Five-Year Colon Surveillance After Screening Colonoscopy
Article Outline
Background & Aims: Outcomes of colon surveillance after colorectal cancer screening with colonoscopy are uncertain. We conducted a prospective study to measure incidence of advanced neoplasia in patients within 5.5 years of screening colonoscopy. Methods: Three thousand one hundred twenty-one asymptomatic subjects, age 50 to 75 years, had screening colonoscopy between 1994 and 1997 in the Department of Veterans Affairs. One thousand one hundred seventy-one subjects with neoplasia and 501 neoplasia-free controls were assigned to colonoscopic surveillance over 5 years. Cohorts were defined by baseline findings. Relative risks for advanced neoplasia within 5.5 years were calculated. Advanced neoplasia was defined as tubular adenoma greater than ≥10 mm, adenoma with villous histology, adenoma with high-grade dysplasia, or invasive cancer. Results: Eight hundred ninety-five (76.4%) patients with neoplasia and 298 subjects (59.5%) without neoplasia at baseline had colonoscopy within 5.5 years; 2.4% of patients with no neoplasia had interval advanced neoplasia. The relative risk in patients with baseline neoplasia was 1.92 (95% CI: 0.83–4.42) with 1 or 2 tubular adenomas <10 mm, 5.01 (95% CI: 2.10–11.96) with 3 or more tubular adenomas <10 mm, 6.40 (95% CI: 2.74–14.94) with tubular adenoma ≥10 mm, 6.05 (95% CI: 2.48–14.71) for villous adenoma, and 6.87 (95% CI: 2.61–18.07) for adenoma with high-grade dysplasia. Conclusions: There is a strong association between results of baseline screening colonoscopy and rate of serious incident lesions during 5.5 years of surveillance. Patients with 1 or 2 tubular adenomas less than 10 mm represent a low-risk group compared with other patients with colon neoplasia.
See editorial on page 1364.
Surveillance with colonoscopy is currently recommended after the detection and removal of colonic neoplasia (defined as nonmalignant adenoma or invasive cancer) because of the risk of newly discovered neoplasia during follow-up, some of which may be malignant.1, 2 In prior studies of adenoma-bearing patients, invasive cancer was found in 0.3%–1.0% within 3 years of baseline colonoscopy.3, 4, 5, 6, 7, 8, 9, 10 However, it is evident that most patients with adenomas will not benefit from surveillance; 30%–50% of adults will develop colon adenomas during their lifetime, and only 6% will develop colorectal cancer.11 Surveillance colonoscopy represents an important and costly component of any colon screening program12 and ideally should be targeted at patients most likely to develop colorectal cancer. Several studies have found an association between findings at the baseline colonoscopy, and subsequent pathology during surveillance, including number of adenomas and adenoma size ≥10 mm.13, 14, 15, 16
Current recommendations call for repeat screening colonoscopy at 10 years for patients with no neoplasia at baseline and surveillance at 5 years or more for patients with 1 or 2 small (<10 mm) tubular adenomas and at 3 years in patients with advanced neoplasia or more than 2 adenomas.1 Recommendations for surveillance intervals after removal of cancer or adenomas with high-grade dysplasia range from 1 to 3 years.1, 2, 17, 18 Few surveillance studies have been performed that include patients with negative baseline colonoscopy,19, 20, 21, 22, 23 and most studies excluded patients with very high-risk lesions.3, 4, 5, 6, 7, 8, 9, 10
In Veterans Affairs Cooperative Study24 No. 380, 3121 asymptomatic participants had screening colonoscopy, all visible polyps were removed, and patients were classified by their most advanced histologic lesion; 1171 (37.5%) individuals had neoplasia and were assigned to follow-up regimens over 5 years. A subset of participants with no neoplasia (n = 501) were assigned to 5-year follow-up, matched by age to patients with adenomas ≥10 mm.
The primary aims of this study were to (1) determine the 5.5-year cumulative incidence rate of advanced neoplasia in patients with and without neoplasia at the baseline screening colonoscopy and (2) determine whether there is an association between baseline endoscopic findings and subsequent risk of advanced neoplasia. This is the first large study of surveillance in patients following screening colonoscopy.
Materials and Methods
Setting and Study Participants
Participants were enrolled in 13 Veterans Affairs Medical Centers between February 1994 and January 1997.24 Centers were selected to achieve geographic and racial diversity. The study protocol was approved by a central Human Rights Committee and by institutional review boards at each participating center. Initial enrollment criteria included individuals aged 50 to 75 years who did not have lower gastrointestinal tract symptoms, prior history of colon disease, or structural examination of the colon within 10 years. Among patients who met the eligibility criteria, 1463 (31.4%) declined to participate, 3196 eligible patients were enrolled, and 3121 had complete colonoscopy examinations to the cecum. Procedures for performing the baseline colonoscopy and evaluation of histology have been described previously.24 Patients were classified by their most advanced histologic lesion at the baseline colonoscopy according to the following hierarchy: no neoplasia, hyperplastic polyp, 1 or 2 tubular adenomas <10 mm, 3 or more tubular adenomas <10 mm, tubular adenoma ≥10 mm, adenoma with villous histology (25% or more), adenoma with high-grade dysplasia, invasive cancer. If patients had examinations within 6 months of the baseline examination, the additional findings were included as part of the baseline findings; 6.0% of patients had more than 1 examination within the first 6 months that was considered part of the baseline examination. The baseline bowel preparation was described as good in 81.2% of patients, fair but adequate in 16.0%, and poor in 2.8%. At enrollment, interviews were conducted to measure purported risk factors for colon neoplasia, including family history of colorectal cancer, past and current tobacco use, and use of nonsteroidal anti-inflammatory drugs.
Interventions
Figure 1 describes the plan for colonoscopy surveillance. Protocols for follow-up were variable, depending on the baseline pathology. Patients with cancer or adenomas with high-grade dysplasia had follow-up based on physician decisions. Patients with villous adenomas or adenomas ≥10 mm were assigned to follow-up at 2 and 5 years after the baseline colonoscopy. Patients with tubular adenomas <10 mm were randomly assigned by concealed allocation to 1 of 2 groups: colonoscopy surveillance at 2 and 5 years or surveillance at 5 years only. Patients with adenomas <1 cm who did not consent to be randomized were followed, and all surveillance examinations were recorded. Five hundred one participants with no neoplasia at baseline were matched by age to patients with adenomas ≥10 mm and assigned to surveillance at 5 years.
Figure 1.
Diagram of study plan with enrollment in each arm of the study.
All pathology was reviewed locally and sent for blinded central pathology review. When there was a discrepancy, a third referee pathologist reviewed the material. Patients with a pathologic interpretation of intramucosal carcinoma or carcinoma in situ were classified in the high-grade dysplasia group. Cancer was defined as invasion of malignant cells beyond the muscularis mucosa. Patients were classified based on the most advanced histologic lesion.
Outcomes and Measurements
In this analysis, we report the cumulative “most advanced” histologic finding at colonoscopy for each patient based on the hierarchy described above. The finding of any advanced neoplasia (defined as tubular adenoma ≥10 mm; villous adenoma; adenoma with high-grade dysplasia; invasive cancer) was the primary end point. Follow-up was censored either at death, diagnosis of cancer or high-grade dysplasia, or at 5.5 years after the baseline examination. Patients assigned to surveillance at 2 years were considered adherent if they had an examination between 21 and 36 months after the baseline examination. Patients assigned to 5-year surveillance were considered adherent if they had an examination between 54 and 66 months of the baseline examination.
Data Analysis
The statistical analysis is based on standard methods of descriptive statistics, the statistical comparison of subgroups, and the assessment of relative risk. Rates and proportions were calculated for categorical data and means and standard deviations for continuous measures. In the analysis of risk factors, comparison subgroups were formed on the basis of several factors that included the following: surveillance status (with/without procedure); assignment status (assigned/not assigned to 5-year follow-up); randomization schedule in the <10 mm subgroup (2 + 5 year/5 year only); and, in the latter randomized study, adherence to assigned schedule (follow-up procedures performed exclusively in the assigned time window). Statistical comparisons for these factors in the respective subgroups of interest were based on t tests for continuous measures and χ2 tests for categorical variables. All P values are 2-sided. Results of colonoscopy surveillance, based on the timing of the first and subsequent surveillance examinations are presented (Table 3). The analysis of relative risk is based on standard log-binomial modeling, and estimates of relative risk, adjusted for age and family history, are provided along with 95% confidence intervals. This modeling approach evaluates risk in terms of relative risk, which is a preferred measure when response outcomes are common.25 For the evaluation of relative risk for the different categories of baseline findings (Table 4), the referent group is the “no neoplasia” (n = 298) subgroup. Other estimates of relative risk based number/size of polyps as provided in the Results section below include definitions of the referent group as appropriate. Data management and all statistical analyses were performed with SAS software (SAS Institute, Cary, NC).
Table 3. Advanced Neoplasia at Follow-up Colonoscopy Based on Timing of Colonoscopy
| 1st f/u <3 yrsa | 1st f/u 3–5.5 yrb | Repeat surveillancec | Cumulative resultd | |||||
|---|---|---|---|---|---|---|---|---|
| With Adv Neo | With Adv Neo | With Adv Neo | With Adv Neo | |||||
| n | n (%) | n | n (%) | n | n (%) | n | n (%) | |
| No neoplasia (n = 298) | 17 | 0 | 281 | 6 (2.1) | 7 | 1(14.3) | 298 | 7(2.4) |
| Tub Ad <10 mm (n = 622) | 314 | 14(4.5) | 308 | 18(5.8) | 213 | 6(2.8) | 622 | 38(6.1) |
| Tub Ad >10 mm (n = 123) | 102 | 9(8.8) | 21 | 0 | 60 | 10(16.7) | 123 | 19(15.5) |
| Villous (n = 81) | 61 | 8(13.1) | 20 | 2(10.0) | 39 | 3(7.7) | 81 | 13(16.1) |
| HGD (n = 46) | 41 | 5(12.2) | 5 | 0 | 35 | 3(8.6) | 46 | 8(17.4) |
| Cancer (n = 23) | 22 | 6(27.3) | 1 | 0 | 16 | 2(12.5) | 23 | 8(34.8) |
| Total | 557 | 636 | 370 | 1193 | ||||
aResults of patients who had their first follow-up colonoscopy <3 years after completion of the baseline colonoscopy. |
bResults of patients who had their first follow-up colonoscopy during years 3–5.5 after completion of the baseline colonoscopy. |
cThese results summarize the endoscopic results in patients who had a second or more surveillance examination(s). |
dCumulative result represents the most advanced lesion found on any colonoscopy during 5.5 years of follow-up. |
Table 4. Relative Risk of Advanced Neoplasia Within 5.5 Years Based on Baseline Finding
| Baseline finding (n with examination) | No advanced neoplasia, n (%) | Advanced neoplasia, n (%) | RRa | 95% CI | P value | Cancer n (%) | HGD/cancer per 1000 person-yr (95% CI) |
|---|---|---|---|---|---|---|---|
| No neoplasia (298) | 291(97.6) | 7(2.4) | 1.00 | 1(0.3) | 0.7(0–2.0) | ||
| Tub Ad <10 mm (622) | 584(93.9) | 38(6.1) | 2.56 | 1.16–5.67 | .02 | 4(0.6) | 1.5(0–2.9) |
| 1 or 2 (496) | 473(95.4) | 23(4.6) | 1.92 | 0.83–4.42 | .13 | 3 | 1.4(0–2.9) |
| >3 (126) | 111(88.1) | 15(11.9)b | 5.01 | 2.10–11.96 | <.001 | 1 | 1.9(0–5.5) |
| Tub Ad >10 mm (123) | 104(84.6) | 19(15.5) | 6.40 | (2.74–14.94) | <.001 | 1(0.8) | 6.4(0–13.5) |
| Villous adenoma (81) | 68(83.9) | 13(16.1) | 6.05 | (2.48–14.71) | <.001 | 1(1.2) | 6.2(0–14.7) |
| HGD (46) | 38(82.6) | 8(17.4) | 6.87 | (2.61–18.07) | <.001 | 2(4.4) | 26.0(3.2–48.8) |
| Cancer (23) | 15(65.2) | 8(34.8) | 13.56 | (5.54–33.18) | <.001 | 5(21.7) | 74.8(14.9–134.7) |
| Number of adenomasc at baseline (n) | |||||||
| 1 or 2 (617) | 577 | 40(6.5) | 7(1.1) | 3.3(1.2–5.5) | |||
| 3 or 4 (145) | 122 | 23(15.9) | 2(1.4) | 6.6(0.1–13.0) | |||
| 5–9 (64) | 53 | 11(17.2) | 3(4.7) | 13.1(0.0–27.9) | |||
| 10+ (8) | 7 | 1(12.5) | 0 | 0.0 |
aAdjusted for age and family history. |
bCompared with patients with 1 or 2 tubular adenomas <10 mm at baseline, patients with 3 or more had a higher rate of advanced neoplasia (P < .002). |
cAdenoma number, irrespective of size and histology; not included in the multivariate model. |
Results
One thousand one hundred seventy-one patients with neoplasia and 501 subjects with no neoplasia at baseline were scheduled to have at least 1 follow-up colonoscopy within 5.5 years (Figure 1). The rates of actual surveillance (defined as at least 1 examination within the 5.5 years of the baseline examination) are shown in Table 1 and Figure 1, and 76.4% of patients with neoplasia and 59.5% with no neoplasia had at least 1 surveillance examination within the 5.5 years after baseline. During surveillance, bowel preparation was described as good in 81.8% of patients, fair in 12.7%, and poor in 5.5%.
Table 1. Characteristics of Study Cohort
| Subjects at baseline | Number | Mean age, yr (SD) | Family history N (%) | Smoking history N (%) | NSAID use N (%) |
|---|---|---|---|---|---|
| No neoplasia (n = 501) | |||||
| With 5-yr procedure | |||||
| Yes | 298 | 63.4(7.0) | 71(23.8) | 228(76.5) | 239(80.2) |
| No | 203 | 63.6(7.0) | 40(19.7) | 149(73.4) | 154(75.9) |
| P value | .787 | .276 | .428 | .246 | |
| Not assigned to follow-up | 1449 | 62.3(7.3) | 136(9.4) | 1063(73.5) | 1130(78.0) |
| Assigned vs not assigned | |||||
| P value | .002 | .001 | .45 | .83 | |
| Tub Ad <10 mm(n = 842) | |||||
| With any 5-yr procedure | |||||
| Yes | 622 | 62.7(7.0) | 95(15.3) | 478(76.9) | 493(79.3) |
| No | 220 | 63.5(7.1) | 30(13.6) | 175(79.9) | 169(76.8) |
| P value | .12 | .56 | .35 | .45 | |
| Randomized assigned to | |||||
| 2 + 5 yr | 300 | 62.4(7.0) | 45(15.0) | 231(77.0) | 233(77.7) |
| 5 yr only | 294 | 62.6(7.2) | 42(14.3) | 229(77.9) | 222(75.5) |
| P value | .76 | .81 | .80 | .54 | |
| Randomized | |||||
| Adherent to protocol | 336 | 62.1(7.1) | 41(12.2) | 258(76.8) | 263(78.3) |
| Advanced neoplasia cohort | |||||
| Tub Ad >10 mm | 155 | 64.5(6.2) | 30(19.5) | 126(81.3) | 113(72.9) |
| Villous adenoma | 93 | 67.5(5.9) | 14(15.1) | 66(71.0) | 67(72.0) |
| Ad with HGD | 51 | 65.1(6.5) | 11(21.6) | 39(76.5) | 34(66.7) |
| Cancer | 30 | 66.1(5.3) | 7(23.3) | 20(66.7) | 23(76.7) |
The mean age (at baseline) and rate of positive family history of colorectal cancer in a first-degree relative are shown for each group in Table 1. Numerical differences in rate of family history were not statistically significant among patients in each baseline group assigned to surveillance. However, among patients with no neoplasia at baseline who had surveillance, positive family history was noted in 23.8% who received surveillance compared with 9.4% of participants not assigned to surveillance (P < .001). The reasons for no surveillance within 5.5 years are presented in Table 2. Death, comorbidity, and refusal were the most common reasons. Five patients with baseline colorectal cancer died within 39 months of the baseline examination with metastatic colorectal cancer. Other causes of death are unknown. We compared demographic factors (age, race) and possible risk factors for advanced neoplasia (family history, smoking, use of nonsteroidal anti-inflammatory drugs) to determine whether the surveillance cohort was similar to patients who did not receive surveillance. In the neoplasia group, the rate of active smoking was higher in patients who had no surveillance compared with those with surveillance (33.8% vs 21.7%, respectively, P < .001). There were no significant differences in the control group.
Table 2. Reasons for No Surveillance Within 5.5 Years of Baseline Colonoscopy
| Baseline group (n at baseline) | Death | Comorbid condition | Reason for no surveillance | Scheduled >5.5 yr | Refusal | Total n (%) | |
|---|---|---|---|---|---|---|---|
| Lost to follow-up | Left VA system | ||||||
| No neoplasia (n = 501) | 41 | 40 | 21 | 24 | 35 | 42 | 203(40.5) |
| Tubular adenoma <10 mm (n = 842) | 55 | 33 | 12 | 18 | 36 | 66 | 220(26.1) |
| Tubular adenoma ≥10 mm (n = 155) | 18 | 2 | 2 | 1 | 1 | 8 | 32(20.6) |
| Villous adenoma (n = 93) | 7 | 1 | 2 | 0 | 0 | 2 | 12(12.9) |
| High-grade dysplasia/cancer (n = 81) | 6 | 2 | 0 | 0 | 0 | 4 | 12(14.8) |
| Total, n (% of total) | 127 (26.5) | 78 (16.3) | 37 (7.7) | 43 (9.0) | 72 (15.0) | 122 (25.5) | 479(100) |
Surveillance results are summarized in Table 3, Table 4. In Table 3, each patient group is defined by the most advanced baseline finding. Results are presented based on the timing of the first and subsequent surveillance examinations: 557 (46.7%) patients had their first surveillance examination within 3 years of the baseline examination; 636 patients had their first surveillance examination between 3 and 5.5 years after baseline; and 370 patients had a second, third, or fourth surveillance examination within the study period. Among patients with advanced neoplasia at baseline, who had a first surveillance examination within 3 years, 12.4% (28 of 226) had advanced neoplasia discovered at the first surveillance examination. Most of these patients had repeat examinations within the study period. The repeat examinations revealed advanced neoplasia in 12.0% (18 of 150). Among patients with tubular adenomas <10 mm at baseline, 314 had their first surveillance in less than 3 years, and 308 patients had their first examination between 3 and 5.5 years. The rates of advanced neoplasia (4.5% and 5.8%, respectively, P = .43) are similar, irrespective of the timing of first surveillance. Among 213 patients with tubular adenomas <10 mm who had a second surveillance examination, 2.8% had advanced neoplasia.
The cumulative result represents the most advanced lesion found on any colonoscopy performed during the 5.5-year study period (Table 4). Among 298 patients with no neoplasia at baseline who had follow-up evaluation, 67 (22.5%) had small tubular adenomas (<10 mm), and 2.4% had advanced neoplasia, including 1 (0.3%) patient with cancer. One thousand four hundred forty-nine patients without neoplasia at screening colonoscopy were not enrolled in the surveillance study. Two of these patients were diagnosed with colorectal cancer during the study period (at 44 and 60 months).
Among 895 patients with neoplasia at baseline who had follow-up evaluation, 325 (36.3%) had small tubular adenomas, 66 (7.4%) had advanced adenomas, 7 (0.8%) had an adenoma with high-grade dysplasia, and 13 (1.5%) had invasive adenocarcinoma of the colon or rectum within 5.5 years of the baseline examination.
The relative risk for interval advanced neoplasia adjusted for age and family history are presented in Table 4. There is a progressive increase in risk, associated with the severity of the baseline findings. Only patients with 1 or 2 tubular adenomas <10 mm at baseline did not have a significant increase in risk compared with the neoplasia-free group (RR, 1.92; 95% CI: 0.83–4.42). Within the small adenoma group, patients with 1 or 2 adenomas had a lower risk of advanced neoplasia compared with patients with 3 or more baseline adenomas (4.6% vs 11.9%, respectively, P < .002). When patients with adenomas ≥10 mm were compared with those with adenomas <10 mm, there was a statistically significant difference in risk of interval advanced neoplasia (RR, 2.68; 95% CI: 1.48–4.84).
We compared 2 surveillance intervals among patients with tubular adenomas <10 mm at baseline. Three hundred patients were randomly assigned to surveillance at 2 + 5 years; 164 (54.7%) patients were adherent to the protocol. Adherence was defined as having had colonoscopy between 21 and 36 months and a second colonoscopy between 54 and 66 months after the baseline examination. One (0.6%) patient had advanced neoplasia at the first examination, and 8 (4.9%) patients had advanced neoplasia at 5.5 years. Two hundred ninety-four patients were assigned to follow-up at 5 years only: 172 (58.5%) were adherent. Adherence was defined as having a single colonoscopy between 54 and 66 months after the baseline examination. Eleven patients (6.4%) had advanced neoplasia at 5.5 years. There was no significant difference in 5.5-year outcome between the groups (P = .55). When we analyzed adherent and nonadherent patients who had any follow-up within 5.5 years (intention to treat), the rate of interval advanced neoplasia was 6.1% and did not differ among the groups.
The most serious outcome was the finding of invasive cancer or high-grade dysplasia. The rates of interval high-grade dysplasia or cancer per 1000 person-years of follow-up are shown in Table 4. The risk of high-grade dysplasia or cancer per 1000 person-years of follow-up was 0.7 with no neoplasia at baseline, 1.5 with tubular adenomas <10 mm, 6.4 with large tubular or villous adenomas, 26.0 with high-grade dysplasia, and 74.8 with cancer. We separately determined rates of incident high-grade dysplasia or cancer based on polyp number at the baseline colonoscopy, irrespective of size or histology. Rates per 1000 person-years were 3.3 with 1 or 2 adenomas, 6.6 with 3 or 4 adenomas, and 13.1 for 5–9 adenomas.
Figure 2 graphically shows the time course for each of 21 patients who developed interval cancer or high-grade dysplasia. Most of these serious lesions (n = 15) were discovered in the first 36 months after baseline colonoscopy. Four of the interval cancers were discovered within 12 months of the baseline examination, all in patients with either high-grade dysplasia or cancer at baseline. Three of the interval cancers were metastatic, and 3 were advanced (T3N0). Three patients with cancers at baseline had interval cancers in the same or adjacent region of the colon as the baseline finding. Seven (33%) of the interval cancers or adenomas with high-grade dysplasia were found in the rectum; 3 (15%) in the sigmoid, descending colon, or splenic flexure; 3 (15%) in the transverse colon; and 8 (40%) in the hepatic flexure, ascending colon, or cecum.
Figure 2.
Interval cancer and adenoma with high-grade dysplasia: time of diagnosis and interval colonoscopy examinations. Legend: The interval between baseline colonoscopy and discovery of new cancer or high-grade dysplasia is graphically depicted in a time graph. Interval colonoscopy examinations are noted when they were performed. Footnote: Six patients who had cancer at baseline developed interval cancer or high-grade dysplasia. The location and stage of the baseline cancers are included.
Four major adverse events (in 1767 examinations) occurred after surveillance colonoscopy. One patient had a colon perforation requiring surgery (n = 1). Three patients had lower gastrointestinal bleeding after polypectomy, resulting in surgery (n = 1), transfusion (n = 1), and hospitalization with endoscopic therapy of polypectomy site (n = 1).
Discussion
This is the first large colon surveillance study in asymptomatic patients who had colorectal cancer screening with colonoscopy. The incidence rate of interval neoplasia after colonoscopy can inform the development of recommendations for appropriate intervals for surveillance.
Our results are consistent with data from other clinical trials in which follow-up colonoscopy was performed. Our study included a cohort of 298 patients who had no neoplasia at baseline screening colonoscopy, representing one of the largest such cohorts to receive follow-up colonoscopy. The rate of any adenoma was 24.8%; the rate of advanced adenoma was 2.4%, including 1 (0.3%) cancer. This cohort provides a unique comparative group to the cohort with neoplasia at baseline. In prior case series, 1129 polyp-free subjects had colonoscopy within 5.5 years of a baseline examination.19, 20, 21, 22, 23 The rates of any adenoma were 16%–41%, and the rates of advanced adenoma were 0.6%–3.4%, which are similar to rates in our study. In patients with neoplasia at baseline, we found incident adenomas in 45.9% and advanced adenomas in 9.6%. These results are similar to other studies in which baseline clearing colonoscopy was performed, and follow-up occurred within 2 to 6 years, with reported interval adenomas in 36%–52% and advanced adenomas in 6.6%–11.3%.3, 4, 5, 6, 7, 8, 9, 10 Our study finds a close relationship between the findings at baseline screening colonoscopy and the relative risk of advanced neoplasia during 5.5 years of follow-up.
Patients with tubular adenomas <10 mm represent the largest proportion of patients discovered to have neoplasia at screening colonoscopy24, 26 and may represent a low-risk group.15, 27 We find that patients with only 1 or 2 tubular adenomas <10 mm at baseline did not have a statistically significant higher rate of interval advanced neoplasia compared with subjects who had no neoplasia at baseline colonoscopy (Table 4). However, the confidence intervals are wide, and this lack of difference may be due to type II error. It is also possible that the true rate of advanced neoplasia in the control group is lower than observed because of self-selection bias. If this were true, the difference between controls and subjects with adenoma <10 mm might be greater. We find that the absolute risk of interval advanced neoplasia, within 5.5 years, in patients with 1 or 2 tubular adenomas at baseline colonoscopy (4.6%) is low compared with patients with advanced neoplasia at baseline (4.6% vs 17.6%, respectively, P < .001). Patients with 3 or more tubular adenomas <10 mm have a higher risk of advanced adenomas compared with patients who had no neoplasia at baseline and to patients who had only 1 or 2 small (<10 mm) tubular adenomas at baseline. Several other studies have made similar observations.13, 14
When our study was initiated, the recommendations for surveillance interval for patients with adenomas less than 10 mm varied from 1 to 5 years after the baseline examination. We tried to compare 2 surveillance programs (2 + 5 years vs 5 years only) by randomly assigning patients with tubular adenomas to one of the regimens, in a manner similar to the National Polyp Study.13 Adherence to the randomization scheme was poor, at less than 60%. Nevertheless, we did not find any significant benefit of early surveillance among adherent subjects.
Our study finds that individuals with either 3 or more adenomas, tubular adenoma ≥10 mm, villous adenoma, or adenoma with high-grade dysplasia at a baseline screening colonoscopy have a similarly higher risk of advanced neoplasia within 5 years compared with patients with no polyps or 1 or 2 small (<10 mm) tubular adenomas. Other reports have suggested that adenoma size ≥10 mm may be associated with a high risk of incident advanced neoplasia.14, 16 Our study also suggests that villous histology may be an important risk factor, confirming an association found in prior sigmoidoscopy studies.15, 28 However, in most cases (74%), the villous adenomas were ≥10 mm. The rates of advanced neoplasia were high in patients who had their first surveillance in less than 3 years after baseline colonoscopy (13.7%) and remained high (11.9%) in patients who had a second surveillance examination. These data suggest that this is indeed a high-risk group. Our results provide support for the current guideline recommendation for surveillance at 3 years and suggest that continued surveillance will be needed in these patients.
We find that patients with invasive cancer at baseline colonoscopy have a very high risk of interval advanced neoplasia during surveillance. Few studies have performed systematic follow-up of patients after curative resection of colorectal cancer.29, 30 Nava and Pagana30 followed 240 patients for 4 years after curative resection of colorectal cancer. They identified 28 (11.7%) patients with cancer during the follow-up. In our high-risk group (baseline high-grade dysplasia or cancer), 11 (15.9%) patients had interval cancer or high-grade dysplasia, 7 of whom were diagnosed within 15 months of the baseline examination (Figure 2).
Our results can be compared with recent chemoprevention studies,4, 5, 6, 7, 8, 9, 10 in which adenoma-bearing patients had follow-up colonoscopy within 3 years after a “clearing” colonoscopy. Interval rates of cancer were 0.3%–0.9% and ranged from 1.7 to 2.4 per 1000 person-years of follow-up. In our study, the rate of interval cancers was 1.7 per 1000 person-years of follow-up among patients with adenomas at baseline.
Many of the interval cancers or adenomas with high-grade dysplasia (15 of 21) were discovered in the first 36 months after baseline colonoscopy, raising issues about the quality of the colonoscopy. A recent analysis of the Polyp Prevention Trial data suggested that some of the interval cancers most likely represented growth of incompletely removed lesions.31 In our study, 6 of 11 patients with baseline cancer or adenoma with high-grade dysplasia had recurrent cancers or high-grade dysplasia in a similar portion of the colon, suggesting the possibility of incomplete removal. Eight patients with adenomas at baseline had interval cancers, 2 of which were in the same portion of the colon as the resected polyp.
It is quite possible that important lesions were not detected at the baseline colonoscopy. Recent large screening studies using computerized tomography colonography have found that colonoscopy may miss 2%–12% of polyps greater than 9 mm.32, 33, 34 Pickhardt et al35 reported that many of the missed adenomatous polyps were behind folds or in the rectum. In our study, 7 (33%) interval cancers or adenomas with high-grade dysplasia were in the rectum. Based on the size, location, and short time interval between baseline examination and newly discovered lesion, it is likely that some of these serious lesions were present but not detected at the baseline examination. These data emphasize the importance of quality in the performance of colonoscopy.36
There are several important limitations of this study. The study cohort was largely limited to men, so these results may not be generalized to surveillance in women. Twenty-two percent of patients with neoplasia at baseline colonoscopy did not have a follow-up colonoscopy within 5 years. Some loss of follow-up was inevitable because of death or intervening illness in this aging population. Although our analysis did not reveal any systematic bias, patients accepting follow-up colonoscopy may differ in important ways from patients who refused. Most patients with baseline advanced neoplasia had colonoscopy within 3 years (Table 3). Detection and removal of small adenomas during early examinations could have reduced the incidence of advanced neoplasia during the surveillance period in these groups. Therefore, our results may underestimate the 5-year rate of advanced neoplasia in patients who received early colonoscopy. Patients censored at death may have had an undetected cancer or advanced neoplasia. Finally, our study was not designed to determine whether nonsteroidal anti-inflammatory drugs, vitamins, tobacco use, or diet would modify the risk of newly discovered advanced neoplasia during surveillance.
In conclusion, these data provide a basis for risk stratification based on findings at the baseline screening colonoscopy. Patients who have cancer detected at baseline colonoscopy need intense follow-up after curative resection to establish that all important lesions were discovered at baseline, that they were completely removed, and that they do not develop new important pathology in a short interval. Our data provide evidence in support of new guidelines2 that recommend repeating colonoscopy within 1 year if cancer is detected at baseline, even if a complete examination was performed at the time of resection. We find that patients with adenomas with high-grade dysplasia, villous histology, size ≥10 mm, or ≥3 tubular adenomas have an intermediate risk of advanced neoplasia during surveillance. The new guidelines call for surveillance at 3 years in these patients, if the initial examination was complete and adequate. Patients with 1 or 2 small adenomas <10 mm have a lower risk of interval advanced neoplasia within 5 years compared with patients with advanced neoplasia. We found no benefit for early surveillance before 5 years in this group. Surveillance is now recommended at 5–10 years in this group.1
We also conclude that it is very likely that important lesions were missed or incompletely removed at the baseline colonoscopy and then discovered during surveillance, even though the examinations were performed by experts with extensive colonoscopic experience. This may represent an important limitation of optical colonoscopy as currently performed. Future advances in optics or the use of chromoendoscopy may enhance the ability to identify neoplasia on the first examination, although this is yet to be proven to improve clinical outcomes. Finally, study is needed to determine whether the risk of advanced neoplasia during surveillance can be modified by chemoprevention.
The authors thank the following: Data Monitoring Board: B. Levin (chairman), C. R. Boland, M. Brown, R. Burt, R. B. D’Agostino, and D.K. Rex. Executive Committee: A. Schatzkin (Bethesda, Maryland), J. F. Collins (Perry Point, Maryland), H. Garewal (Tucson, Arizona), S. Prindiville (Bethesda, Maryland). Planning Committee: J. Selby and C. Quesnberry (Oakland, California). Veterans Affairs Cooperative Program Office: J. R. Feussner, D. Deykin, and P. Huang. Staff of the study chairman’s office: T. Johnston, M. Sutton, and M. Fouts. Staff at the Veterans Affairs Cooperative Studies Program Coordinating Center: B. Calvert, Dr. J. Collins, C. Crigler, M. Lee, M. Rhoades, and E. Spence.
References
- Guidelines for colonoscopy surveillance after polypectomy: a consensus update by the US Multi-Society Task Force on colorectal cancer and the American Cancer Society. Gastroenterology. 2006;130:1872–1885
- Guidelines for colonoscopy surveillance after cancer resection: a consensus update by the American Cancer Society and the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology. 2006;130:1865–1871
- . The Funen Adenoma Follow-up Study. Scand J Gastroenterol. 1993;28:869–874
- Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas. N Engl J Med. 2000;342:1149–1155
- Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas. N Engl J Med. 2000;342:1156–1162
- A randomized trial of aspirin to prevent colorectal adenomas. N Engl J Med. 2003;348:891–899
- Colorectal cancer in patients under close colonoscopic surveillance. Gastroenterology. 2005;129:34–41
- Celecoxib for the prevention of sporadic colorectal adenomas. N Engl J Med. 2006;355:873–884
- Celecoxib for the prevention of colorectal adenomatous polyps. N Engl J Med. 2006;355:885–895
- A randomized trial of rofecoxib for chemoprevention of colorectal adenomas. Gastroenterology. 2006;131:1674–1682
- Colorectal cancer screening and surveillance: clinical guidelines and rationale—update based on new evidence. Gastroenterology. 2003;124:544–560
- . Cost-effectiveness model for colon cancer screening. Gastroenterology. 1995;109:1781–1790
- Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. N Engl J Med. 1993;328:901–906
- Adenoma size and number are predictive of adenoma recurrence: implications for surveillance colonoscopy. Gastrointest Endosc. 2000;51:433–437
- . Long-term risk of colorectal cancer after excision of rectosigmoid adenomas. N Engl J Med. 1992;326:658–662
- Adenoma characteristics as risk factors for recurrence of advanced adenomas. Gastroenterology. 2001;120:1077–1083
- Recommended colorectal cancer surveillance guidelines by the American Society of Clinical Oncology. J Clin Oncol. 1999;17:1312
- . American Cancer Society guidelines for the early detection of cancer, 2005. CA Cancer J Clin. 2005;55:31–44
- A normal initial colonoscopy after age 50 does not predict a polyp-free status for life. Am J Gastroenterol. 1994;89:1156–1159
- Incidence and recurrence rates of colorectal adenomas: a prospective study. Gastroenterology. 1995;108:402–408
- Five-year incidence of adenomas after negative colonoscopy in asymptomatic average-risk persons. Gastroenterology. 1996;111:1178–1181
- New occurrence and recurrence of neoplasms within 5 years of a screening colonoscopy. Am J Gastroenterol. 2002;97:1524–1529
- Five-year risk of colorectal neoplasia among persons with no neoplasia on baseline screening colonoscopy. Gastroenterology. 2004;126:A-36
- Use of colonoscopy to screen asymtomatic adults for colorectal cancer. N Engl J Med. 2000;343:162–168
- Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol. 2003;157:940–943
- Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N Engl J Med. 2000;343:169–174
- Treatment of small colorectal polyps: a population-based study of the risk of subsequent carcinoma. Mayo Clin Proc. 1984;59:305–310
- Pathologic features of initial adenomas as predictors for metachronous adenomas of the rectum. J Natl Cancer Inst. 1998;90:1661–1665
- Cumulative incidence of metachronous colorectal cancer. Dis Colon Rectum. 1993;6:388–393
- . Postoperative surveillance of colorectal carcinoma. Cancer. 1982;49:1043–1047
- Analysis of colorectal cancer occurrence during surveillance colonoscopy in the dietary polyp prevention trial. Gastrointest Endosc. 2005;61:385–391
- Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med. 2003;349:2191–2200
- Computed tomographic colonography (virtual colonoscopy): a multicenter comparison with standard colonoscopy for detection of colorectal neoplasms. JAMA. 2004;291:1713–1719
- Analysis of air contrast barium enema, computed tomographic colonography and colonoscopy: prospective comparison. Lancet. 2005;365:305–311
- The adenoma miss rate at optical colonoscopy: a novel assessment using virtual colonoscopy as a separate reference standard. Ann Intern Med. 2004;141:352–359
- Quality in the technical performance of colonoscopy and the continuous quality improvement process for colonoscopy: recommendations of the US Multi-Society task force on colorectal cancer. Am J Gastroenterology. 2002;97:1296–1308
Supported by the Department of Veterans Affairs Cooperative Study Program (No. 380) and The Department of Veterans Affairs with statistical support for the study (to D.G.W.)
The Department of Veterans Affairs approved final submission of the manuscript.
Conflicts of interest: None relevant to this manuscript.
PII: S0016-5085(07)01306-6
doi:10.1053/j.gastro.2007.07.006
© 2007 AGA Institute. Published by Elsevier Inc. All rights reserved.
Refers to article:
- Toward Risk Stratification for Screening and Surveillance of Colorectal Neoplasia: One Small Step for the Colonoscopist
