Screening for colorectal cancer (CRC) has proven effectiveness. Colorectal Cancers Detected After Colonoscopy Frequently Result From Missed Lesions

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8: Colorectal Cancers Detected After Colonoscopy Frequently Result From Missed Lesions HEIKO POHL*,,, and DOUGLAS J. ROBERTSON*,, *Outcomes Group and Department of Gastroenterology, VA Medical Center, White River Junction, Vermont; Dartmouth Medical School, Hanover, New Hampshire; and Department of Gastroenterology, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire See related article, Singh H et al, on page 1128 in Gastroenterology. BACKGROUND & AIMS: Colorectal cancers (CRCs) that are detected in patients who have received colonoscopies (interval cancers) arise from missed lesions, incomplete resections of adenomas, or de novo. We estimated rates of interval cancer from missed lesions. METHODS: Adenoma miss rates, cancer prevalence among patients with adenoma (based on size), and rates of adenoma-to-cancer transitions were estimated from the literature. We used a model to apply these risk estimates to a hypothetical average-risk population that received screening colonoscopies. We calculated the proportion of individuals with tumors missed at the baseline colonoscopy and tumors that arose from missed adenomas during a 5-year follow-up period. Sensitivity analyses were performed to assess robustness. RESULTS: We found that 0.7 per 1000 persons undergoing a screening colonoscopy had a cancer that was missed at the baseline colonoscopy and an additional 1.1 per 1000 subsequently developed cancer from a missed adenoma. Therefore, the expected rate of individuals with CRC, based on missed adenomas, was 1.8 per 1000 persons within 5 years. By using the most conservative assumptions a low miss rate and low prevalence of cancer in adenomas 0.5 per 1000 persons would have a detectable CRC within 5 years after a screening colonoscopy. In contrast, using the highest reported miss rates and cancer prevalence, CRCs from missed lesions would occur in 3.5 per 1000 screened persons. CONCLUSIONS: A significant number of patients undergoing a screening colonoscopy that did not detect cancer actually have a malignant lesion or adenoma that could progress in a short interval. Most interval cancers might reflect missed rather than new lesions. Improving adenoma detection could reduce the rate of interval cancers. Keywords: Colonoscopy; Colorectal Cancer Screening; Miss Rate; Interval Cancer. Screening for colorectal cancer (CRC) has proven effectiveness 1,2 and is widely recommended by organizations such as the American Cancer Society 3 and the United States Preventive Services Task Force. 4 Although these 2 organizations recommend a panel of CRC screening options, colonoscopy is believed to be the most effective screening modality because it allows both the detection of early stage cancer and removal of precancerous adenomatous polyps across the entire length of the bowel before cancer development. Despite these unique operating characteristics, the occurrence of cancer after a recent colonoscopy that did not detect one, also referred to as interval cancer, 5 is well described across a number of different study designs. Observational studies using administrative data found that between 5% and 9% of those diagnosed with CRC underwent colonoscopy within the prior 6 to 36 months, suggesting that these were missed cancers. 6,7 Clinical trials, mostly in adenoma-bearing populations that underwent complete clearing colonoscopy, report CRC rates between 3 and 9 per 1000 persons within 3 to 6 years after a colonoscopy Explanations for the occurrence of these cancers shortly after a colonoscopy include missed lesions (eg, large adenomas and cancer), incomplete resection of adenomas, and de novo cancer development. 8 Several groups have used algorithms to classify interval cancers based on the finding on the colonoscopy before the one that detected cancer. 8,13,14 By using this methodology, it has been estimated that at least 50% of interval cancers are owing to a missed lesion at the index colonoscopy. However, this approach has significant limitations including the need to make fixed rules (eg, based on the time since last colonoscopy) to determine which cancers are new and which are related to missed lesions. Such approaches also do not distinguish between missed adenomas that progress to cancer and missed cancers at baseline, simply grouping the two together under missed lesions. We hypothesized that an estimate of the rate of expected interval cancers could be derived from factors such as adenoma miss rates, the risk of cancer present in adenomas, and the annual risk of transition from adenoma to cancer. The primary aim of the study was to calculate the expected rate of interval cancers related to missed lesions in a screening population and, secondarily, to assess the rate of interval cancers in an adenoma-bearing population. Methods Study Population For the base-case analysis we analyzed a hypothetical cohort of 1000 average-risk adults undergoing a screening colonoscopy. In a secondary analysis we evaluated the risk of 1000 individuals who had at least one adenoma at a screening colonoscopy. We developed a mathematical model using Excel (Microsoft Office 2008; Microsoft Corporation, Redmond, WA) to evaluate both the expected rate of cancers missed at index Abbreviation used in this paper: CRC, colorectal cancer by the AGA Institute /$36.00 doi: /j.cgh

2 October 2010 MISSED LESIONS AND INTERVAL COLORECTAL CANCER 859 Table 1. Probabilities Applied in the Base-Case Analysis and Range Used in a One-Way Sensitivity Analysis Base case Range Reference Adenomas per person screened, mean ,23 Adenomas per person with any adenoma, mean ,16,18 20,23 Distribution of adenoma by size Diminutive (1 5 mm) 60% 50% 70% Small (6 9 mm) 26% 22% 30% Large ( 10 mm) 14% 8% 20% Adenoma miss rate 22,24,28 Diminutive (1 5 mm) 26% 20% 30% Small (6 9 mm) 13% 10% 15% Large ( 10 mm) 6% 2% 12% Cancer prevalence in adenoma, by size 23,26,29 31 Diminutive (1 5 mm) 0.1% 0% 0.1% Small (6 9 mm) 0.3% 0% 0.5% Large ( 10 mm) 7.0% 4% 8.7% Annual transition rate from adenoma to cancer by size 32 34,36,38,41 From diminutive to small From small to large From large to cancer colonoscopy and the rate of incident cancers that developed from missed adenomatous polyps during 5 years of follow-up evaluation. Probabilities Probability estimates were extracted from the literature. Base-case values and ranges used in sensitivity analyses are summarized in Table 1. Prevalence data. We estimated adenoma prevalence in the screening cohort based on the mean number of adenomas per person in the screening cohort. Autopsy studies on patients representing a screening cohort reported a mean number of 0.8 to 1.2 adenomas per person This estimate is supported by colonoscopy studies that reported a range of adenoma detection by endoscopists. Assuming that high adenoma detectors have a very low miss rate, the true adenoma prevalence should be similar to the upper range of reported adenoma detection rates of between 0.5 to With respect to polyp size distribution, literature estimates for large ( 10 mm) adenomas vary widely between 8% and 25%, with more recent studies reporting lower rates relative to older studies. 25,26 Our base-case estimates are congruent with the polyp distribution chosen in recent microsimulation models on CRC screening strategies. 27 We applied the range in size distribution in the sensitivity analysis (Table 1). Miss rate. Adenoma miss rates are derived primarily from tandem colonoscopy studies or computer tomography colonography studies that used segmental unblinding. 22,28 Miss rates from tandem colonoscopy studies have been summarized in a recent meta-analysis. 24 Although most studies used 3 size categories (diminutive, small, and large adenomas), the cut-off values differed. Specifically, the small size category in various studies included 5- to 10-mm polyps, to 9-mm polyps, 29 or 6- to 9-mm polyps. 23,30 For this analysis we used the following size categories: diminutive adenomas (1 5 mm), small adenomas (6 9 mm), and large adenomas ( 10 mm) and assigned reported data accordingly. Cancer prevalence in adenomas. Several prospective clinical colonoscopy trials have determined the rate of cancer in adenomas based on size, 23,26,29,30 and these rates approximate estimates from autopsy series that have been summarized in a systematic review. 31 For the base rate we chose a prevalence of 0.1% for diminutive adenomas, 0.3% for small adenomas, and 7.0% for large adenomas. Transition rates. Although the precise rate of transition from adenoma to cancer is unknown, cost-effectiveness analyses have calculated a rate of transition based both on studies examining the natural progression of adenomas (eg, using serial barium enema) 37,38 and autopsy series. 16,39,40 Annual transition rates for advanced adenomas also have been estimated using data from a national colonoscopy registry based on the prevalence of advanced adenomas and the observed cancer incidence in the population. 41 Analysis The primary outcome of interest was the rate of interval cancers resulting from a missed lesion at baseline colonoscopy in an average-risk screening population. This rate is the sum of all individuals with cancer present but missed at index colonoscopy and cancer that progressed from missed adenomas during 5 years of follow-up evaluation. The analysis is summarized in Figure 1. We first calculated the expected number of adenomas based on the mean number of adenomas per screened person in a screening population of 1000 persons and computed the proportion of adenomas in each size category. We then estimated the probability (p) of missing a lesion and applied published data on the prevalence of cancer in adenomatous polyps in each size category to obtain the rate of missed cancer (p [miss-rate] p [cancer prevalence]). To assess the rate of CRCs arising from a missed adenoma during follow-up evaluation, we applied the published transition rates (p[trans]) to missed adenomas by size. The rate of missed adenomas that transition into the next size category or to cancer follows an exponential function: p [miss-rate] (1 EXP [ p (trans) time]). 42 We chose a cycle length of 1 year and followed the model through 5 cycles. Adenomas could only transition once into the next size category per year (from diminutive to small and from small to large). Regression in

3 860 POHL AND ROBERTSON CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 Figure 1. Model of the analysis to evaluate the risk of interval CRCs as a result of missed lesions in a hypothetical cohort of 1000 persons undergoing a screening colonoscopy. Shown are probabilities applied in the base-case analysis. adenoma size was not allowed. Similar to prior studies on natural progression of adenomas and cost-effectiveness studies, 31,34,36 adenomas could only transition to cancer once they were large. We calculated the rate of missed cancers and the rate of cancers that arose from missed lesions individually for each size category, which also provided an overall proportion of interval cancers in each size category. Our secondary outcome of interest was the expected rate of CRCs as a result of missed lesions in an adenoma-bearing population of 1000 patients. We repeated the earlier-described analysis using the prevalence of adenomas in this population. This information was extracted from studies that provided information on the number of adenomas per patient with any adenoma. 15,16,18 20,23 Our approach calculated the expected rate of cancer on a per adenoma basis. To simplify the model we only allowed that each cancer that resulted from a missed lesion would only occur in one person. Sensitivity Analysis To calculate a range around the base estimate we selected a combination of 2 variables that would affect the rate of interval cancers most: the miss rate and the cancer prevalence. We applied the extremes of reported ranges for both variables simultaneously to calculate the most conservative estimate (low range) and the least conservative estimate (upper range). We further varied all estimates over a broad range of plausible assumptions in a one-way sensitivity analysis. Results Expected Rate of Missed Colorectal Cancers at Baseline Colonoscopy in a Screening Population In the base-case analysis, at index colonoscopy a polyp containing cancer was missed in 0.7 per 1000 persons (Table 2). Seventy percent of missed cancers were attributed to missed lesions of 10 mm or larger (0.5 per 1000 persons). When applying the most conservative assumptions, a low miss rate (2% for adenomas 10 mm) and low cancer prevalence in adenomas (4% for adenomas 10 mm), the rate of missed cancers would decrease to 0.1 per 1000 persons. In that scenario all cancer cases would be owing to missed large adenomas. In contrast, applying the least conservative assumptions, a high miss rate (12% for adenomas 10 mm) and a high cancer prevalence (8.7% for adenomas 10 mm), would result in a missed cancer rate of 1.5 per 1000 persons, with 80% of cases (1.2 per 1000 persons) related to missed large lesions. Table 2. Persons With Missed Colorectal Cancers and Colorectal Cancers That Progressed From a Missed Adenoma Within 5 Years After a Screening Colonoscopy Missed CRCs per 1000 persons CRCs progressing from a missed adenoma per 1000 persons CRCs owing to a missed lesion per 1000 persons Screening population, base case (range) 0.7 ( ) 1.1 ( ) 1.8 ( ) Adenoma-bearing population, base case (range) 2.0 ( ) 3.2 ( ) 5.1 ( ) NOTE. Presented are the minimum and maximum estimates based on varying the adenoma miss rate and the prevalence of cancer in adenomas.

4 October 2010 MISSED LESIONS AND INTERVAL COLORECTAL CANCER 861 Table 3. One-Way Sensitivity Analysis for Key Variables Range Missed CRCs per 1000 persons CRCs that progressed from missed adenomas per 1000 persons CRCs owing to a missed lesion per 1000 persons Adenoma per person, mean Adenoma size distribution Diminutive, 50% 70% Small, 22% 30% Large, 8% 20% Adenoma miss rate Diminutive, 20% 30% Small, 10% 15% Large, 2% 12% Cancer prevalence in adenoma Diminutive, 0% 0.1% Small, 0% 0.5% Large, 4.0% 8.7% Annual transition rate Diminutive to small, 1% 4% Small to large, 1% 4% Large to cancer, 2% 10% NOTE. Presented are the minimum and maximum estimates of missed CRCs at the index screening colonoscopy and CRCs that progressed from missed adenomas within 5 years. Expected Rate of Colorectal Cancer Arising From Missed Adenomas in a Screening Population The rate of cancers that arose from missed adenomas during 5 years of follow-up evaluation was 1.1 per 1000 persons. Applying the most conservative assumptions this rate decreased to 0.4 per 1000 persons. By using the least conservative assumptions, the rate of incident cancers increased to 2.1 per 1000 persons. Again, missed large adenomas at baseline accounted for the majority of these cancers (85%; 0.9 per 1000 persons). For the most and the least conservative risk estimates, missed large adenomas accounted for 71% and 91% of cancers, respectively. Expected Rate of Interval Colorectal Cancer as a Result of Missed Lesions in a Screening Population At the end of the 5-year follow-up period, 1.8 per 1000 persons screened (approximately 1 in 600) would be expected to harbor a CRC related to a missed lesion. Even with the most conservative assumptions 0.5 per 1000 persons undergoing a screening colonoscopy would be predicted to harbor a CRC at the end of the observation period. In contrast, when applying the least conservative parameters, 3.5 per 1000 persons would be expected to harbor a cancer 5 years after a screening colonoscopy. Expected Rate of Interval Colorectal Cancers as a Result of a Missed Lesion in an Adenoma-Bearing Population When adapting the model for a cohort of entirely adenoma-bearing persons, the rate of missed cancers at colonoscopy was 2.0 in 1000 persons (Table 2). An additional 3.2 per 1000 persons were expected to progress to cancer from a missed adenoma during follow-up evaluation. The combined rate of individuals with interval CRC attributed to a missed lesion at index colonoscopy was 5.1 per 1000 persons. In other words, in the adenoma-bearing population approximately 1 in 200 persons were expected to harbor a cancer related to either a missed cancer or an adenoma that transitioned to cancer within 5 years of follow-up evaluation. The presented range ( per 1000) is based on applying the most and least conservative risk parameters (as described earlier). Sensitivity Analysis The main results of our analysis remained robust when varying individual probabilities over a broad range of plausible assumptions (Table 3). As expected, the rate of interval cancers was most affected by the adenoma miss rate. By using the minimum published miss rate, the rate of expected interval cancers as a result of missed lesions was about half when compared with the base case. However, within the tested range for the miss rate, the proportion of persons with interval cancers quadrupled. Varying the remaining variables had little effect on the minimum estimate but broadened the range toward a higher estimate. Discussion Our model determined the expected rate of interval CRC after a screening colonoscopy resulting either from a missed cancer at baseline or an adenoma that transitioned to cancer in a short interval. We used available information on adenoma miss rates and the prevalence of cancer in adenomas by size and made plausible assumptions about transition rates from adenoma to cancer during the follow-up period. We calculated that a total of 1.8 per 1000 persons undergoing a screening colonoscopy would harbor a CRC as a result of a missed lesion within 5 years. This rate more than doubles (5.1 per 1000 persons) if colonoscopy is applied to an entirely adenoma-bearing population. To put the results of our model in context, it would be important to compare our expected rate of cancer secondary to missed lesions with the observed rate of interval cancer in a screening and a surveillance population. The screening population probably most relevant to our estimates has been studied by Kahi et al. 10 In 715 average-risk persons 2 interval cancers

5 862 POHL AND ROBERTSON CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 10 were observed within 5 years after a screening colonoscopy (2.8 per 1000 persons). 10 Based on our work (base-case estimate, 1.8 per 1000 persons), more than half of cancers would be explained by missed lesions. The only other study reporting on interval cancers after a screening colonoscopy by Lieberman et al 9 followed up 3121 veterans after a screening colonoscopy and found 16 cancers within 5.5 years (5.1 per 1000 persons). Participants were predominantly male, of older age, and more likely smokers, therefore representing a higher than average risk population. When using the high range estimate of our model (3.5 per 1000 persons) more than half of these observed interval cancers might have been related to missed lesions. Several studies have reported on cancers detected after a colonoscopy in adenoma-bearing populations. 9,11 14,43 48 Studies with more than 90% colonoscopy follow-up report between 4.7 and 11.1 interval CRCs per 1000 patients within 3 to 6 years of follow-up evaluation. The largest study pooled 8 trials with 9167 subjects, 8 including several studies that have individually reported on interval cancers previously. 9,11,12,14,46 Cancer was detected in 58 patients (6.3 per 1000 patients) within 4 years of the baseline examination. A similar rate was seen in a Danish study by Jorgensen et al, 44 which reported 12 CRC cases in 1881 patients within 4 years after a colonoscopy and adenoma resection (6.4 per 1000 patients). When relating our calculated rate of expected interval cancers owing to a missed lesion in the adenoma-bearing population (5.1 per 1000 patients within 5 years) to the observed rate of interval cancers in the earlierdescribed studies, it could be suggested that between 70% and 80% of interval cancers might be attributable to either missed cancers or missed adenomas that transitioned to cancer in a short interval. This estimate is higher than the prior estimate of 50%, which was derived by applying an a priori algorithm to determine the cause of interval cancers. 8,14 Conversely, our results suggest that few cancers are caused by a fast-growing (de novo) tumor without a detectable lesion or are the result of an incompletely resected adenoma at a prior colonoscopy. Although our model suggests that many interval cancers are related to missed lesions it does not exclude the fact that some may be the result of fast growth (de novo cancers). Recently, studies have found a disproportionately high prevalence of microsatellite instability in interval cancers, 5 suggesting a different and perhaps more rapid pathway to cancer. Although some of these cancers may be fast growers without a detectable lesion at baseline, it is also possible that early microsatellite instability lesions have a different morphology that makes them harder to detect. In our analysis we attempted to understand the expected rate of interval cancers exclusively related to missed lesions detectable by endoscopy. We did not account for de novo cancers without a detectable lesion at prior colonoscopy, and did not distinguish between flat and polypoid lesions or tubular and serrated adenomas. Similarly, our analysis does not provide information on the contribution of incompletely resected adenomas to interval cancer. One of the more striking findings from our model is the marked difference in expected missed interval cancer rates when the population was changed from a screening population to an adenoma-bearing population. In fact, the literature generally supports a much lower rate of interval cancer in those with a normal baseline colonoscopy. In a prospectively designed study in the United States 49 no cancers were detected in 1256 individuals within 5 years after a normal colonoscopy, and in a cross-sectional study from Germany 50 no cancers were observed in 553 persons within 10 years of a prior normal colonoscopy. Our results would generally be consistent with this finding. We assumed that the adenoma prevalence in the screening cohort or the proportion of large adenomas would have a large effect on the risk for interval cancers. Although the rate of interval cancers varied within the plausible ranges for these factors, it is important to note that even when applying a low adenoma prevalence, in 1000 persons would harbor a CRC within 5 years after a screening colonoscopy. Strengths of our approach relative to prior algorithm-driven analysis, is that we did not rely on arbitrary rules to adjudicate cancer cause. Rather, we used available clinical data from multiple sources regarding the frequency of missed lesions and the likelihood that lesions of a certain size contain cancer. The quality of this data is generally good and comes from multiple independent sources. Therefore, our estimates of missed prevalent cancers are likely accurate. In contrast, a major limitation of our work was the limited data on the risk of transition from adenomas to cancer. Because most cancers in our hypothetical cohort progressed from missed large adenomas, the rate of incident cancers is affected primarily by transition rates from large adenomas to cancer. Given the importance of this transition rate to our model, we have estimated it conservatively using a transition rate (3%) that is in the lower range of reported transition rates ,36,41 It is important to note that our approach basing our estimates on adenoma size is similar to published cost-effectiveness analyses of CRC screening strategies. 34,36 These analyses calculated transition rates based on autopsy studies and studies on the history of adenomas that included primarily larger polyps of 10 mm or greater. 16,31,38 40 To further account for the potential variance in the transition rate we have performed sensitivity analyses and reported those accordingly. Our expected rate of cancers as a result of a missed lesion may seem high because we did not distinguish between symptomatic and asymptomatic cancers. Most interval cancers are of an early stage and found in asymptomatic subjects. 8 10,14 Therefore, it is possible that some of the predicted missed cancers in our population would never be clinically relevant (eg, the person died of some other cause). However, colonoscopy for screening or surveillance is recommended for healthy persons, and most of the missed lesions likely would become clinically significant at some point. Our report has implications for clinical practice and research. It highlights the importance of missed adenomas as a contributor to incident cancer. Based on our model and a comparison of our results with the literature described earlier, a majority of interval cancers appear to be the result of missed lesions. Reported adenoma miss rates vary and are dependent on factors related to the patient (eg, preparation), the examiner (experience, withdrawal time), and available technique. In fact, we now have good evidence that endoscopists with a low adenoma detection perform colonoscopies that are associated with a higher risk of interval cancers. 51 This is very consistent with our findings. In summary, our model provides an estimated rate of interval CRC owing to missed lesions at baseline colonoscopy. It suggests that most interval cancers are the result of a missed cancer at baseline colonoscopy or of a missed adenoma that progresses to cancer in a short interval. Measures to improve

6 October 2010 MISSED LESIONS AND INTERVAL COLORECTAL CANCER 863 adenoma detection are most likely to have the biggest impact on improving the effectiveness of colonoscopy. References 1. Hewitson P, Glasziou P, Irwig L, et al. Screening for colorectal cancer using the faecal occult blood test, Hemoccult. Cochrane Database Syst Rev 2007;1:CD Atkin WS, Edwards R, Kralj-Hans I, et al. Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010;375: Levin B, Lieberman DA, McFarland B, et al. Screening and surveillance for the early detection of colorectal cancer and adenomatous polyps, 2008: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology. CA Cancer J Clin 2008; 58: US Preventive Services Task Force. Screening for colorectal cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 2008;149: Sawhney MS, Farrar WD, Gudiseva S, et al. Microsatellite instability in interval colon cancers. Gastroenterology 2006;131: Baxter NN, Sutradhar R, Forbes S, et al. Can quality indicators for colonoscopy be derived from administrative data? Identifying predictors of new or missed cancers after colonoscopy. Gastroenterology 2009;136:A Gupta R, Steinbach M, Ballman KV, et al. Colorectal cancer despite colonoscopy: critical is the endoscopist, not the withdrawal time. Gastroenterology 2009;136:A Robertson DJ, Lieberman DA, Winawer SJ, et al. Interval cancer after total colonoscopy: results from a pooled analysis of eight studies. Gastroenterology 2008;134:A Lieberman DA, Weiss DG, Harford WV, et al. Five-year colon surveillance after screening colonoscopy. Gastroenterology 2007;133: Kahi CJ, Imperiale TF, Juliar BE, et al. Effect of screening colonoscopy on colorectal cancer incidence and mortality. Clin Gastroenterol Hepatol 2009;7: , quiz Alberts DS, Martinez ME, Roe DJ, et al. Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas. Phoenix Colon Cancer Prevention Physicians Network. N Engl J Med 2000;342: Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup. N Engl J Med 1993;329: Leung K, Pinsky P, Laiyemo AO, et al. Ongoing colorectal cancer risk despite surveillance colonoscopy: the Polyp Prevention Trial Continued Follow-up Study. Gastrointest Endosc 2010;71: Pabby A, Schoen RE, Weissfeld JL, et al. Analysis of colorectal cancer occurrence during surveillance colonoscopy in the dietary Polyp Prevention Trial. Gastrointest Endosc 2005;61: Eide TJ, Stalsberg H. Polyps of the large intestine in Northern Norway. Cancer 1978;42: Williams AR, Balasooriya BA, Day DW. Polyps and cancer of the large bowel: a necropsy study in Liverpool. Gut 1982;23: Rickert RR, Auerbach O, Garfinkel L, et al. Adenomatous lesions of the large bowel: an autopsy survey. Cancer 1979;43: Rex DK, Helbig CC. High yields of small and flat adenomas with high-definition colonoscopes using either white light or narrow band imaging. Gastroenterology 2007;133: Barclay RL, Vicari JJ, Greenlaw RL. Effect of a time-dependent colonoscopic withdrawal protocol on adenoma detection during screening colonoscopy. Clin Gastroenterol Hepatol 2008;6: Soetikno RM, Kaltenbach T, Rouse RV, et al. Prevalence of nonpolypoid (flat and depressed) colorectal neoplasms in asymptomatic and symptomatic adults. JAMA 2008;299: Barclay RL, Vicari JJ, Doughty AS, et al. Colonoscopic withdrawal times and adenoma detection during screening colonoscopy. N Engl J Med 2006;355: Pickhardt PJ, Choi JR, Hwang I, et al. Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 2003;349: Rex DK, Overhiser AJ, Chen SC, et al. Estimation of impact of American College of Radiology recommendations on CT colonography reporting for resection of high-risk adenoma findings. Am J Gastroenterol 2009;104: van Rijn JC, Reitsma JB, Stoker J, et al. Polyp miss rate determined by tandem colonoscopy: a systematic review. Am J Gastroenterol 2006;101: Winawer S, Fletcher R, Rex D, et al. Colorectal cancer screening and surveillance: clinical guidelines and rationale update based on new evidence. Gastroenterology 2003;124: Gschwantler M, Kriwanek S, Langner E, et al. High-grade dysplasia and invasive carcinoma in colorectal adenomas: a multivariate analysis of the impact of adenoma and patient characteristics. Eur J Gastroenterol Hepatol 2002;14: Zauber AG, Lansdorp-Vogelaar I, Knudsen AB, et al. Evaluating test strategies for colorectal cancer screening: a decision analysis for the U.S. Preventive Services Task Force. Ann Intern Med 2008;149: Pickhardt PJ, Nugent PA, Mysliwiec PA, et al. Location of adenomas missed by optical colonoscopy. Ann Intern Med 2004;141: Shinya H, Wolff WI. Morphology, anatomic distribution and cancer potential of colonic polyps. Ann Surg 1979;190: Butterly LF, Chase MP, Pohl H, et al. Prevalence of clinically important histology in small adenomas. Clin Gastroenterol Hepatol 2006;4: Koretz RL. Malignant polyps: are they sheep in wolves clothing? Ann Intern Med 1993;118: Frazier AL, Colditz GA, Fuchs CS, et al. Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 2000;284: Hur C, Chung DC, Schoen RE, et al. The management of small polyps found by virtual colonoscopy: results of a decision analysis. Clin Gastroenterol Hepatol 2007;5: Chen CD, Yen MF, Wang WM, et al. A case-cohort study for the disease natural history of adenoma-carcinoma and de novo carcinoma and surveillance of colon and rectum after polypectomy: implication for efficacy of colonoscopy. Br J Cancer 2003;88: Pickhardt PJ, Hassan C, Laghi A, et al. Small and diminutive polyps detected at screening CT colonography: a decision analysis for referral to colonoscopy. AJR Am J Roentgenol 2008;190: Pickhardt PJ, Hassan C, Laghi A, et al. Cost-effectiveness of colorectal cancer screening with computed tomography colonography: the impact of not reporting diminutive lesions. Cancer 2007;109: Hofstad B, Vatn MH, Andersen SN, et al. Growth of colorectal polyps: redetection and evaluation of unresected polyps for a period of three years. Gut 1996;39: Stryker SJ, Wolff BG, Culp CE, et al. Natural history of untreated colonic polyps. Gastroenterology 1987;93: Vatn MH, Stalsberg H. The prevalence of polyps of the large intestine in Oslo: an autopsy study. Cancer 1982;49:

7 864 POHL AND ROBERTSON CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No Clark JC, Collan Y, Eide TJ, et al. Prevalence of polyps in an autopsy series from areas with varying incidence of large-bowel cancer. Int J Cancer 1985;36: Brenner H, Hoffmeister M, Stegmaier C, et al. Risk of progression of advanced adenomas to colorectal cancer by age and sex: estimates based on 840,149 screening colonoscopies. Gut 2007;56: Bronshtein IN, Semendyayev KA, Musiol G, et al. Handbook of mathematics. 5th ed. New York: Springer, 1997: Citarda F, Tomaselli G, Capocaccia R, et al. Efficacy in standard clinical practice of colonoscopic polypectomy in reducing colorectal cancer incidence. Gut 2001;48: Jorgensen OD, Kronborg O, Fenger C, et al. Influence of long-term colonoscopic surveillance on incidence of colorectal cancer and death from the disease in patients with precursors (adenomas). Acta Oncol 2007;46: Loeve F, van Ballegooijen M, Boer R, et al. Colorectal cancer risk in adenoma patients: a nation-wide study. Int J Cancer 2004; 111: Robertson DJ, Greenberg ER, Beach M, et al. Colorectal cancer in patients under close colonoscopic surveillance. Gastroenterology 2005;129: Meagher AP, Stuart M. Does colonoscopic polypectomy reduce the incidence of colorectal carcinoma? Aust N Z J Surg 1994;64: Bertario L, Russo A, Sala P, et al. Predictors of metachronous colorectal neoplasms in sporadic adenoma patients. Int J Cancer 2003;105: Imperiale TF, Glowinski EA, Lin-Cooper C, et al. Five-year risk of colorectal neoplasia after negative screening colonoscopy. N Engl J Med 2008;359: Brenner H, Haug U, Arndt V, et al. Low risk of colorectal cancer and advanced adenomas more than 10 years after negative colonoscopy. Gastroenterology 2010;138: Kaminski MF, Regula J, Kraszewska E, et al. Quality indicators for colonoscopy and the risk of interval cancer. N Engl J Med 2010; 362: Reprint requests Address requests for reprints to: Heiko Pohl, MD, Section of Gastroenterology, VA Medical Center, 215 North Main Street, White River Junction, Vermont heiko.pohl@dartmouth.edu; fax: (802) Acknowledgments The authors thank Douglas Rex from the Indiana University School of Medicine, Indianapolis, IN, for providing additional details on polyp histology from his study published in The authors thank the following members of the VA outcomes groups at the VA Medical Center in White River Junction, Vermont, for their support: H. Gilbert Welch, Brenda Sirovich, Robin Larson, Lisa Schwartz, and Steven Woloshin. Conflicts of interest The authors disclose no conflicts. Funding This material is the result of work supported with resources and the use of facilities at the VA Medical Center, White River Junction, Vermont. Dr Robertson s work is supported by a VA HSR&D Career Development Award. The contents of this article do not represent the views of the Department of Veterans Affairs or the United States Government.