Objectives To assess detection rate and predictive factors of sessile serrated polyps (SSPs) in organised colorectal cancer (CRC) screening programmes based on the faecal immunochemical test (FIT).
Design Data from a case series of colonoscopies of FIT-positive subjects were provided by 44 Italian CRC screening programmes. Data on screening history, endoscopic procedure and histology results, and additional information on the endoscopy centre and the endoscopists were collected, including the age-standardised and sex-standardised adenoma detection rate (ADR) of the individual endoscopists. The SSP detection rate (SSP-DR) was assessed for the study population. To identify SSP-predictive factors, multilevel analyses were performed according to patient/centre/endoscopist characteristics.
Results We analysed 72 021 colonoscopies, of which 1295 presented with at least one SSP (SSP-DR 1.8%; 95% CI 1.7% to 1.9%). At the per-patient level, SSP-DR was associated with males (OR 1.35; 95% CI 1.17 to 1.54) and caecal intubation (OR 3.75; 95% CI 2.22 to 6.34), but not with the FIT round. The presence of at least one advanced adenoma was more frequent among subjects with SSPs than those without (OR 2.08; 95% CI 1.86 to 2.33). At the per-endoscopist level, SSP-DR was associated with ADR (third vs first ADR quartile: OR 1.55; 95% CI 1.03 to 2.35; fourth vs first quartile: OR 1.89; 95% CI 1.24 to 2.90).
Conclusion The low prevalence of SSPs and the lack of association with the FIT round argue against SSP as a suitable target for FIT-based organised programmes. Strict association of SSP-DR with the key colonoscopy quality indicators, namely caecal intubation rate and high ADR further marginalises the need for SSP-specific quality indicators in FIT-based programmes.
- COLORECTAL CANCER SCREENING
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Organised colorectal cancer screening programmes with faecal immunochemical test have been implemented in Europe.
Adenoma detection rate and caecal intubation rate are the main indicators of the quality of colonoscopy.
Serrated polyps have been suggested to play a role in colorectal carcinogenesis.
Prevalence of serrated polyps and its association with main colonoscopy quality indicators have not been addressed in organised screening programmes.
What are the new findings?
In 72 021 colonoscopies for a positive faecal immunochemical test, only 1.8% of the subjects were diagnosed with a sessile serrated polyp (SSP), and no association with the screening round was detected.
The detection of SSPs was strictly associated with the two main quality indicators for colonoscopy, namely the adenoma detection and caecal intubation rates. A 12-fold gradient in SSP detection was shown when classifying patients according to both quality indicators.
SSPs were more frequently detected among patients with advanced adenomas, supporting an association between the two types of lesions.
How might it impact on clinical practice in the foreseeable future?
Serrated polyps are unlikely to represent a suitable target for organised screening programmes based on the faecal immunochemical test.
Available quality indicators already incorporate the detection level of serrated lesions.
Colorectal cancer (CRC) is a major cause of morbidity and mortality.1 ,2 CRC screening with a biannual faecal occult blood test (FOBT) has been shown to reduce CRC mortality,3 ,4 and faecal immunochemical tests (FITs) have been shown to be more accurate than guaiac-FOBT (g-FOBT) for advanced neoplasia.5 ,6 An organised screening programme with FIT has been implemented in Italy,7 as well as in other European countries and Australia.8 The efficacy of an FOBT-based screening programme ultimately depends on the accuracy of the post-FOBT colonoscopy. In a previous analysis, we have shown a substantial variability in the adenoma detection rate (ADR) that, in turn, has been associated with the risk of post-colonoscopy interval cancer.9–14
A ‘serrated pathway’ has been proposed as an alternative to the adenoma-carcinoma sequence, and considered partially responsible of the risk of interval CRC, especially in the proximal colon.5 ,6 Thus, the detection of serrated lesions may result in an additional reduction of CRC incidence, as compared with the simple detection of adenomas. Moreover, patients with serrated lesions considered to be at higher risk of CRC—such as large sessile serrated polyps (SSPs) or those with dysplasia—are usually included in post-polypectomy surveillance programmes in order to further reduce the CRC risk.15–18
The prevalence of serrated lesions within an organised screening programme with FIT is unclear, as well as its association with the screening round. In addition, the association between the main quality indicators for colonoscopy and the detection of such lesions in a FIT programme has not been addressed, generating uncertainty regarding the need to include these lesions within quality assurance programmes.19–21 Since these lesions may be overlooked during a colonoscopy due to their flat and thin appearance, it is likely that endoscopist-related features may be relevant in explaining the expected variability in the detection of these lesions.
Our aim was to assess the SSP detection rate (SSP-DR) in an organised screening programme with FIT as well as its association with patient and endoscopist characteristics.
Materials and methods
Organised CRC screening programme in Italy
The organised screening programme with FIT in Italy is performed at a regional level. Screening centres in each region are responsible for inviting eligible subjects. Regional programmes invite people (mostly aged 50–69 years) to perform a single-sample FIT on a biennial basis. Due to heterogeneity in resources, the screening programme has been mainly implemented in the north-central Italian regions.7 Individuals with a positive FIT result (cut-off=20 µg HB/faecal g) are offered a colonoscopy. Subjects attending a colonoscopy are required to give pre-procedural consent and are given explicit instructions on bowel preparation. Sedation and bowel preparation are performed in accordance with local guidelines, which vary among screening centres. All demographic, colonoscopic and pathological data are recorded by the screening centre staff or by the endoscopy nurses, depending on local organisational arrangements.
There is no nationwide established accreditation programme, or examination and/or audit for an endoscopist to practise in the organised screening programme.
All the organised CRC screening programmes in Italy were invited to participate in this study. Programmes were required to provide individual data about all colonoscopies carried out within the regional screening programmes during 2010. Depending on data availability, individual programmes or regions could provide additional data referring to other calendar years. Demographic information, screening history and colonoscopic procedure and histology results were collected for each patient on a standard precoded electronic form. Information concerning the endoscopy centre (EC), and the endoscopists was collected based on a standard form (table 1).
In order to reduce variation in the baseline prevalence of lesions, and to avoid double-counting of partially removed lesions, only first colonoscopies following a positive FIT were included in the analysis.
As long as monitoring of the screening activity (including documenting and reporting performance indicators at the centre and operator level) is part of a planned evaluation activity, each centre does not need specific authorisation from an ethics committee to conduct follow-up analyses of screening outcomes for people enrolled in the local programmes. Moreover, the consent signed by screening participants includes the handling of personal data by screening authorities for quality monitoring purposes. For the purposes of this multicentre analysis, all data were anonymised, following Italian regulations.
At the patient level, individual colonoscopy data routinely included the following information: (1) patient demographics, (2) number of previous FIT examinations, (3) quality of bowel preparation (endoscopist's judgement, aggregated as adequate/inadequate), (4) sedation (conscious/deep/none), (5) caecal intubation and (6) number and characteristics of diagnosed polyps/masses, including morphology, localisation, size and histology. In detail, due to the heterogeneity of the bowel preparation scales adopted in different programmes, participating centres were asked to record bowel cleansing, according to a 4-category scale (excellent, good, fair, poor) that was subsequently aggregated as adequate (excellent/good) and inadequate (fair/poor).
At the endoscopist level, we collected (1) demographics, (2) specialty, (3) number of years’ experience as endoscopist (≤5, 6–9, ≥10-year experience), (4) number of screening colonoscopies performed in 2010 (≤100, 101–180, >180), (5) overall number of colonoscopies performed in 2010 (≤300, 301–600, >600) and (6) the quartile of age-standardised and sex-standardised ADR, as previously defined22 (this parameter was calculated only for endoscopists whose number and case-mix of colonoscopies allowed standardisation).
At the centre level, we collected the following indicators: (1) use of dedicated screening sessions for FIT-positive colonoscopies (ie, scheduled outside daily non-screening endoscopy), (2) use of sedation (<33%, 33%–66% and >66% of colonoscopies) and (3) adoption of routine quality assurance procedures (ie, monitoring of post-procedural complications, documentation of caecal intubation, scale used to assess quality of bowel preparation). Based on the 2010 data, we also calculated the number of post-FIT-positive colonoscopies performed in 2010 (≤300, 301–600 and >600).
Study end points and statistical analysis
Advanced adenomas were defined as adenomas showing any of the following features: size ≥10 mm, or high-grade dysplasia, or villous component >20%.
Histopathological classification of serrated lesions within the organised screening programme is in line with the European Guidelines, and all the participating programmes had implemented formal training for the pathologists with updated classification of the serrated lesions.23 In particular, serrated lesions were mainly divided into hyperplastic and SSPs, and the degree of dysplasia and size were recorded. For the purpose of our analysis, only SSPs were considered since the role of hyperplastic polyps in CRC carcinogenesis is supposed to be marginal.21 However, due to the suboptimal pathologist interobserver agreement in the classification of serrated lesions, we also performed a sensitivity analysis by summing up SSPs with ≥10 mm hyperplastic polyps. A second sensitivity analysis was performed by limiting the analysis only to screenees with SSPs without coexisting adenomas in order to exclude possible confounding resulting from the association between the two lesions.
In order to identify possible SSP predictors, we evaluated the association between several variables and SSPs. These variables were classified as characteristics of (1) patient, (2) EC and (3) endoscopist (table 1). Sedation was considered an organisation modality of the EC.
We initially explored the relationship between each group of variables and SSP detection by univariate analysis. The variables with a significant association with SSPs were included in a multivariate logistic model with SE correction. Finally, the variables that remained associated with SSP detection were included in a final multilevel model (with Laplace approximation), that accounted for the intra-level and inter-level variability at the following levels: region, EC and endoscopist.
In order to explore a possible association between SSPs and advanced adenomas, we finally compared SSP prevalence between subjects with and without advanced adenomas.
All reported p values are two-sided. A p value of less than 0.05 was considered statistically significant. All analyses were performed using the Stata (V.10.0) statistical package.
This study included 72 021 colonoscopies performed for positive FIT (mean age: 61.3 years; men: 57%). Additional details on the study cohort are provided in a previous publication.22
Overall, 41 960 patients (58.3%) were at the first FIT examination, and those remaining had one or more previous negative FIT examinations (table 2). Sedation was used in 83.3% of the cases. When reported (n=24 041, 40.3%), it was conscious sedation in most cases (89.9%). Caecal intubation was achieved in 93.5% of the patients. Colon cleansing was rated adequate in 93% of the patients.
The endoscopist's characteristics were available for 71% of the endoscopists (354/501), accounting for 89.6% of the colonoscopies (64 686/72 201). In the remaining cases, the endoscopists’ identification code was recorded, but their characteristics were not provided. Of the examinations where the information was reported, 76.5% colonoscopies were performed by male endoscopists, and 43.8% by endoscopists between 50–59 years of age (table 3). The majority of the colonoscopies (74.8%) were performed by gastroenterologists, with the remaining by surgeons or other specialties. Distribution of endoscopists/colonoscopies according to experience/volume of procedures is provided in table 3. More than 69% of the endoscopists had more than 10 years of experience.
Endoscopic centre characteristics
EC information was available for 76 centres, accounting for 91.5% of the study colonoscopies (66 088/72 201). In 84.2% of the 76 centres, colonoscopy for positive FITs was performed in dedicated sessions, accounting for 90% of the procedures (table 4). Sedation was used in ≥66% of colonoscopies in 60 ECs (78.9%), accounting for 55 946 procedures (84.7%). Sedation was performed in 33%–66% and <33% of colonoscopies in nine (11.8%) and seven centres (9.2%), respectively. Distribution of centres according to the volume of yearly screening colonoscopies is reported in table 4.
Sessile serrated polyp detection rate
At a per-patient analysis, at least one SSP was detected in 1295 patients, corresponding to 1.8% of the procedures. Distribution and main characteristics of the SSPs are provided in table 5.
Detection rates of SSPs according to the study variables are provided in table 6. The SSP-DR varied across regions (range: 0.22%–6.95%) and showed a trend toward an increase over time in those programmes providing data spanning over several years (range: 1.62%–2.09%).
At a multilevel analysis (table 7), SSP-DR was associated only with the following factors: patient-related, male sex (OR 1.35; 95% CI 1.17 to 1.54) and caecal intubation (OR 3.75; 95% CI 2.22 to 6.34); endoscopist-related, the two highest quartiles of ADR as compared with the first quartile (third quartile: OR 1.55; 95% CI 1.03 to 2.35; fourth quartile: OR 1.89; 95% CI 1.24 to 2.90). The distribution of SSP-DR according to ADR and caecal intubation rate is described in figure 1.
When adding—at a sensitivity analysis—the large (≥10 mm) hyperplastic polyps (n=1029), or excluding patients with coexisting SSPs and adenomas, the main results of the multilevel analysis were confirmed (see online supplementary appendix 1–3).
Association between SSPs and advanced adenomas
At a per-patient analysis, the presence of at least one advanced adenoma was more frequent among subjects with an SSP than those without (593/1295, 46% vs 20 401/70 726, 29%, p<0.001; OR 2.08; 95% CI 1.86 to 2.33).
In an organized CRC screening program with FIT, the detection of sessile serrated polyps was strictly associated with two key quality indicators of colonoscopy, namely the adenoma detection rate and caecal intubation rate, and it occurred in less than 2% of the FIT+ subjects. We also showed an association between SSPs and advanced adenomas. The results of our analysis are relevant for the following reasons.
First, we showed a strict association between a high ADR and SSP-detection rate. Among the 354 endoscopists included in our analysis, those distributed in the two higher quartiles of ADR showed a significant increase in the detection of SSPs. The ADR-endoscopist distribution is likely to be accurate when considering the very high number of patients with at least one adenoma – i.e. 32,265/72,021 patients – in our enriched-disease population, and the minimization of the variability in the expected prevalence of neoplasia among FIT positive subjects, given the high PPV of the test, consistent across different programs.
The fact that endoscopist's ADR is a good proxy for SSP detection simplifies the quality assurance process. Indeed, the utilization of the SSP detection rate as a quality indicator in a screening would be hampered by the very low prevalence of these lesions (which would require a high number of colonoscopies per endoscopist to achieve precision for individual endoscopist estimates), as well as by the observed inter-pathologist variability in SSP diagnosis.24 On the other hand, the feasibility of endoscopist's ADR as an auditable quality indicator has already been shown in several countries.22 ,25 ,26 Inter-pathologist variability in adenomatous diagnosis is also very low, ensuring the comparability of such a quality indicator among the endoscopists.27 A suboptimal ADR is related to the post-colonoscopy interval cancer risk,13 ,14 ,28 ,29 and a suboptimal endoscopist's ADR is subject to improvement with training.30 Thus, the finding of a strict association between ADR and SSP detection rates in our series supports the use of ADR as a proxy for SSP detection.
A higher SSP-DR was reported in a previous study conducted in a small (N=389) gFOBT population, with any significant serrated polyps (including small proximal hyperplastic polyps) as the main target.31 However, we do not feel that it affects the main conclusions of our study. Our report is based on a large database of routine screening programs, with a highly specific (i.e., restricted to SSP) definition of the outcome of interest, and it is therefore likely to be reproducible and generalizable in a European setting.
Also, as expected, the SSPs prevalence in our study varied across different regions, ranging between 0.22% and 6.95%, and it tended to be higher in the most recent series, whenever several years' data were available. However, the highest observed prevalence was still lower than 7%, which in fact remains consistent with our conclusion suggesting that FIT does not increase the SSP population to an extent similar to what FIT does for advanced neoplasia.
Although our analysis did not address the reasons underlying the association between ADR and SSP-DR, the most plausible hypothesis is that a meticulous exploration of the colorectal mucosa and the ability to recognize non-polypoid lesions by experienced endoscopists would simultaneously result in higher detection of both adenomatous and serrated lesions.32 Of note, a long withdrawal time has recently been associated with an increase in both adenoma and serrated polyp detection 33. In addition, the findings of our sensitivity analysis, showing the same trend for the association between the endoscopist's ADR and the SSP-DR also when considering only patients without coexisting adenomas, would confirm that the higher quality of the examination technique, rather than the observed association between SSPs and adenomas, can explain the increased detection both of adenomas and of SSPs. The implications of a higher SSP detection rate among endoscopists with a high ADR are intriguing: the observed association of a high ADR in a primary endoscopic screening with a reduction in CRC incidence and mortality13 ,14 ,28 ,29 may be partly explained by the higher detection of SSPs. Indeed, the association between advanced neoplasia and SSPs coupled with the impressive enrichment of neoplasia in FIT positive may potentiate the efficacy of screening in those with advanced neoplasia, when considering that a putative role for these lesions in the natural history of interval CRC has been consistently shown at a molecular level.21 The strong association between the caecal intubation rate and SSPs was not unexpected. The SSPs in our series – as well as in previous ones32 ,34 – tend to be more frequently located in the proximal colon. Since incomplete colonoscopies are frequently limited to the distal colon,35 a large decrease in SSP prevalence between the incomplete and complete colonoscopy groups appears plausible. Similarly to ADR, the association between caecal intubation and SSP detection is clinically relevant. The caecal intubation rate is already considered a universal key indicator of colonoscopy quality, being also associated with the risk of post-colonoscopy interval cancer.12 ,36 ,37 Thus, any improvement in such an indicator is likely to favourably affect the SSP detection rate.
Overall, there was a strong association for SSP detection between ADR and caecal intubation rate, a 12-fold difference being observed between those in the first ADR quartile with incomplete colonoscopy and those in the fourth ADR quartile with successful caecal intubation. Such a large difference underlines the impact of our findings in the clinical practice.
Second, the very low prevalence of SSPs argues against SSPs as a relevant target of FIT programs. Our SSP prevalence is similar to previous estimates obtained in primary colonoscopy screening or unselected settings,38 ,39 as well as with a large cohort trial comparing FIT with stool DNA, where SSPs were considered to be false positive results.40
Thus, the role of FIT-related SSP detection in explaining the efficacy of FIT in reducing CRC incidence and mortality41 ,42 is likely to be negligible. Indeed, while FIT+ subjects represent a selected sub-group of the enriched-disease population with prevalence of 45% and 29% for adenomas and advanced adenomas, respectively22 – much higher than those measured in a primary screening colonoscopy setting43 ,44 – no enrichment of SSP was observed in our FIT+ population, which suggests that FIT-related SSP detection may represent a random event.
The lack of association between FIT and SSP detection is also supported by the following two observations. First, we showed a similar prevalence of SSA in subjects at 1st or ≥2nd FIT rounds – 1.9% vs 1.7%. This finding is markedly different from the substantial decrease in the prevalence of advanced adenomas in subsequent FIT rounds as compared to the first round, due to the higher sensitivity of FIT for prevalent advanced neoplasms. Secondly, in order to evaluate the contribution of SSPs to FIT positivity, we assessed the prevalence of advanced adenomas in subjects with and without SSPs. Assuming an independent contribution of SSPs to FIT positivity, a lower prevalence of advanced adenomas would have been expected in subjects with SSPs. Contrary to this assumption, we found an increased prevalence of advanced adenomas in screenees detected with SSPs, which would confirm the association between SSPs and advanced adenomas already shown in primary screening settings,38 ,45 ,46 ,47 ,48 ,49 suggesting that these two lesions may share environmental or genetic risk factors.
Thus, if SSPs are to be considered a major target for a population-based screening program, FIT would appear to be a suboptimal choice, especially when compared with a primary colonoscopy screening.32 ,50 However, further data on the natural history of SSPs – and of serrated lesions more generally – are needed before considering SSPs as a major target for CRC screening population programs. Estimates of SSP prevalence are also useful for measuring the burden related with post-polypectomy surveillance. We have shown that nearly half the SSPs were associated with at least one adenoma, in these cases surveillance being likely driven by the latter lesion. Thus, the actual burden related with post-polypectomy surveillance of only SSPs in a FIT program is limited to less than 1% of the FIT+ population, marginalizing the impact of more intensive post-polypectomy surveillance intervals, as advocated by endoscopy societies.15 ,16
There are limitations in our analysis.
Pathological assessment was not centralized. However, all the participating programmes had implemented formal training for the pathologists with updated classification of the serrated lesions, while the very large sample size of our analysis is likely to minimize the effect of pathology variability. For instance, despite a high inter-observer variability having been shown in the diagnosis of advanced adenomas,27 the detection rates of advanced neoplasia in different screening programmes appear to be similar,51 suggesting that the under- and over-diagnosis rates by the pathologists tend to be similar across the programmes. Moreover, at a sensitivity analysis, the inclusion of ≥10 mm hyperplastic polyps – that could have been misclassified as SSPs – did not affect the main result of our study, confirming the robustness of our conclusions.
We excluded small hyperplastic polyps from our analysis and we did not distinguish between SSPs and traditional serrated adenomas (TSA). The former choice was justified by the unlikely role of small indolent hyperplastic lesions in colorectal carcinogenesis,21 while the very low prevalence of TSAs argues against any consideration of the DR of these lesions as a relevant quality indicator.20 ,32 ,34
Finally, the extent of missing data was relevant for a number of variables related to endoscopists' and endoscopy centers' characteristics. However, those items were not associated with SSP-DR, while less than 0.5% missing values affected two out of the three variables included in the final multilevel multivariate analysis (i.e., patients' gender and caecal intubation). Even if the third variable (i.e., endoscopist's ADR) was missing for 5075 records (7%), information about endoscopist's Id and whether an adenoma was detected were available for all the study colonoscopies. Endoscopists' ADR was calculated only for the endoscopists whose number and case-mix of colonoscopies allowed for standardization, thus only cases which could not contribute meaningful information were excluded from the final model.
In conclusion, we showed that, due to their low prevalence, serrated lesions are unlikely to represent a suitable target for FIT programmes, and that the adoption of new indicators specific for such lesions is not advised, when considering the strict association between such already widely adopted key quality indicators as ADR and the caecal intubation rate, and the detection of such lesions.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Collaborators The Equipe working Group: Giuseppe Feliciangeli (ASUR Marche AV3, Macerata, Italy); Cinzia Campari (Azienda Unità Sanitaria Locale Reggio Emilia, Italy); Fabio Falcini (Romagna Cancer Registry, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (IRST) IRCCS, Meldola, Forlì, Italy). Orietta Giuliani (Romagna Cancer Registry, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori (IRST) IRCCS, Meldola, Forlì, Italy). Omero Triossi (Azienda Unità Sanitaria Locale Ravenna, Italy). Vincenzo Matarese (Unit of Gastroenterology, Departments of General Surgery, S Anna University Hospital of Ferrara, Italy). Chiara Fedato (Settore promozione e sviluppo igiene e sanità pubblica, Regione Veneto, Venezia, Italy). Anna Turrin (Settore promozione e sviluppo igiene e sanità pubblica, Regione Veneto, Venezia, Italy). Susanna Baracco (Veneto Tumour Registry, Padua, Italy). Fabio Monica (Gastroenterology Unit, Azienda Ospedaliero—Universitaria Ospedali Riuniti, Trieste, Italy). Beatrice Mallardi (Clinical and Descriptive Epidemiology Unit, ISPO—Istituto per lo Studio e la Prevenzione Oncologica, Florence, Italy). Francesco Quadrino (Agency for Public Health, Lazio Region, Rome, Italy. Ivana Valle (S.S.D. Popolazione a rischio—Dipartimento di Prevenzione, ASL3-Genovese, Genova, Italy). Patrizia Landi (Gastroenterology Unit, P.O. Bellaria AUSL Bologna, Italy). Elisabetta Buscarini (Gastroenterology Unit, Maggiore Hospital, Crema, Italy). Richard Nienstedt (Gastroenterology Unit, Ospedale S. Chiara, Trento, Italy). Pietro Occhipinti (Gastroenterology Unit, AOU Maggiore della Carità, Novara, Italy). Arrigo Arrigoni (Gastroenterology Unit, AOU Città della Salute e della Scienza, Torino, Italy). Mario Grassini (Gastroenterology Unit, Ospedale Cardinal Massaia, Asti, Italy).
Contributors MZo, CS and CH: study concept and design; interpretation of results; drafting of the manuscript. MZo and FDR: acquisition of data and statistical analysis. All the authors: critical revision of the article for important intellectual content and final approval of the article.
Funding The project was financed by the Italian Ministry of Health—CCM (Centre for Disease Control).
Competing interests DKR: Olympus—research support.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.