CT colonography for detection and characterisation of synchronous proximal colonic lesions in patients with stenosing colorectal cancer
- Seong Ho Park1,
- Ju Hee Lee1,
- Seung Soo Lee1,
- Jin Cheon Kim2,
- Chang Sik Yu2,
- Hee Cheol Kim3,
- Byong Duk Ye4,
- Mi-Jung Kim5,
- Ah Young Kim1,
- Hyun Kwon Ha1
- 1Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea
- 2Department of Surgery, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea
- 3Department of Surgery, Sungkyunkwan University School of Medicine, Samsung Medical Center, Seoul, Korea
- 4Department of Gastroenterology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea
- 5Department of Pathology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea
- Correspondence to Dr Seong Ho Park, Department of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Asanbyeongwon-gil 86, Songpa-gu, Seoul 138-736, Korea;
Contributors Study concept and design: SHP, JHL, AYK, HKH. Acquisition of data: SHP, JHL, SSL, JCK, CSY, HCK, BDY, M-JK, AYK. Analysis and interpretation of data: SHP, JHL, BDY, M-J K. Drafting of the manuscript: SHP, JHL. Critical revision of the manuscript for important intellectual content: SHP, JHL, SSL, JCK, CSY, HCK, BDY, M-JK, AYK, HKH. Statistical analysis: SHP, JHL. Administrative, technical, or material support: SSL, JCK, CSY, HCK, BDY, M-JK, AYK. Supervision: SHP, HKH. Final approval of the version: SHP, JHL, SSL, JCK, CSY, HCK, BDY, M-JK, AYK, HKH.
- Revised 11 October 2011
- Accepted 14 October 2011
- Published Online First 23 November 2011
Objective To investigate CT colonography (CTC) performance for detecting and characterising synchronous lesions proximal to a stenosing colorectal cancer and to suggest patient management strategies according to the CTC findings.
Methods 411 consecutive patients underwent CTC for proximal colonic evaluation after failed colonoscopy past a newly diagnosed stenosing colorectal cancer. Pathological examination of colectomy specimen and/or postsurgical colonoscopy with pathological confirmation of the proximal synchronous lesions to serve as reference standards existed in 284 patients. Per-patient and per-lesion diagnostic performance measures of CTC for diagnosing proximal synchronous lesions ≥6 mm analysed by histopathological categories were obtained for the 284 patients. Per-lesion sensitivity and positive predictive value (PPV) of various CTC lesion size criteria and lesion size combined with other CTC findings for diagnosing cancer in the proximal colon were determined.
Results Both per-patient and per-lesion CTC detection sensitivities for proximal synchronous cancers were 100% (6/6 patients and 8/8 lesions; 95% CI 64.3% to 100% and 70.7% to 100%, respectively) with the corresponding per-patient negative predictive value (NPV) of a negative CTC of 100% (194/194 patients; 95% CI 98.3% to 100%). Per-patient NPV of a negative CTC for advanced neoplasia (ie, advanced adenomas and colorectal cancers) was 97.4% (189/194 patients; 95% CI 93.9% to 99.1%). A lesion size ≥15 mm on CTC as the criterion to specifically diagnose proximal cancer yielded 87.5% (7/8 lesions; 95% CI 50.8% to 99.9%) per-lesion sensitivity, rendering one 8-mm submucosal cancer mischaracterised as a non-cancerous lesion, and 70% (7/10 lesions; 95% CI 39.2% to 89.7%) per-lesion PPV. Additional CTC findings did not improve the sensitivity.
Conclusion CTC is highly sensitive in detecting synchronous cancers proximal to a stenosing colorectal cancer. CTC has limited capability in differentiating advanced adenomas from colorectal cancer and this compromises the PPV of CTC for the presence of proximal cancer.
- Colorectal cancer
- CT colonography
- preoperative evaluation
- computer tomography
- hepatobiliary radiology
- inflammatory bowel disease
Significance of this study
What is already known on this subject?
Synchronous cancers are reported to be present in 1–7% of patients diagnosed with colorectal cancer.
Complete evaluation of the entire colon in patients diagnosed with colorectal cancer is important for proper surgical/post-surgical management.
CT colonography (CTC) may have a high sensitivity for detecting synchronous cancers proximal to a stenosing colorectal cancer.
What are the new findings?
Excellent sensitivity of CTC for detecting synchronous cancers proximal to a stenosing colorectal cancer is confirmed.
CTC sensitivity for detecting proximal synchronous adenomas, including advanced adenomas, is moderate.
CTC ability to characterise proximal synchronous lesions detected on CTC (ie, specific diagnosis of cancer vs non-cancerous lesion) is limited; this makes the actual CTC sensitivity to specifically diagnose proximal cancer mildly compromised in order to maintain positive predictive value (PPV) for the cancer diagnosis at a reasonable level.
How might it impact on clinical practice in the foreseeable future?
CTC can be used effectively to screen the proximal colon for synchronous cancers in patients who have a stenosing colorectal cancer and, considering the limited capability of CTC in differentiating advanced adenomas from colorectal cancer, a conservative criterion such as lesion size ≥15 mm would be appropriate for the cancer diagnosis on CTC to balance sensitivity and PPV. A more robust guideline for patient management according to the CTC findings to maximise patient benefits while minimising adverse results will be further defined.
Complete evaluation of the entire colon in patients diagnosed with colorectal cancer is important because identification of synchronous cancers, reported to be present in 1–7% of these patients,1 ,2 may modify the extensiveness of the surgical procedure, and failure to diagnose them may result in second surgery or failure of curative treatment. CT colonography (CTC) is a recently developed technique recommended by some authorities for evaluation of the proximal colon in patients with stenosing colorectal cancer that prevents colonoscopic examination past the cancer.3 ,4 However, the recommendations are based on several preliminary studies that merely suggested high CTC sensitivity for detecting proximal synchronous cancers,5–10 in contrast to the thorough validation of CTC in screening for colorectal cancer.11–15 Among the limitations of these earlier studies were mixed study populations including patients with and without stenosing cancer, small numbers of relevant subjects (9–67 patients), inclusion of only a few patients with large proximal synchronous cancers, and/or lack of systematic analysis of CTC performance according to lesion histology. Moreover, while correct characterisation of the proximal synchronous colonic lesions detected on CTC (ie, specific diagnosis of cancer vs non-cancerous lesion) is also crucial for appropriate patient management, because a lack of correct lesion characterisation would still lead to either insufficient or unnecessarily extensive surgery despite accurate lesion detection, the previous studies only investigated lesion detection capability of CTC. Therefore, it is as yet unclear how accurately CTC can specifically suggest proximal synchronous cancer versus a non-cancerous lesion instead of merely detecting the lesion, as well as how to manage individual proximal synchronous lesions when detected by CTC—that is, surgical resection by increasing the surgical extent versus less invasive postsurgical colonoscopic removal. We therefore investigated the performance of CTC to detect and characterise synchronous lesions proximal to a stenosing colorectal cancer and sought to suggest patient management strategies according to the CTC findings.
Materials and methods
The Institutional Review Board of Asan Medical Center approved this retrospective study and waived patient informed consent.
Between April 2007 and December 2009, 427 consecutive patients who had failed colonoscopic examination of the proximal colon past a newly diagnosed advanced colorectal cancer, without an acutely severe colonic obstruction requiring immediate colonic decompression (see figure 1), underwent CTC with intravenous contrast enhancement for evaluation of the proximal colon as well as extracolonic organs (ie, staging workup) at Asan Medical Center, a tertiary referral centre. The reasons for failed colonoscopic examination of the proximal colon were the inability to advance the colonoscope past the lesion (n=405), and the stenosing cancer-associated severely impaired bowel preparation (n=19), restricted manipulation of colonoscope (n=2), and both (n=1) in the proximal colon. Of these, 16 patients turned out to have cancers located in the proximal ascending colon and ileocaecal valve, resulting in the almost absence of a proximal colonic segment, and were excluded from the study. The remaining 411 patients (mean age±SD, 59.9±11.5 years; 237 men and 174 women) constituted the study population. The stenosing colorectal cancers were located in the rectum (n=105), sigmoid colon (n=184), descending colon (n=30), transverse colon (n=43), and ascending colon (n=49).
Technical assessment of CTC included all 411 patients, whereas analysis of colonic findings on CTC included 284 patients (59.4±11 years; 165 men and 119 women) according to the inclusion and exclusion criteria. The inclusion criteria were complete evaluation of the proximal colon by pathological examination of colectomy specimen and/or postsurgical colonoscopy and pathological confirmation of any proximal colonic lesions present to serve as the reference standards (see Reference standards section). The exclusion criterion was the presence of numerous polyps (ie, polyposis) in the proximal colon, which made lesion-by-lesion matching between the reference standards and CTC impossible.
Cathartic purgation, faecal tagging, and colonic distention used standard methods widely adopted (see online appendix). Both prone and supine scans were performed, one without intravenous contrast and the other at a portal phase after intravenous contrast enhancement, using a 16-detector CT scanner (Somatom Sensation 16; Siemens Medical Solutions, Erlangen, Germany) (see online appendix).
Blinded CTC interpretation
Performance measures of CTC for diagnosing lesions in the proximal colon were evaluated using our original clinical reading which was prospective and blinded in nature as it preceded the reference standards and was standardised. Images were interpreted by one of the two experienced board-certified radiologists (who had an approximate 1000- and 500-CTC case experience, respectively, at the beginning of the study period), using primary three-dimensional endoluminal navigation with two-dimensional problem solving at a CTC workstation (Xelis; Infinitt, Seoul, Korea), except for patients with excessive amounts of residual faecal matter, in which a primary two-dimensional approach was used (n=51). Technical adequacy for the proximal colon was assessed according to the CTC Reporting and Data System; inadequate examinations (C0) where the presence of polyps ≥1 cm could not be excluded were recorded along with the cause of each problem.16 All lesions ≥6 mm in maximal diameter were reported.16 Lipomas were generally disregarded.16 Segmental location and lesion morphology were recorded.16 The readers did not suggest malignancy or benign nature for the detected lesions in the proximal colon except for apparently malignant-looking large masses with strong suspicion of adjacent tissue invasion. Extracolonic evaluation was performed as routine but the results were not included in this study. Any examination-related complications were recorded.
Reference standards for the proximal colonic findings and CTC reference standards correlation
Complete evaluation of the proximal colon by pathological examination of colectomy specimens and/or postsurgical colonoscopy and pathological confirmation of the existing colonic lesions served as the reference standards for analysis of colonic findings on CTC and were available in 284 patients, including 79 evaluated with colectomy specimens only and 205 evaluated with colectomy specimens and postsurgical colonoscopy, but excluding seven patients with numerous (at least several dozens of) polyps in the proximal colon (figure 1). Ninety of 132 (68.2%) CTC-positive patients (ie, presence of lesions ≥6 mm in the proximal colon on CTC) and 194 of 279 (69.5%) CTC-negative patients (ie, absence of lesions ≥6 mm in the proximal colon on CTC) were included in the colonic findings analysis cohort, indicating no significant difference between the two proportions (p=0.820). In the majority of the patients without reference standards for the proximal colon, this was due to medical reasons that rendered further proximal colonic evaluation unbeneficial (figure 1), leaving only 41 unspecified patients who did not undergo colonoscopy after curative colorectal surgery likely related to the absence of synchronous lesions (n=30) and the presence of only small (6–9 mm) lesions (n=11) in the proximal colon on preoperative CTC. The interval between CTC and the reference standards in the 284 patients ranged from 0 days to 24 months (median 23 days). A long interval mostly represented postoperative surveillance colonoscopy in patients with negative CTC findings. In patients with proximal synchronous cancers, the interval ranged from 3 to 21 days (median 12 days).
Colonoscopy was performed by experienced board-certified gastroenterologists or colorectal surgeons, who had at least a 1000-colonoscopy case experience, using a video colonoscope (CF 260; Olympus Optical Co., Tokyo, Japan). Segmental location, size and morphology of detected lesions were recorded, and lesions were removed or biopsied for pathological examination. Pathological examination of the colectomy specimens was performed by experienced board-certified pathologists, who had at least a 2-year non-trainee experience in the relevant field, and reported in the same way. Adenocarcinomas were defined as malignant lesions which had invaded into the mucosal lamina propria or deeper, whereas malignant cells that were confined by the glandular basement membrane (ie, intraepithelial carcinomas) were considered as high-grade dysplasia.17 Segmental unblinding of the CTC results13 was not used and complete CTC results were made available to the colonoscopists and pathologists prior to their examinations. Lesion matching between reference standards and CTC was performed by consensus between an experienced board-certified radiologist (with an approximate 300-CTC case experience) and an experienced board-certified gastroenterologist (with an approximate 8000-colonoscopy case experience), both of whom had not been involved in performing the CTC or colonoscopy examinations of the study patients or handling of the study data, and required lesion location within the same or adjacent segment and lesion size within 50% of the reference findings.12 ,13 Those lesions missed during the blinded CTC interpretation were retrospectively evaluated and plausible reasons for missing these lesions were determined.
Retrospective CTC review for analysis of CTC criteria to suggest proximal synchronous cancer
Regarding the CTC criteria for characterising the proximal colonic lesions detected by CTC as cancer versus non-cancerous lesion, we focused on the characterisation of lesions ≥10 mm measured on CTC because, considering the very low rate of cancer among colonic lesions less than 10 mm,13 ,18–22 it would be impractical to suggest such a small lesion to be a cancer that needs to be concomitantly removed by surgery. Therefore, all proximal colonic lesions ≥10 mm or greater detected on CTC, including both true (identified either prospectively or retrospectively) and false-positive lesions, were marked on the CTC data. Then, additional lesional/peri-lesional characteristics on CTC, including lesion surface contour (even and smooth vs irregular, lobular or centrally depressed), pericolic abnormality (infiltration or nodular advancing margin), and enlarged pericolic lymph nodes (short axis ≥5 mm), were retrospectively evaluated by an experienced board-certified radiologist (with an approximate 200-CTC case experience and a 10-year non-trainee experience in abdominal CT) who had not been involved in the clinical CTC interpretation of the study patients or the lesion matching and was blinded to the lesion identity. We did not perform quantitative assessment of lesion enhancement as it was shown to be unrelated to lesion histology.23
Performance measures of CTC for diagnosing lesions ≥6 mm in the colon proximal to stenosing cancer included per-patient sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) as well as per-lesion sensitivity and PPV. These performance measures were assessed according to lesion histological categories including all types of lesions, adenomas or colorectal cancers, advanced neoplasia, and colorectal cancers. Colonic lipomas were excluded from the analysis.16 Advanced neoplasia represented both advanced adenomas (adenomas ≥10 mm in size or with a substantial villous component or high-grade dysplasia15 ,24) and colorectal cancers. For the per-patient analysis, a patient with positive CTC result (ie, presence of one or more lesions ≥6 mm in the proximal colon at CTC) and with one or more lesions ≥6 mm of a particular histological category at the reference standards, was considered a true positive for that category.12 The denominator for specificity was the number of patients who had no lesions ≥6 mm of each histological category in the proximal colon at the reference standards. Per-lesion sensitivities of CTC for detecting individual histological types of lesions, that is, non-neoplastic lesions, non-advanced adenomas, advanced adenomas, and cancers, were also separately obtained. Regarding the analysis of CTC criteria to suggest proximal synchronous cancer on CTC, per-lesion sensitivities and PPVs of various CTC findings for diagnosing proximal synchronous cancer were obtained using the retrospective CTC image review results. The 95% CIs were calculated for proportional data.
CTC examination quality and procedural complications
CTC examination quality in the proximal colon was inadequate in 15 of 411 (3.6%) patients due to inadequate distension (n=8), a large amount of weakly tagged faecal matter (n=6) or both (n=1). All patients tolerated the examination well. There was no incidence of colonic perforation or other significant complications related to the CTC procedure.
Summary of synchronous colonic lesions proximal to stenosing cancer
Of the 284 patients with complete reference standards, 78, 73, 44 and 6 (2.1%) patients had one or more lesions ≥6 mm of any histology, adenoma or colorectal cancer, advanced neoplasia (either advanced adenoma or colorectal cancer), and colorectal cancer, respectively, in the colon proximal to stenosing colorectal cancer. Six patients, who did not have proximal synchronous cancers, were found to have synchronous cancers distal to the stenosing cancer resulting in an overall 4.2% (12/284) rate of synchronous cancers in the study population.
A total of 152 colonic lesions ≥6 mm were identified in the proximal colon at the reference standards, including 108 sessile, 30 pedunculated, 12 flat and 2 mass (>3 cm) lesions (table 1). A total of 175 lesions ≤5 mm were also present in the proximal colon but were excluded from further analysis. All the lesions ≤5 mm were either non-neoplastic lesions or non-advanced adenomas (table 1). There were a total of eight synchronous colorectal cancers in the proximal colon, all confirmed to be adenocarcinomas by surgical resection: a 19-mm and a 22-mm mucosal cancer, both with tumour invasion into the mucosal lamina propria; an 8-mm and a 15-mm cancer with submucosal tumour invasion deep to 500 μm and 3175 μm from the muscularis mucosa, respectively; a 27-mm, a 34-mm and a 52-mm T3 cancer; and a 28-mm T4 cancer. Five of the eight cancers were outside the extent of low anterior or anterior resections for stenosing cancer, whereas three were within the extent of right hemicolectomy for stenosing cancer.
CTC performance for diagnosing synchronous lesions in the proximal colon
Of the 284 patients with complete reference standards, 90 and 194 (68.3%) patients had positive (presence of lesion ≥6 mm in the proximal colon on CTC) and negative (absence of lesion ≥6 mm in the proximal colon on CTC) findings on CTC, respectively. The per-patient CTC sensitivity for detecting patients with colorectal cancer and advanced neoplasia in the proximal colon was 100% (6/6 patients) and 88.6% (39/44 patients), respectively. The corresponding per-patient NPV was high, 100% (194/194 patients) for colorectal cancers and 97.4% (189/194) for advanced neoplasia.
The per-lesion sensitivity of CTC for detecting colorectal cancer was also 100% (8/8 lesions). Per-lesion CTC sensitivities for detecting the three other histological categories and each individual histological type were lower and are summarised in tables 2 and 3. CTC missed 13 of 57 advanced adenomas (size of the missed lesions, 10–15 mm), 25 of 73 non-advanced adenomas (size of the missed lesions, 6–9 mm), and eight of 14 non-neoplastic lesions (6 hyperplastic and 2 inflammatory lesions; size of the missed lesions, 6–9 mm). The probable causes for missing these were flat morphology (n=9, including 7 retrospectively unidentifiable and 2 identifiable lesions); size discrepancy between CTC and the reference standards (ie, easily identifiable in retrospect but ≤5 mm on CTC and ≥6 mm on the reference standards; n=7); obscuration by faecal matter and/or suboptimal distention (n=6) and by respiration (n=1); simple perception error (n=5); and no explicit causes (ie, retrospectively unidentifiable without obvious reasons) (n=18). Of 157 lesions reported at CTC, 51 were false-positive lesions, resulting in per-lesion PPV of 67.5% (106/157) for all histological types of lesion. Forty-six false-positive lesions were 6–9 mm, three were 10 mm, and one each was 11 mm and 17 mm. Twelve false-positive lesions were due to size discrepancy between CTC and the reference standards, that is a matched lesion was present but ≤5 mm on the reference standards.
Analysis of CTC criteria to suggest proximal synchronous cancer
The per-lesion sensitivity and PPV of CTC for specifically diagnosing synchronous proximal cancer varied according to different CTC imaging criteria. A lesion size criterion of ≥15 mm on CTC (figure 2) yielded 87.5% (7/8 lesions) sensitivity with moderate PPVs of 70% (7/10 lesions). A combined criterion of lesion size ≥15 mm and presence of uneven lesion surface (figure 2), pericolic abnormality or pericolic lymphadenopathy, resulted in 87.5% (7/8 lesions) sensitivity and 77.8% (7/9 lesions) PPV (table 4).
We found that CTC had a 100% per-patient and per-lesion sensitivity for detecting synchronous cancers in the colon proximal to stenosing cancer, and a negative CTC finding in the proximal colon of 100% NPV for synchronous cancer. This result was consistent with previous small studies.5–9 In accordance with the markedly high sensitivity of CTC for detecting proximal synchronous cancers shown in our current study as well as in the previous studies, a recent meta-analysis also showed the high sensitivity of CTC for detecting colorectal cancers across a broad spectrum of indications, protocols and techniques (overall sensitivity of 96.1%; 398/414 cancers) although the study did not directly address detection of synchronous cancers proximal to a stenosing colorectal cancer.25
CTC sensitivities for lesions other than cancer in the proximal colon shown in our study were slightly lower than those reported for screening CTC.11–14 The lower sensitivity was likely due to technical difficulties associated with stenosing cancer, including difficulties in optimising proximal colonic examination (ie, 3.6% rate of technical inadequacy compared with 1.1% (12 of 1110 patients) reported in a screening CTC programme26), the inability to fully utilise both two- and three-dimensional CTC review techniques in the presence of a large amount of fluid/faeces, and the frail nature of the patients. However, missing some non-cancerous proximal lesions on CTC is unlikely to be a substantial immediate clinical concern in patients with stenosing colorectal cancer as diagnosing them would not affect the immediate treatment plan. Moreover, the CTC sensitivity for advanced neoplasia was still fairly high (per-patient sensitivity of 88.6%), contributing to the considerably high 97.4% NPV. Therefore, a negative CTC finding in the proximal colon, which is expected to account for the majority of patients with stenosing colorectal cancer (68.3% in our study population), may virtually exclude not only the need for surgical resection of the proximal colon but also the need for early postsurgical colonoscopic polypectomy for the proximal colon.
Compared with the excellent detection of proximal synchronous cancers by CTC, correct diagnosis of the proximal cancers detected on CTC as cancers, while minimising incorrect characterisation of CTC-detected non-cancerous lesions, was less straightforward. Lesion size ≥10 mm on CTC as the criterion for diagnosing cancers in the proximal colon had misdiagnosed one of the eight proximal cancers and yet only yielded 17.5% per-lesion PPV. Use of more strict criteria for the diagnosis of proximal cancer improved the PPV but reciprocally decreased CTC sensitivity for diagnosing proximal synchronous cancers. While colonoscopy can analyse lesion surface pit patterns at high-magnification and can directly assess lifting patterns after submucosal injection for lesion characterisation,27–30 CTC parameters for lesion characterisation are more restricted, particularly for less advanced intramural colonic lesions. Among the CTC findings analysed in our study, we found that a lesion size ≥15 mm on CTC might be used as a practical, albeit limited, criterion to suggest proximal synchronous cancers, since this was the most strict size criterion that did not mischaracterise any proximal cancers except for one 8-mm submucosal cancer and had a moderate PPV of 70% (7/10 lesions) for the cancer diagnosis. The presence in our patients of an 8-mm submucosal cancer and the occasional occurrence of tiny cancers19–22 suggest that this size-based approach carries a statistically minuscule but potential risk to ignore small synchronous cancers that need to be surgically removed. However, lowering the size threshold in order not to misdiagnose such small cancers would not be appropriate as it would rapidly decrease PPV for the cancer diagnosis substantially by increasing mischaracterisation of CTC-detected benign lesions as well as false-positive lesions because most CTC false-positive lesions were small. This would generate an unacceptably high rate of unnecessarily extensive surgery with associated increased risks and morbidity. Alternatively, considering the potential risk of neglecting small cancers on CTC, early colonoscopy after resection of the stenosing cancer should be performed in patients who had small synchronous proximal lesions detected on preoperative CTC but unremoved by surgery.
Uneven lesion surface did not specifically suggest malignancy but its absence in a lesion ≥10 mm on CTC could exclude cancer, mildly helping lesion characterisation. Pericolic infiltration/nodular advancing margin was indicative of cancer with PPV of 100%. However, the finding is limited in that it basically distinguishes T3/4 cancers from T1/2 lesions with the reported sensitivity and specificity of 42–100% and 75–95%, respectively,8 ,31 ,32 and is observed in already fairly large lesions. Therefore, the finding may not add much to lesion size-based characterisation but may be used to reconfirm cancer. Similarly, pericolic lymphadenopathy was specific for diagnosing cancer with PPV of 100%, but was not sensitive. Consequently, the ability of CTC to specifically diagnose synchronous cancer in the colon proximal to stenosing cancer was only slightly improved by combining other CTC findings with lesion size.
This study had limitations. First, not all patients who underwent CTC were assessed by the reference standard methods. Nevertheless, in the majority of patients not assessed by the reference methods, this was due to medical reasons, which rendered further proximal colonic evaluation unbeneficial, and would have been excluded even in a prospective study. Therefore, in essence, most of the patients who were eligible for further proximal colonic evaluation using the reference standard procedures in both CTC-positive and CTC-negative groups, had the reference standards allowing for adequate assessment. However, considering that the 41 patients who were eligible but did not undergo the reference standard procedures had negative or only mildly positive CTC findings, excluding them from the analysis of CTC diagnostic performance may have slightly overestimated the sensitivity of CTC (ie, partial verification bias33). Second, there was a long time interval between CTC and the reference methods in some patients, predominately those with negative CTC findings. However, the time interval was within 21 days in those who had proximal synchronous cancer which was the primary concern. If anything, the prolonged interval would presumably slightly underestimate the sensitivity and NPV of CTC for non-cancerous lesions, since some ‘missed’ lesions could have conceivably developed or increased in size since the time of CTC. Third, despite the large size of the study cohort, the number of proximal synchronous cancers was small. However, the number was still much larger than that in previous studies. The rate of synchronous cancers in our study population was approximately the average of that in previous reports1 ,2 and the number of proximal synchronous cancers of our study represents their natural prevalence. Given the low prevalence of proximal synchronous cancers, a future multi-institutional prospective study to define a robust guideline for management of patients with stenosing colorectal cancer according to CTC findings to maximise patient benefits while minimising adverse results would be needed.
In conclusion, CTC is highly sensitive in detecting synchronous cancers in the colon proximal to a stenosing colorectal cancer, enabling a negative CTC result to confidently exclude proximal synchronous cancers. However, the limited capability of CTC for characterising the lesions detected on CTC makes the actual CTC sensitivity to specifically diagnose proximal cancers mildly compromised in order to maintain PPV for the cancer diagnosis at a reasonable level and may necessitate early postsurgical colonoscopy in patients who had small synchronous proximal lesions detected on preoperative CTC but unremoved by surgery.
SHP and JHL are co-first authors.
Competing interests None.
Ethics approval Ethics approval was provided by the IRB of Asan Medical Center.
Provenance and peer review Not commissioned; externally peer reviewed.