Results of Repeat Sigmoidoscopy 3 Years After a Negative Examination FREE

Limited evidence supports a recommended frequency of endoscopic colorectal cancer screening after a negative examination. One option recommended by consensus guidelines is a screening colonoscopy every 10 years.¹ The data supporting this interval stem largely from a case-control study of sigmoidoscopy² that found a higher rate of sigmoidoscopy in controls compared with distal colorectal cancer cases. In that study, the negative association of sigmoidoscopy with mortality due to cancer persisted for up to 10 years. In contrast, a Veteran's Administration case-control study of endoscopy use demonstrated an effect of endoscopy for only up to 6 years.^3- 4

The recommended interval for repeat screening flexible sigmoidoscopy (FSG) after a negative examination is 5 years.^1,5 A number of small case series demonstrate widely varying rates of abnormal findings on follow-up after a negative examination, up to 7% for detection of adenomas and 1% to 2% for detection of advanced adenomas after as little as 1 year.^6- 11 Only a few small colonoscopy studies have reported the results of repeat examination after a negative examination, and all comprise selected patient populations.^12- 14

Determining the incidence of adenoma after a negative examination is complicated by the inherent limitations in endoscopic detection of prevalent adenomas. Several studies using tandem colonoscopy, or colonoscopy by one practitioner followed immediately by another colonoscopy by a second, have shown that adenomas, especially small adenomas, are easily missed. Miss rates have ranged from 13% for adenomas smaller than 1 cm¹⁵ to 27% for adenomas smaller than 5 mm.¹⁶ In 1 tandem study, 6% of adenomas sized at least 1 cm were missed.¹⁶ Whether lesions are missed or represent new growth, an understanding of the expected yield from subsequent examinations is critical to establishing informed guidelines for follow-up. To our knowledge, no study has reported large-scale results of repeat screening after a negative examination.

The Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) is a randomized, controlled community-based study evaluating the effectiveness of cancer screening tests on site-specific cancer mortality. Sponsored by the National Cancer Institute, the trial is being conducted in 10 screening centers in the United States.¹⁷ For the colorectal cancer end point, the trial is evaluating the effect of 60-cm FSG on colorectal cancer mortality. In the initial trial design, sigmoidoscopy was performed in the intervention group at inception and again at 3 years. In keeping with prevailing practice, the interval between examinations was subsequently extended to 5 years. In this investigation, we examine the yield of a repeat FSG examination 3 years after an initial negative examination.

Participants in this study were taken from the intervention arm of the PLCO trial. Randomization began in November 1993 and was completed in July 2001, with more than 154 000 people aged 55 to 74 years enrolled. Those randomized before December 1995 received FSG at enrollment and 3 years later. The 10 PLCO centers are located in the following cities: Washington, DC; Detroit, Mich; Salt Lake City, Utah; Denver, Colo; Honolulu, Hawaii; Minneapolis, Minn; Marshfield, Wis; Pittsburgh, Pa; St Louis, Mo; and Birmingham, Ala. Participants in the intervention arm undergo periodic cancer screening tests, including chest radiograph, FSG, digital rectal examination and prostate-specific antigen screening (for men), and cancer antigen 125 screening and vaginal ultrasound (for women). A baseline questionnaire recorded personal sociodemographic characteristics, medical history, cancer screening history within the previous 3 years, and family history. Participants were recruited into the PLCO trial through mass mailings and met the following eligibility criteria: (1) age 55 to 74 years; (2) no current treatment for cancer except basal cell or squamous cell skin cancer; (3) no known prior cancer of the colon, rectum, prostate, lung, or ovaries; (4) no surgical removal of the colon, lung, ovary, or prostate; (5) no participation in another cancer screening or cancer prevention trial; (6) no finasteride use (in men) or no tamoxifen use (in women) in the past 6 months; (7) provision of informed consent; (8) no more than 1 prostate-specific antigen test in the past 3 years (for men randomized after April 1995); and (9) no colonoscopy, sigmoidoscopy, or barium enema in the past 3 years (for individuals randomized after April 1995).

Flexible sigmoidoscopy examinations were performed by trained nurses or certified physicians. Examiners used depth of insertion, adequacy of bowel preparation, and primary visual findings to place each sigmoidoscopy examination into 1 of 4 mutually exclusive hierarchical result categories. The abnormal suspicious result category signified any finding of polyp or mass, regardless of size or technical adequacy of the examination. The inadequate result category signified less than a 50-cm depth of insertion or estimated visualization by the examiner of less than 90% of the mucosal surface due to inadequate bowel preparation. The abnormal not suspicious result category signified any abnormality other than a mass or polyp (eg, hemorrhoids or diverticulosis) in a technically adequate examination. Finally, the negative result category signified a technically adequate examination without a polyp, mass, or incidental abnormality.

An FSG examination was considered positive if the examiner noted a polypoid lesion or mass. At sigmoidoscopy, the examiner recorded the location and shape and estimated the size of each of the 4 largest lesions. Lesions were usually not biopsied or removed. Individuals were referred to their personal physicians for evaluation of screening-detected abnormalities and were tracked to determine the results from subsequent diagnostic workup. Data on diagnostic follow-up with repeat FSG or colonoscopy were collected using trained medical record abstractors, who recorded the pathologic findings, size, and location of each lesion found. The anatomical site of lesions was recorded as rectum, sigmoid colon, descending colon, splenic flexure, transverse colon, hepatic flexure, ascending colon, or cecum, and, when available, the distance in centimeters to the lesion was documented.

For classification of abnormalities found on subsequent colonoscopy, a lesion in the rectum, sigmoid colon, or descending colon was considered distal. If the anatomical site was not recorded and the distance was, then a lesion at a distance of less than 50 cm into the colon was considered distal. Any lesion with a recorded anatomical location from the splenic flexure to the cecum or a lesion with no recorded location but a distance of at least 50 cm into the colon was considered proximal. Pathologic results were obtained from the local community pathologist report. An adenoma was defined as advanced if it contained villous features (villous or tubulovillous adenomas), was large (≥1 cm as estimated by the endoscopist), or had severe dysplasia. Carcinoma in situ was classified as severe dysplasia.

Individuals included in the current study were randomized to the screening intervention, had an initial FSG that showed no polypoid abnormality or mass, and, by protocol, underwent repeat screening FSG 3 years after the initial examination. Individuals with other abnormalities at initial FSG (eg, diverticulosis or hemorrhoids) are included, as are individuals with inadequate examinations due to inadequate preparation or depth of insertion of less than 50 cm.

To determine whether a lesion detected as a result of repeat sigmoidoscopy was due to increased depth of insertion or improved preparation at the second examination, the depth of insertion at the initial examination, the depth of insertion at the year 3 examination, and the size and location of the abnormal finding at the year 3 examination were recorded and compared.

Lesions were classified as detected possibly because of increased depth of insertion or improved preparation at the year 3 examination or detected in a previously examined area. All lesions found at the year 3 examination in participants with an inadequate examination at baseline due to poor preparation were considered detected because of improved preparation. When the depth of insertion was greater in year 3 than at baseline, the anatomical location of the lesion was reviewed. The colon was classified into the following regions: rectum, 0 to 10 cm; sigmoid, 10 to 30 cm; and descending colon/splenic flexure, 30 cm or more. If the lesion was in an anatomical region (eg, rectum, sigmoid) that was not completely examined (or not examined at all) at the initial examination, and the year 3 examination extended the coverage of that region, then the lesion was considered to be detected possibly because of increased depth of insertion. If the lesion was detected in a region of the colon that had been previously completely examined with an adequate preparation, then the lesion was considered to be detected in a previously examined area. For example, if the depth of insertion were 50 cm at baseline and 55 cm at the year 3 examination, and if a descending colon polyp were detected at year 3, then the polyp was considered as detected because of increased depth of insertion because more of the descending colon was examined at year 3. However, in the same example, if the polyp at the year 3 examination were detected in the sigmoid colon, then the lesion was considered detected in a previously examined area because at the baseline examination the sigmoid colon had, by definition, been fully examined.

Institutional review boards at each of the participating institutions approved the PLCO protocol and participants provided written informed consent. The publications subcommittee of the PLCO steering committee approved the publication of this article.

P values for frequency comparisons between 2 groups were generated based on the χ² test or the McNemar test when year 0 vs year 3 results on the same participants were compared. For all comparisons, P<.05 indicates statistical significance. Multivariable logistic regression was used to assess the effect of various factors on the likelihood of having any distal adenoma or advanced distal adenoma. Factors included in the model were sex, age, body mass index, education, smoking history, family history of colorectal cancer, previous endoscopy, and result of the baseline screening. Individuals with a positive year 3 screening examination who did not undergo diagnostic follow-up were included in overall results but were excluded from analyses for which distal adenoma status was required. Confidence intervals (CIs) for odds ratios (ORs) were calculated based on Wald statistics. SAS statistical software, version 8.2 (SAS Institute Inc, Cary, NC), was used for all analyses.

Of 11 583 individuals without a polypoid mass or lesion on initial FSG and eligible for repeat screening 3 years later, 9317 (80.4%) returned (Table 1). The mean (SD) age at randomization was 65.7 (4.0) years and 61.6% were men. Nearly 87% were white and 73% had education beyond high school. Only 5.9% were current smokers and 10.1% reported a family history of colorectal cancer in a first-degree relative. Slightly more than 40% self-reported undergoing an examination of the lower bowel with sigmoidoscopy, colonoscopy, or barium enema in the 3 years prior to study entry. At the initial FSG, 89.8% had insertion of the sigmoidoscope to 50 cm or beyond and 57.3% had findings such as internal hemorrhoids or diverticulosis. The characteristics of those who returned in comparison with the 2266 individuals (19.6%) who did not return were similar except that nonreturnees were moderately less educated (36.6% were college graduates or beyond vs 41.3% in returnees), more likely to be current or former smokers (56.1% vs 50.6% in returnees), and more likely to have had an inadequate FSG at baseline (22.6% vs 11.2% in returnees).

The results of the repeat FSG examination, 3 years after the initial examination, are shown in Table 2, and a schematic diagram of the outcome of the sample is shown in Figure 1. Of 9317 returnees, 1292 (13.9%) had a repeat FSG that was classified as abnormal suspicious, signifying that a polyp or mass was detected. Men were more likely than women to have a polyp or mass found at repeat screening (15.6% vs 11.1%; P<.001). Inadequate examinations comprised 12.3% of the repeat screens and were more frequent in women than in men (18.1% vs 8.6%; P<.001). Among men, the rate of inadequate examinations was similar at the 2 points (8.6% at repeat FSG vs 8.8% at initial FSG) but women had an increased rate of inadequate examinations at year 3 (18.1% at repeat FSG vs 14.9% at initial FSG; P<.001). The depth of insertion at the year 3 examination was similar to that at the baseline examination, with 86.7% having an insertion to 50 cm or beyond. Of those with an abnormal suspicious year 3 examination, 64.4% had 1 abnormal lesion identified, 19.7% had 2, 7.8% had 3, and 8.1% had 4 or more abnormal lesions found. The largest polyp size, estimated at the time of sigmoidoscopy by the examiner, was at least 1 cm in 5.5%, between 0.5 and 0.9 cm in 21.6%, and 0.5 cm or smaller in 73.0%.

Of the 1292 individuals with an abnormal suspicious year 3 screening sigmoidoscopy, diagnostic follow-up was available for 951 (73.6%). Eight hundred forty-seven underwent colonoscopy and 104 underwent repeat FSG (Figure 1). The findings in the distal colon on sigmoidoscopy or colonoscopy are presented in Table 3. Overall, 30.1% were found to have an adenomatous polyp in the distal colon and 0.6% had cancer in the distal colon (n = 6). Men were more likely than women to have adenomas (32.4% vs 25.3%; P = .03). A total of 22.1% had only hyperplastic polyps detected and 27.3% had no distal lesion identified at colonoscopy or on repeat FSG. Of the 286 who had a distal adenoma, 15.4% had an adenoma that was 1 cm or more in size, 14.0% had an adenoma with tubulovillous or villous histology, and 2.4% had an adenoma with high-grade dysplasia. Overall, 25.2% of individuals with a distal adenoma had an advanced distal adenoma.

The cumulative number of individuals yielding adenoma, advanced adenoma, or cancer in the distal and proximal colon is summarized in Table 4. In the distal colon, 292 (3.1%) of 9317 had an adenoma or cancer. The yield of an advanced adenoma or cancer in the distal colon was 78 (0.8%). Data in the proximal colon are available for 847 individuals who underwent colonoscopy after an abnormal year 3 FSG. The yield for adenomas and advanced adenomas in the proximal colon was lower than that for the distal colon. The cumulative yield for adenoma in the entire colon was 4.1%, with an advanced adenoma or cancer rate of 1.3%. Seven cancers were detected, of which 6 were in the distal colon.

A detailed review of individuals with advanced distal adenomas was performed to assess whether the lesions detected as a result of repeat sigmoidoscopy at year 3 were detected because of increased depth of insertion at the second examination or improved preparation, in contrast with detection in a portion of the colon that had been previously examined. Of 72 with advanced distal adenomas, only 14 (19.4%) of the diagnoses could possibly be attributed to either increased depth of insertion or better preparation at the second examination.

Multivariable logistic regression of the factors associated with detection of distal adenoma and distal advanced adenoma at 3-year follow up are presented in Table 5. Men (OR, 1.7; 95% CI, 1.3-2.2; P<.001) and current smokers (OR, 1.6; 95% CI, 1.0-2.5; P = .05) were more likely to have distal adenomas at repeat examination. Family history of colorectal cancer in a first-degree relative was associated with increased risk of advanced distal adenomas (OR, 2.0; 95% CI, 1.1-3.7; P = .02), as was an inadequate examination at baseline (OR, 2.7; 95% CI, 1.4-5.4; P = .004). A colon examination with sigmoidoscopy, colonoscopy, or barium enema within 3 years prior to study entry was associated with a decreased risk at the year 3 examination of any distal adenoma (OR, 0.7; 95% CI, 0.6-0.9; P = .01) and an advanced distal adenoma (OR, 0.4; 95% CI, 0.3-0.7; P = .002).

Table 6 presents a detailed listing of the 7 cancers detected as a result of the year 3 sigmoidoscopy. The detected cancers fall into 3 classifications: (1) cancer detected because the depth of insertion on repeat sigmoidoscopy extended beyond that of the initial sigmoidoscopy (case 3); (2) cancer detected because an abnormality was detected at the second sigmoidoscopy (a distal marker) in a region previously examined at the initial sigmoidoscopy, triggering a colonoscopy that led to detection of a cancer (cases 1 and 7); and (3) cancer detected in a region that had been previously examined at initial sigmoidoscopy (cases 2, 4, 5, and 6). Cancers in cases 1, 3, and 7 might have been detectable with a colonoscopy at the time of initial screening, although cases 3 and 7 had small lesions at the time of detection. Similarly, in cases 4, 5, and 6, the cancers were small and may represent lesions that grew to detectable size during the 3-year period between examinations, but alternative explanations, such as evolution from a flat adenoma or a missed lesion at initial sigmoidoscopy, cannot be excluded.

The incidence of adenoma or cancer in the distal colon 3 years after a negative FSG was 3.1%. One quarter of the individuals with adenomas had advanced adenomas, and 6 distal colon cancers were detected, for an advanced distal lesion detection rate of 0.8%. Although it cannot be determined whether these lesions were missed at the initial FSG or whether they developed over the 3-year observation period between examinations, 80.6% of those with advanced distal adenomas had lesions found in a portion of the colon that had been adequately examined at the initial sigmoidoscopy. Regardless of whether these were missed or new lesions, to our knowledge these data provide the first representative estimate of what can be expected on repeat examination 3 years after a negative FSG.

Screening examinations were performed in the context of the PLCO cancer screening trial, a national randomized trial with broad geographic representation. Participants were community-based volunteers who were asymptomatic at enrollment. By protocol, they underwent repeat screening 3 years later, so repeat examination was not dependent on clinical circumstances. Flexible sigmoidoscopy was performed by trained nurses, certified internists, or specialists. The quality of the examinations was good, as demonstrated by a depth of insertion of the sigmoidoscope to more than 50 cm in more than 90% of the population at the baseline examination and in more than 86% at the year 3 examination. The high rate of return for repeat screening¹⁸ and the high levels of patient satisfaction with sigmoidoscopy, when formally assessed,¹⁹ further attest to the standards with which examinations were performed. Preliminary data further suggest that nurses and physicians performed sigmoidoscopy with equivalent efficacy.²⁰

These data have great importance in determining optimal screening intervals and understanding the limitations of screening. The yield of cancer in the distal colon at the baseline PLCO examination for individuals enrolled in the same period as participants included in these analyses was 27 per 10 000 (data not shown) compared with a yield of cancer in the distal colon at the year 3 examination of 6 of 9317 (6.4/10 000); thus, the ratio of cancer yields at the 2 examinations was about 4.2:1.0. The yield of advanced distal adenomas at the baseline evaluation was 2.5% (data not shown) vs a yield of 0.8% at 3 years after a negative sigmoidoscopy, giving a ratio of advanced adenoma yield of 3.1:1.0. Thus, the yield for cancer and advanced adenomas 3 years after a negative sigmoidoscopy was one fourth to one third of what was detected at the initial examination 3 years prior.

Recent studies of screening colonoscopy suggest that 1% to 2% of individuals will have a proximal advanced adenoma or cancer detected by colonoscopy that would have been missed with FSG.^21- 22 The prevalence of advanced proximal neoplasia has fueled interest in screening colonoscopy. However, colonoscopy as a screening examination is expensive, and one of the variables that most influences its cost is the frequency of its use.²³ Strategies using colonoscopy as a primary screening mode have advocated its use less frequently (eg, once every 10 years).¹ Although some organizations have touted colonoscopy every 10 years as the "preferred" colorectal cancer screening strategy,²⁴ there are no observational data examining the yield of a second colonoscopy 10 years after a negative examination. Previous studies examining the yield after negative sigmoidoscopy or colonoscopy are limited by small, unrepresentative samples^{6- 9,12- 14} or variable follow-up based on clinical symptoms,^10,14 making it difficult to generalize the findings to colorectal cancer screening.¹¹ It is difficult to predict how colonoscopy would have performed had it been used instead of sigmoidoscopy. Colonoscopy would have detected additional individuals with advanced proximal neoplasia, some of which have not come to clinical presentation. Some of the distal cancers and advanced adenomas detected at the year 3 examination might have been detected at baseline with a longer initial depth of insertion or a better bowel preparation. However, our analysis suggests that the vast majority of the advanced lesions detected at year 3 would not have been detected at baseline, even if colonoscopy had been used, because more than 80% of these lesions were detected within reach of the initial sigmoidoscopy.

Our data are of concern because the 0.8% yield of advanced lesions on repeat sigmoidoscopy after a negative examination is only moderately reduced compared with the 1% to 2% yield of advanced proximal lesions that has encouraged the use of screening colonoscopy. Furthermore, the only factor associated with protection from detection of an advanced distal adenoma at 3-year follow-up (OR, 0.4) was a lower gastrointestinal tract examination within the 3 years prior to study entry. The implication is that more frequent examinations will detect and prevent subsequent advanced lesions and thereby lessen morbidity. If the yield on an interim examination after 3 years is relatively high, then a longer interval between examinations, such as 5 or 10 years, may result in more substantial morbidity and mortality. Studies of repeat examinations after longer time intervals are needed. Whether these findings are translatable to what would be found after a negative colonoscopy is not known. If they are, then individuals undergoing colonoscopy once every 10 years may be at increased risk for advanced findings if they wait 10 years for the next screening examination.

We also found an increased likelihood of advanced distal lesions with inadequate baseline examinations due to either poor preparation or less-than-optimal depth of insertion. Our data support consideration for colonoscopy in patients with inadequate sigmoidoscopy.

It is possible that patients might benefit from interval examinations with FSG as opposed to less frequent examinations with colonoscopy. This may be especially true for individuals between ages 50 and 65 years, whose risk of proximal neoplasia is lower than that of those older than 65 years.^25- 26 Alternatively, screening methods such as molecular stool testing^27- 28 or virtual colonoscopy²⁹ may be required during the long intervals between colonoscopic screening to improve effectiveness. In our sample, as in others,^21- 22 the prevalence rate for advanced lesions in the proximal colon was a fraction of that in the distal colon. In the screening colonoscopy study in the Veteran's Administration population, 80% of the cancers were in the distal colon.²¹ This further supports consideration of more frequent screening of the distal colon.

One limitation of this study is that participants were drawn from a cancer screening trial that is overrepresented by more educated, more affluent, and nonminority participants.³⁰ As a result, these data may not be applicable to specific minority or ethnic groups. Our estimates for findings in the distal colon are conservative because only 951 of 1292 people with an abnormality underwent follow-up diagnostic testing. However, 83.4% of the individuals who did not undergo follow-up testing had lesions of 5 mm or smaller detected at the repeat FSG and, thus, are at low risk of both advanced distal and proximal lesions.³¹ More than 27% of individuals with a positive sigmoidoscopy had no distal lesion on follow-up colonoscopy. These screens can be viewed as either false-positive sigmoidoscopy or false-negative colonoscopy examinations. Sigmoidoscopy positivity rates vary across examiners^32- 33 because of variability in examiner threshold for deciding whether a bump should be considered a polyp or not. For example, in a study of back-to-back FSG, 18.5% had a polyp identified at the second examination when none was identified at the first.³⁴ Similarly, studies of tandem colonoscopy show that lesions sized 5 mm or smaller can be missed up to 27% of the time.¹⁶ Thus, variability in findings at sigmoidoscopy or on follow-up colonoscopy after sigmoidoscopy, especially for small lesions, is expected.

In conclusion, our results show that 3 years after a negative FSG, there is a 0.8% incidence of advanced adenomas or cancer detectable in the distal colon. Although the overall percentage with detected abnormalities is modest, these data raise concern about the impact of a prolonged screening interval after a negative examination.