The utility of flexible sigmoidoscopy after a computerized tomographic colonography revealing only rectosigmoid lesions

Authors


Dr P. E. Young, Division of Gastroenterology, 8901 Wisconsin Avenue, Bldg 9, Bethesda, MD 20889, USA.
E-mail: peyoungmd@hotmail.com

Summary

Background  Identifying polyps by computerized tomographic colonography typically prompts colonoscopy, increasing its cost, risk and inconvenience. Many polyps are confined to the rectosigmoid and theoretically amenable to resection via flexible sigmoidoscopy.

Aim  To determine the prevalence of advanced proximal colonic neoplasia when computerized tomographic colonography reveals only rectosigmoid polyps, and characterize the yield of polypectomy via flexible sigmoidoscopy in such patients.

Methods  Subjects underwent computerized tomographic colonography and colonoscopy with segmental unblinding. Patients with only rectosigmoid findings by computerized tomographic colonography were identified retrospectively. Flexible sigmoidoscopy findings were estimated by including lesions distal to the descending/sigmoid colon junction during colonoscopy. Proximal lesions were also reviewed. Advanced lesions were defined as: adenocarcinoma, tubular adenoma >1 cm, ≥3 tubular adenomas, tubulovillous histology or high-grade dysplasia.

Results  By computerized tomographic colonography, 15% (203 of 1372) had only rectosigmoid polyps. Concomitant lesions in the proximal colon were seen in 32% (64 of 203) during colonoscopy. Advanced proximal neoplasia occurred in 2% (three of 203) with only rectosigmoid polyps on computerized tomographic colonography.

Conclusions  Using flexible sigmoidoscopy to follow-up computerized tomographic colonography demonstrating only rectosigmoid polyps would eliminate 15% of subsequent colonoscopies. This strategy carries a small risk of missed proximal advanced neoplasia. This miss rate appears comparable to that of colonoscopy alone. Further study on the cost-effectiveness of this approach is warranted.

Introduction

While colorectal cancer (CRC) is a prototypical disease amenable to primary prevention through screening, it remains the second leading cause of cancer death in the United States with an incidence of nearly 150 000 new cases and 56 000 deaths per year.1 Barriers to adequate screening are numerous and include both systematic and patient-based issues. Systematic barriers include a lack of universal insurance coverage in the US, the costs associated with the various CRC screening tests in nations with government sponsored healthcare systems, and the limited availability of same-sex endoscopists for women.2 Patient-based barriers include, but are not limited to, fear of cancer detection, fear of discomfort, avoidance of the unpleasant preparation that is involved in many CRC screening tests and a lack of knowledge regarding the need for screening.3 Some of these barriers have been overcome by advances in technology such as computerized tomographic colonography (CTC).4 First described in 1994, CTC has rapidly evolved and is being used in some locations as a primary CRC screening tool.5, 6 Since CTC’s inception, numerous developments including the use of multidetector scanners, three-dimensional ‘fly-through’ views, digital subtraction of contrast-labelled colonic debris and increased interpreter education have increased the accuracy of the examination. CTC holds the promise of providing another method of extending CRC screening to a larger population.7

A question that has arisen in centres where CTC is used for primary CRC screening is how to manage patients in whom CTC reveals only rectosigmoid (RS) findings. At present, most endoscopists faced with this scenario advocate a complete colonoscopy to remove any RS polyps and to re-examine the colon for lesions which may have been missed on CTC. This approach is similar to that taken with a flexible sigmoidoscopy (FS) in which adenomatous lesions are identified.

Findings from multiple studies support the contention that simply screening the left colon is inadequate to predict the risk for proximal neoplasia.8–11 One important limitation of these studies, however, is that the initial screening examination, sigmoidoscopy, did not permit proximal colonic examination. CTC, in contrast, examines the entire colon and thus it could be argued that FS with polypectomy would constitute sufficient screening in patients in whom no proximal polyps are seen on CTC. FS can, on average, reach an insertion depth of 52 cm in women and 55 cm in men.12 This would place the tip of the examining endoscope at or above the junction of the descending and sigmoid colon in the vast majority of patients.13 FS has several advantages over colonoscopy in this setting. FS requires less time than colonoscopy in terms of preparation and performance, is not associated with limitations on driving and working, and avoids the potential risks of chemical sedation.14, 15 The colonic perforation rate during FS is also substantially less than that during colonoscopy. As per the best available data, perforation occurs in one per 50 000 cases with FS, whereas the rate for colonoscopy is closer to one per 3000 for gastroenterologists and one per 1500 for general surgeons.16, 17

One crucial piece of knowledge that needs to be obtained before using FS for polypectomy for patients with isolated RS findings on CTC is the determination of the risk of missing proximal colonic neoplasia. Our aim in the proposed study was to determine, in patients in whom CTC has revealed only RS polyps, the prevalence of advanced and non-advanced proximal colonic lesions overlooked by CTC.

Methods

The study is a retrospective analysis of data collected as part of a prospective trial comparing the diagnostic accuracy of CTC to colonoscopy for the detection of colorectal polyps. Both investigations were approved by our institutional review board. All study subjects were between the ages of 50 and 79 years who were at average risk for colorectal neoplasia and had no symptoms concerning for organic gastrointestinal disease. All patients underwent both CTC and colonoscopy on the same day. The specifics of the prospective trial have been previously published.18 In brief, all participants underwent a colonoscopy by an experienced colonoscopist on the same day as their CTC. All detected polyps were measured with a calibrated linear probe prior to removal. The endoscopists were initially blinded to the CTC results and segmental unblinding with respect to the CTC findings was performed during the colonoscopy. Because of the retrospective nature of the study, limited demographic data were available including gender and age but not race.

All subjects whose CTC revealed only RS lesions were included in our analysis. We excluded all subjects for whom CTC was normal or revealed polyps proximal to the sigmoid colon. For the included subjects, we examined (i) the colonoscopic findings for each subject and (ii) the specific histological pathology for all resected polyps.

As our question involved the reliability of FS in detecting polyps in patients with only RS findings at CTC, we needed a reliable surrogate for FS. We chose the findings at colonoscopy up to and including the sigmoid colon–descending colon junction as a reasonable anatomic delineation. The location of this anatomic region was determined by the endoscopist using a standard colonoscope. Several studies have shown that the average endoscopist is able to examine reliably the colon up to the sigmoid–descending colon junction with FS.14, 15 We defined the number of polyps that would be missed in the colon proximal to the descending sigmoid junction by both CTC and FS as our primary end point. Secondly, we examined the histology of the ‘missed’ lesions. Polyps were defined as advanced if they met any of the following criteria: adenocarcinoma, tubular adenoma ≥1 cm, more than three tubular adenomas, tubulovillous histology or high-grade dysplasia. All other findings including small tubular adenomas, hyperplastic polyps and normal mucosa were deemed non-advanced.

The justification for this schema is as follows: both normal mucosa and sporadic hyperplastic polyps are thought to have negligible malignant potential and thus there is little clinical importance to not detecting them with either CTC or FS.19 Both tubulovillous adenomas and large or multiple (three or more) tubular adenomas have increased malignant potential if not detected and removed, and are predictive of more advanced subsequent colonic neoplasia.20 Missing a frank adenocarcinoma would probably have significant clinical consequences.

The following is a representative example of our data extraction process: subject A1 underwent CTC which suggested a single 1-cm rectal polyp. Examination of colonoscopic findings for subject A1 revealed that a 1-cm tubular adenoma was identified and removed in the sigmoid as well as a 6 mm hyperplastic polyp in the caecum. Thus, subject A1 was noted to have a lesion by colonoscopy that was not detected by CTC and would have been missed had he/she undergone only an FS in response to the CTC findings. The lesion is deemed to be of low significance based on our prespecified criteria.

Sample size and data analysis

The aim of the proposed study was to determine, in patients in whom CTC has revealed only RS polyps, the prevalence of colonic neoplasia (advanced and non-advanced) discovered via follow-up colonoscopy that would not have been discovered by follow-up FS alone. The current literature suggests that the rate of ‘missed’ colonic polyps with colonoscopy is between 6% and 11%.21, 22 It is our hypothesis that CTC and FS will be able to detect distal colonic polyps ≥ 6 mm with similar accuracy. We wanted to estimate the true proportion to within 5% with a 95% confidence interval. On the basis of these parameters and using the formula provided below, a sample size of 200 was deemed sufficient.

image

The rate will be reported descriptively with a 95% confidence interval. The prevalence of proximal neoplasia was calculated as a point prevalence as follows:

image

Secondary analysis of the histological nature of each polyp was extracted from the database and is reported descriptively. All analyses are on a per-patient basis to reflect clinical relevance.

Results

A total of 1372 patients underwent both CTC and colonoscopy. The mean age of the patients was 57.8 years and was not statistically different between those who did or did not have right-sided lesions. Forty-one per cent of those studied were women. A description of CTC and colonoscopy findings is illustrated in Figure 1.

Figure 1.

 Breakdown of computerized tomographic colonography patients.

Two hundred and three patients (15%) had only RS polyps on CTC. The distribution of males and females was similar to the group at large (Table 1). All 203 patients with RS polyps seen on CTC underwent colonoscopy and 37% (76 of 203) were confirmed to have only RS polyps. Of these lesions, 50% were either hyperplastic polyps or normal colonic mucosa, 40% were diminutive adenomas, and 11% were advanced adenomas (Table 2). Sixty-three patients with RS only findings on CTC, 31% (63 of 203), had no polyps found on colonoscopy and were thus presumably false positives. Colonoscopy detected polyps proximal to the descending sigmoid junction that were not seen on CTC in 32% (64 of 203). The histology of these polyps is shown in Table 3. Among the 64 patients with proximal polyps on colonoscopy, 32 (50%) had hyperplastic polyps, 29 (45%) had non-advanced adenomas and three (5%) had advanced adenomas. The average size of the non-advanced lesions was 4 mm, whereas the advanced lesions averaged 11 mm. Thus, the rate of missing advanced proximal adenomas by utilizing FS after CTC revealed only RS polyps was 2% (three of 203).

Table 1.   Demographics
 Only rectosigmoid VC findings (= 203)Proximal VC findingsNO polyps on VC (= 716)Total number (= 1372)
  1. Average size of right-sided lesions missed by VC: 4.0 mm.

  2. Average size of right-sided advanced lesions missed by VC: 11.1 mm.

Age58.958.457.257.8
Gender
 Female84173300557
 Male119280416815
    1372
Table 2.   OC and VC findings by location
 Only rectosigmoid VC findingsProximal VC findingsNo VC findingsTotal number of patients
Corresponding polyps found on OC761820258
 Advanced adenomas842050
 Diminutive adenomas30930123
 Benign polyps3847085
No corresponding polyps found on OC63214448725
 Advanced adenomas0909
 Diminutive adenomas031031
 Benign polyps028028
 Other63146448657
Polyps found in other segment only on OC1521268304
 Advanced adenomas1034
 Diminutive adenomas98113130
 Benign polyps513152170
Corresponding polyps and other segment polyps found on OC4936085
 Advanced adenomas29011
 Diminutive adenomas2022042
 Benign polyps275032
Total2034537161372
Table 3.   Missed VC polyps (by worst histology)
 Distal polyps onlyProximal polypsNo polypsTotal
Hyperplastic (mm)3246152230
 ≤52433132189
 6–9581528
 ≥1035513
Tubular adenoma (mm)3072114216
 ≤5203596151
 6–99261752
 ≥10111113
Tubulovillous adenoma (mm)0527
 ≤50000
 6–90000
 ≥100527
Villous adenoma (mm)0000
 ≤50000
 6–90000
 ≥100000
Carcinoma (mm)2204
 ≤50000
 6–90000
 ≥102204
All or no corresponding polyps found139328448915
Total2034537161372

Discussion

Colorectal cancer remains the second leading cause of cancer-related mortality in the United States. Unfortunately, the current ability to provide endoscopic screening for primary CRC prevention is insufficient to screen the eligible population. CTC represents an additional tool to provide total colonic evaluation. Recently, CTC standards for integration into gastroenterological practice were published by the American Gastroenterological Association.23 According to these standards, CTC is recommended for evaluation of the colon proximal to an obstruction and after failed colonoscopy. It may also be considered in patients ‘unwilling to undergo other primary screening modalities’ as well, but is not endorsed as a primary screening modality. These standards were published prior to the release of the American College of Radiology Imaging Network (ACRIN) 6664 trial results. The ACRIN trial is a multicentre trial of 2531 patients comparing CTC to colonoscopy and represents the largest study if CTC as a primary screening modality to date. Preliminary results of this study demonstrated that CTC is highly accurate for the detection of intermediate and large polyps with an accuracy similar to colonoscopy.24 One factor which may account for the favourable performance of CTC in this trial is that all the participating radiologists in this trial underwent a rigorous training process prior to starting the study. As imaging and processing technology improves and becomes more readily available, understanding the optimal management of CTC findings will be paramount for maximizing the potential benefit of this modality. Our study addresses the management of a specific group of patients undergoing CTC, those with only RS findings by CTC.

For those patients who undergo FS without further right-sided colonic evaluation, it is clear that the presence or absence of lesions in the RS is a poor predictor of more proximal findings.10, 11, 25 Based on recently published data from the PLCO trial, the yield in adenoma detection for distal polyps (5–9 mm) detected at FS was 15% for women and 16% for men [number needed to screen (NNS) of 7 and 6 respectively].26 The incremental change in adding right-sided evaluation via colonoscopy was low, with the NNS being lowered to 6 for women and 5 for men. In patients who have had total colonic evaluation with CTC, it would be appealing to tailor our endoscopic approach towards the anatomic distribution of polyps identified via CTC as the information regarding the proximal colon is included in the baseline examination. Colonoscopy remains the preferred method for reducing the risk of CRC development given its ability both to detect and to remove colonic neoplasia. Even in expert centres, however, the miss rate for advanced colonic neoplasia has been reported to be 6% with colonscopy.22 The results of our study reveal that the potential miss rate (prevalence) of advanced proximal colonic neoplasia is 2% for patients with isolated RS polyps identified via CTC that is followed up with FS alone. This miss rate appears to be superior to what has been observed with the currently accepted gold standard, colonoscopy.

Based largely on cohort data, colonoscopy has been held up as the gold standard for CRC screening and prevention.27 Certainly when performed to the standards outlined in the combined ACG/ASGE guidelines, colonoscopy probably represents the most effective CRC screening method currently available.28 Recently, however, the appropriateness of this standard as applied in general clinical practice has been called into question.29 A number of variables have been shown to influence colonoscopy completion rates and thus the yield of screening colonoscopy. Patient factors include female gender, age, the presence of diverticuli, difficult anatomy, inadequate preparation and patient discomfort.30, 31 Other factors which decrease caecal intubation rates are procedures performed in the afternoon, and those performed in private offices.32, 33 Endoscopist features, particularly annual colonoscopic volume of over 200, also appear to enhance caecal intubation rates.34

The second key to optimal CRC screening is polyp detection. A recent systematic review of tandem colonoscopy studies suggests that colonoscopy misses 5–10 mm adenomatous polyps in approximately 13% of cases.35 A recent population-based study found a 4% miss rate for adenocarcinoma, although this study included only right-sided lesions.36 Several factors positively influence polyp detection. These include bowel preparation performed on the day of the procedure, a withdrawal time of >6 min, and perhaps technological enhancements such as narrow band imaging or chromoendoscopy.37, 38 Other potential methods of enhancing polyp detection such as the use of a 170° colonoscope or performing colonoscopy in retroflexion, have not proven efficacious.39, 40

A variety of modelling studies have evaluated the cost-effectiveness of colonoscopy as a cancer screening modality.41 The cancer miss rate for colonoscopy used in these models ranged from 3% to 5%, whereas the miss rate for advanced polyps was 5–15%. In our analysis, the miss rate for proximal advanced neoplasia achieved with a combination of CTC followed by FS when only RS findings are noted was lower than these estimates. Our results indicate that this management approach is superior to the observed miss rates of proximal advanced neoplasia for FS in combination with FOBT, which is considered an acceptable screening modality by multiple professional societies and is included as an option in the current consortium guideline.42 A recently published article by Kim and colleagues demonstrated a similar rate of detection for advanced adenomas whether CTC or colonoscopy was used as the primary screening modality.43

The calculated miss rate for non-advanced adenomas in our study was somewhat higher at 14%. The practical implications of this are less clear than for polyps with more advanced features.44 It is possible that small adenomas without high-risk features such as villous histology grow sufficiently slowly to warrant a longer surveillance interval. Previous studies have suggested that lesions <1 cm have a slow rate of growth, and in some cases may actually regress, over at least a 3-year period.45, 46 A more recent study showed no evidence of polyp regression in polyps <1 cm and a variable growth rate which was not entirely predicated on polyp histology.47 All these investigations were limited by small sample size – <100 patients in each study. We hope, as the number of patients who have undergone ‘watchful waiting’ for diminutive colon polyps with CTC surveillance increases, we will have a better idea of the natural history of these lesions.

Our present study has several limitations. First, CTC is not yet widely available. Secondly, even in those centres where CTC is performed, the sensitivity varies widely.9 At least a part of this variation may be explained by differences in collimation, type of scanner, imaging mode and interpreter experience, all of which are modifiable parameters.48 Thirdly, the retrospective nature of this study precludes several analyses, which might prove worthwhile. For example, the lack of certain demographic data prevents honing our conclusions further as they may apply to specific populations. Nonetheless, this analysis provides additional data as we attempt to optimize our strategies for effective and efficient CRC detection and prevention.

Acknowledgement

Declaration of personal and funding interests: None.

Disclaimer

The opinions and assertions contained herein are the sole views of the authors and should not be construed as official or as representing the views of the US Navy, Department of Defense or Department of Veterans Affairs.