Abstract
Background & Aims
Limited data exist regarding the actual risk of developing advanced adenomas and cancer following polypectomy or the factors that determine risk.
Methods
We pooled individual data from 8 prospective studies comprising 9167 men and women aged 22 to 80 with previously-resected colorectal adenomas to quantify their risk of developing subsequent advanced adenoma or cancer as well as identify factors associated with development of advanced colorectal neoplasms during surveillance.
Results
During a median follow-up of 47.2 months, advanced colorectal neoplasia was diagnosed in 1082 (11.8%) of the patients, 58 of whom (0.6%) had invasive cancer. Risk of a metachronous advanced adenoma was higher among patients with 5 or more baseline adenomas (24.1%; SE=2.2) and those with an adenoma 20 mm in size or greater (19.3%; SE=1.5). Risk factor patterns were similar for advanced adenomas and invasive cancer. In multivariate analyses, older age (P <0.0001 for trend) and male sex (odds ratio [OR], 1.40; 95% confidence interval [CI] 1.19–1.65) were significantly associated with increased risk of metachronous advanced neoplasia, as were the number and size of prior adenomas (P <0.0001 for trend), the presence of villous features (OR, 1.28; 95% CI 1.07–1.52), and proximal location (OR, 1.68; 95% CI 1.43–1.98). High-grade dysplasia was not independently associated with metachronous advanced neoplasia after adjustment for other adenoma characteristics.
Conclusions
Occurrence of advanced colorectal neoplasia is common following polypectomy. Factors that are most strongly associated with risk of advanced neoplasia are patient age and the number and size of prior adenomas.
In the United States, colorectal cancer is the third most common cancer in both men and women and the second leading cause of cancer death, accounting for nearly 150,000 new cases and 50,000 deaths annually1. Colonoscopy screening with removal of adenomas is an effective strategy for reducing colorectal cancer incidence and mortality,2, 3 and it is one of several accepted screening modalities in the United States4. Among individuals who have one or more adenomas removed at colonoscopy, 20 to 50% will be found to have a metachronous lesion when undergoing follow-up (surveillance) colonoscopy within three to five years5, 6, 7, 8, 9. As many as 20% of patients diagnosed with neoplasms during surveillance have advanced adenomas, defined as those with a diameter 10 mm or larger, having greater than 25% villous features, or having high-grade dysplasia5, 6, 7, 9, 10, and a small, though not negligible, proportion are newly diagnosed with invasive colorectal cancer7, 11, 12, 13, 14, 15, 16.
Professional groups have developed guidelines for surveillance colonoscopy intervals after polypectomy3, 17 with the goal of more efficient detection of early cancers or advanced adenomas, the latter carrying a presumed higher risk of progressing to invasive cancer. These guidelines use the characteristics of neoplasia at the most recent colonoscopy to classify patients as either higher or lower risk. Knowledge of the relation between patient characteristics or features of the baseline adenoma with risk of metachronous neoplasms, however, is based on a small body of scientific evidence18. Relatively few reports, including one meta analysis19, have addressed this issue, and none has been large enough to provide precise estimates of risk of advanced adenomas or invasive cancer. To clarify these issues, we pooled data from eight prospective studies to estimate absolute risks of metachronous advanced adenoma, colorectal cancer, and their combination (advanced colorectal neoplasia) and to identify patient characteristics and adenoma features that are independently associated with risk of these outcomes.
Materials and Methods
Study Population
The pooled analyses used patient-level data from eight North American studies6, 7, 10, 12, 20, 21, 22, 23 of patients with sporadic colorectal adenoma (Table 1). Six studies were randomized controlled trials 6, 7, 12, 20, 21, 22, and in two of these 7, 20 a modest degree of efficacy of the intervention was demonstrated on the primary endpoint. Eligibility criteria for the chemoprevention trials were generally similar. However, study populations in the non-intervention studies 10, 23 included average-risk individuals with a first-time diagnosis of adenomatous polyps. We aimed to include studies that met the following criteria: 1) 800 or more study participants; 2) study protocol requiring complete baseline colonoscopy with removal of one or more adenomas and removal of all visualized lesions; 3) a specified schedule of surveillance follow-up colonoscopies; and 4) availability of end-point data regarding the number, size, and histopathology of adenomas and colorectal cancers detected in follow-up examinations. To our knowledge, we included all studies meeting these criteria that had reported findings by June 2005. Of the 10,021 men and women who were enrolled in the individual studies, we excluded patients who had a colorectal cancer present at baseline (n=27) and those who did not have a follow-up colonoscopy performed after the first six months of study (n=827), since these likely were individuals who were not under typical post-polypectomy surveillance. Thus, data for 9167 (91.5%) patients remained for inclusion in our pooled analyses.
Table 1.
Description of Studies Included in the Pooled Analysesa
| APPS | NPS | CPPS | PPT | WBF | VA | AFT | UDCA | |
|---|---|---|---|---|---|---|---|---|
| Number enrolled | 864 | 1418 | 930 | 2079 | 1429 | 3121/895b | 1121 | 1285 |
| Study design | 4-arm trial | 2-arm endoscopy study | 2-arm trial | 2-arm trial | 2-arm trial | 2-arm endoscopy study | 3-arm trial | 2-arm trial |
| Entry criteria | Any recent adenoma | First adenoma found at screening | Any recent adenoma | Any recent adenoma | Any recent adenoma | First adenoma found at screening | Any recent adenoma | Any recent adenoma |
| Recruitment period | 1984–88 | 1980–1990 | 1991–1998 | 1991–1998 | 1990–1998 | 1994–1997 | 1994–1998 | 1995–1999 |
| Age range (years) | 25–79 | 22–88 | 27–80 | 35–89 | 40–80 | 50–75 | 29–79 | 40–80 |
| Number of centers | 6 | 7 | 7 | 8 | 1 | 13 | 9 | 2 |
| Follow-up colonoscopy schedule | Years 1 and 4 | Years 1 and 3 or Year 3 | Years 1 and 4 | Years 1 and 4 | Years 1 and 3 for 889 patients; only year 3 for 415 patients | Years 2 and 5 for patients with large adenomas; only year 5 for patients with small adenomas | Year 3 | Year 3 |
| Number (%) completedc | 837 (96.9) | 939 (81.0) d | 913 (98.2) | 2024 (97.4) | 1304 (91.3) | 871 (97.3) | 1086 (96.9) | 1193 (92.8) |
| Study Pathologist Review (%) | 100 | 96 | 0 (baseline) 100 (follow-up) |
99 | 0|| | 100 | 0 (baseline) 100 (follow-up) |
0e |
APPS, Antioxidant Polyp Prevention Study21; NPS, National Polyp Study8; CPPS, Calcium Polyp Prevention Study20; WBF, Wheat Bran Fiber study;12 VA, Veterans Affairs Cooperative Study23; AFT, Aspirin Folate Trial7; UDCA, Ursodeoxycholic Acid study22.
3121 includes participants with and without polyps; 895 is population with adenomas at baseline that were included in the pooling analyses.
Indicates number of participants who underwent one or more follow-up colonoscopies.
Percent based on patients who were on study long enough to be due for one or more surveillance colonoscopies.
Study pathologist review was included for missing data.
Study Endpoints
We considered as endpoints all colorectal neoplasms that were diagnosed during an interval beginning six months after the baseline examination and ending on the date of the last protocol- specified colonoscopic examination. We excluded neoplasms found at colonoscopies conducted in the first six months following the baseline colonoscopy, since these examinations appeared to have been done primarily to address problems identified during the baseline colonoscopy and were not performed as surveillance procedures. In the course of each of the parent studies, staff reviewed the endoscopy and pathology reports to abstract data on number, size, and location of all neoplasms. For our analyses, we categorized anatomic location of lesions as proximal (cecum, ascending colon, hepatic flexure, transverse colon, and splenic flexure), or distal (descending colon, sigmoid colon, and rectum). Histological features of colorectal neoplasms were determined from pathology reports generated by either an independent study pathologist or a community pathologist, depending on the individual study protocol (see Table 1). Definitions for adenomas were: tubular (25% or less villous component), tubulovillous (26 to 75% villous component), or villous (greater than 75% villous component). We considered advanced adenomas to be those that had one or more of the following features: 10 mm in diameter or larger, presence of high-grade dysplasia, or greater than 25% villous features (also classified as tubulovillous or villous histology). We further combined advanced adenomas and invasive cancer into an endpoint of “advanced colorectal neoplasia”. Villous histology was included in our definition of advanced adenomas, although this was not the case in each of the original studies10. We tested whether villous features, high-grade dysplasia, and adenoma size were predictive of each of the features used to define advanced adenomas. The associations were similar across each of the features used to define advanced metachronous adenoma, suggesting that the inclusion of villous histology in our definition of advanced adenoma did not significantly influence the observed associations.
Risk Factor Data
The studies had used self-administered questionnaires to obtain data on sociodemographic variables (age, sex, and race), cigarette smoking, family history of colorectal cancer in first- degree relatives, and history of polyps or adenomas before the baseline examination. Cigarette smoking status was classified as never, former, or current. Patient weight and height had been measured or self-reported according to study protocol. We calculated body mass index (BMI) by dividing weight (kg) by the square of height (m). Weight, height, and cigarette smoking were not ascertained in the National Polyp Study10. History of adenomas or polyps diagnosed before the baseline examination had been assessed by self report; the Veterans Affairs Cooperative Study23 and the National Polyp Study10 only included individuals whose first adenoma was diagnosed at the baseline examination.
Data on baseline adenoma characteristics (number, size, location, histology, and high-grade dysplasia) had been ascertained from review of clinical and pathology reports; there were 957 patients whose pathology reports did not specify the presence or absence of villous features; these were included in the tubular adenoma group since their association with outcomes of interest were similar. The Antioxidant Polyp Prevention Study24 and the Calcium Polyp Prevention Study20 did not collect data regarding the presence or absence of high-grade dysplasia in baseline adenomas.
Statistical Analysis
We generated summary descriptive data for baseline characteristics and endpoints for each study and for the total population. For each baseline characteristic, we estimated the absolute risk of colorectal neoplasia during the stipulated follow-up period for each specific outcome: non- advanced adenoma(s), one or more advanced adenoma(s), and invasive cancer. We also compared the absolute risk of advanced adenoma and cancer among patients defined as being at higher or lower risk according to the most recent surveillance guidelines3. In this categorization, the low-risk group included patients with one or two small (less than 1 cm), tubular adenoma(s) with low-grade dysplasia; the high-risk group included patients with three or more adenomas, or any adenomas 1 cm or larger, or with greater than 25% villous features, or with high-grade dysplasia.
We estimated pooled adjusted odds ratios for the study endpoints using logistic regression models that controlled for study, age, sex, race, smoking status, BMI, family history of colorectal cancer, history of polyp or adenoma prior to the baseline examination, and the baseline adenoma characteristics. There were too few cancers diagnosed in the study population to permit a full multivariate analysis of this outcome separately. Heterogeneity between study-specific risk estimates was assessed for each exposure variable separately by conducting a log likelihood ratio test25 comparing the multivariate model presented in the tables with a multivariate model including additional interaction terms between each study indicator variable and the exposure variable of interest. Heterogeneity was assumed to be present when the likelihood ratio test yielded a P-value <0.05. Given the lack of evidence of heterogeneity based on this criterion (P=0.09 for high-grade dysplasia and >0.30 for all other variables), pooled odds ratios (OR) and 95% confidence intervals (CI) computed from all individual data in unordered polytomous logistic regression models are presented for two principal outcomes, non-advanced and advanced neoplasms (advanced adenomas and invasive cancer).
We also conducted random-effects modeling using a two-stage method in which study-specific ORs and 95% CIs were first estimated, followed by estimation of summary ORs, weighted by the inverse of their variance26. These results were not materially different from those of the fixed effects models presented. We considered the likelihood that differential follow-up time across studies might confound the observed associations. However, inclusion of a variable for follow- up time in the multivariate models did not appreciably alter the results. Tests for trend were conducted in separate analyses in which the exposure variable was modeled as an ordinal variable with values corresponding to the categorization level of that variable in the original analyses.
If a study, by design, had not collected information for a specific variable, we excluded it from the analyses that pertained to that factor (for example, the National Polyp Study data were not included in analyses of risk associated with BMI and smoking). In the case of data that were missing/unknown for other reasons, we first created a missing data indicator variable for missing responses and then compared the model results to those where the missing data observations were dropped from the analyses. The overall results were not appreciably different between models using these two approaches, and we present here the results from the models that included the missing data indicator variable.
Results
The study population of 9167 patients was predominantly male (71.2%) and Caucasian (89.1%), with a mean age of 62.0 years (Table 2). The median follow-up for the entire group was 47.2 months (range of medians among studies 36.9 to 59.0), and 46.7% of patients had at least one neoplasm diagnosed during this period (Table 3). Advanced adenomas were found during follow-up in 1024 patients (11.2% of the total population), and 58 (0.6%) had invasive colorectal cancer.
Table 2.
Baseline Characteristics of Patients
| APPSa (N=837) | NPS (N=939) | CPPS (N=913) | PPT (N=2024) | WBF (N=1304) | VA (N=871) | AFT (N=1086) | UDCA (N=1193) | Total (N=9167) | |
|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||
| Age, mean (SD), y | 61.1±8.3 | 61.0±9.9 | 61.0±9.1 | 60.6±10.0 | 65.7±8.8 | 63.0±6.9 | 57.4±9.6 | 65.8±8.5 | 62.0±9.5 |
|
| |||||||||
| Sex (%) | |||||||||
| Male | 79.2 | 71.5 | 72.3 | 64.8 | 66.8 | 98.0 | 63.5 | 67.5 | 71.2 |
| Female | 20.8 | 28.5 | 27.7 | 35.2 | 33.2 | 2.0 | 36.5 | 32.5 | 28.8 |
|
| |||||||||
| Race (%) | |||||||||
| White | 85.0 | 90.5 | 85.1 | 89.5 | 96.1 | 83.2 | 85.6 | 93.0 | 89.1 |
| Black | 6.8 | 7.0 | 8.0 | 8.2 | 0.8 | 9.6 | 5.8 | 0.7 | 5.7 |
| Other | 8.2 | 2.4 | 6.9 | 2.3 | 3.1 | 7.1 | 8.6 | 6.4 | 5.2 |
|
| |||||||||
| Family history of colorectal cancerb (%) | |||||||||
| No | 78.5 | 77.6 | 67.6 | 69.8 | 75.8 | 84.5 | 51.4 | 71.3 | 71.4 |
| Yes | 19.4 | 22.4 | 21.2 | 24.8 | 17.0 | 15.5 | 30.7 | 27.8 | 22.8 |
| Unknown | 2.2 | 0.0 | 11.2 | 5.5 | 7.2 | 0.0 | 18.0 | 0.9 | 5.8 |
|
| |||||||||
| Cigarette smoking status (%) | |||||||||
| Never | 30.9 | N/Ac | 33.4 | 38.3 | 34.2 | 23.0 | 42.4 | 30.0 | 34.1 |
| Former | 45.4 | N/A | 47.4 | 48.2 | 52.1 | 55.1 | 42.5 | 56.3 | 49.6 |
| Current | 21.5 | N/A | 19.2 | 13.4 | 13.6 | 21.8 | 15.0 | 11.9 | 15.8 |
| Unknown | 2.2 | N/A | 0.0 | 0.0 | 0.2 | 0.1 | 0.1 | 1.8 | 0.5 |
|
| |||||||||
| Body mass index (kg/m2)d (%) | |||||||||
| <25 | 31.6 | N/A§ | 31.3 | 26.4 | 34.2 | 13.4 | 30.7 | 29.7 | 28.4 |
| 25–<30 | 48.9 | N/A | 44.2 | 46.6 | 45.0 | 45.2 | 46.8 | 44.6 | 45.9 |
| 30+ | 19.6 | N/A | 24.5 | 26.9 | 20.8 | 41.4 | 22.5 | 25.7 | 25.7 |
|
| |||||||||
| Previous polype (%) | |||||||||
| No Yes | 75.4 | 100.0 | 78.0 | 81.8 | 53.8 | 100.0 | 69.7 | 56.6 | 75.7 |
| Unknown | 23.5 | 0.0 | 21.5 | 18.2 | 34.7 | 0.0 | 29.9 | 43.4 | 22.4 |
| 1.1 | 0.0 | 0.5 | 0.0 | 11.6 | 0.0 | 0.4 | 0.0 | 1.8 | |
|
| |||||||||
| Mean no. of adenomasf | 1.9±1.4 | 1.9±1.5 | 1.9±1.4 | 1.7±1.2 | 1.5±0.9 | 2.1±1.9 | 1.6±1.0 | 1.7±1.1 | 1.7±1.3 |
|
| |||||||||
| Location (%) | |||||||||
| Distal colorectum | 58.7 | 60.4 | 52.8 | 52.3 | 46.8 | 39.8 | 40.0 | 37.2 | 48.4 |
| Proximal only | 19.1 | 14.7 | 24.1 | 28.3 | 31.8 | 34.7 | 34.5 | 36.7 | 28.6 |
| Proximal and distal | 22.2 | 21.4 | 22.9 | 17.4 | 15.3 | 25.5 | 13.7 | 19.7 | 19.1 |
| Unknown | 0.0 | 3.5 | 0.2 | 2.0 | 6.1 | 0.0 | 11.8 | 6.4 | 3.9 |
|
| |||||||||
| Size of largest adenoma, mean (SD)g, mm | 8.7±6.0 | 11.2±7.3 | 7.4±5.6 | 8.3±7.1 | 8.3±5.9 | 7.7±6.5 | 7.3±5.3 | 8.9±5.9 | 8.5±6.4 |
|
| |||||||||
| Adenoma histology (%) | |||||||||
| Tubularh | 57.2 | 81.8 | 81.1 | 79.6 | 78.7 | 87.8 | 85.8 | 79.3 | 79.3 |
| Tubulovillous/villous | 42.8 | 18.2 | 18.9 | 20.4 | 21.3 | 12.2 | 14.2 | 20.7 | 20.7 |
|
| |||||||||
| High-grade dysplasia (%) | |||||||||
| No | 96.2 | 91.1 | N/Ai | 92.1 | 79.0 | 94.7 | N/Ai | 92.6 | 90.5 |
| Yes | 3.8 | 8.9 | N/A | 7.9 | 21.0 | 5.3 | N/A | 7.4 | 9.5 |
APPS, Antioxidant Polyp Prevention Study21; NPS, National Polyp Study8; CPPS, Calcium Polyp Prevention Study20; WBF, Wheat Bran Fiber study;12 VA, Veterans Affairs Cooperative Study23; AFT, Aspirin Folate Trial7; UDCA, Ursodeoxycholic Acid study22.
History of colorectal cancer in one or more parent, sibling or child.
Excludes 15 patients with missing data.
No data available for NPS study patients.
History of previous polyps or adenomas prior to qualifying colonoscopy; NPS and VA study included only participants with a first-time adenomatous polyp.
Excludes 55 participants with missing data.
Excludes 353 participants with missing data.
Includes 957 participants with adenomas where histology was not specified, unknown, or other histologies.
No data available for CPPS and AFT participants.
Table 3.
Patient Summary Endpoints During Surveillance Follow-upa
| APPSb (N=837) | NPS (N=939) | CPPS (N=913) | PPT (N=2024) | WBF (N=1304) | VA (N=871) | AFT (N=1086) | UDCA (N=1193) | Total (N=9167) | |
|---|---|---|---|---|---|---|---|---|---|
| Median follow-up, months (range) | 49.1 (11.4–75.8) | 36.9 (6.1–57.0) | 48.6 (10.9–91.4) | 52.1 (6.5–84.5) | 39.1 (6.7–88.6) | 59.0 (7.8–66.0) | 36.9 (11.5–66.4) | 38.0 (6.4–88.1) | 47.2 (6.1–91.4) |
| Median number of colonoscopies (range) | 2.0 (1.0–8.0) | 1.0 (1.0–6.0) | 2.0 (1.0–7.0) | 2.0 (1.0–7.0) | 2.0 (1.0–6.0) | 1.0 (1.0–7.0) | 1.0 (1.0–5.0) | 1.0 (1.0–5.0) | 2.0 (1.0–8.0) |
| Any adenoma during follow-upc, n (%) | 432 (51.6) | 324 (34.5) | 428 (46.9) | 1077 (53.2) | 641 (49.2) | 395 (45.4) | 476 (43.8) | 507 (42.5) | 4280 (46.7) |
| Large adenomad, n (%) | 53 (6.3) | 60 (6.4) | 55 (6.0) | 159 (7.9) | 137 (10.5) | 66 (7.6) | 66 (6.1) | 115 (9.6) | 711 (7.8) |
| Tubulovillous/villous histologye, n (%) | 110 (13.1) | 15 (1.6) | 103 (11.3) | 82 (4.1) | 82 (6.3) | 23 (2.6) | 76 (7.0) | 89 (7.5) | 580 (6.3) |
| High-grade dysplasia, n (%) | 4 (0.5) | 6 (0.6) | 2 (0.2) | 23 (1.1) | 12 (0.9) | 6 (0.7) | 0 (0.0) | 1 (0.1) | 54 (0.6) |
| Advanced adenomaf, n (%) | 128 (15.3) | 64 (6.8) | 120 (13.1) | 195 (9.6) | 177 (13.6) | 69 (7.9) | 108 (9.9) | 163 (13.7) | 1024 (11.2) |
| Colorectal cancer, n (%) | 5 (0.6) | 3 (0.3) | 8 (0.9) | 13 (0.6) | 8 (0.6) | 8 (0.9) | 6 (0.6) | 7 (0.6) | 58 (0.6) |
Refers to period beginning with six months after baseline colonoscopy to last colonoscopy included in the analysis.
APPS, Antioxidant Polyp Prevention Study21; NPS, National Polyp Study8; CPPS, Calcium Polyp Prevention Study20; WBF, Wheat Bran Fiber study;12 VA, Veterans Affairs Cooperative Study23; AFT, Aspirin Folate Trial7; UDCA, Ursodeoxycholic Acid study22.
Patients with one or more adenoma(s) or cancer detected at follow-up.
Presence of one or more adenoma(s) 10 mm or larger in diameter.
Presence of one or more adenoma(s) with greater than 25% villous features.
Presence of one or more advanced adenoma(s), defined as lesions having one or more of the following features: 10 mm or larger in diameter, greater than 25% villous features, or high-grade dysplasia.
The proportions of patients diagnosed during follow-up with advanced adenoma or invasive cancer increased with age, male sex, and history of polyps or adenoma prior to the baseline examination (Table 4). Risk varied only slightly according to race, family history of colorectal cancer, smoking history, or BMI. Both advanced adenoma and invasive cancer occurred more often in patients with baseline adenomas that were larger, more numerous, were located proximally, had villous features, or had high-grade dysplasia. Risk of a metachronous advanced adenoma approached 20% in patients who had four or more baseline adenomas, or whose largest baseline adenoma was 20 mm in size or greater (Table 4). Risk of invasive cancer was 1.2% in patients who had larger (20 mm or greater) baseline adenomas and 1.3% in those who had prior lesions with high-grade dysplasia. Analyses that assessed risk according to surveillance guidelines’ risk stratification categories3 showed a moderate degree of discrimination. Compared to patients who could be categorized as low-risk, those deemed high-risk were more often diagnosed with both advanced adenoma (15.5% vs. 6.9%) and colorectal cancer (0.8% vs. 0.5%). There was essentially no difference in the occurrence of non-advanced neoplasms.
Table 4.
Risk of New Neoplasia at Follow-up, according to Baseline Patient and Adenoma Characteristics
| Characteristic | Number (%) | Non-advanced adenoma % (95% CI) | Advanced adenoma % (95% CI) | Cancer % (95% CI) |
|---|---|---|---|---|
|
| ||||
| Age, y | ||||
| < 40 | 154 (1.7) | 18.8 (12.7–25.0) | 3.9 (0.8–7.0) | 0.0 |
| 40–49 | 804 (8.8) | 27.7 (24.6–30.8) | 6.3 (4.7–8.0) | 0.1 (0.0–0.4) |
| 50–59 | 2397 (26.1) | 34.9 (33.0–36.8) | 8.6 (7.5–9.8) | 0.3 (0.1–0.6) |
| 60–69 | 3676 (40.1) | 36.0 (34.4–37.5) | 12.2 (11.1–13.2) | 0.6 (0.3–0.8) |
| 70–79 | 2074 (22.6) | 36.9 (34.9–39.0) | 14.5 (13.0–16.0) | 1.3 (0.8–1.7) |
| 80+ | 62 (0.7) | 33.9 (22.1–45.7) | 17.7 (8.2–27.3) | 1.6 (0.0–4.7) |
|
| ||||
| Sex | ||||
| Female | 2642 (28.8) | 29.0 (27.3–30.7) | 9.8 (8.7–10.9) | 0.3 (0.1–0.5) |
| Male | 6525 (71.2) | 37.3 (36.1–38.4) | 11.7 (10.9–12.5) | 0.8 (0.6–1.0) |
|
| ||||
| Race | ||||
| White | 8166 (89.1) | 35.0 (34.0–36.1) | 11.4 (10.7–12.0) | 0.6 (0.4–0.8) |
| Black | 527 (5.7) | 37.6 (33.4–41.7) | 10.4 (7.8–13.0) | 0.8 (0.0–1.5) |
| Other | 474 (5.2) | 29.1 (25.0–33.2) | 8.9 (6.3–11.4) | 1.3 (0.3–2.3) |
|
| ||||
| Family history of colorectal cancera | ||||
| No | 6547 (71.4) | 34.5 (33.4–35.7) | 11.0 (10.3–11.8) | 0.6 (0.4–0.8) |
| Yes | 2089 (22.8) | 36.7 (34.6–38.7) | 11.6 (10.3–13.0) | 0.6 (0.3–0.9) |
| Unknown | 531 (5.8) | 32.2 (28.2–36.2) | 11.1 (8.4–13.8) | 1.3 (0.3–2.3) |
|
| ||||
| Cigarette smoking statusb | ||||
| Never | 2805 (34.1) | 33.6 (31.9–35.4) | 11.1 (9.9–12.2) | 0.5 (0.2–0.8) |
| Former | 4081 (49.6) | 36.9 (35.4–38.4) | 12.1 (11.1–13.1) | 0.9 (0.6–1.2) |
| Current | 1299 (15.8) | 36.8 (34.2–39.4) | 11.8 (10.0–13.5) | 0.2 (0.0–0.5) |
| Unknown | 43 (0.5) | 34.9 (20.6–49.1) | 9.3 (0.6–18.0) | 2.3 (0.0–6.8) |
|
| ||||
| Body mass index (kg/m 2) c, d | ||||
| <25 | 2332 (28.4) | 32.2 (30.3–34.1) | 11.7 (10.4–13.1) | 0.5 (0.2–0.8) |
| 25–<30 | 3771 (45.9) | 36.6 (35.0–38.1) | 11.6 (10.6–12.7) | 0.7 (0.4–1.0) |
| 30+ | 2110 (25.7) | 38.1 (36.0–40.2) | 11.6 (10.2–12.9) | 0.8 (0.4–1.2) |
|
| ||||
| Previous Polype | ||||
| No Yes | 6941 (75.7) | 33.4 (32.3–34.5) | 9.8 (9.1–10.5) | 0.6 (0.4–0.8) |
| Unknown | 2057 (22.4) | 39.5 (37.4–41.6) | 15.2 (13.7–16.8) | 0.8 (0.4–1.2) |
| 169 (1.8) | 38.5 (31.1–45.8) | 16.6 (11.0–22.2) | 0.6 (0.0–1.7) | |
|
| ||||
| Adenoma Numberf | ||||
| 1 | 5465 (60.0) | 30.2 (29.0–31.4) | 8.6 (7.8–9.3) | 0.5 (0.4–0.7) |
| 2 | 2054 (22.5) | 38.3 (36.2–40.4) | 12.7 (11.3–14.1) | 0.5 (0.2–0.9) |
| 3 | 890 (9.8) | 45.4 (42.1–48.7) | 15.3 (12.9–17.6) | 1.1 (0.4–1.8) |
| 4 | 326 (3.6) | 45.4 (40.0–50.8) | 19.6 (15.3–23.9) | 1.2 (0.0–2.4) |
| 5+ | 377 (4.1) | 51.2 (46.1–56.2) | 24.1 (19.8–28.5) | 0.8 (0.0–1.7) |
|
| ||||
| Adenoma location | ||||
| Distal colorectum | 4434 (48.4) | 30.4 (29.1–31.8) | 8.5 (7.7–9.3) | 0.4 (0.2–0.6) |
| Proximal only | 2620 (28.6) | 37.3 (35.4–39.1) | 11.8 (10.6–13.1) | 0.8 (0.5–1.2) |
| Proximal and distal | 1754 (19.1) | 44.2 (41.9–46.6) | 17.5 (15.7–19.3) | 1.0 (0.6–1.5) |
| Unknown | 359 (3.9) | 26.7 (22.2–31.3) | 8.1 (5.3–10.9) | 0.0 |
|
| ||||
| Adenoma sizeg | ||||
| <5mm | 2540 (28.8) | 36.4 (34.5–38.3) | 7.7 (6.6–8.7) | 0.5 (0.2–0.8) |
| 5–<10 mm | 3115 (35.3) | 36.8 (35.1–38.5) | 8.7 (7.7–9.7) | 0.5 (0.2–0.7) |
| 10–<20mm | 2487 (28.2) | 31.4 (29.6–33.3) | 15.9 (14.5–17.4) | 0.8 (0.5–1.2) |
| 20+ mm | 672 (7.6) | 31.8 (28.3–35.4) | 19.3 (16.4–22.3) | 1.2 (0.4–2.0) |
|
| ||||
| Adenoma histology | ||||
| Tubularh | 7268 (79.3) | 35.1 (34.0–36.2) | 9.7 (9.0–10.4) | 0.6 (0.4–0.7) |
| Tubulovillous/villousi | 1899 (20.7) | 33.9 (31.8–36.0) | 16.8 (15.1–18.5) | 0.9 (0.5–1.4) |
|
| ||||
| High-grade dysplasiaj | ||||
| No | 6485 (90.5) | 35.3 (34.2–36.5) | 10.6 (9.8–11.3) | 0.5 (0.4–0.7) |
| Yes | 683 (9.5) | 35.9 (32.3–39.5) | 16.0 (13.2–18.7) | 1.3 (0.5–2.2) |
|
| ||||
| Low-risk groupk | 4644 (50.7) | 34.5 (33.1–35.8) | 6.9 (6.2–7.6) | 0.5 (0.3–0.7) |
| High-risk group | 4523 (49.3) | 35.3 (33.9–36.7) | 15.5 (14.5–16.6) | 0.8 (0.5–1.0) |
Percent of patients with a particular outcome reported during surveillance + standard error.
History of colorectal cancer in one or more parent, sibling, or child.
No data available for National Polyp Study participants (n=939)
Excludes 15 participants with missing data.
History of previous polyps or adenomas prior to qualifying colonoscopy; National Polyp Study and VA Cooperative Study included only participants with a first-time adenoma (n=1810).
Number of adenomas detected at qualifying examination; excludes 55 participants with missing data.
Size of largest adenoma; excludes 353 participants with missing data.
Includes participants with adenomas where histology was not specified, unknown, or other histologies (n=957).
One or more adenoma(s) with greater than 25% villous features.
One or more adenoma(s) with high-grade dysplasia. No data available for 1999 Calcium Polyp Prevention Study and Aspirin Folate Trial participants.
Defined according to the most recent surveillance guidelines3. The low-risk group includes patients with 1 to 2 small (less than 1 cm), tubular adenoma(s) with low-grade dysplasia; the high-risk group includes patients with 3 or more adenomas, or any adenomas 1 cm or larger in size, or with greater than 25% villous features, or with high-grade dysplasia.
In light of the similar risk factor patterns for advanced adenoma and invasive cancer (Table 4), we combined these two outcomes into a category of advanced colorectal neoplasms for subsequent analyses. Results of the multivariate models (Tables 5) indicate that after adjustment for other variables, older age was strongly associated with risk of advanced neoplasms (P- trend<0.0001). The odds of advanced neoplasms was modestly higher in men (OR, 1.40; 95% CI, 1.19–1.65) and in patients who reported a history of previous polyps (OR, 1.76; 95% CI, 1.48–2.09); it was not clearly associated with race, family history of colorectal cancer, current cigarette smoking, or obesity (defined as BMI of 30 kg/m2 or greater). Adenoma number was strongly and independently associated with the odds of both advanced and non-advanced metachronous neoplasms (P-trend <0.0001). Patients with larger adenomas were also more likely to develop advanced neoplasms. Relative to patients with a baseline adenoma less than 5 mm, the adjusted odds for advanced neoplasms for those with an adenoma 10 to 19 mm was 2.27 (95% CI, 1.84–2.78), and for those with at least one 20 mm or larger it was 2.99 (95% CI, 2.24–4.00). Presence of any proximal adenoma at baseline was significantly associated with metachronous advanced neoplasms (adjusted OR, 1.68; 95% CI, 1.43–1.98). The strong association observed in the unadjusted statistical model between baseline villous histology and metachronous advanced neoplasia was attenuated in the multivariate analysis, and the association with baseline high-grade dysplasia essentially disappeared once adenoma size, histology, and location were included in the multivariate model. Study-specific analyses for baseline adenoma characteristics and advanced neoplasms showed consistent results across studies (Figure 1), a finding that supports the lack of statistically significant heterogeneity between study-specific risk estimates. Including in our multivariate models a variable for treatment assignment for those studies that entailed an intervention did not materially alter any of our results
Table 5.
Pooled Odds Ratios of Colorectal Neoplasia for Baseline Patient and Adenoma Characteristics
| Crude OR (95% CI) | Adjusted O | Ra (95% CI) | ||
|---|---|---|---|---|
|
| ||||
| Characteristic | Non-advanced | Advanced | Non-advanced | Advanced |
|
| ||||
| Age | ||||
| < 40 | 0.39 (0.26 – 0.59) | 0.32 (0.14–0.73) | 0.47 (0.31–0.72) | 0.41 (0.18–0.94) |
| 40–49 | 0.68 (0.57 – 0.81) | 0.61(0.45–0.85) | 0.72 (0.60–0.87) | 0.67 (0.48–0.93) |
| 50–59 | 1.00 | 1.00 | 1.00 | 1.00 |
| 60–69 | 1.13 (1.01–1.26) | 1.56 (1.31–1.86) | 1.10 (0.98–1.24) | 1.39 (1.16–1.68) |
| 70–79 | 1.25 (1.10–1.43) | 2.09 (1.72–2.52) | 1.21 (1.05–1.38) | 1.72 (1.40–2.11) |
| 80+ | 1.16 (0.66–2.05) | 2.59 (1.30–5.15) | 1.24 (0.69–2.25) | 2.70 (1.31–5.57) |
| P-trend<0.0001 | P-trend<0.0001 | |||
|
| ||||
| Sex | ||||
| Female | 1.00 | 1.00 | 1.00 | 1.00 |
| Male | 1.56 (1.41–1.72) | 1.50 (1.29–1.74) | 1.45 (1.30–1.62) | 1.40 (1.19–1.65) |
|
| ||||
| Race | ||||
| White | 1.00 | 1.00 | 1.00 | 1.00 |
| Black | 1.11 (0.92–1.34) | 0.97 (0.73–1.30) | 1.12 (0.92–1.37) | 1.08 (0.79–1.47) |
| Other | 0.72 (0.59–0.89) | 0.74 (0.54–1.01) | 0.83 (0.67–1.03) | 0.83 (0.60–1.16) |
|
| ||||
| Family history of colorectal cancerb | ||||
| No | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.12 (1.00–1.24) | 1.11 (0.94–1.29) | 1.15 (1.03–1.29) | 1.17 (0.99–1.38) |
|
| ||||
| Cigarette smoking statusc | ||||
| Never | 1.00 | 1.00 | 1.00 | 1.00 |
| Former | 1.20 (1.08–1.33) | 1.23 (1.05–1.43) | ||
| Current | 1.17 (1.02–1.35) | 1.11 (0.90–1.38) | 1.07 (0.96–1.20) | 1.08 (0.92–1.27) |
| 1.16 (1.00–1.35) | 1.13 (0.90–1.42) | |||
|
| ||||
| Body mass index d (kg/m2) | ||||
| <25 | 1.00 | 1.00 | 1.00 | 1.00 |
| 25–<30 | 1.23 (1.10–1.38) | 1.09 (0.93–1.29) | 1.10 (0.98–1.24) | 1.00 (0.84–1.19) |
| 30+ | 1.32 (1.16–1.51) | 1.13 (0.94–1.36) | 1.23 (1.08–1.41) | 1.13 (0.93–1.38) |
| P-trend=0.003 | P-trend=0.226 | |||
|
| ||||
| Previous Polype | ||||
| No | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.50 (1.34–1.66) | 1.95 (1.68–2.26) | 1.37 (1.21–1.55) | 1.76 (1.48–2.09) |
|
| ||||
| Adenoma number | ||||
| 1 | 1.00 | 1.00 | 1.00 | 1.00 |
| 2 | 1.58 (1.42–1.77) | 1.81 (1.54–2.14) | 1.46 (1.30–1.64) | 1.39 (1.17–1.66) |
| 3 | 2.38 (2.04–2.79) | 2.85 (2.30–3.54) | 2.05 (1.73–2.42) | 1.85 (1.46–2.34) |
| 4 | 2.70 (2.09–3.48) | 4.11 (2.99–5.63) | 2.23 (1.71–2.92) | 2.41 (1.71–3.40) |
| 5+ | 4.30 (3.33–5.56) | 6.94 (5.12–9.40) | 3.63 (2.76–4.78) | 3.87 (2.76–5.42) |
| P-trend < 0.0001 | P-trend < 0.0001 | |||
|
| ||||
| Adenoma locationf | ||||
| Distal colorectum | 1.00 | 1.00 | 1.00 | 1.00 |
| Any proximal | 1.78(1.62–1.95) | 2.27 (1.98–2.60) | 1.29 (1.16–1.44) | 1.68 (1.43–1.98) |
|
| ||||
| Size of largest adenoma | ||||
| <5 mm | 1.00 | 1.00 | 1.00 | 1.00 |
| 5–<10 mm | 1.03 (0.92–1.16) | 1.15 (0.95–1.39) | 1.01 (0.90–1.14) | 1.17 (0.95–1.42) |
| 10–<20 mm | 0.92 (0.82–1.04) | 2.18 (1.82–2.62) | 0.94 (0.82–1.08) | 2.27 (1.84–2.78) |
| 20+ mm | 1.02 (0.84–1.23) | 2.92 (2.28–3.73) | 1.00 (0.80–1.25) | 2.99 (2.24–4.00) |
| P-trend= 0.4944 | P-trend < 0.0001 | |||
|
| ||||
| Adenoma histology | ||||
| Tubularg | 1.00 | 1.00 | 1.00 | 1.00 |
| Tubulovillous/villoush | 1.09 (0.97–1.22) | 1.96 (1.69–2.27) | 1.05 (0.92–1.20) | 1.28 (1.07–1.52) |
|
| ||||
| High-grade dysplasia | ||||
| No | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.16 (0.97–1.38) | 1.77 (1.41–2.22) | 1.04 (0.86–1.26) | 1.05 (0.81–1.35) |
Adjusted for age, sex, race, family history of colorectal cancer, previous polyp, cigarette smoking, body mass index, baseline number of adenoma, adenoma size, location, histology, high-grade dysplasia, and study.
History of colorectal cancer in one or more parent, sibling, or child. Excludes 531 patients with unknown or missing data.
No data available for NPS study participants (n=939); excludes 43 participants with unknown or missing data from other studies.
No data available for NPS study participants (n=939); excludes 15 participants with missing data from other studies.
History of previous polyps or adenomas prior to qualifying colonoscopy. Excludes 169 participants with unknown or missing data.
Excludes 359 participants with unknown or missing data on location.
Includes 957 participants with adenomas where histology was not specified, unknown, or other histologies.
Adenomas with greater than 25% villous features.
Figure 1. Study-Specific and Pooled Multivariate Odds Ratios of Advanced Colorectal Neoplasia (Advanced Adenoma and Cancer) for Baseline Adenoma Characteristics.
Odds ratios (ORs) are adjusted for age, sex, race, smoking status, BMI, family history of colorectal cancer, history of polyp or adenoma prior to the baseline examination, and the baseline adenoma characteristics. ORs for number and size of adenomas represent an increase in risk per adenoma and per 10mm increment, respectively. The black squares and horizontal lines correspond to the study-specific multivariate ORs and 95% confidence intervals (CIs), respectively. The area of the black square reflects the study-specific weight (inverse of the variance). The pooled ORs are based on data from 8814 patients due to missing data on baseline adenoma size and number for 353 individuals. The diamond represents the pooled multivariate OR and 95% CIs. The solid line represents an OR of 1.0. The abbreviations for the studies are listed in Table 1.
Discussion
This pooled analysis of post-polypectomy patients showed that after a median of four years of follow-up surveillance, roughly one in ten patients was diagnosed with an advanced colorectal neoplasm, and approximately one in 150 was diagnosed with colorectal cancer. Development of metachronous advanced colorectal neoplasia was associated with the number, size, location, and histological features of prior adenomas, as well as with the age and sex of patients. Of these factors, patient age and adenoma size and number had the strongest association with risk.
Prior reports have indicated that risk of metachronous advanced adenoma and cancer is increased with adenoma multiplicity9, 10, 27, 28, 29, 30, 31, villous features9, 29, 31, 32, high-grade dysplasia9, 27, 29, 31, 33, 34, and adenoma size5, 9, 28, 29, 30, 31, 34, 35. The presence of proximal adenomas has also been associated with a higher likelihood of metachronous advanced colorectal neoplasia in some studies5, 27, 32. However, reports from individual studies have been based on relatively small numbers of patients with metachronous advanced neoplasms and the estimates of risk have been imprecise. A report from one meta-analysis19 also indicated that prior adenoma number and high-grade dysplasia were predictive of subsequent advanced adenoma, but the investigators did not use subject-level data, and their conclusions were based on only four studies for adenoma number and two studies for high-grade dysplasia.
The large number of patients and endpoints in our pooling study allowed us to generate more precise risk estimates than have previously been reported and to perform multivariate analyses to assess independent contribution of various factors to risk of advanced neoplasia. In our data, both patient age and number of prior adenomas were strong predictors of risk; the proportion of patients who developed advanced neoplasia increased almost linearly with increases in these factors. Our multivariate analyses confirmed the independent importance of these two factors, as well as prior adenoma size and proximal location, in determining risk. We found that once other factors were accounted for, the presence of villous histology in a prior adenoma was only modestly associated with subsequent advanced neoplasia. We observed no independent association between family history and subsequent advanced colorectal neoplasia among post- polypectomy patients, unlike the stronger associations typically seen for family history in epidemiological studies of colorectal cancer incidence36. Further, there were no clear associations between metachronous advanced neoplasia and current cigarette smoking or obesity; thus, these factors seem unlikely to add much information in assessing risk.
In our univariate analyses, high-grade dysplasia was clearly associated with risk; however, this association was largely explicable by other factors (e.g., prior adenoma size, location, and villous features). We are uncertain why our results in this regard differ from those in some9,29, 34, 35, though not all27, 33, prior reports of an increased risk of metachronous advanced neoplasms associated with high-grade dysplasia. The apparent discordance in results may reflect differences in the extent to which other risk factors were taken into account in the various reports. Other possible explanations include differences in study populations or in the definition of high-grade dysplasia, which in our data was derived from reports produced by pathologists who were based in both academic and community hospital settings. Lastly, in our data only three percent of patients who had small, tubular adenomas also had high-grade dysplasia; the small number of these patients did not permit us to assess whether an isolated finding of high- grade dysplasia might be associated with increased risk.
Our pooled analyses were based on patients who were followed prospectively for four years on average, according to well-defined protocols, and with relatively few losses to follow-up, thereby lessening the likelihood that differential follow-up intensity could bias the associations we observed. However, eligibility criteria for all the studies included in our analyses required an adequate colonic preparation and a full examination to the cecum; thus we were unable to assess the potential effects of variations in the quality of baseline colonoscopy examinations. Colonoscopy quality may be an important factor in routine clinical practice, and missed and incompletely excised lesions are thought to be likely explanations for some of the advanced adenomas and cancers found during surveillance 11, 15, 16, 32, 37, 38. Missing data for some variables also posed a challenge in our analyses; although we used two different methods to address this issue, we cannot exclude the possibility of bias resulting from these approaches 39.
The goal of surveillance is to minimize (not eliminate) the risk of interval cancer. Current risk stratification guidelines3 are based on small individual studies. Our pooled data represent the largest cohort of patients who have had systematic surveillance colonoscopy. These data show relatively good discrimination between low and high-risk groups (6.9 vs. 15.5%) using current risk-stratification guidelines. Thus, our results strengthen the concept of risk stratification and should lead to improved physician adherence to these guidelines. Enhanced adherence would have an improved impact on saving resources from inappropriate intensive surveillance in the lower risk group, which constitutes approximately 75% of those under surveillance40. Nevertheless, the guidelines are not perfect; 7.4% of patients in our analyses who would be categorized as low-risk by the guidelines developed an advanced adenoma or invasive cancer during follow-up. Our data warrant consideration of additional risk factors, such as proximal location, sex, and age in the next iteration of the surveillance guidelines. Lastly, a formal prediction modeling analysis would be helpful in determining which combination of factors would maximally distinguish risk of recurrent advanced adenomas and cancers.
Results of our pooled data show that occurrence of colorectal neoplasia is common following colonoscopic polypectomy. Several patient characteristics and features of the prior adenoma are associated with advanced colorectal neoplasia at surveillance colonoscopy, the most important of these being the number and size of prior adenomas and patient age. Although current guidelines help stratify patients into high and low-risk groups, it may be possible to further refine estimates of risk for prediction risk modeling; our analyses point to a promising set of characteristics that might be considered in these analyses.
Acknowledgments
Work was supported by Public Health Service grants CA-41108, CA-23074, CA95060, CA37287, CA104869, CA23108, CA 59005, CA 26852 from the National Cancer Institute. Dr. Jacobs is supported by a KO7 Career Development Award (CA106269) from the National Cancer Institute. Funding for the Veteran’s Affairs Study was supported by the Cooperative Studies Program, Department of Veterans Affairs. We are indebted to all the patients, staff, and clinical investigators at the individual study sites. We thank Erin Ashbeck and Dr. Donna Spiegelman for their contributions in the preparation of the manuscript.
Footnotes
No conflicts of interest exist.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA A Cancer J Clin. 2008;58:71–96. doi: 10.3322/CA.2007.0010. [DOI] [PubMed] [Google Scholar]
- 2.Pignone M, Rich M, Teutsch SM, et al. Screening for colorectal cancer in adults at average risk: A summary of the evidence for the U.S. Preventive services task force. Ann Intern Med. 2002;137:132–141. doi: 10.7326/0003-4819-137-2-200207160-00015. [DOI] [PubMed] [Google Scholar]
- 3.Winawer SJ, Zauber AG, Fletcher RH, et al. Guidelines for colonoscopy surveillance after polypectomy: A consensus update by the US Multi-Society Task Force on Colorectal Cancer and the American Cancer Society. CA Cancer J Clin. 2006;56:143–159. doi: 10.3322/canjclin.56.3.143. quiz 184–145. [DOI] [PubMed] [Google Scholar]
- 4.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 A Cancer J Clin. 2008;58:130–160. doi: 10.3322/CA.2007.0018. [DOI] [PubMed] [Google Scholar]
- 5.Martinez ME, Sampliner R, Marshall JR, et al. Adenoma characteristics as risk factors for recurrence of advanced adenomas. Gastroenterology. 2001;120:1077–1083. doi: 10.1053/gast.2001.23247. [DOI] [PubMed] [Google Scholar]
- 6.Schatzkin A, Lanza E, Corle D, et al. Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas. Polyp Prevention Trial Study Group. N Engl J Med. 2000;342:1149–1155. doi: 10.1056/NEJM200004203421601. [DOI] [PubMed] [Google Scholar]
- 7.Baron JA, Cole BF, Sandler RS, et al. A randomized trial of aspirin to prevent colorectal adenomas. N Engl J Med. 2003;348:891–899. doi: 10.1056/NEJMoa021735. [DOI] [PubMed] [Google Scholar]
- 8.Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup [see comment citation in medline] N Engl J Med. 1993;329:1977–1981. doi: 10.1056/NEJM199312303292701. [DOI] [PubMed] [Google Scholar]
- 9.Lieberman DA, Weiss DG, Harford WV, et al. Five-year colon surveillance after screening colonoscopy. Gastroenterology. 2007;133:1077–1085. doi: 10.1053/j.gastro.2007.07.006. [DOI] [PubMed] [Google Scholar]
- 10.Winawer SJ, Zauber AG, O’Brien MJ, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. N Engl J Med. 1993;328:901–906. doi: 10.1056/NEJM199304013281301. [DOI] [PubMed] [Google Scholar]
- 11.Robertson DJ, Greenberg ER, Beach M, et al. Colorectal cancer in patients under close colonoscopic surveillance. Gastroenterology. 2005;129:34–41. doi: 10.1053/j.gastro.2005.05.012. [DOI] [PubMed] [Google Scholar]
- 12.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;342:1156–1162. doi: 10.1056/NEJM200004203421602. [DOI] [PubMed] [Google Scholar]
- 13.Bertagnolli MM, Eagle CJ, Zauber AG, et al. Celecoxib for the prevention of sporadic colorectal adenomas. N Engl J Med. 2006;355:873–884. doi: 10.1056/NEJMoa061355. [DOI] [PubMed] [Google Scholar]
- 14.Arber N, Eagle C, Spicak J, et al. Celecoxib for the prevention of colorectal adenomatous polyps. N Engl J Med. 2006;355:885–895. doi: 10.1056/NEJMoa061652. [DOI] [PubMed] [Google Scholar]
- 15.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;61:392–394. doi: 10.1016/s0016-5107(04)02765-8. [DOI] [PubMed] [Google Scholar]
- 16.Farrar WD, Sawhney M, Nelson DB, et al. Colorectal cancers found after a complete colonoscopy. Lancet. 2006;4:1259–1264. doi: 10.1016/j.cgh.2006.07.012. [DOI] [PubMed] [Google Scholar]
- 17.Lin OS. Clinical update: Postpolypectomy colonoscopy surveillance. The Lancet. 2007;370:1674–1676. doi: 10.1016/S0140-6736(07)61707-3. [DOI] [PubMed] [Google Scholar]
- 18.Imperiale TF, Sox HC. Gudelines for surveillance intervals after polypectomy: Coping with the evidence. Ann Intern Med. 2008;148:477–479. doi: 10.7326/0003-4819-148-6-200803180-00010. [DOI] [PubMed] [Google Scholar]
- 19.Saini SD, Kim HM, Schoenfeld P. Incidence of advanced adenomas at surveillance colonoscopy in patients with a personal history of colon adenomas: A meta-analysis and systematic review. Gastrointest Endosc. 2006;64:614–626. doi: 10.1016/j.gie.2006.06.057. [DOI] [PubMed] [Google Scholar]
- 20.Baron JA, Beach M, Mandel JS, et al. Calcium supplements for the prevention of colorectal adenomas. N Engl J Med. 1999;340:101–107. doi: 10.1056/NEJM199901143400204. [DOI] [PubMed] [Google Scholar]
- 21.Greenberg ER, Baron JA, Tosteson TD, et al. A clinical trial of antioxidant vitamins to prevent colorectal adenoma. N Engl J Med. 1994;331:141–147. doi: 10.1056/NEJM199407213310301. [DOI] [PubMed] [Google Scholar]
- 22.Alberts DS, Martinez ME, Hess LM, et al. Phase III trial of ursodeoxycholic acid to prevent colorectal adenoma recurrence. J Natl Cancer Inst. 2005;97:846–853. doi: 10.1093/jnci/dji144. [DOI] [PubMed] [Google Scholar]
- 23.Lieberman DA, Weiss DG, Bond JH, et al. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. Veterans Affairs Cooperative Study Group 380. N Engl J Med. 2000;343:162–168. doi: 10.1056/NEJM200007203430301. [DOI] [PubMed] [Google Scholar]
- 24.Greenberg ER, Sporn MB. Antioxidant vitamins, cancer, and cardiovascular disease. N Engl J Med. 1996;334:1189–1190. doi: 10.1056/NEJM199605023341810. [DOI] [PubMed] [Google Scholar]
- 25.Hosmer DW, Lemeshow S. Applied logistic regression. John Wiley & Sons, Inc; 2000. [Google Scholar]
- 26.DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–188. doi: 10.1016/0197-2456(86)90046-2. [DOI] [PubMed] [Google Scholar]
- 27.Bonithon-Koop C, Piard F, Fenger C, et al. Colorectal adenoma characteristics as predictors of recurrence. Dis Colon Rectum. 2004;47:323–333. doi: 10.1007/s10350-003-0054-1. [DOI] [PubMed] [Google Scholar]
- 28.Nusko G, Mansmann U, Kirchner T, et al. Risk related surveillance following colorectal polypectomy. Gut. 2002;51:424–428. doi: 10.1136/gut.51.3.424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Atkin WS, Morson BC, Cuzick J. Long-term risk of colorectal cancer after excision of rectosigmoid adenomas. N Engl J Med. 1992;326:658–662. doi: 10.1056/NEJM199203053261002. [DOI] [PubMed] [Google Scholar]
- 30.Noshirwani KC, van Stolk RU, Rybicki LA, et al. Adenoma size and number are predictive of adenoma recurrence: Implications for surveillance colonoscopy. Gastrointest Endosc. 2000;51:433–437. doi: 10.1016/s0016-5107(00)70444-5. [DOI] [PubMed] [Google Scholar]
- 31.Bertario L, Russo A, Sala P, et al. Predictors of metachronous colorectal neoplasms in sporadic adenoma patients. Int J Cancer. 2003;105:82–87. doi: 10.1002/ijc.11036. [DOI] [PubMed] [Google Scholar]
- 32.Loeve F, van Ballegooijen M, Boer R, et al. Colorectal cancer risk in adenoma patients: A nation-wide study. Int J Cancer. 2004;111:147–151. doi: 10.1002/ijc.20241. [DOI] [PubMed] [Google Scholar]
- 33.van Stolk RU, Beck GJ, Baron JA, et al. Adenoma characteristics at first colonoscopy as predictors of adenoma recurrence and characteristics at follow-up. The Polyp Prevention Study Group. Gastroenterology. 1998;115:13–18. doi: 10.1016/s0016-5085(98)70359-2. [DOI] [PubMed] [Google Scholar]
- 34.Yang G, Zheng W, Sun Q-R, et al. Pathologic features of initial adenomas as predictors for metachronous adenomas of the rectum. J Natl Cancer Inst. 1998;90:1661–1651. doi: 10.1093/jnci/90.21.1661. [DOI] [PubMed] [Google Scholar]
- 35.Bertario L, Russo A, Sala P, et al. Risk of colorectal cancer following colonoscopic polypectomy. Tumori. 1999;85:157–162. doi: 10.1177/030089169908500302. [DOI] [PubMed] [Google Scholar]
- 36.Butterworth AS, Higgins JPT, Pharoah P. Relative and absolute risk of colorectal cancer for individuals with a family history: A meta-analysis. Eur J Cancer. 2006;42:216–227. doi: 10.1016/j.ejca.2005.09.023. [DOI] [PubMed] [Google Scholar]
- 37.Barclay RL, Vicari JJ, Doughty AS, et al. Colonoscopic withdrawal times and adenoma detection during screening colonoscopy. N Engl J Med. 2006;355:2533–2541. doi: 10.1056/NEJMoa055498. [DOI] [PubMed] [Google Scholar]
- 38.Rex DK. Maximizing detection of adenomas and cancers during colonoscopy. Am J Gastroenterol. 2006;101:2866–2877. doi: 10.1111/j.1572-0241.2006.00905.x. [DOI] [PubMed] [Google Scholar]
- 39.Vach W, Blettner M. Logistic regression with incompletely observed categorical covariates - investigating the sensitivity against violation of the missing at random assumption. Stat in Med. 1995;14:1315–1329. doi: 10.1002/sim.4780141205. [DOI] [PubMed] [Google Scholar]
- 40.Mysliwiec PA, Brown ML, Klabunde CN, et al. Are physicians doing too much colonoscopy? A national survey of colorectal surveillance after polypectomy. Ann Intern Med. 2004;141:264–271. doi: 10.7326/0003-4819-141-4-200408170-00006. [DOI] [PubMed] [Google Scholar]