PSC), and primary biliary cirrhosis (PBC)) are well recognised although their frequency is not well established. We have reviewed our cohort of patients with AIH to determine how frequently a diagnosis of PSC was subsequently made and whether this might have been predicted

Aims & Methods: We reviewed clinical and laboratory data on all patients in whom a diagnosis of AIH was made at our Clinic from January 1979 to January 2007 (n = 118). We identified those in whom a subsequent diagnosis of PSC had been made. Those patients attending the Clinic whose liver function tests (LFTs) had an elevated alkaline phosphatase (ALP) or gamma glutamyl transpeptidase (GGT) (n = 16) underwent magnetic resonance cholangiogram (MRC). The features of this “cholestatic” group were compared with a randomly selected group of 25 patients who did not have cholestatic LFTs.

Abstract 209

Results: Eight patients already had an additional diagnosis of PSC which was made one to 22 years from initial presentation. They were more commonly male (not statistically significant (NS)), younger (p<0.05), more often had ulcerative colitis (UC) (NS), had lower AST (p<0.05), and higher ALP (NS) at presentation than the controls. Four of the 16 patients with cholestatic LFTs had an MRC which showed PSC (table). These 16 tended to have lower AST at presentation than the control group (NS) but there were no other differences. The International Autoimmune Hepatitis Group (IAHG) scores (for autoimmune hepatitis) were calculated retrospectively. There were no significant differences between the groups.

Conclusion: At least 12 (10%) of 118 patients with an initial diagnosis of AIH subsequently developed evidence of PSC. IAHG scores at presentation did not predict those who would develop PSC. AIH patients who show cholestatic LFTs during follow-up should have MRC performed.

DIABETES AND ARTHROPATHY AS SURROGATE MARKERS OF IRON OVERLOAD AND LIVER FIBROSIS IN GENETIC HAEMOCHROMATOSIS

H. G. Jarvis, K. Hird, S. F. Stewart. Liver Unit, Freeman Hospital, Newcastle upon Tyne, UK

Introduction: Diabetes and arthropathy are classical features of haemochromatosis but debate exists over their association with the disease and the role of iron in their pathogenesis. It is also not clear whether these features are more prevalent in more advanced disease and if they could be useful in determining which patients to biopsy and which should undergo surveillance for the complications of cirrhosis.

Aims & Methods: We aimed to determine whether diabetes and arthropathy at diagnosis related to the degree of iron overload and liver fibrosis in a regional cohort of patients with genetic haemochromatosis. Case notes on 108 C282Y homozygotes or C282Y/H63D compound heterozygotes were analysed and data collected on ferritin at diagnosis, venesections and liver biopsy results. Univariate relationships between categorical variables were analysed by the χ2 test and differences in means by the Student’s t test.

Abstract 210

Results: 17/108 (16%) were found to be diabetic at diagnosis. Diabetic patients had higher ferritins (3100.2±1571.9 vs 1673.4±328; p = 0.008) and required more venesections (39.23±20.2 vs 20.74±4.4; p = 0.008) to de-iron than non-diabetic patients. Although there was a trend towards more fibrosis this was not significant. 43/108 (40%) had arthropathy at diagnosis. Patients with arthropathy required more venesections to de-iron (29.43±10.2 vs 19.24±4; p = 0.04) but did not have statistically higher ferritins. There was a strong correlation between the presence of arthropathy at diagnosis and advanced fibrosis on liver biopsy (p = 0.03).

Conclusion: Diabetes in haemochromatosis relates to the degree of iron overload but does not help to determine which patients have advanced fibrosis. In contrast, patients with arthropathy do not have higher ferritins but are more likely to have advanced fibrosis. It may be reasonable to consider arthropathy as one of the factors influencing the decision to perform liver biopsy in these patients.

THE ROLE OF HEV TESTING IN DRUG INDUCED LIVER INJURY

M. Joseph1, H. R. Dalton1, H. Fellows1, W. Stableforth2, P. S. Thurairajah3, U. Warshow3, S. Hazeldine4, R. Remnarace5, S. Ijaz6, S. H. Hussaini1, R. P. Bendall7. 1Cornwall Gastrointestinal Unit, Royal Cornwall Hospital, Truro, UK, 2The New Zealand Liver Transplant Unit, Auckland City Hospital, Auckland, New Zealand, 3Gastroenterology, Derriford Hospital, Plymouth, 4Gastroenterology, Royal Devon and Exeter Hospital, Exeter, 5Gastroenterology, Torbay General Hospital, Torbay, 6Virus Reference Department, Centre for Infections, Health Protection Agency, London, 7Microbiology, Royal Cornwall Hospital, Truro, UK

Introduction: An accurate diagnosis of drug-induced liver injury (DILI) involves the application of the standard criteria for drug-induced liver injury.1 These criteria are based on a temporal relationship between the drug therapy and biochemical evidence of liver injury and the exclusion of alternative diagnoses. We have previously shown that locally acquired hepatitis E is an emerging infection in developed countries, is much more common than previously recognised, has a predilection for elderly males and can be misdiagnosed as DILI.2

Aims & Methods: To study the role of HEV testing in DILI. Retrospective review of a cohort of patients with suspected DILI (n = 69). The standard criteria for DILI were applied to this group. Patients with suspected DILI who met these criteria were retrospectively tested for HEV on stored sera taken at the time of presentation. Patients with confirmed DILI (HEV negative) were compared to a group of patients with locally acquired hepatitis E (n = 45) to determine variables that predicted either diagnosis.

Results: 47/69 had criterion-referenced DILI. 22/47 were HEV negative and thus had confirmed DILI. 6/47 were HEV positive and thus did not have DILI, but had hepatitis E infection. Compared to patients with confirmed DILI, patients with hepatitis E were significantly more likely to be male (OR 3.09, CI 1.05 to 9.08); less likely to present in November and December (0.03, CI 0.01 to 0.52); have lower serum bilirubin (p = 0.015); and higher serum alanine aminotransferase (p<0.001) and ALT/ALKP ratio (p<0.001).

Conclusion: The diagnosis of DILI is not secure without testing for HEV, particularly in elderly male patients with a predominant transaminitis.

ABNORMAL LIVER FUNCTION TESTS IN PATIENTS WITH TYPE 1 DIABETES MELLITUS: SEROPREVALENCE AND CLINICAL CORRELATIONS

J. S. Leeds1, E. M. Forman1, S. Tesfaye2, D. S. Sanders1. 1Gastroenterology, 2Diabetes, Royal Hallamshire Hospital, Sheffield, UK

Introduction: Abnormal liver function is well recognised in Type 2 diabetes mellitus and is associated with non-alcoholic fatty liver disease (NAFLD). There are no large studies examining the prevalence of abnormal liver function and its correlations in patients with Type 1 diabetes mellitus. The aim of this study was to determine the prevalence of abnormal liver function tests (particularly ALT) in a large cohort of patients with Type 1 diabetes and examine the clinical correlations and possible causes.

Aims & Methods: Patients were recruited from the Diabetes centre. Height and weight were measured and body mass index (BMI) calculated. Detailed drug history and other medical comorbidities were noted. Blood was taken for HbA1c, bilirubin, ALT, alkaline phosphatase, albumin and renal function.

Results: 614 patients were included in the study mean age 43.3 years (283 females). The prevalence of abnormal LFTs were bilirubin 29/614 (4.7%), ALT 13/614 (2.1%), ALP 43/614 (7.0%) and albumin 43/614 (7.0%). Males had higher ALT than females (p<0.001) but comparable age and HbA1c. There was no relationship between ALT and age group or BMI on univariate analysis but a significant correlation between ALT and male sex (p = 0.001), increasing HbA1c (p = 0.001) and creatinine (p = 0.006). Multivariate analysis using backwards stepwise logistic regression showed that ALT was significantly associated with creatinine (p = 0.027), male sex (p = 0.006) and HbA1c (p<0.001). Investigation of these patients revealed fatty liver on ultrasound scan in 5 patients (38.5%) and 1 patient admitted to alcohol excess. No biopsies were performed.

Conclusion: Abnormal liver function tests were common in Type 1 diabetes mellitus. Abnormal ALT appears to be associated with worse glycaemic control, male sex and higher creatinine levels. Age and BMI were not related to ALT in this cohort, however in those with raised ALT fatty liver was common.

THE SYSTEMIC INFLAMMATORY RESPONSE TO ACUTE LIVER FAILURE PREDICTS THE DEVELOPMENT OF RENAL FAILURE

J. A. Leithead, J. W. Ferguson, C. M. Bates, J. Davidson, A. J. Bathgate, P. C. Hayes, K. J. Simpson. Scottish Liver Transplant Unit, Royal Infirmary of Edinburgh, Edinburgh, UK

Introduction: Renal dysfunction is a common complication of acute liver failure. The current hypothesis is that the pathophysiological mechanisms are similar to those of hepatorenal syndrome in cirrhosis. However, we hypothesise that the systemic inflammatory response to acute liver failure plays a key role.

Aims & Methods: Our aim was to identify predictive factors for the development of renal failure in acute liver failure and, in particular, to assess the relationship with the systemic inflammatory response. A single-centre study of 454 patients admitted with acute liver failure between 01/92 and 06/07. Renal failure was defined as the need for renal replacement therapy (RRT). The 4 systemic inflammatory response syndrome (SIRS) components are temperature <36or>38°C, heart rate >90 bpm, WCC<4or>12×109/l, and PaCO2 <4.3 kPa.

Results: Mean age at time of presentation was 39.2+/−14.4 years (M:F = 1:1.3). The most common causes of liver injury were paracetamol (70%), viral (11.9%), idiosyncratic drug reactions (6.4%). 236 patients (52%) developed renal failure. The estimated 28-day survival for patients who did and did not develop renal failure was 15% and 58% respectively (p<0.001). The renal failure patients had a faster heart rate (p<0.001) and higher neutrophil count (p<0.001) at the time of hospital admission and were more likely to have a temperature <36or>38°C (p = 0.004). 41% of patients with renal failure had ⩾3 components of SIRS and only 20% of the non renal failure group (p<0.001). An increasing number of components of SIRS was associated with an increased probability of renal failure; 33%, 46%, 52%, 68% and 70% of the patients with 0, 1, 2, 3 and 4 components respectively required RRT (p = 0.004). The renal failure group were more likely to demonstrate infection (p = 0.001). Nevertheless, in both those with (p = 0.017) and those without infection (p = 0.002) the number of components of SIRS on admission was associated with the need for renal support. On multivariate analysis the factors associated with renal failure were fulfilled King’s College Hospital poor prognostic criteria (OR 4.84; 95% CI 2.62 to 8.93, p<0.001), paracetamol as the cause of liver injury (OR 2.75; 95% CI 1.42 to 5.33, p = 0.003), infection (OR 3.18; 95% CI 1.70 to 5.93, p<0.001), hypotension (OR 5.12; 95% CI 2.84 to 9.24, p<0.001) and the number of SIRS components on admission (OR 1.41; 95% CI 1.06 to 1.87, p = 0.017). A patient with 4 components of SIRS was 2.6 times more likely than a patient with no SIRS components to require RRT.

Conclusion: The systemic inflammatory response to acute liver failure is a predictor of renal failure requiring RRT. This supports previous studies linking the systemic inflammatory response with encephalopathy and poor prognosis, and the need for future targeted therapeutic intervention.

ESTIMATED GLOMERULAR FILTRATION RATE IS A PREDICTOR OF DEATH ON THE WAITING LIST FOR LIVER TRANSPLANTATION

J. A. Leithead, J. W. Ferguson, P. C. Hayes. Scottish Liver Transplant Unit, Royal Infirmary of Edinburgh, Edinburgh, UK

Introduction: Serum creatinine is a component of the MELD score, which in the USA is used to prioritise patients on the liver transplant waiting list. However, creatinine is influenced by gender, age and race and, as a result, MELD may be less accurate in some subgroups. Estimated glomerular filtration rate (eGFR) calculated using the MDRD4 equation takes into account these variables and may be a more useful prognostic indicator.

Aims & Methods: Our aim was to compare eGFR and creatinine as predictors of prognosis in patients awaiting liver transplantation. A single centre study of 464 patients listed for liver transplantation between 11/92 and 06/07. Patients listed for HCC, acute liver failure or combined renal transplant were not included. Logistic regression was used to assess the association of variables with 6-month waiting list mortality. The accuracy of traditional MELD (regression coefficients altered for our model) and MELD with eGFR substituted for creatinine in predicting mortality were assessed by the area under the ROC curve.

Results: Mean age at the time of listing was 55.4+/−11.5 years (M:F = 1:1). The main indications for transplantation were PBC (29%), alcoholic liver disease (24%), sclerosing cholangitis (15%). 65 patients (14%) died prior to transplantation. Median time to transplant was 64 (range 1–499) days. Median eGFR of the entire cohort at listing was 69 (IQ range 56–82) ml/min/1.73m2 and median creatinine was 90 (IQ range 77–109) μmol/l. 31% of patients had an eGFR<60 ml/min/1.73 m2 and only 10% had a creat>133 μmol/l. Despite having a lower eGFR than men (median 66 vs 74, p<0.001) women had a lower creatinine (median 82 vs 98, p<0.001) and lower MELD score (median 16 vs 17, p = 0.028). LogeGFR (p = 0.002) and logecreat (p<0.001) were associated with 6-month waiting list mortality on univariate analysis. The presence of an eGFR<60 was predictive of death in both men (p = 0.021) and women (p = 0.003), while creat>133 was not predictive in either sex (men p = 0.088, women p = 0.615). In the entire cohort the relative risk of death by 6 months post transplant in patients with an eGFR<60 was 3.01 (95% CI 1.08 to 8.36, p = 0.035). This association was independent of MELD and logesodium. The area under the ROC curve for predicting 6-month waiting list mortality based on MELD in men was 0.770 (95% CI 0.651 to 0.889) and in women was 0.740 (95% CI 0.593 to 0.887). When MELD was recalculated with logeGFR substituted for logecreat the c-statistic remained unchanged (0.771; 95% CI 0.653 to 0.889) and increased (0.789; 95% CI 0.659 to 0.918) in males and females respectively.

Conclusion: An eGFR<60 ml/min/1.73 m2 is an independent predictor of 6-month liver transplant waiting list mortality. eGFR substituted for creatinine may improve the accuracy of MELD in predicting death.

ANG 1–7 FAILS TO POTENTIATE ACETYLCHOLINE VASODILATATION IN THE PRECONSTRICTED IN SITU PERFUSED CIRRHOTIC RAT LIVER

J. S. Lubel, C. Herath, Z. Jia, L. M. Burrell, P. W. Angus. Department of Medicine, University of Melbourne, Melbourne, Australia

Introduction: Angiotensin 1-7 (Ang-(1-7)) is a vasoactive peptide of the renin-angiotensin system with properties that oppose that of Angiotensin II. In-vitro experiments have shown that Ang-(1-7) potentiates acetylcholine induced dilatation of aorta segments form cirrhotic rats.1 The aim of this study was to determine whether Ang-(1-7) could significantly alter the vaso-responsiveness of the cirrhotic liver to acetylcholine.

Aims & Methods: Male Sprague Dawley rats underwent bile duct ligation for 4 weeks. In-situ perfusion of livers was performed using oxygenated Krebs-Henseleit solution supplemented with 1% bovine serum albumin and 0.1% dextrose delivered at a flow rate of 30 ml/min. Animals were than randomly assigned to vehicle or Ang-(1-7) group (n = 4 each group). Liver viability was assessed by hepatic oxygen extraction, macroscopic appearance and stability of portal pressure. Following stabilisation, livers were perfused with either vehicle or 10-5 mol/l Ang-(1-7) for 20 minutes. Preconstriction was achieved using 10-4 mol/l Methoxamine. Portal pressure was measured using a manometer equilibrated to atmospheric pressure and recorded following cumulative bolus doses of acetycholine (10-8 mol/l–10-4 mol/l).

Results: There was no significant difference in body weight, spleen or liver weight between the two groups. The baseline and maximal portal pressures following methoxamine were similar in both groups. The portal pressures dropped in both groups following the initial two doses of Acetylcholine (10-8 mol/l and 10-7 mol/l), however subsequent doses were associated with increases in portal pressure reflecting altered epithelial responsiveness to these higher doses of acetycholine. Importantly, there was no evidence of an Ang-(1-7) enhanced vasoresponsiveness to acetycholine in cirrhotic livers as has previously been demonstrated in aortas of cirrhotic animals.

Conclusion: Although Ang-(1-7) has antifibrotic activity in the liver there is no demonstrable vasodilatory effect in cirrhotic rat livers either at basal conditions2 or preconstricted with methoxamine. Hepatic endothelial dysfunction is the probable reason for failure of a significant vasodilatory response.

POINT PREVALENCE AND CHARACTERISTICS OF CHRONIC HEPATITIS B IN THE SCREENED POPULATION OF PREGNANT WOMEN IN A CITY WITH A HIGH ETHNIC MINORITY

S. Mahmood1, F. Younis1, J. Rehman1, S. Moreea1, G. Chudasama2, V. Beckett2. 1Gastroenterology, 2Obstetric and Gynaecology, Bradford Royal Infirmary, Bradford, UK

Introduction: Bradford has a South Asian population of 18.5% (2001 census), with an unknown hepatitis B (HBV) prevalence. The Pakistan National Consensus Conference on Guidelines for Hepatitis B and C, Karachi, 2003, described the prevalence of HBsAg at 2.5% in pregnant women, of which 17% were eAg +ve and 61% anti HBe +ve1. As all pregnant women are screened for HBV we surmised that this would approximate the current prevalence of HBV in our local population.

Aims & Methods: The aim was to determine the prevalence and characteristics of HBV in pregnant women in relation to ethnicity between June 2005–September 2006. The method used was a retrospective analysis of the HBV databases from the Obstetrics/Gynaecology and Hepatology departments to obtain the following: demographics, percentage of women with a positive HBsAg test as well as their eAg/biochemical/virological status. Their attendance to clinic and requirement for anti-HBV treatment was also studied.

Results: During the study period there were 4817 pregnancies–53% Asians, 42% white Caucasians, 3% Africans and 2% unknown. 99% accepted screening for HBV. 42 women (0.88%) were HBsAg positive—29 Asians (prevalence in Asians = 1.1%), 8 white Caucasians (prevalence 0.4%), 5 Africans (prevalence 3%). All 42 HBV positive women (8 eAg +ve, 34 eAg –ve (81%)) were referred to the hepatology clinic. 33 (79%) women attended clinic, their characteristics are as follows: 8 (24%) eAg +ve, 2 of whom were immunotolerant and 3 had ALT>40 IU/ml and viral load (VL) >20 000 IU/ml. There were 25 (76%) eAg -ve, only 4 had a VL >2000 IU/ml, 1 of whom had an ALT>40. 2/33 (4.8%) patients are on treatment with Lamivudine. 11/29 (38%) of the Asian HbSAg +ve patients were born in Pakistan.

Conclusion: This study suggests a lower prevalence of HBV in our Asian population presumably due to a mixture of first and second generation women and some of who may have been immunised against HBV. The prevalence of HBV in African women is high due to the fact that they are mostly a first generation population. The prevalence of HBV in the Caucasian population is low as expected. There is a high attrition rate to the hepatology clinic. This study gives an insight into the approximate prevalence and characteristics of HBV in the local population and may help other areas with a high ethnic minority plan HBV services.

BLOOD GLUCOSE AND GLYCOSYLATED HAEMOGLOBIN ARE INSENSITIVE DIAGNOSTIC INDICATORS OF GLUCOSE INTOLERANCE IN PATIENTS WITH CIRRHOSIS

K. J. Rhee1, K. A. Chandarana1, R. Ruparelia1, E. Woolman2, S. Kanagalingam1, M. Thomas2, M. Y. Morgan1. 1Centre for Hepatology, Department of Medicine, Royal Free Campus, Royal Free and University College Medical School, 2Department of Clinical Biochemistry, Royal Free Hospital, Royal Free Hampstead NHS Trust, London, UK

Introduction: The prevalence of impaired glucose tolerance and diabetes mellitus in patients with cirrhosis ranges from 43–80% and 10–29% respectively, although these data are based on relatively small series of patients.1 The presence of diabetes has a significantly negative effect on survival, particularly following orthotopic liver transplantation.2 Currently there are no guidelines for the diagnosis of impaired glucose metabolism in this patient population.

Aims & Methods: The aims of this study were to determine (1) the frequency of impaired glucose tolerance in patients with cirrhosis and (2) the sensitivity and specificity of random/fasting blood glucose concentrations and glycosylated haemoglobin (HbA1C) for the diagnosis of impaired glucose metabolism in this patient population. All patients with biopsy-proven cirrhosis attending a hepatology out-patient clinic over one three month period were recruited for the study; patients with diabetes mellitus were excluded. Fasting/random blood glucose and HbA1c levels were measured. An extended glucose tolerance test (GTT) was undertaken after consumption of a 75 g glucose load (394 ml original Lucozade); blood glucose concentrations were measured at baseline and at 30 minute intervals for 3 hours.

Abstract 217 Prevalence of abnormal glucose tolerance in patients with cirrhosis

Results: The total population comprised 55 individuals; eight (15%) were known to have diabetes. The remaining 47 patients (28 women, 19 men; mean age 56 (range 39–80) years) were assessed per protocol. The majority of patients showed one or more abnormalities of glucose tolerance,3 viz: diabetes (10; 21%); impaired glucose tolerance (10; 21%); impaired fasting glucose (8; 17%; late hypoglycaemia (15; 32%). Only the minority (15; 32%) had normal glucose tolerance. Routinely used screening blood tests were not sufficiently sensitive for diagnostic purposes (table).

Conclusion: Patients with cirrhosis show significant abnormalities of glucose metabolism which can not be diagnosed reliably using conventional blood tests. These patients should undergo intermittent oral GTTs and a low threshold for treatment should be adopted.

SLOW-RELEASE CARBOHYDRATES: A RATIONAL, PATIENT-ACCEPTABLE APPROACH FOR THE NUTRITIONAL MANAGEMENT OF PATIENTS WITH CIRRHOSIS?

K. A. Chandarana1, S. J. Eaton2, M. Y. Morgan1. 1Centre for Hepatology, Department of Medicine, Royal Free Campus, Royal Free and University College Medical School, 2Unit of Paediatric Surgery, Institute of Child Health, Royal Free and University College Medical School, London, UK

Introduction: Patients with cirrhosis progress to a state of early, accelerated starvation even after short periods of fasting. The European Society for Parenteral and Enteral Nutrition (ESPEN) has, therefore, recommended that these individuals should consume small frequent meals throughout the day, together with a late-night snack.1 However, this frequent feeding requirement is disruptive and could result in the consumption of high fat, high carbohydrate snacks rather than a balanced redistribution of food portions. Therefore, other, more patient-acceptable approaches of modulating fuel use are required.

Aims & Methods: The aim of this study was to determine if use of corn-starch, a slow-release carbohydrate, might provide a more convenient and sustainable way of preventing early, accelerated starvation in patients with cirrhosis. Eight patients (6 men: 2 women; mean (range) age, 56 (46–64) years; mean BMI, 23.3 (17–27)) with biopsy-proven alcohol-related cirrhosis and normal random blood glucose and HbA1c levels were selected for study; none had consumed alcohol for at least 3 months. Seven healthy volunteers (all men; mean age 32 (21–57) years; mean BMI, 25.6 (21–31)), served as controls. Subjects were provided with 50 g of corn starch dissolved in 150 ml water to drink over 30 s. Blood glucose concentrations were measured immediately before, and at 30 min intervals for up to 7 h post corn starch ingestion. Corn starch use was assessed by measurement, at similarly timed intervals, of breath 13CO2 using isotopic ratio mass spectrometry.

Abstract 218 Blood glucose-time profile and corn-starch use

Results: There were no significant differences between the patients and the healthy volunteers in the corn starch release pharmacokinetics. However, corn starch ingestion was associated with greater variability in blood glucose concentrations and in blood-glucose time profiles in the patients (table).

Conclusion: In patients with cirrhosis corn starch is used adequately but results in more sustained fluctuations in blood glucose levels than in healthy volunteers, even in the absence of obvious glucose intolerance. This may limit its usefulness, and that of other sustained release carbohydrates, as nutritional supplements in this patient population.

SYSTEMATIC REVIEW OF THE DIAGNOSTIC PERFORMANCE OF SERUM MARKERS OF LIVER FIBROSIS IN ALCOHOLIC LIVER DISEASE

J. Parkes1, I. Guha2, S. Harris1, W. M. C. Rosenberg3, P. J. Roderick1. 1Public Health Sciences and Medical Statistics, 2Department of Hepatology, University of Southampton, Southampton, 3Institute of Hepatology, University College London, London, UK

Introduction: Alcoholic liver disease (ALD) is an important consequence of hazardous drinking, and a significant cause of death and morbidity in many countries. Detection of liver fibrosis at an early stage could provide opportunities for more optimal management. Serum markers of liver fibrosis offer an attractive alternative to liver biopsy. Evidence of the diagnostic performance of such biomarkers of liver fibrosis in ALD is needed to assess the effectiveness of such tests. To provide such evidence, a systematic review of diagnostic studies of serum markers was conducted.

Aims & Methods: MEDLINE, EMBASE, and reference lists from retrieved articles were searched. Studies were included if they evaluated paired samples of liver biopsy and serum, included >30 participants, and presented data as sensitivity, specificity, or receiver operator characteristic curves.

Results: 12 studies were eligible for inclusion. There were 14 separate markers—5 evaluated as single markers and the rest as components of 6 panels of markers. Studies were heterogeneous, and were generally smaller and older than similar studies in other liver diseases. Serum markers were able to identify/exclude those people with severe fibrosis/cirrhosis with good diagnostic accuracy at the thresholds presented, performance was less good at detection of less severe levels. Both panels of markers and hyaluronic acid (the best single marker but in few studies) performed well to identify cirrhosis/severe fibrosis, although there was no consistency in threshold of HA reported by studies.

Conclusion: In ALD there are few small studies that have evaluated serum markers ability to identify fibrosis on biopsy. These showed that they are promising in the identification of cirrhosis/severe fibrosis. Due to the paucity of studies and lack of direct comparisons of single markers and panels it is not possible to recommend any one specifically. Markers have lower diagnostic accuracy at lesser stages of fibrosis where panels tended to perform better than single markers. More research needed into detecting early ALD.

HIGH FREQUENCY OF TISSUE TRANSGLUTAMINASE ANTIBODIES IN PATIENTS WITH ALCOHOLIC LIVER DISEASE

C. A. Salmon, J. S. Leeds, R. J. Atkinson, D. S. Sanders, D. C. Gleeson, M. A. Karajeh. Liver and Gastroenterology Unit, The Royal Hallamshire Hospital, Sheffield, UK

Introduction: It has been suggested that a link exists between chronic liver disease and coeliac disease. Earlier reports have shown high rates of false positive assays for antigliadin and tissue transglutaminase (tTG) antibodies in patients with chronic liver disease. Human anti tTG has been found to be more specific than guinea pig tTG, however there are few data specifically assessing the relationship between human tTG and severe alcoholic liver disease (ALD).

Aims & Methods: The aim was to assess the reliability of human tTG in patients with severe ALD. We also assessed for correlation between tTG and the severity of the liver disease as measured by albumin and bilirubin. We identified patients who had coeliac serology performed (for various clinical indications) from a database of patients with severe ALD defined as portal hypertension, previous decompensation or biopsy proven cirrhosis. Patients with inflammatory bowel disease (IBD) and irritable bowel syndrome (IBS) were age and sex matched to provide two control groups. All patients had coeliac serology with anti-tTG using human recombinant tTG linked gliadin specific peptides and IgA endomysial antibody (EmA). The patients with ALD also had bilirubin and albumin as a measure of their liver disease. Patients with positive tTG were offered duodenal biopsy.

Results: We identified 74 ALD patients (mean age 54.2 years, 47 males). These were age and sex matched to the same number of patients in the IBD and IBS groups. tTG was positive in 32/74 (43.2%) with ALD, 6/74 (8.1%) with IBD and 9/74 (12.2%) with IBS (p<0.0001). Median tTG (IQR) levels in those with ALD, IBD and IBS were 13 (17.25), 4 (6) and 3 (6) respectively (Kruskal-Wallis, p<0.0001). EmA was positive in 1/74 with ALD, 1/74 with IBD and 4/74 with IBS. Of the patients with a positive tTG, 27 of 33 (81.8%) ALD, 4 of 6 (66.7%) IBD and 8 of 9 (88.9%) IBS had duodenal biopsies. These revealed no cases of coeliac disease in the ALD group, whereas 2 patients in the IBD group and 4 patients in the IBS group had histological changes of coeliac disease. There was a positive correlation between tTG and bilirubin (Spearman’s rho 0.39, p = 0.001) and a negative correlation between tTG and albumin (Spearman’s rho, −0.38, p = 0.001).

Conclusion: In patients with severe ALD there is a high frequency of false positive tTG. This was not related to the presence of a positive EmA or coeliac disease on biopsy. This suggests that tTG in isolation is not a good screening tool for coeliac disease in patients with ALD. There does seem to be a correlation between the level of tTG and the severity of liver disease raising the possibility of an aetiopathological role which merits further exploration.

LONG-TERM FOLLOW-UP OF BONE MINERAL DENSITY FOLLOWING ORTHOTOPIC LIVER TRANSPLANTATION

D. A. Sheridan1, L. Smith1, R. M. Francis2, M. Hudson1. 1Liver and Transplant Unit, 2Musculosketetal Unit, Freeman Hospital, Newcastle upon Tyne, UK

Introduction: With increasing survival following orthotopic liver transplantation (OLT), data regarding the long-term impact on bone mineral density (BMD) are needed. We conducted an audit of 108 patients attending the liver transplant annual review clinic to determine assessment and treatment of reduced BMD up to 13 years following OLT.

Aims & Methods: A retrospective review of notes from the annual review clinic collected age at time of OLT, diagnosis, immunosuppression, body mass index and treatment with calcium/vitamin D3 and bisphosphonates. Results of total lumbar spine and total hip BMD by DEXA scanning at initial assessment and intervals up to 13 years following OLT were recorded. Comparison was made between those whose primary indication for OLT was cholestatic liver disease (PBC + PSC; n = 31) to those with hepatitic liver disease (alcoholic liver disease, hepatitis B/C and autoimmune hepatitis; n = 50). 27/108 patients had acute liver failure with no pre-transplant assessment of BMD.

Abstract 221 Bone mineral density before and after liver transplantation

Results: Overall those with cholestatic liver disease had significantly lower pre-transplant BMD of the total lumbar spine and total hip than the hepatitic group (p = 0.033). Steady recovery in bone density of the lumbar spine is then seen but a decline in bone density of the hip at 1 year post OLT in the cholestatic group. Subgroup analysis showed this decline to be attributable to those maintained on long-term steroids. There was no significant difference in long-term bone density between those on tacrolimus compared to cyclosporin A at the lumbar spine (0.977±0.191 and 0.972±0.158 g/cm2; p = 0.963) or at the hip (0.791±0.109 and 0.822±0.146 g/cm2; p = 0.649). Timing of DEXA scans post OLT and use of bone prophylaxis was variable.

Conclusion: Short-term fracture risk in patients coming to liver transplantation is greatest in the first year in those with cholestatic liver disease. Long-term fracture risk declines in the majority following liver transplantation as bone density of the lumbar spine recovers. However those on long-term steroids have a fall in total hip BMD which may increase long term hip fracture risk. This audit highlights that evidence based guidelines for the long-term assessment of bone density and bone prophylaxis following liver transplantation are needed and should be targeted to high risk individuals.

THE INFLUENCE OF CHRONIC HEPATITIS C INFECTION AND HOST APOLIPOPROTEIN E GENOTYPE ON SERUM LIPID PROFILES

D. A. Sheridan1, D. A. Price2, P. Donaldson1, G. Toms1, D. Neely3, M. Bassendine1. 1Liver Research Group, Medical School, 2Department of Medicine, 3Department of Clinical Biochemistry, Royal Victoria Infirmary, Newcastle upon Tyne, UK Kingdom

Introduction: Hepatitis C virus (HCV) utilises pathways of lipoprotein metabolism within the liver for viral replication and infectivity. Host ApoE genotype is a major determinant of serum LDL cholesterol levels and correlates with cardiovascular risk. ApoE genotype may influence outcome in acute HCV infection.1 It has also been suggested that LDL cholesterol level may influence response to anti viral therapy.2

Aims & Methods: In the present study we assessed the influence of HCV genotype and host ApoE genotype on serum total cholesterol, non HDL cholesterol, triglycerides and HDL in HCV infected patients. These parameters were compared in those achieving a sustained virological response (SVR) and non-responders. Paired pre and 24 weeks post treatment lipid profiles were retrospectively obtained in 100 patients who achieved a sustained virological response (SVR) and 66 non-responders (NR) to antiviral therapy. Each individual thereby acted as their own control for age, sex and BMI. ApoE genotype was compared in 72 SVRs and 57 non-responders and 288 healthy controls.

Results: Those with HCV genotype 3 had significantly lower pre treatment total cholesterol levels than HCV genotype 1 (3.97 mmol/l±0.14 and 4.63 mmol/l±0.12 respectively, p<0.001). This effect was reversible in the SVR group but persisted in the non responders post treatment. In both SVRs and non-responders, non-HDL cholesterol was also significantly lower in HCV genotype 3 (2.80 and 2.52 mmol/l) than genotype 1 (3.63 and 3.31 mmol/l; SVR and NR respectively; p<0.001). Triglyceride level significantly increased in those with HCV genotype 3 achieving a SVR but there was no difference in HDL between HCV genotypes, SVR’s or non responders. In the HCV infected patients, the Apo ∊2/∊3 genotype was found in only 3.5% of non-responders compared to 9.7% in SVR’s and 11.1% in a healthy control group. In those infected by HCV genotype 1, cholesterol levels correspond to expected levels for ApoE genotype. In contrast, in HCV genotype 3 infection, cholesterol levels were lower than expected in ∊3/∊3 and ∊3/∊4 genotypes.

Conclusion: The influence of host ApoE genotype should be considered in investigating the association of serum cholesterol levels and treatment outcome in HCV infection. We found that HCV genotype 3 lowers total and non-HDL cholesterol levels, and exerts a greater influence than HCV genotype 1 or host ApoE genotype.

BACTERIAL TRANSLOCATION AND INTESTINAL DYSFUNCTION IN CIRRHOSIS

U. Thalheimer1, F. De Iorio1, F. Capra1, M. Lleo2, V. Zuliani1, V. Ghidini2, M. Tafi2, G. Caburlotto2, M. Gennari2, I. Vantini1. 1Dipartimento di Scienze Biomediche e Chirurgiche, Policlinico GB Rossi, 2Dipartimento di Patologia, Sezione di Microbiologia, Universitá di Verona, Verona, Italy

Introduction: Bacterial infections are common in cirrhotic patients and seem to play an important role in the development of the complications of cirrhosis.1 Bacterial translocation seems to be the initial step preceding the occurrence of overt bacterial infection in the cirrhotic patient. The presence of bacterial DNA in blood and ascites has been reported to correlate well with the presence of bacterial translocation in rats. A recent study reported the presence of bacterial DNA in serum and ascites in patients with advanced cirrhosis without signs of infection.2

Aims & Methods: Our aim was to search for bacterial DNA in a range of cirrhotic patients both with and without ascites, and to study its correlation with any abnormality of intestinal motility or permeability and the presence of bacterial overgrowth. We studied 7 cirrhotic patients with or without ascites. Exclusion criteria were age less than 18 years, presence of the systemic inflammatory response syndrome (SIRS), overt infection or positive blood or ascitic fluid cultures, upper GI bleeding or antibiotics within the preceding 2 weeks, intestinal disorders including major abdominal surgery, use of any drug with an effect on small bowel motility such as beta blockers, lactulose or prokinetics, presence of HCC or portal vein thrombosis, acute alcoholic hepatitis, pregnancy and refusal of consent. Serum and ascites samples were obtained on day 1, and serum samples were taken twice daily for the following 3 days. Bacterial DNA was assayed by PCR using universal primers for rRNA 16s. Oro-caecal transit time and bacterial overgrowth were assessed with H2 breath testing. Intestinal permeability was assessed by determining urinary lactulose and mannitol excretion with HPLC.

Results: Six patients were male and 1 female, age range was 42–78 years, aetiology was alcohol in 4, HCV in 2, HBC in 1. Ascites was present in 4 and Child Pugh grade was A in 4 and B in 3. All patients had increased intestinal permeability, 6 had decreased transit time and 1 had bacterial overgrowth. In only one patient (with ascites), PCR was positive for bacterial DNA both in ascites and serum for all 4 days on which samples were taken.

Conclusion: Bacterial translocation as assessed by presence of bacterial DNA is less common in less severe cirrhosis and seems to be anticipated by changes in intestinal motility and permeability.

INCREASED MORTALITY AND LENGTH OF STAY RELATED TO ABNORMAL LIVER FUNCTION TESTS ON ADMISSION TO THE GENERAL INTENSIVE CARE UNIT: A 6 MONTH RETROSPECTIVE ANALYSIS

S. J. Thomson1, M. L. Cowan2, S. Musa3, P. Collinson4, M. Grounds5, T. M. Rahman3. 1Gastroenterology and Intensive Care, 2Gastroenterology, 3Gastroenterology and Intensive care, 4Chemical Pathology, 5Intensive Care, St George’s Hospital, London, UK

Introduction: Abnormalities of liver enzymes and albumin are common in the critically ill patient for a number of reasons. Liver function tests (LFT) are routinely analysed at the time of admission to the Intensive Care Unit (ICU) but the significance of abnormal results is not clear. We sought to define the incidence of abnormal LFT and effect on length of stay and 30 day mortality.

Aims & Methods: The results of LFT of patients undergoing first admission to the General ICU of a large teaching hospital in South London between 1 October 2006 and 31 March 2007 were obtained from Chemical Pathology. Mortality statistics were obtained from the hospital records.

Abstract 224 Length of stay and mortality in 661 ICU patients according to LFT on admission

Results: A total of 661 patients had a first admission to the ICU during the 6 months studied. The average age was 59.7 years (SD 18.1), 58% were male and the mean length of stay was 4.1 days (range 1–45). Mortality rate at 30 days was 22.2% (147/661). On the day of admission serum albumin was <35 g/dl in 614 patients (93%; median 22 g/dl). Excluding albumin, only 333 (50%) had entirely normal LFT. Average length of stay (LOS) was greater in those with an abnormal bilirubin, or albumin below 22 g/dl on admission. Abnormalities in each of ALT, AKP, GGT and Albumin below 22 g/dl on admission, were associated with greater 30 day mortality.

Conclusion: Abnormality of LFT is common in the critically ill patient admitted to the General ICU. Albumin is almost invariably below the normal range. Relatively minor elevations of liver enzymes are associated with increased LOS and 30 day mortality. Lower levels of albumin at the time of admission are likely to reflect chronic illness and therefore result in prolonged stay and mortality. The cause of these abnormalities is likely to be multifactorial and further prospective studies are needed to elucidate the cause.

REVIEW OF LIVER BIOPSIES IN A HEPATOLOGY CENTRE: HOW OFTEN AND WHY IS THE DIAGNOSIS CHANGED?

J. I. Wyatt. Histopathology Department, St James’s University Hospital, Leeds, UK

Introduction: 160 liver biopsies/year are sent to Leeds for review, at request of the local pathologist (PR, x/year) or because the patient is referred to Leeds and there are uncertainties about the original report (CR, y/year). To determine the proportion of medical biopsies where review had affected patient management, original and referral reports for 100 consecutive cases were reviewed.

Aims & Methods: Inclusion criteria: adult medical liver biopsies referred to Leeds were original pathology report was available for review. The original pathologist’s diagnosis was compared retrospectively with that in the centre histology report (JIW) which included a note of any CPC discussion. Original clinical information included in the referred report was noted, together with further information from correspondence and at CPC discussion. Cases were categorised as major change in diagnostic category (for example, chronic hepatitis to chronic biliary disease, possible on basis of original clinical information and histology slides alone), minor change in interpretation within category (for example, most likely aetiology, change in stage, second diagnosis, including information from CPC), or no change.

Results: Biopsy review and clinical discussion let to a major change in 21% and minor change in 27% diagnoses. Contributory factors included misleading clinical information (2) misinterpretation of special stains (2). Review most often resulted in a major change in chronic biliary disease (9/25), acute hepatitis (3/12), cholestases (2/6) and vascular outflow (2/5). Incorrect interpretation of the Shikata stain for copper associated protein contributed to 7/9 instances in chronic biliary disease. Minor changes were usually a result of insights from additional clinicopathological correlation, and showed less clustering in diagnostic categories. No more specific diagnoses was achieved in 8 cases originally reported as normal/non-specific.

Conclusion: Biopsy review led to revised diagnosis in 48% cases. A major diagnosis not recognised initially in 21%; 11/21 may have been avoided with correct clinical information and appreciation of special stains. A minor change in 27, more often due to specialist interpretation in the light of clinical pathological discussion. This audit has highlighted areas of liver histology CPD helpful to DGH pathologists.

AUDIT OF CHANGE IN NEEDLE BIOPSY GAUGE FOR ULTRASOUND GUIDED MEDICAL LIVER BIOPSIES: ROOM FOR FURTHER IMPROVEMENT

J. I. Wyatt, S. Karthik, D. Treanor. Histopathology Department St James’s University Hospital, Leeds, UK

Introduction: Histopathological diagnosis in hepatology depends on the features of diffuse liver disease being adequately represented in the biopsy specimen. Larger biopsies give more diagnostic information. Pragmatically biopsies with >6 portal tracts (PTs) are considered adequate1 although >10 portal tracts are necessary to avoid underscoring chronic hepatitis C.2 To reduce inadequate biopsies, we changed from 18 g to 16 g needles for routine ultrasound guided outpatient medical liver biopsies during summer 2006. This is a blinded retrospective audit of the size and diagnostic value of tissue sections obtained from liver biopsies during this transition period.

Aims & Methods: 40 consecutive patients biopsied during May–July 2006 and their needle gauge (18 g or 16 g) were identified from radiology records. Slides were retrieved and studied retrospectively, blinded to needle gauge. Objective analysis: Slides were scanned (imagescope) and the length, width and area of the best level were measured using virtual slide viewing software. Subjective analysis: Slides were reviewed by one pathologist (JIW) for: the number of PTs in the best level; assessment of biopsies as inadequate (no useful diagnosis possible), insufficient (for example, evidence of chronic biliary disease? ductopaenia; hepatitis C, insufficient for scoring), sufficient for full diagnosis.

Results: Two of 22 18 g cases were excluded because of multiple passes. 16 g needles (n = 18) produced significantly wider sections than 18 g needles (n = 20) (mean 0.96 and 0.85 mm respectively; p<0.037) but there was no difference in length (mean 11.16 vs 11.19 mm) or tissue area (mean .8.23 mm2 vs 7.86 mm2) Median PT counts were 9 (4–17) for 16 g and 7 (4–12)for 18 g (p = NS); on average there were 7.8PT/10 mm for 16 g biopsies and 5.5PT/10 mm for 18 g biopsies. Only one biopsy from each group was inadequate for any useful diagnosis. Biopsies assessed as sufficient for full diagnosis increased from 45% (18 g) to 67% (16 g) (p = 0.31).

Conclusion: Wider biopsy needles provide significantly wider specimens; these have more portal tracts, although the range is large and the difference did not reach significance. Subjectively, they are more often sufficient for full histological diagnosis. However with 16 g needle this is still only achieved in 67% biopsies, and a broader needle or second pass would be needed to improve the diagnostic value of medical liver biopsies further.

Nutrition posters

IGG ANTIBODIES TO FOODS ARE OF NO RELEVANCE TO IGE MEDIATED FOOD ALLERGY

I. Seerat1, N. M. Croft1, A. Hickey2, H. Longhurst2. 1Digestive Diseases Clinical Academic Unit, 2Immunopathology, Barts and the London, London, UK

Introduction: IgG food antibodies have been claimed to be important in both enhancement and suppression of IgE mediated food allergic reactions. They are often measured in commercial laboratories with little in the way of clinical evidence to justify this.

Aims & Methods: The aim of this study was to examine the serum concentrations of IgG antibodies to casein, egg and peanut in children with IgE mediated allergy to dairy, egg and peanut. Patients with food allergy and controls were recruited from the allergy and paediatric gastroenterology clinics. Controls were children having bloods taken in the paediatric gastroenterology clinic with no atopy or inflammatory bowel disease. Total serum IgE, and specific IgE and IgG antibodies to casein, egg and peanut using the Pharmacia Unicap system were measured. Medians were compared using the Mann-Whitney U test.

Results: A total of 94 children were recruited. 62 had food allergies requiring dietary avoidance (mean age 6 years), with raised RAST or SPT and/or acute onset within one hour of ingestion. 21 reacted to dairy products, 25 to egg and 27 allergic to peanut. There were 32 controls, with a median age of 11 years. As expected the allergic group had higher total IgE and RAST concentration to casein, egg and peanut than the controls. All three food antibody IgG levels were higher for the food allergy group as a whole than the controls (p<0.05). However in the food allergic patients there were no significant differences found in the food IgG to which they react (for example, casein IgG in dairy allergy) when compared to allergic children without that specific allergy.

Conclusion: Although food allergic children have higher levels of food specific IgG antibodies than controls, these data suggest that IgG food antibodies in serum are not related to the presence (or absence) of IgE mediated food allergic reactions. Any direct relevance to the pathogenic process remains unlikely.

MUSCLE CRAMPS ARE THE COMMONEST SIDE EFFECT OF HOME PARENTERAL NUTRITION

D. A. Elphick1, M. Baker2, J. P. Baxter3, J. Nightingale2, T. Bowling4, M. E. McAlindon1. 1Department of Gastroenterology, Royal Hallamshire Hospital, Sheffield, 2Nutrition Support Team, Leicester Royal Infirmary, Leicester, 3Nutrition Support Team, Ninewells Hospital, Dundee, 4Nutrition Unit, Queen’s Medical Centre, Nottingham, UK

Introduction: Quality of life is adversely affected by the need for home parenteral nutrition (HPN), partly because of complications of feed delivery. The major complications of catheter related sepsis, venous thrombosis and liver disease are well recognised but infrequent.

Aims & Methods: This study aimed to determine if there were other minor, but common complications that caused patient distress. All patients from four HPN centres were asked if they suffered any side effects of parenteral nutrition and whether these side effects related to the timings of the feed or required specific intervention.

Results: 45 patients (24 male, 21 female, median age 47 (23–75) years) were included in the study. The patients had been on HPN for median 2 (0.3–16) years, and administered feed on median 5 (2–7) days per week. Reported side effects from HPN were muscle cramps (n = 7), nausea (n = 3), headache (n = 2) hand/foot swelling (n = 1), palpitations (n = 1), weakness (n = 1), ache at Hickman line site (n = 1) and thirst (n = 1). Hot flushes/sweating were reported by four patients, but this is likely to be a physiological response due to thermogenesis. Of the seven patients reporting cramps as a side effect of their HPN, six had tried quinine and/or baclofen, and 2 had slowed the rate of administration of their feeds to reduce the symptoms. The underlying pathologies leading to HPN requirement in these 7 patients were Crohn’s disease (n = 2), refractory coeliac disease, sclerosing mesenteritis, small bowel intussusception, mesenteric ischaemia and intestinal dysmotility. When asked specifically whether they suffered muscle cramps in relation to their HPN, a further 5 patients (small bowel dysmotility (n = 4), mesenteric ischaemia (n = 1)) reported this to be the case; 3 of whom had tried baclofen or quinine to reduce symptoms but none had altered their feed administration. A further 11 patients suffered muscle cramps at times unrelated to parenteral nutrition administration. 22 did not suffer from muscle cramps. Serum electrolytes showed no consistent difference between those with cramps and those without.

Conclusion: Muscle cramps were the commonest minor side effect of HPN. Although not life-threatening, the cramps were of sufficient severity to warrant pharmacological intervention in 9 of 12 patients who had cramps in relation to feeds, and parenteral nutrition administration was slowed in 2 of the 12. Muscle cramps are therefore likely to affect the quality of life of these patients.

A NEW ERA IN ENTERAL NUTRITION PRACTICE?

S. Khair, S. Karanth, H. Lawrence, S. Galbraith, S. McClean, C. Hames, Q. Tuffnell, S. Cottee, J. Woodward. Gastroenterology and Clinical nutrition, Addenbrooke’s Hospital, Cambridge, UK

Introduction: Tube feeding into the stomach via nasogastric or percutaneous gastrostomy tubes revolutionised the practice of enteral alimentation in the latter half of the twentieth century. Paradoxically, greater use of gastrostomies resulted in higher mortality rates—with one month mortality rates as high as 20% reported by 2002. This study aimed to determine the effects on enteral nutrition practice of modern enteral feeding techniques and patient selection.

Aims & Methods: Retrospective review of records of all patients receiving gastrostomy tubes during the same period over the years 2000–1, 2003–4 and 2006–7 from one institution. Indications and outcomes were recorded. Nasal bridles were routinely placed from 2003 for patients who experienced frequent nasogastric tube displacement. A “feeding issues multidisciplinary team” (FIMDT) was established in 2006 to discuss all gastrostomy requests and other contentious nutrition issues.

Results: (1) Fewer gastrostomy tubes were placed in each time period—89 (2000), 80 (2003) and 69 (2006). (2) The proportion of gastrostomies placed for cerebrovascular disease declined—54 (61%), 30 (38%) and 9 (13%). (3) The proportion of gastrostomies placed for head and neck cancer increased—11 (12%), 8 (10%) and 26 (38%). (4) A dramatic reduction in 30-day mortality rates was observed: 15 (17%), 6 (8%) and 1 (1.4%). In 2000, 9/15 deaths within one month of gastrostomy insertion were in patients following stroke. (5) The use of nasal bridles to secure nasogastric tubes increased from 11 in 2003 to 63 in 2006. 22% of nasal bridles were used in stroke patients. Of 14 patients in 2006 requiring a bridled nasogastric tube, only one required a gastrostomy for long-term feeding.

Conclusion: The last five years have seen a remarkable change in the practice and outcome of enteral nutrition due to the improved delivery of nasogastric feeding and better patient selection for gastrostomy tube placement. In most reported series, stroke patients comprise the majority requiring gastrostomy placement. These patients are at high risk of dying within one month of the procedure. Careful selection of patients by discussion at the FIMDT and the use of nasal bridle fixation has resulted in a reduction of gastrostomy placements in these patients with resultant improvement in overall outcome.

ARE WE RECOGNISING AND APPROPRIATELY TREATING MALNUTRITION AMONG HOSPITAL IN-PATIENTS?

C. Lamb, J. Parr, M. Warren. Gastroenterology, North Tyneside General Hospital, Tyne and Wear, UK

Introduction: The UK National Institute for Health and Clinical Excellence (NICE) released guidance on “Nutrition support in adults” in February 2006, recommending that all in-patients be screened for malnutrition on admission to hospital. Appropriate and immediate referral to nutrition and dietetic services should then be undertaken.

Aims & Methods: The study was carried out on a single day in 2007 and based in a hospital serving a population of 350 000 in North East England. A screening tool was in use within the hospital. Local policy dictates its use on all patients at admission and referral to nutrition and dietetics services of those identified as suffering from malnutrition. Compliance with this policy was measured by prospective audit of both the screening tool use and whether those identified as malnourished were referred for nutritional support. Patients provided with nutritional support—for example, food supplements without appropriate referral were also recorded. The tool was then applied to all those not assessed on admission to accurately measure malnutrition on the day of study. All adult patients admitted for more than 24 h were included.

Results: All 328 (100%) patients meeting inclusion criteria were assessed (185 female, 143 male, mean age 71.1 years). Median length of admission was 8 days. In 226 (68.9%) cases, the screening tool had been completed and 74 patients (22.6%) were malnourished. Only 26 of these patients (35.1% of total identified) were appropriately referred to dietetic services. No action was taken in 37 (50%) patients identified as malnourished. 11 patients (14.9%) received dietary intervention without formal dietetic referral. Assessment of the remaining 102 “unscreened” patients revealed a prevalence of malnutrition across the whole cohort of 31.1% (102/328) thus 27.5% (28/102) of malnourished patients had gone unrecognised and untreated.

Conclusion: Using a “NICE-compliant” malnutrition screening tool the prevalence of malnutrition was 31.1% in our in-patient population. National guidance states that all patients should be screened for malnutrition on admission. In practice this occurred in only two thirds (68.9%) of patients thus over a quarter (27.5%) of the total malnourished cohort went unrecognised. Moreover only a third (35.1%) of the identified group were appropriately referred to dietetic services and a further 15.1% were provided nutritional support without appropriate referral. We conclude therefore that even with a locally developed screening tool and policy in place, current levels of recognition and treatment of malnutrition are low and opportunities for intervention are being missed. In the face of recent NICE guidance, hospitals must ensure that staff are properly trained and dietetics services equipped to meet the demand that will arise from the increased identification of malnourished patients.

UNDERSTANDING MORTALITY AFTER GASTROSTOMY TUBE INSERTION: USE OF A SURVIVAL MODEL BASED ON COMORBIDITY

G. R. Longcroft-Wheaton1, P. Marden2, B. Colleypriest3, D. Gavin3, G. Taylor4, M. Farrant5. 1Gastroenterology, St Richards Hospital, Chichester, 2Gastroenterology, Bristol Royal Infirmary, Bristol, 3Gastroenterology, 4Research and Development Support Unit, Royal United Hospital, 5Gastroenterology, Royal United Hospital, Bath, UK

Introduction: There has been considerable discussion concerning percutaneous gastrostomy tube (PEG) feeding and the high mortality seen after insertion. This study aims to improve our understanding of the causes of mortality of in-patients receiving a PEG by comparison with a survival curve predicted from a model proposed by Levine et al.1

Aims & Methods: A retrospective study of patients receiving a PEG over an 18-month period was performed. 55 cases were analysed. 44 were found to be eligible for inclusion. A Levine score was calculated for this cohort. A survival curve after PEG was produced and compared to the Kaplan-Meier curve predicted by the Levine model.

Results: The mortality at 1, 3, 6 and 12 months was 16%, 20%, 25% and 28% respectively. This matched the predicted death rate from the Levine model closely (Pearsons = 0.96). A χ2 was performed, which confirmed no statistically significant difference in survival between the 2 groups (p = 0.35).

Conclusion: We found that the mortality of patients receiving a PEG followed that predicted for a similar cohort of patients without PEGs in the Levine et al model. It would suggest that the deaths observed were due to underlying comorbidities. This can provide a baseline for mortality targets for PEG services and is useful information for patients regarding the risks and benefits of the procedure. It demonstrates that PEG does no harm. However, it also shows that most deaths occur within the first month of placement, and would support arguments for delaying placement until outcome from the underlying condition is more predictable.

TEN YEARS’ EXPERIENCE OF PEG-RELATED MORTALITY FROM A DISTRICT GENERAL HOSPITAL

J. Mannath, H. Hart, N. J. Trudgill, I. Mohammed. Gastroenterology, Sandwell General Hospital, West Bromwich, UK

Introduction: PEG tube placement is common practice, however the NCEPOD enquiry 2004 showed high 7-day mortality. A multidisciplinary approach and careful patient selection was recommended.

Aims & Methods: To review the PEG related mortality over a period from 1997–2006 and compare mortality and PEG insertion rates after implementing trust guidelines in 2001. Data on all PEG insertions at Sandwell Hospital were collected prospectively by specialist nurse and mortality obtained through Patient Administration System.

Results: A total of 387 PEGs were done, of which 57% were males and 43% females. Mean age was 70 (17–96). Median number of PEGs per year was 42. CVA (66%) and other neurological diseases (13%) were the commonest indications. Only 4% dementia patients received PEG. Since implementation of trust guidelines in 2001, there has been a reduction in the number of total PEG insertions due to a reduction of PEGs for CVA. The cumulative 7-day and 30-day mortality over the 10-year period was 7.5% and 23% respectively. There was an absolute reduction in 7-day as well as 30-day mortality over the past 5 years, although not statistically significant (p = 0.3). This was achieved by a combination of careful patient selection and education of other medical teams by our specialist nurse. Older patients (>75) had significantly higher 30-day mortality (70.5% vs 29.5%, p = 0.021).

Conclusion: Our long term PEG-related mortality compares favourably to the NCEPOD enquiry. There is a reduction in absolute mortality and number of PEG insertions over past 5 years due to careful patient selection after review by the specialist nurse and discussion in MDT. Elderly patients (>75) have a significantly higher mortality following PEG insertion.

RESPONSE TO 5 G OR LESS OF LACTOSE IN PATIENTS WITH LACTOSE INTOLERANCE

L. Harrington, J. F. Mayberry. Department of Digestive Diseases, University Hospitals of Leicester, Leicester, UK

Introduction: It is estimated that some 5–10 million people in the UK may have lactose intolerance, with many remaining undiagnosed although they experience symptoms. Researchers however disagree about the amount of lactose needed to cause symptoms in those who are lactose intolerant. The consensus is that at least 12 g of lactose is needed to cause mild symptoms, although some papers and case-studies suggest that less than 6 g, or even the amount included as a “filler” in medication, may be enough to provoke symptoms. At the same time, lactose is being used in greater quantities commercially during food preparation.

Aims & Methods: To investigate the effect of 5 g and less of lactose in lactose-intolerant patients to determine whether the quantities of lactose “hidden” in food and medication are sufficient to cause symptoms. Lactose-intolerant patients participated in a double-blind randomised controlled study. They were restricted to a lactose-free diet for four days and then took one of four doses of 5 g or less of lactose. Their response to the dose was recorded on a study-designed questionnaire that measured total system response. This process was repeated with a second dose of lactose.

Abstract 233 Percentage of subjects who reacted to various doses of lactose

Results: Of the doses that contained lactose, symptoms were recorded for 47.1% of these doses. All participants who recorded a response, including those with a low total symptom rating, indicated that they would avoid that dose in the future. It is estimated that 75% of patients with lactose-intolerance will react to approximately 4.3 g of lactose, but the amount of lactose contained in medication is not sufficient to cause symptoms of lactose intolerance.

Conclusion: Lactose-intolerant patients react to doses of lactose less than 5 g and, when asked, state this is not tolerable. They should be told that they may react to the amount of lactose found in less than half a glass of milk—lower amounts of lactose than acknowledged in current literature. When people are lactose intolerant they need clear, correct, information about foods and drinks that they should avoid. However, these patients should be informed that if their medications contain lactose, this is unlikely to cause problems. This may increase medication compliance.

10-YEAR AUDIT OF PARENTERAL NUTRITION SERVICE IN A DGH: IS COMPLIANCE WITH ALL NICE GUIDELINES ESSENTIAL TO PROVIDE A HIGH QUALITY SERVICE?

K. M. J. Robinson1, J. McKiernan1, G. Uprichard2, S. A. Kapadia3, B. C. McKaig3. 1Dietetics, 2Pharmacy, 3Gastroenterology, Royal Wolverhampton Hospitals NHS Trust, Wolverhampton, UK

Introduction: The nutrition support team (NST) at New Cross Hospital was established in 1997. Prospective data have been collected annually on all adult patients referred to the NST. BAPEN1 and NICE2 have published recommended guidelines for PN practice, a key recommendation being the need for a nutrition support nurse. We compare our 10-year data audit against standards from these guidelines.

Aims & Methods: Prospective audit data has been collected on 887 patients (>8500 PN days) referred to the adult NST from 1997–2007. Analysis of this data is compared to national guidelines and to previously published PN outcomes of established centres.

Results: Despite identifying a need for a support nurse, the NST has functioned with dietetic, medical and pharmacy input alone. Over the 10-year period, there has been a substantial reduction in the number of patients exposed to PN (126 in 1997 to 48 in 2007) and total PN days (1422 in 1997 to 400 in 2007). Our current figures are now comparable to national standards. The use of enteral tube feeding (ng/nj) has correspondingly increased. The number of patients receiving PN inappropriately has fallen from 64% to 10%. Following trust initiated infection prevention measures the number of infectious complications has fallen from 23% to 9% since 2001. This also coincides with the rationalisation of catheter usage, in particular the withdrawal of PIC lines. In our series, infection, inappropriate use of PN and interruption of PN was associated with patients in whom PN was administered via IC lines. Patient assessment and monitoring is compliant with NICE guidelines and to date we have had no metabolic complications from PN therapy.

Conclusion: A consistent approach to promoting enteral feeding can reduce the need for inappropriate PN. Basic infection prevention measures are effective in reducing infective complications of PN. Although nurse support is desirable, we have demonstrated that a safe, appropriate PN service can be provided by a dedicated team without specialist nurse support.

OSTEOPOROSIS IN PATIENTS OVER 50 FOLLOWING RESTORATIVE PROCTOCOLECTOMY FOR ULCERATIVE COLITIS, IS DEXA SCREENING WARRANTED?

S. D. McLaughlin1, Z. L. Perry-Woodford2, M. W. Johnson3, S. K. Clark2, P. P. Tekkis1, P. J. Ciclitira3, R. Nicholls1. 1Department of Biosurgery and Surgical Technology, Imperial College, 2Department of Surgery, St Mark’s Hospital, 3Department of Gastroenterology, Nutritional Sciences Research Division, King’s College, London, UK

Introduction: Ulcerative colitis (UC), corticosteroids and age are risk factors for osteopenia and osteoporosis. The current BSG guidelines1 recommend performing DEXA in high risk patient groups. There are no studies of bone density in patients over 50 following restorative proctocolectomy. We hypothesised that these patients would be at high risk since they are likely to have had worse disease and received higher doses of corticosteroids than an average population of UC patients. We sought to establish whether DEXA should be routinely performed in this group.

Aims & Methods: We sought to establish whether DEXA should be routinely performed in patients over 50 following restorative proctocolectomy for UC. All RPC patients over 50 seen in an outpatient clinic from August 2006 to August 2007 underwent a DEXA bone scan unless recently performed. The notes were reviewed.

Results: See table. Osteoporosis was found in 18% of patients.

Abstract 235

Conclusion: Males are at increased risk; this is in keeping with studies in UC.2 RPC may increase bone density; those patients with reduced bone density had undergone RPC more recently than normals. Osteopenia was very common, this is important and clinicians should ensure RPC patients over 50 take adequate dietary calcium and vitamin D or consider supplementation to prevent osteoporosis. This study suggests that osteoporosis in RPC patients over 50 is no more common that in normal UC patients; routine DEXA is not warranted and the standard BSG guidelines should be followed.

RE-FEEDING SYNDROME AND RE-FEEDING-LIKE ABNORMALITIES IN PATIENTS WITH ALCOHOLIC LIVER DISEASE: ARE WE VIGILANT ENOUGH?

S. Menon, J. Dillon, A. Bohan, A. Brind. Gastroenterology, University Hospital of North Staffordshire, Stoke on Trent, UK

Introduction: Patients with alcoholic liver disease represent a group that are vulnerable to re-feeding syndrome (RS) when fed, due to pre-existing undernourishment and electrolyte abnormalities. Clinical signs of RS are often overlooked due to these factors.

Aims & Methods: To evaluate the incidence of RS and re-feeding-like abnormalities (RL) in patients with alcoholic liver disease admitted to an acute Hepatology ward. All patients with a diagnosis of alcoholic liver disease admitted over a period of 4 months (December 2006 to March 2007) were prospectively assessed for electrolyte abnormalities. Severe hypophosphataemia (SH) was defined as serum phosphate <0.5 mmol/l and mild to moderate hypophosphataemia (MH) was defined as serum phosphate 0.5–0.8 mmol/l. Hypomagnesemia and hypokalemia were defined as serum magnesium <0.75 mmol/l and serum potassium <3.5 mmol/l respectively. All patients with electrolyte abnormalities were treated.

Abstract 236 Results

Results: 36 patients were admitted over a 4-month period (22 male, 14 female: mean age 52 years, 16/44, 44% cirrhotics). All patients were commenced on oral nutritional supplements (mean 540 kcal/day) and 13 patients received supplemental nasogastric (NG) feeding. 11/36 (30%) patients developed MH and one patient developed SH (normal initial serum phosphate). Of these, 8/11 (73%) were on NG feeding (see table). 5/11 (45%) of patients with MH had concurrent hypokalemia and hypomagnesemia and of these, 4 were on NG feeding. 9/11 (81%) of patients developing MH had symptomatic liver disease. Two patients with MH died due to hepatorenal syndrome.

Conclusion: RS and RL are common in patients with alcoholic liver disease. Clinical signs are often subtle and a high index of clinical suspicion must be exercised to detect and aggressively treat these abnormalities in this vulnerable population.

A TESTING TIME . . . ARE PATIENTS AT RISK OF RE-FEEDING SYNDROME BEING ADEQUATELY MONITORED?

J. Parr, C. Lamb, M. Warren. Gastroenterology, North Tyneside General Hospital, North Tyneside, UK

Introduction: The UK National Institute of Health and Clinical Excellence (NICE) have recently published a guideline entitled “Nutrition support in adults”(February 2006) containing recommendations on the management of malnourished patients. It is known that a significant proportion of adult patients admitted to hospital are suffering from malnutrition and are therefore at risk of developing re-feeding syndrome when nutritional support is introduced.

Aims & Methods: Our study took place in a District General Hospital in the North East of England. All 328 adult inpatients were screened on a single day (01/05/2007) for malnutrition using MUST (Malnutrition Universal Screening Tool) scores. We aimed to determine whether those patients identified as being malnourished at the time of admission to hospital, as defined by MUST score of 2 or greater (and therefore at potential risk of the re-feeding syndrome) had received adequate biochemical and electrolyte monitoring as recommended by NICE. Data regarding blood results were obtained retrospectively from the computerised biochemistry results system.

Results: 100% of adult patients (n = 328) were screened. All adult speciality wards were included. 105(32%) were identified as being at risk of malnutrition using the MUST score. 35 males: 70 females. Average age 75 years (range 17–98). Average inpatient length of stay at the time of the study was 15.5 days (range 1–90). All high risk patients had potassium levels checked on admission. Only 24% (n = 25) had phosphate levels and 9.5% (n = 10) had magnesium levels checked within 24 h of admission. In only 30 patients of 105 overall (28%), were both phosphate and magnesium checked at some stage during their stay and in 52 patients (50%) neither phosphate nor magnesium had been checked at any stage. In those who had stayed 3 days or more both phosphate and magnesium levels had been checked in only 13 of 92 (14%) over the initial 3 day period.

Conclusion: Malnutrition is common among hospital in-patients and emphasis is being put on early recognition and treatment. Our study shows that full and early monitoring of the biochemical markers of the re-feeding syndrome is not currently being performed in high-risk patients. Early recognition, monitoring and appropriate treatment of those at risk are essential in order to reduce the potentially life-threatening complications and electrolyte imbalances associated with the condition.

SIGNIFICANT NUMBERS OF COELIAC PATIENTS ON A GLUTEN-FREE DIET DO NOT MEET RNI TARGETS FOR MICRONUTRIENTS

G. G. Robins, D. Wild, P. D. Howdle. Academic Unit of Medicine, University of Leeds, Leeds, UK

Introduction: A strict gluten-free diet (GFD) is the only recognised treatment for patients diagnosed with coeliac disease (CD). However, there is little information on the quality of the GFD, especially with regard to micronutrients. The main source of selenium in a typical diet is bread and cereals.

Aims & Methods: The aim of this study is to identify whether patients on a GFD are meeting the recommended nutrient intake (RNI) of a number of micronutrients. Over a 9-month period, all patients seen in a weekly, specialist gastroenterology clinic in a large UK teaching hospital, with histologically confirmed CD and on a strict GFD for at least 6 months were invited to complete prospectively a validated 5-day food diary. 65 patients were identified, of whom 46 returned completed usable diaries. Completed diaries were inputted into “Microdiet for Windows v2.52”, a computerised nutrient databank. Data for gluten-free foods not in the nutrient databank were obtained from direct communication with the manufacturers.

Results: Mean age of respondents was 54.0 years (range 21–79). 32 out of the 46 patients were female. Although self-reported, no food diary included in the analysis contained gluten. Results are shown in the table.

Abstract 238 Number (%) of patients on GFD meeting RNI guidelines for micronutrients

Conclusion: Significant numbers of CD patients on a GFD fail to meet RNI targets across a number of micronutrients. This is particularly true with regards to selenium, magnesium and calcium recommendations for patients with CD.

FEMALE COELIAC PATIENTS ON A GLUTEN-FREE DIET CONSUME A DISPROPORTIONATE AMOUNT OF NON-EXTRINSIC MILK SUGARS

G. G. Robins, D. Wild, P. D. Howdle. Academic Unit of Medicine, University of Leeds, Leeds, UK

Introduction: Previous local research1 has shown that female patients with coeliac disease (CD) on a gluten-free diet (GFD) consume significantly more calories, although percentage of energy obtained from the major food groups did not differ significantly from the representative population. Current guidelines recommend only 10% of all energy is derived from non-milk extrinsic sugars (NMES).

Aims & Methods: The aims are to confirm or refute the above findings in a larger group and to identify which subgroups of foods are responsible for the increase in calorie consumption. Over a 9-month period, all patients seen in a weekly, specialist gastroenterology clinic in a large UK teaching hospital, with histologically confirmed CD and on a strict GFD for at least 6 months were invited to complete prospectively a validated 5-day food diary. 65 patients were identified, of whom 46 returned completed usable diaries. Completed diaries were inputted into “Microdiet for Windows v2.52”, a computerised nutrient databank. Data for gluten-free foods not in the nutrient databank were obtained from direct communication with the manufacturers. Results obtained were compared against the “National Diet and Nutrition Survey of Adults 19–64” published in July 2003 by the Office for National Statisitics. Student’s unpaired t test was used to test for statistical significance.

Results: Mean age of respondents was 54.0 years (range 21–79). 32 out of the 46 patients were female. Mean BMI for females was 24.7 kg/m2 and for males 27.7 kg/m2. Overall, for women 36% of energy was obtained from fats, 48% from carbohydrates, (including 15% from NMES) and 15% from protein. This compares with 34% from fats, 49% from carbohydrates (including 12% from NMES) and 17% from protein in a female representative population from Northern England. Corresponding figures for males were 35%, 48%, 15% and 15% (vs 35%, 48%, 13% and 17%). These differences were not significant. Mean (SD) energy intake for the males was 2373 (588) kcal/day, vs 2259 (533) kcal/day) in the representative population and 1912 (366) kcal/day, vs 1618 (440) kcal/day for the females (p<0.001). For females on a GFD, mean carbohydrate intake was 244 (59) g per day vs 202 (58) g in a representative population. Increased fibre intake accounted for only 1 g of this, whereas non-milk extrinsic sugars accounted for 23 g (p<0.0001), equivalent to 29% of the extra calories ingested by these patients.

Conclusion: Female CD patients on a strict GFD do have a significantly higher energy intake, a large percentage of which is accounted for by simple sugars, reflecting the sugary nature of a lot of GFD foods. This implies less sugary snack foods should be more readily available for those on a GFD.

CALCIUM INTAKES AND OSTEOPOROSIS SCREENING IN PATIENTS WITH COELIAC DISEASE, CROHN’S DISEASE AND ULCERATIVE COLITIS

J. E. Dart1, M. Abraham2, V. Bara2, S. P. L. Travis3, D. P. Jewell3, K. Whelan2. 1Department of Nutrition, Alfred Hospital, Melbourne, Australia, 2Nutritional Sciences Division, King’s College London, London, 3Gastroenterology Unit, John Radcliffe Hospital, Oxford, UK

Introduction: The increased risk of osteoporosis in patients with coeliac disease, Crohn’s disease and ulcerative colitis (UC) is multifactorial and includes alterations in calcium absorption and metabolism as well as steroid treatment in Crohn’s and UC. While the recommended calcium intake for healthy adults is 700 mg/d (Department of Health), guidelines for these patients recommend intakes of 1000 mg/d (ECCO guidelines), 1000–1200 mg/d (2007 BSG guidelines) or 1500 mg/d (2000 BSG guidelines).

Aims & Methods: The aim of this study was to measure dietary and supplemental calcium intakes and osteoporosis screening in these patients and to investigate whether differences exist between disease groups. Ninety one patients with coeliac disease (n = 31), Crohn’s disease (n = 36) or UC (n = 24) were recruited from gastroenterology outpatient clinics at the John Radcliffe Hospital, Oxford, UK. Exclusion criteria were patients with osteoporosis or previous fractures. Dietary calcium intake was measured using a validated food-frequency questionnaire and supplemental calcium intake was measured using prescription data and patient reports. Details of DEXA scans were recorded from medical notes.

Abstract 240 Calcium intakes and proportions of patients meeting guideline intakes

Results: Median dietary calcium intake was 1079 mg/d (min 285, max 3098) and median total calcium intake was 1150 mg/d (min 285, max 3598). Supplemental calcium was prescribed to 18 (58%) patients with coeliac disease, 8 (22%) patients with Crohn’s and only one (4%) patient with UC (p<0.001). Dietary (p = 0.042), supplemental (p = 0.002) and total (p = 0.001) calcium intakes differed significantly between patient groups. Of those patients who should have had a DEXA scan (2000 BSG guidelines), 28 out of 31 (90%) with coeliac disease, 8 out of 20 (40%) with Crohn’s and none out of 16 (0%) with UC actually had one performed (p<0.001, coeliac vs Crohn’s).

Conclusion: Dietary, supplemental and total calcium intakes in patients with coeliac disease, Crohn’s disease and UC are highly variable and differ between patient groups. Although patients with UC had the lowest dietary calcium intakes, they were least likely to be prescribed calcium supplements and undergo DEXA scan.

Endoscopy posters

IMPROVING VIEW QUALITY AND TRANSIT TIME IN CAPSULE ENDOSCOPY

L. Muntane1, V. Gummer2, J. O’Brien2, M. K. McStay2. 1Barts and the London Medical School, London, 2Gastroenterology, Southend University Hospital, Southend, UK

Introduction: Capsule endoscopy is a new diagnostic method allowing painless visualisation of the entire small bowel. Standard preparation consists of clear fluids from 12:00pm the day before the test, and then nil-by-mouth from midnight. However, the test may be limited by poor luminal views and by incomplete small bowel transit in up to 25% of cases. Current data on the use of bowel preparation and pro-kinetics in order to improve the quality and completeness of the examination are limited and inconsistent.

Aims & Methods: We assessed whether magnesium citrate (Citramag) and senna bowel preparation taken with oral metoclopramide as a pro-kinetic agent resulted in superior luminal views and completion rates compared to standard preparation or Citramag alone. Patients were prospectively allocated to one of the following groups: (1) standard preparation (N), (2) Citramag (2 sachets in 2 l water) taken on the afternoon/evening prior then nil-by-mouth from midnight (C), 3) Citramag and senna (1 sachet) taken as above +10 mg metoclopramide 15 minutes pre-procedure (CSM). All patients were given simethicone with water when the capsule was swallowed to reduce bubble artefact. The studies were reviewed by a single investigator blinded to the preparation given. The view quality was graded using a previously validated 5-point scale based on the degree of obscured mucosal views (0 = 0–20%, 1 = 20–40%, etc).1 The score was applied to 5-minute segments of video from entry into the duodenum and every 10% of the small bowel transit time thereafter (max score = 44). The gastric transit times (GTT), small bowel transit times (SBTT), completion rates, significant findings and adverse events were recorded.

Results: View quality was significant improved and SBTT significantly reduced in the CSM group compared to standard preparation (median score 34 vs 30, p = 0.04; median SBTT 180 vs 241 min, p = 0.008; n = 18). View quality was better and SBTT significantly reduced in the CSM group compared to the C group (median score 30, p>0.05; 180 vs 235 min, p = 0.03; n = 18). There was no difference in view quality or SBTT between N and C. Completion rates were equivalent in all three groups: 94%. There was no significant difference in GTT between the groups. There was no difference between groups in patient demographics or the number of findings seen. There were no adverse events.

Conclusion: Bowel preparation with Citramag and senna used together with metoclopramide pre-procedure results in significantly better small bowel visualisation and reduced small bowel transit times than standard preparation or Citramag alone.

MANAGEMENT OF EARLY PERI-AMPULLARY NEOPLASIA: SURGICAL VERSUS ENDOSCOPIC RESECTION

R. J. Mead1, M. D. Duku1, P. Goggin1, R. D. Ellis1, D. N. Poller1, D. Cowlishaw1, N. Pearce2, C. Johnson3, S. Sadek4, M. Walters4, P. Bhandari1. 1Gastroenterology, Portsmouth Hospitals Trust, Cosham, 2Upper GI Surgery, 3Upper GI Surgery, Southampton University Hospital, Southampton, 4Upper GI Surgery, Portsmouth Hospitals Trust, Cosham, UK

Introduction: Despite centralisation of surgical expertise, and dedicated meticulous surgical selection pancreaticoduodenectomy remains a procedure with considerable morbidity and mortality. In recent years we have used endoscopic papillectomy for further assessment and cure, in conjunction with more traditional surgery or palliation. We present our outcomes to date.

Aims & Methods: All patients with early neoplasia of the ampulla of vater were included in our study. These included T1 tumours, adenomas with dysplasia, and other benign lesions. All patients were either treated with Whipples, endoscopic papillectomy or best palliative care, on their own or in combination. Decisions were made on the basis of likely pathology, technical feasability, and medical comorbidity. Diagnostic workup consisted of chromendoscopy, CT scan, and EUS. Patients with endoscopic appearance at chromendoscopy or biopsy suggestive of invasive cancer were offered surgery. Patients with a benign appearance and biopsy, or who were unfit for surgery had a papillectomy. Patients unfit for either were palliated. We aim to compare the survival, morbidity, mortality, and hospital stay outcomes between surgical and endoscopic groups.

Abstract 242 Early results

Results: We identified early neoplasia between 2005–7. 10 men and 14 women met the criteria. 7 were palliated. 1 patient underwent a papillectomy and subsequent Whipples.

Conclusion: Surgery is an ultimate curative option but is associated with significant morbidity and mortality. Endoscopic treatment is challenging, but safe and effective. All patients decisions require very careful assessment and expertise.

TECHNICAL FEASIBILITY AND SAFETY OF ENDOSCOPIC MUCOSAL RESECTION OF LARGE AND FLAT COLONIC POLYPS

A. K. Muddu1, R. Mead1, D. Poller2, D. Cowlishaw2, A. Parvaiz3, M. Thompson3, D. O’Leary3, T. Trebble1, A. Quine1, P. M. Goggin1, P. Bhandari1. 1Department of Gastroenterology, 2Department of Pathology, 3Department of Surgery, Queen Alexandra Hospital, Portsmouth, UK

Introduction: With the advent of bowel cancer screening an increasing number of asymptomatic individuals are being diagnosed with large and flat colonic polyps. Traditionally, these would be surgically resected due to concerns about the efficacy and safety of endoscopic resection and risks of underlying cancer. However, with improved experience and techniques in endoscopic mucosal resection (EMR), these may be endoscopically resected and we report our experience of EMR of large and flat colonic polyps.

Aims & Methods: The aim was to examine the technical feasibility and outcome of EMR of large colonic polyps (>25 mm). We reviewed the data of all patients who had colonic polyps resected by EMR and ESD (endoscopic submucosal dissection) using an IT+ needle knife since December 2006. Data were collected on polyp size and type, resectability, histology, complications and completion of excision.

Results: 55 colonic polyps above 25 mm (range 25 mm–70 mm) in diameter from 51 patients were considered for endoscopic mucosal resection with 9 Is (sessile polypoid type) Paris type polyps (mean diameter 33 mm) and 46 LST (laterally spreading tumours), (mean diameter 44 mm). 5/55 polyps were deemed unsuitable for endoscopic resection due to endoscopic suspicion (based on non lifting sign and type V pit pattern) of invasive carcinoma in four polyps and the lesion crossing the dentate line in one polyp. These polyps were then surgically resected with three found to have invasive cancer and two found to have high grade dysplasia. Out of the 50 polyps resected endoscopically 44/50 (88%) polyps had complete resection in one sitting. 6/50 (12%) polyps required two procedures to achieve complete resection. The average time taken for endoscopic resection was 45 min (30–120 min). Significant bleeding occurred in one patient at 48 h requiring further endoscopy and haemostatic treatment. No perforations occurred in our series and no one required emergency surgery or had EMR-related mortality.

Conclusion: This series shows that it is technically feasible to endoscopically resect and cure more than 96% of large and flat colonic polyps. It is safe to perform endoscopic resection of high risk lesions with no mortality and minimum morbidity. Accurate endoscopic assessment of the lesion prior to EMR is essential.

A COMPARISON OF SCOPEGUIDE WITH CONVENTIONAL COLONOSCOPY

P. J. Mullen. Department of Gastroenterology, The Princess Elizabeth Hospital, Guernsey, UK

Introduction: The movement of electromagnetic transmission coils, built into the insertion tube of the colonoscope, within the magnetic field of a receiving dish enables the Olympus ScopeGuide system to provide a real time 3D image of the position and configuration of the colonoscope inside the patient. This system has several potential advantages but is primarily designed to reduce the difficulty of colonoscopy by allowing for early identification and prevention/resolution of loop formation.

Aims & Methods: In order to evaluate these proposed advantages, the completion rate and insertion time of 390 colonoscopies performed with an Olympus CFQ260L colonoscope using the ScopeGuide technology were compared with the preceding 420 performed with a conventional, variable-stiffness colonoscope (Olympus CF240AL). One experienced operator undertook all the procedures and data were collected prospectively.

Results: The 2 groups were equivalent as regards age, gender, previous colon resection and use of electrocautery. There were 12 failures (8 non-operator dependent) in the ScopeGuide group, giving overall and adjusted completion rates of 378/390 (96.6%) and 378/382 (99.0%). These were significantly better (p<0.005) than the conventional group (381/420, 90.7% and 381/401, 95.5%). Contrary to expectation, insertion times were the same in each group, although there was a non-significant trend with ScopeGuide for there to be fewer prolonged procedures (⩾13 min), particularly in females. Only 1 of the 4 true failures in the ScopeGuide group was classified as being due to excessive patient discomfort, compared with 10 of 20 in the conventional group.

Conclusion: ScopeGuide can improve the completion rate of colonoscopy even for experienced operators. Insertion times in this study were not shortened but there was evidence of procedures being less painful.

THE EFFECT OF MUSIC ON ANXIETY LEVELS OF PATIENTS ATTENDING FOR ENDOSCOPY

H. El-Hassan, K. McKeown, A. F. Muller. Department of Gastroenterology, The Kent and Canterbury Hospital, Canterbury, UK

Introduction: Music has been recognised in some studies to have a calming effect and may relieve stress. Patients attending for gastrointestinal endoscopy are often anxious and worried. This study examined whether the provision of music before endoscopic procedures altered anxiety levels in these patients.

Aims & Methods: Prospective randomised controlled trial. Patients fulfilling the following five inclusion criteria were invited to participate: (a) first ever endoscopy; (b) aged ⩾18 years; (c) ability to read and comprehend the information sheet; (d) competent to give consent; (e) ability to listen to music. Patients were randomised into two groups—Control (C) and Music (M); anxiety levels were measured using the State-Trait Anxiety Inventory (STAI)1 on arrival in the Unit and immediately before the endoscopy procedure, after 15 minutes listening to self-selected music therapy (M) or no music for the same period (C). Group scores ranged from a minimum of 20 to a maximum of 80 points. The effect of age (⩽ or >51 years) and procedure (gastroscopy or flexible sigmoidoscopy/colonoscopy) on anxiety levels was also assessed. Ethical approval was obtained from the local research and ethics committee prior to commencement.

Results: 180 patients (M: F 81: 99) mean age (±SD) of C: 59.18±17.80 and M: 60.73±15.24 years. There was no difference in anxiety levels at baseline based on age (<51 years, n = 56: 42.21±9.18; >51 years, n = 124: 39.99±10.13 (p = 0.15) or procedure: Gastroscopy, n = 87: 39.43±9.9, flexible sigmoidoscopy/colonoscopy: n = 93: 41.86±9.75 (p = 0.98). There was no difference in anxiety scores for group C (n = 88) at baseline (38.92±9.87) and immediately pre-endoscopy (38.182±11.40, p = 0.243). There was a significant reduction in anxiety scores in group M (n = 92) from 42.37±9.62 to 26.45±6.82 (p<0.001).

Conclusion: Anxiety levels in patients attending for endoscopy are not influenced by age or procedure, but are significantly reduced by listening to music compared to controls. The availability of music within the endoscopy unit is a simple strategy that will improve the well being of patients.

A PROSPECTIVE SINGLE CENTRE STUDY OF THE CLINICAL SIGNIFICANCE OF ERYTHEMA ON COLONOSCOPY

S. C. Ng, M. Shariff, G. Holdstock, D. Datta. Gastroenterology, The Hillingdon Hospital, London, UK

Introduction: Erythema in the absence of other signs is a finding commonly visualised on colonoscopy but the significance of this non-specific feature is unclear. Whether mucosal biopsies reveal clinically important histological abnormalities has not been formally studied.

Aims & Methods: The aim of this study was to assess the clinical and histological significance of erythema as the only endoscopic finding in patients referred for colonoscopy of all indications. Consecutive patients referred for colonoscopy found to have erythema were prospectively included. Patients with other endoscopic variables including loss of vascular pattern, granular mucosa, friable mucosa, ulceration, erosions, pseudopolyps, visual dysplasia, strictures or cancers were excluded. Between two to four biopsies were taken from the site of erythema and approximately two further biopsies were obtained from normal looking mucosa. Procedures were performed by two experienced endoscopists (more than 400 colonoscopies per year) and one supervised trainee endoscopist.

Results: Forty patients (22 male, 18 female) with a mean age of 50 (range 18–85), were included. Subjects were prepared for colonoscopy with sodium picosulphate. Indications for colonoscopy were altered bowel habits (n = 12), rectal bleeding (n = 9) and diarrhoea (n = 19). Erythema was identified in the rectum in 20 cases, left colon in 14 cases and right colon in six cases. Of the targeted biopsies, abnormal histology was reported in 13 cases (33%). Clinically significant pathology was found in seven cases (20%). There were four cases of non-specific colitis, three cases of inflammatory bowel diseases (2 ulcerative colitis, 1 Crohn’s disease), four cases of melanosis coli and two cases of mucosal prolapse. Non-targeted samples were normal apart from one case of collagenase colitis. The remaining 27 patients (67%) had normal biopsies. Of the seven cases of colitis, five patients presented with diarrhoea and two had rectal bleeding.

Conclusion: Clinically important pathology can exist in significant number of cases in spite of non-specific colonoscopic finding like erythema. This suggests that erythema in the absence of other signs may represent an early stage in the evolution of chronic non specific inflammatory bowel disease and biopsies are important, especially when clinical suspicion is high.

ADENOMA DETECTION RATES ARE HIGHER THAN PREDICTED AND CORRELATE WITH COLONOSCOPIC WITHDRAWAL TIMES IN THE BOWEL CANCER SCREENING PROGRAMME

D. L. Nylander1, M. Rutter2, G. Clifford3, J. Singh3, L. Hurst2, J. Painter4, C. Rees5. 1South of Tyne Bowel Cancer Screening Centre, Sunderland Royal Hospital, Sunderland, 2Tees Bowel Cancer Screening Centre, North Tees Hospital, Stockton, 3South of Tyne Bowel Cancer Screening Centre, Queen Elizabeth Hospital, Gateshead, 4South of Tyne Bowel Cancer Screening Centre, Sunderland Royal Hospital, Sunderland, 5Tees Bowel Cancer Screening Centre, South Tyneside General Hospital, South Shields, UK

Introduction: One aim of the bowel cancer screening programme (BCSP) is to detect and remove adenomas by colonoscopy, hoping that this will reduce the incidence of future colorectal cancer (CRC). In planning the service, we estimated from the first phase screening pilot, that adenomas would be detected in 40% of individuals undergoing screening colonoscopy.

Aims & Methods: The South of Tyne and Tees BCS centres (total population 1.5 million) began screening colonoscopies in 3/2007. We analysed data collected during all procedures in the first 6 months (extracted from BCSP central database) to assess the following: (1) adenoma detection rate (ADR); (2) colonoscopy withdrawal times (CWT), to see if endoscopists complied with the suggested mean of 6 minutes or more1; (3) if CWT correlated with ADR; (4) if endoscopists with similar ADR in BCSP have similar ADR in “symptomatic” service. Procedures in which there were no polyps, neoplastic lesions or pathology needing biopsy were identified and used to calculate CWT for each endoscopist. We also obtained data from the symptomatic service of endoscopist numbers 1–3 performed over the same period.

Abstract 247

Results: Five endoscopists performed total of 515 procedures. Adenomas were detected in 272 (53%), significantly more than the 40% predicted in the pilot study (p<0.0001). There was significant correlation between mean CWT and ADR (Spearman r = 0.9747; p = 0.0167). In their symptomatic service, endoscopists 1, 2 and 3 had ADRs of 10%, 10% and 18% respectively. There was no significant difference between these ADRs (p = 0.1 for endoscopist 3 vs 2; p = 0.1 for endoscopist 3 vs 1).

Conclusion: ADR was significantly higher in our centres than in the pilot study. This has impact on workload and is important for those planning to set up BCSP. All endoscopists had mean withdrawal time of 6 min or more, but one had an ADR lower than the rest. Despite all endoscopists being compliant with CWT 6 min or more, strong correlation exists between mean CWT and ADR, supporting importance of slow, methodical extubation. Endoscopists who have similar ADRs within the BCSP also have similar ADRs in their symptomatic service sugesting no difference in practice between these services.

Research submitted on behalf of the Northern Region Endoscopy Group (NREG).

USE OF THE SX-ELLA SELF-EXPANDING MESH METAL STENT IN THE MANAGEMENT OF VARICEAL HAEMORRHAGE: INITIAL EXPERIENCE IN EIGHT PATIENTS

M. Reynolds, A. K. Burroughs, D. W. Patch, J. P. O’Beirne. Sheila Sherlock Department of Hepatobiliary-pancreatic Medicine and Liver Transplantation, Royal Free Hospital, London, UK

Introduction: Patients with variceal bleeding that is refractory to endoscopic or pharmacotherapy often undergo salvage procedures such as Sengstaken-Blakemore (SBT) or transjugular intrahepatic stent shunt (TIPS) insertion. While highly effective these procedures have important limitations. SBT is a temporary therapy and often requires the patient to be intubated while TIPS is associated with a high mortality in patients with advanced liver disease. The recently described SX-Ella DANIS stent is a self expanding covered mesh metal stent that is deployed in the lower oesophagus achieving haemostasis by compressing oesophageal varices. The stent may be left in situ for up to 2 weeks and is removable.

Aims & Methods: We evaluated the use of the SX-Ella stent in the management of patients with endoscopically proven or suspected variceal bleeding who were refractory to standard therapy and who had contraindications to TIPS or where SBT could not be inserted. Stents were deployed as per the manufacturers instructions. The delivery system was placed in the distal stomach either over an endoscopically placed guidewire or blindly. A balloon was then inflated to anchor the distal end of the stent at the cardia before deployment of the stent. The balloon was then deflated and the delivery system removed. Satisfactory stent deployment was confirmed by endoscopic examination and radiography.

Results: Eight patients (all Childs-Pugh Class C) underwent stent insertion. Indications for stent insertion were: contraindication to TIPS (n = 5), failed SBT insertion (n = 2) and oesophageal perforation due to SBT (n = 1). Stent insertion was successful in 7/8 (87.5%). In patients with successful stent deployment, haemostasis was achieved in 5/7 (71%). 3 patients survived the acute bleeding episode (37.5%) and had the stents removed endoscopically at 10 (4–14) days post insertion. Causes of death were exsanguination in 2 (proven subsequently to be bleeding from gastric varices), terminal HCC (n = 1) and progressive multiple organ failure (n = 2). There were no complications related to stent insertion or removal and no stent migration was observed.

Conclusion: In patients with oesophageal variceal haemorrhage, insertion of the SX-Ella DANIS stent offers an alternative to other methods of salvage therapy. Stent insertion is easy and appears safe. Haemostasis rates are good (71%) but mortality is affected primarily by severity of the underlying liver disease.

GASTROSCOPY AFTER COMMUNITY TESTING FOR H PYLORI: WHAT DO WE FIND?

J. Parr, A. Saeed. Gastroenterology, Queen Elizabeth Hospital, Gateshead, UK

Introduction: H pylori testing is a key part of dyspepsia investigation in the UK, as patients without alarm symptoms may receive eradication therapy before considering further investigation. The H pylori stool antigen test (HPSA) is used for all H pylori testing in our region. All results from the area are collected centrally. We examined gastroscopy (OGD) reports of patients who had an HPSA test within the previous year, to discover if significant pathology was detected.

Aims & Methods: We aimed to find out the proportion of patients who have HPSA tested and proceed to having an OGD within 1 year of follow-up. In addition, we obtained information regarding the mode of referral, age and prevalence of significant OGD findings in this group. We used the HPSA database to identify all patients having an HPSA test over a 13 month period. Temporary residents were excluded. We cross referenced this against our endoscopy database to identify patients in whom an OGD had been scheduled within 1 year from the date of HPSA testing.

Abstract 249 Gastroscopy findings

Results: 2171 patients were included. There were 713 positive HPSA tests and 1458 negative. 139 patients(19.5%)who tested positive were scheduled for OGD within 1 year. 60% of these were in patients >55 years old. This compares to 278 (19%) of those testing negative(51% were >55 years old).The average time to OGD was 4.8 months in both groups. The findings at OGD were divided into normal/mild abnormalities(inflammation only) and more serious abnormalities (cancers, ulcers, strictures, Barrett’s). The results are shown in the table. In 4 cancers detected under the 2 week rule, time to OGD from intitial HPSA test was less than 6 weeks. The other cancers detected took longer in the HPSA positive group (3, 7, 10 months) than in the HPSA negative group (2, 2, 4, 7 months) to come to OGD. Only 1 out of 11 cancers occurred in a patient under the age of 55 years (oesophageal) and the OGD was performed less than 4 weeks after HPSA negative result, due to the presence of alarm symptoms. Cancer pick up rates in patients proceeding to OGD were 0.4% in the under 55 years age group compared to 4.4% in patients >55 years. This gives a cancer detection rate of 2.6% in patients referred for further investigation. Cancers were diagnosed endoscopically in 0.5% of all patients having an HPSA test.

Conclusion: A significant number of patients proceed to OGD in the year following HPSA testing (19%). In our study, cancer and other serious abnormalities were uncommon, regardless of methods used for referral. In patients under 55 with no alarm symptoms, endoscopy appears to add little to patient management over a test and treat policy.

PRECUT SPHINCTEROTOMY: IS IT WORTH THE RISK?

K. K. Peddi1, L. Wilbraham2, R. England2, D. F. Martin2. 1Gastroenterology, 2GI Radiology, South Manchester University Hospital NHS Trust, Wythenshawe, UK

Introduction: Precut Sphincterotomy by needle knife technique is a method of gaining deep biliary access when standard methods fail and its safety is widely debated. Expert endoscopists argue that it is a very effective way of gaining deep biliary access and the benefits outweigh the risks if carefully performed.

Aims & Methods: We analysed a large database of all needle knife papillotomies performed at our tertiary centre over a five year period (December 2000 to July 2006)to see whether it was worth the risk. They were performed by either of the two very experienced endoscopists in the centre. All the cases were analysed retrospectively using a proforma.

Results: A total of 2800 ERCPs were performed in this period and a precut sphincterotomy using needle knife was required in 312 patients (11.4%) with an intention to gain deep biliary access. 131 were males and 181 females. 155 of 312 were referred from other hospitals. Patients age ranged from 20 to 98 years (mean age 71.9). Sedation was with Midazolam (1–25 mg) along with either Pethidine (12.5–100 mg) or Fentanyl (10–150 μg). Flumazenil was required in 7 patients for reversal. 3 patients required general anaesthetic. Indications were jaundice (129), dilated ducts (40), abnormal LFTs (44), cholangitis (22), stones (118), acute pancreatitis (34), stent review (3), biliary leak (6), pseudocyst (2), pain (21), previously failed (52), tumour (15), PSC (1). Deep biliary cannulation was achieved in 209 patients at first (67%), 55 out of 65 patients (84.61%) at second and 2 patients at 3rd attempt, making overall successful cannulation in 266 out of 312 (85.25%). Each patient was assessed immediately after procedure, 24 hours later and then followed up at 30th day by a specialist nurse. Complications were defined and graded according to the consensus criteria.1 There were 52 “adverse events” reported during first attempt with needle knife papillotomy (14.4%) and 18 of these met the definition of “complication” (5.76%). One mild and one moderate bleeds reported (0.64%). There were 2 mild and 3 moderate pancreatitis (1.6%). Two mild and one moderate perforations occurred (0.96%). Seven cases of mild cholangitis reported (2.24%). One death occurred due to pancreatitis (0.3%). Other events include submucosal injection (10), abdominal pain (14), sore eye and rhinitis at 30 days.

Conclusion: Precut sphincterotomy using needle knife papillotomy can result in mild complications even in expert hands but severe/fatal complications are very rare. It is worth taking the risk in selected patients if performed carefully by expert endoscopists.

DO ALARM FEATURES CONFER AN INCREASED RISK OF UPPER GASTROINTESTINAL CANCER IN PATIENTS AGED UNDER 55 YEARS WITH DYSPEPSIA?

N. Powell, J. Mawdsley, J. Teare, H. Thomas, R. Negus, J. Hoare, T. Orchard. Gastroenterology, Imperial College Healthcare NHS Trust, London, UK

Introduction: To expedite the diagnosis of upper gastrointestinal (GI) cancers, the National Institute for Health and Clinical Excellence (NICE) and the British Society of Gastroenterology recommend urgent endoscopy for patients of all ages with dyspepsia, if alarm features such as weight loss coexist. However, upper GI cancers are predominantly diseases of the elderly and the merit of endoscopy in young patients, even those with alarm features, has not been well defined.

Aims & Methods: We identified all patients undergoing upper GI endoscopy for dyspepsia in our unit over a 23-year period. We also recorded the presence of 4 NICE defined alarm features (weight loss, anaemia, dysphagia and vomiting) concomitantly listed as indications for endoscopy. The prevalence of upper GI cancers diagnosed in patients aged <55 years undergoing endoscopy for dyspepsia in both the presence and absence of alarm features was calculated. For comparison we also included patients aged >55 years with dyspepsia and alarm signs.

Results: From our dyspepsia population we identified 1371 patients aged <55 years with alarm features, 11 672 aged <55 years with no alarm features and 1663 patients aged >55 years with alarm features. The cancer prevalence was significantly higher in patients aged <55 years in the presence of alarm features (2.1%) compared to those without (0.4%, OR 5.34 (95% CI 2.29 to 12.47), p<0.0001). The presence of 2 alarm features was associated with significantly more cancers (10.6%) than the presence of 1 alarm feature (1.7%, p<0.0001). The alarm feature most significantly associated with cancer was weight loss (OR 12.01, 95% CI 2.01 to 71.84, p<0.0001), followed by vomiting (OR 5.08, 95% CI 1.34 to 19.27, p<0.0001). In older patients (aged >55 years) with dyspepsia and alarm features the prevalence of cancer (7.2%) was significantly higher than in younger patients (<55 years) with (OR 3.62, 95% CI 2.44 to 5.36, p<0.0001) or without (OR 19.32, 95% CI 8.99 to 41.50, p<0.0001) alarm features. The number of endoscopies and cost (based on Department of Health tariffs) required to diagnose 1 cancer was 14.0 (£5558) in patients >55 years with alarm features, 47.3 (£18,778) in patients aged <55 years with alarm features and 259.4 (£102,982) in patients aged <55 years with no alarm features.

Conclusion: The presence of alarm features significantly increases the diagnostic yield of upper GI cancer in young patients with dyspepsia, but is still very low with cancers only present in 2.1% of patients. Consequently offering urgent endoscopy to this population is a high cost strategy. The presence of particular alarm features such as weight loss or vomiting, or the presence of two or more coexistent alarm signs might offer improved determination of which young patients should undergo urgent endoscopy.

VALIDATION OF BOWEL PREPARATION QUALITY SCALES

M. Rahman, N. Van Someren, K. Besherdas, R. D’Souza. Gastroenterology, Chase Farm, Enfield, Middlesex, UK

Introduction: It is vital that colonoscopy completion rates in the UK improve before a screening programme for colon cancer can be successfully implemented. A completion rate of 90% is considered acceptable and has been accepted by the UK Endoscopy community. The main reason for failure to complete colonoscopy in our unit were poor patient intolerance or bowel preparation or severe diverticular disease. Adequate preparation of the colon is essential for proper colonoscopic visualisation. At our unit a non-validated 3-point scoring system based on subjective items is used to evaluate colonic cleansing.

Aims & Methods: Several scoring systems for bowel preparation have been developed. In this study our objective was to compare the Aronchick and Ottawa scoring system with our current standard system. A total of 100 patients undergoing routine diagnostic colonoscopy were included in the study. This study was performed over a 5-month period from May to September 2007 and carried out at two hospitals: Edgeware Community Hospital and Chase Farm Hospital. Of these 100 patients 50 had Moviprep at one centre and 50 Picolax at the other centre. Patients response regarding quality of bowel preparation were assessed prior to the colonoscopy and compared with the endoscopists. We compared three different scoring system for colonoscopy preparation—that is, Standard, Aronchick and Ottawa scale.1, 2 Before performing the Ottawa scale we carried out a calibration exercise. Standard: (1) Good; (2) Satisfactory; (3) Poor.

Results: The inter-observer agreement was assessed statistically using Pearson’s correlation coefficient. Picolax was found to be superior in elderly patients compared to Moviprep; while there was no significant difference between Picolax and Moviprep in younger patients (<45 years of age). Younger patients had better score for the colonic cleansing compared to older patients (p<0.01). Patients were unreliable judges of the quality of bowel preparation leading to overestimation of the cleanliness of their colon. The Aronchick and Ottawa scoring systems were superior to the standardised scoring system (p<0.01). However no difference was found between these two colonic preparation scaling system (p = NS).

Conclusion: Picolax was superior to Moviprep in older patients. Younger age was a good predictor for adequate bowel preparation. Patients are unreliable judges of the quality of bowel preparation tending to over estimate the cleanliness of their colon. The Ottawa scale gives much more specific information about which area of bowel is poorly prepared, whereas Aronchick outlines the whole picture. Both the Aronchick and Ottawa quality scale are easy to perform and give much more information than the standard scoring system and should be implemented into current endoscopy practice.

A UK AUDIT INTO THE RADIATION SAFETY AWARENESS AND PRACTICE IN ERCP

N. Rajoriya, A. Mee. Gastroenterology, Royal Berkshire Hospital, Reading, UK

Introduction: Endoscopic retrograde cholangiopancreatography (ERCP) relies on the use of ionising radiation, however risks and awareness of the safety aspects of the procedure are often not appreciated by the ERCP’ist. Guidelines have been published (Ionising Radiation Regulations 1999) with recommended levels of annual exposure.

Aims & Methods: The aim of this audit was to assess the safety practices and awareness of the radiation aspect of ERCP by doctors in the UK carrying out the procedure. An electronic questionnaire was sent (via email) to physicians, surgeons and radiologists registered with the British Society of Gastroenterology (BSG). Professors, Consultants and trainees (Specialist Registrars (SPRs)) were included. Emails were also sent to doctors in the ERCP community (provided from list previously used in a national audit of ERCP). The audit questionnaire was returned electronically to a base website (www.surveymonkey.com) and the information collated thereafter.

Results: 172 responses (92.4% male) were collected over a 2-month period. 93.5% were consultants, 4.8% trainees, and 1.8% professors. 51.5% were between the ages of 45–65, 42.7% between 36–45 and the rest below 36 years of age. 85.5% were gastroenterologists, 9.3% surgeons and 5.2% radiologists. Most (48%) have been practicing ERCP for 11–20 years, 21.7% >20 years, and 30.4% <10 years. 73.4% do 1 session per week, 17% do 2 sessions, 1.2% do 3 sessions and 7.6% do on average less than 1 session on a weekly basis. In the last 12 months 2.4% have done >250 ERCPs, 7.6% between 200–250, 19.4% between 150–200, 37.6% between 100–150, and 32.9%<100 ERCPs. With regard to protection: 37.2% always wear a thyroid shield, 8.1% >75% of the time, 6.4% 50–75% of the time, 6.4% <50% of the time, and 41.9% never wear a thyroid shield. 95.3% do not wear protective lead eyewear (only 1.7% do 100% of the time). Of those who did wear protective eyewear, 60% (10% prescription lenses) were provided free by the NHS trust/hospital. With regards to radiation exposure badges, 32% did not wear one. The majority (42.7%) wear a badge on the trunk (under the lead apron), 18.1% on the pelvic area (under apron) and 3.5% on the trunk and pelvis (under the apron). 61.8% of doctors have never been informed on their annual radiation exposure and only 11.8% regularly informed on an annual basis. 92.1% were unsure of their last year’s exposure. Only 29.1% had been on an ERCP course (involving safety lectures) in the last 5 years with 53% of 157 consultants not been on a course since becoming an SPR.

Conclusion: This audit highlights a wide variation of practice and awareness of UK ERCP’ists regarding safety in ERCP. We recommend that radiation protection officers have more input, and assess ERCP’ists practice to ensure that UK ERCP’ists are exposed to the lowest doses of radiation as possible to carry out the service provided.

A STEP-WISE ASSESSMENT OF COLONOSCOPIC BOWEL PREP EFFECTIVENESS

E. Hall1, J. Lattaway1, J. Vasani2, M. D. Rutter3. 1Endoscopy, 2Gastroenterology, University Hospital of Hartlepool, Hartlepool, 3Gastroenterology, University Hospital of North Tees, Stockton-on-Tees, UK

Introduction: Effective and well-tolerated bowel preparation is essential for optimal colonoscopic imaging and adenoma detection. There are almost as many different regimens as there are endoscopy units, involving different types and timing of medication and differing diets and restrictions. Some are more effective than others, and our unit felt its bowel prep was suboptimal.

Aims & Methods: The aim was to assess the effect of incremental adjustments in the bowel prep regimen. Bowel prep effect is entered onto the hospital endoscopy reporting system at the time of the procedure, graded as good, satisfactory or poor. Although the categorisation is subjective, the hospital’s colonoscopist population remained unchanged during the study, reducing this potential bias. In April 2007, we changed our original bowel prep (low residue diet 3 days and fluids only 1 days before procedure; 4 senna tablets and 2 Picolax sachets taken the day before the procedure) by increasing the senna dose to 10 tablets (as used in our sister hospital). To minimise potential bias, we did not publicise the change widely to colonoscopists. Two-sided Fisher’s exact analysis was performed, comparing “good” prep versus “not good” (a combination of those recorded as satisfactory or poor). As no difference was noted, a further change was made to the regimen in September 2007, reverting to 4 tablets of senna but bringing the diet and Picolax timing in line with our sister hospital (low residue diet 2 days and fluids only 1 days before procedure; 4 senna tablets and 1 Picolax sachets taken the day before the procedure, second Picolax taken on day of procedure). Further statistical analysis was performed.

Results: See table.

Abstract 254 Bowel prep changes

Conclusion: Increasing senna dose from 4 to 10 tablets did not improve the quality of bowel prep. Adjusting the timing of Picolax and liquids only aspect of the diet resulted in a statistically significant improvement in bowel prep quality. This improvement was achieved despite a shorter duration of altered diet prior to the colonoscopy, thus impacting less on the quality of life of our patients.

A SURVEY OF RIGID SIGMOIDOSCOPY USE AMONG UK GASTROENTEROLOGY TRAINEES IN TWO DEANERIES

G. J. Sadler, P. Ellul, B. C. McKaig. Gastroenterology, New Cross Hospital, Wolverhampton, UK

Introduction: Studies have shown rigid sigmoidoscopy (RS) causes more patient discomfort, has a lower diagnostic yield and poorer ease of use for procedures such as biopsy and polypectomy when compared to flexible sigmoidoscopy (FS). Visualisation of the rectosigmoid junction and sigmoid colon at RS is successful in only 40–70% of examinations. In addition, disposable instruments may pose an infection risk by potentially harbouring organisms in the bellows or light head. For these reasons, and due to the increasing availability of FS in out-patient clinic settings, the utility of RS is being called into question. However, there is no doubt it remains an effective, cheap and widely available diagnostic tool.

Aims & Methods: We conducted a questionnaire survey of registrars in the West Midlands and North Western Deaneries about RS use in their current working practice.

Results: 47 trainees responded—70% male and 30% female. 15% had never used RS. Only 32% of those who had used RS had been trained to do so; 89% felt they had received insufficient training. More than two thirds of trainees do not use RS in clinical practice, for reasons including the lack of available facilities (47%), its use being actively discouraged (16%) and feeling incompetent to perform the procedure (19%). 38% cannot take biopsies during RS. Reasons a trainee thought it may be appropriate to consider RS were: rectal bleeding (57%), change in bowel habit (17%), colitis assessment (45%), tenesmus (30%), pruritus ani (9%), proctalgia (19%), and cancer/polyp follow-up and weight loss (0%). Only 57% thought RS was a skill all gastroenterologists should be competent in. When asked if they had rectal symptoms requiring investigation, 79% said they would rather defer RS for an elective FS.

Conclusion: For a variety of reasons, the vast majority of trainees across these two Deaneries do not use RS in clinical practice, and believe insufficient training is given in its use. Surprisingly few would even consider using it in patients with worrying rectal symptoms such as bleeding, tenesmus and proctalgia. This may simply reflect a preference to defer investigation to FS which is increasingly available in the out-patient clinic setting, and is also the investigation, given the choice of RS or FS, most trainees would prefer to undergo themselves. It therefore seems unlikely, especially considering just over half believe RS is a gastroenterological skill, that many of these trainees will use RS in future clinical practice or even become competent in its use—if the same is true across the rest of the UK, is the “passing of the rigid sigmoidoscope” something for us to be concerned about?

COMPARATIVE SHORT AND LONG-TERM OUTCOMES OF ENTERAL STENTING IN PALLIATING GASTRIC-DUODENAL OUTLET OBSTRUCTION IN ADVANCED GASTRIC AND NON-GASTRIC MALIGNANT DISEASE

S. Sagi, K. Bodger, S. Sarkar. Digestive Disease Centre, University Hospitals Aintree, Liverpool, UK

Introduction: Enteral stenting is recognised as an effective method of palliation in malignant gastric and duodenal outlet obstruction with success rates similar to surgical palliation. Most published data are from case series and there is a lack of comparative data in efficacy of palliative enteral stenting between gastric and non-gastric malignant disease.

Aims & Methods: To compare the outcomes of enteral stenting in palliating gastric-duodenal obstruction in gastric and non-gastric malignant disease. All patients who underwent endoscopic metal stent placement for symptomatic malignant obstruction between 10/98 and 12/05 were identified using the Endoscribe database. Patients with proximal (oesophageal, junctional or proximal gastric) lesions were excluded. Data were abstracted from case notes using a standard Proforma and, pre- and post-procedure oral intake was assessed using a simple scale (1 = None/Inadequate; 2 = Liquids/Thickened Only; 3 = Semisolids; 4 = Unmodified diet).

Results: 32 patients, underwent palliative enteral stenting for malignant gastric (n = 16) and non-gastric (n = 16) disease. Non-gastric diagnoses included; pancreatic primary (n = 9), duodenal primary (n = 3) and other malignancies (n = 4). 14 patients in the gastric and 15 in the Non-gastric groups had metastatic disease. Patients with gastric disease were older (age 75 vs 67 years; p = 0.048) but ASA scores were similar (3.8 vs 3.9: gastric and non-gastric respectively) and the doses of sedation used were identical (mean Midazolam 4.6 mg and Fentanyl 39 μg). Oral intake showed a mean improvement by 1.1 and 1.3 points for gastric and non-gastric disease respectively. Discharges: 72% were successfully discharged; which was slightly higher for those with gastric disease (81% vs 66% for gastric and non-gastric disease respectively). Mean survival in days (SD) were similar at 107 (163) and 93 (110) days (p = 0.8) for gastric and non-gastric disease respectively. 30-day mortality was high with a higher trend for non-gastric disease at 50% compared to 25% for gastric disease (p = 0.1). Complication rates were high in both groups at 50% for gastric and 62.5% for non-gastric disease as was the re-intervention rate at 44% for gastric and 37.5% for non-gastric disease. 25% in each group required either re-stenting or by-surgery.

Conclusion: Enteral stent offers reasonable palliation of symptoms with successful hospital discharge in the majority. However, complications, re-intervention and 30-day post-procedure mortality rates were high with a possible potential benefit towards gastric disease despite the older patient population. All these factors should be considered in an MDT setting before stenting these terminally ill patients.

DO ENDOSCOPY AUDIT TOOLS OVERESTIMATE POLYP DETECTION RATES?

N. Jawad1, T. Bhagat1, S. Murray1, F. Jawad2, R. J. Makins1, E. W. Seward1. 1Endoscopy Unit, Whipps Cross University Hospital, 2Final Year Student, Barts and the London Medical School, London, UK

Introduction: Adenoma detection is taken as a proxy marker of the quality of colonoscopy. The accepted benchmark varies, and many professional societies have judged that ten or even 20% is the expected level for practising colonoscopists. This has attained more importance due to the welcome focus on colonoscopy quality in the UK at present. The majority of commercially available endoscopy recording systems allow the endoscopist to audit their practice very easily by maintaining a database of procedures that can be regularly interrogated. However, unless they possess a means of being updated by access to histology reports, there is a real risk that the adenoma detection rate will represent an overestimate—as self-reported “adenomas” may actually represent alternative diagnoses such as hyperplastic polyps, inflammatory polyps and cancers. Updating the database is currently laborious and time-consuming.

Aims and Methods: We proposed to audit our actual adenoma detection rate (as determined by histology) against our polyp detection rate (as determined by our endoscopy database) to establish if there is a significant difference. Every colonoscopy performed at our institution between July-December 2006 was reviewed and if polyps were identified, the histology was obtained. Confirmatory histology was required to record it as an adenoma. Each independent endoscopist was assessed, significance measured by χ2 test. A significant value was taken to be p<0.05.

View this table: