|Home | About | Journals | Submit | Contact Us | Français|
Among randomized trials evaluating flexible sigmoidoscopy (FSG) for its effect on colorectal cancer mortality, only the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial screened its participants more than one time. We report outcomes from the PLCO screening FSG program and evaluate the increased yield produced by a second FSG.
Participants were screened by 60-cm FSG in 10 regional screening centers at study entry and 3 or 5 years later, depending on the time of random assignment. Results from subsequent diagnostic intervention were tracked and recorded in a standardized fashion, and outcomes were compared according to sex and age. The protocol discouraged repeat FSG in persons with colorectal cancer or adenoma diagnosed after the initial FSG.
Of 77447 enrollees, 67073 (86.6%) had at least one FSG and 39443 (50.9%) had two FSGs. Diagnostic intervention occurred in 74.9% after a positive first FSG and in 78.7% after a positive repeat FSG. The second FSG increased the screening yield by 32%: Colorectal cancer or advanced adenoma was detected in 37.8 per 1000 persons after first screening and in 49.8 per 1000 persons after all screenings. The second FSG increased the yield of cancer or advanced adenoma by 26% in women and by 34% in men. Of 223 subjects who received a diagnosis of colorectal carcinoma within 1 year of a positive FSG, 64.6% had stage I and 17.5% had stage II disease.
Repeat FSG increased the detection of colorectal cancer or advanced adenoma in women by one-fourth and in men by one-third. Screen-detected carcinomas were early stage (stage I or II) in greater than 80% of screened persons. Colorectal cancer mortality data from the PLCO, as the definitive endpoint, will follow in later publications.
Limited empirical data suggest that repeated colonoscopic screening detects a greater number of colorectal adenomas and carcinomas than a single screen alone. Among the goals of the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial is to determine whether repeated flexible sigmoidoscopy (FSG) screenings are beneficial and at what time interval.
Study participants were screened by FSG at study entry and 3 or 5 years later. Of 67073 screened enrollees, 27630 (41.2%) received only one FSG and 39443 (58.8%) were rescreened at 3 or 5 years.
Overall 11867 (74.9%) of 15839 participants who had a positive first screen and 7184 (78.7%) of 9133 participants who had a positive second screen received a diagnostic intervention. Colorectal cancer or advanced adenoma was found in 3.78% of persons after the first screen and in 4.98% of persons after both screens, such that the second FSG increased the screening yield by 32%. There were 177 colorectal carcinomas detected by the first screen and 46 more by the second screen; 183 (82.1%) of 223 were early (stage I or II).
The data indicate that repeated screening by FSG increases the yield of detected colorectal adenomas and carcinomas.
Data regarding the effect of repeated FSG screening on colorectal cancer incidence and mortality are not yet available. Data regarding the completion rate of a second FSG does not differentiate between ineligibility, unavailability, and nonadherence. Yields from repeated FSG may not reflect potential yields from repeated colonoscopy. Analysis of the relative clinical outcomes when screening was repeated at a 3-year interval vs a 5-year interval will be reported separately.
From the Editors
Endoscopic methods detect colorectal cancer and adenoma with greater sensitivity than fecal occult blood testing (1,2). However, endoscopy is expensive. Because the cost-effectiveness of endoscopic colorectal cancer screening depends strongly on the age groups screened and on the frequency of screening (3), it is important to determine whether repeated screenings are beneficial and at what time interval they should be conducted. It is possible that protection from the first endoscopic screen may be sustained over time, thereby diminishing the importance of repeat screenings. Alternatively, protection from the first endoscopic screen may wane rapidly enough to justify relatively frequent repeat screenings.
Limited empirical data are available to guide clinicians and policy makers about the efficacy of repeated endoscopic screening, particularly in persons with a history of negative screenings (4). Colonoscopies that are repeated 5–6 years after a negative colonoscopy detect advanced adenomas in 1.3%–4.5% of individuals (5–7). In the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial, screening flexible sigmoidoscopy (FSG) that was repeated 3 years after a negative screen detected advanced adenoma or cancer in 0.8% of individuals (8).
Several recent studies challenge the notion of the superior value of colonoscopy vs sigmoidoscopy in terms of cancer prevention. In case–control studies, a negative result from sigmoidoscopy has been associated with decreased colorectal cancer mortality risk for 6 to 10 years or longer (9–11). In a population-based study from Manitoba, persons who received a negative result from colonoscopy had reduced colorectal cancer incidence for up to 10 years (12). There were two relevant population-based studies from Ontario. In a case–control study that included patients who died of colorectal cancer and matched control subjects who did not (13), colonoscopy was associated with a greater reduction in cancer mortality in the distal colon than in the proximal colon. In a cohort study that compared individuals who had received a negative colonoscopy with a control group of individuals who had not received colonoscopy (14), there was a greater reduction in the incidence of tumors in the distal colon than in the proximal colon. The United Kingdom Flexible Sigmoidoscopy Trial observed a 36% lower incidence of distal (rectum and sigmoid) colorectal cancer, but a statistically insignificantly (2%) lower incidence of proximal colorectal cancer, in persons aged 55–64 years who were randomly assigned to a one-time FSG intervention group compared with a noncontact control group (15). These studies provide strong justification for sigmoidoscopy screening as a means to protect against distal colorectal cancer and challenge the view that colonoscopy offers an incremental mortality or morbidity benefit relative to FSG.
Randomized studies in Norway (16,17), the United Kingdom (18), and Italy (19) are currently evaluating one-time FSG as a method to prevent colorectal cancer incidence or mortality. In the United States, the PLCO cancer screening trial is evaluating a program that includes two FSG examination cycles, with the second examination offered 3–5 years after the first. In 2005, the PLCO investigators reported outcomes from the first FSG examination cycle (20). Herein, we report the programmatic yield (ie, the yield achieved) from the overall PLCO screening effort, with a focus on contributions from a second FSG screening.
The PLCO Cancer Screening Trial (ClinicalTrials.gov Identifier: NCT00002540, http://clinicaltrials.gov/ct2/show/NCT00002540), a multicenter randomized clinical trial sponsored by the National Cancer Institute, is testing the effectiveness of early prostate, lung, colorectal, and ovarian cancer detection with 1) digital rectal examination and blood testing for prostate-specific antigen (PSA); 2) chest x-ray; 3) 60-cm FSG; and 4) transvaginal ultrasound, blood CA-125 testing, and physical examination vs usual care. Ten PLCO screening centers contribute data (Birmingham, AL; Denver, CO; Detroit, MI; Honolulu, HI; Marshfield, WI; Minneapolis, MN; Pittsburgh, PA; Salt Lake City, UT; St Louis, MO; and Washington, DC). Detailed descriptions of recruitment methods, eligibility criteria, and data collection activities have been reported (21). Screening centers obtained written informed consent from each subject, and institutional review boards at each of the participating institutions approved the PLCO research protocol.
Enrollment began on November 8, 1993, and ended on July 2, 2001. On January 1, 1996, PLCO changed the age requirement for eligibility from 60–74 to 55–74 years (22). Subject eligibility criteria, reported in detail earlier (20), excluded persons who reported any history of colorectal cancer and persons who reported any lower gastrointestinal procedure (proctoscopy, sigmoidoscopy, barium enema, or colonoscopy) in the 3 years before study enrollment. Most study subjects were recruited by mailing informational brochures and letters of invitation to conveniently available age-eligible persons who were identified on public, commercial, or screening center–specific mailing lists that were not to our knowledge directed disproportionately to any colorectal cancer risk group. A baseline questionnaire, administered at study entry, recorded personal sociodemographic characteristics (age, race, sex, marital status, and education), cancer family history, personal medical history (including history of colorectal polyp), cigarette smoking history, and cancer screening history within 3 years.
Participants in the PLCO trial were initially offered an FSG at study entry and a repeat FSG 3 years later. A protocol modification (implemented on December 7, 1998) changed the follow-up interval to 5 years. To accommodate appropriate surveillance colonoscopy, the PLCO protocol discouraged repeat screening in persons with colorectal cancer or adenoma diagnosed after study entry.
Physician and nurse examiners, all centrally registered, followed standardized procedures to perform and record results from the 60-cm FSG examination. Examiners used depth of insertion, adequacy of bowel preparation, and primary visual findings to assign each sigmoidoscopy examination to one of three mutually exclusive categories as results. A result was defined as positive by any finding of a polyp or mass. Results were considered to be inadequate if there was less than 50 cm depth of insertion or if visual inspection was limited to less than 90% of the mucosal surface due to inadequate bowel preparation, without detection of a polyp or mass. Finally, results were considered to be negative if a technically adequate examination did not detect a polyp or mass. Subjects who were given a technically inadequate sigmoidoscopy due to incomplete bowel preparation could return at a later date for a second procedure. In these instances, we classified subjects according to the second examination in our analyses.
According to the PLCO trial protocol, subjects with screen-detected abnormalities were referred to their personal physicians for a diagnostic intervention. The PLCO trial staff communicated sigmoidoscopy test results to participants at the end of a screening encounter. Later, letters and reports were mailed to each participant and his or her personal physician of record to communicate all PLCO test results, including detailed sigmoidoscopy findings, and to reinforce recommendations for follow-up. The PLCO investigators did not formulate, as a matter of protocol, study-wide standards or guidelines for the diagnostic evaluation of screen-detected abnormalities. However, study participants and their personal physicians were granted access to screening center consultants who could respond to questions and deliver advice regarding diagnostic approaches. The PLCO tracked subjects for at least 12 months after each FSG screening and identified, collected, and abstracted medical records pertaining to subsequent diagnostic work-ups. Information abstracted from medical records included the occurrence and date of each diagnostic FSG and/or colonoscopy intervention and the anatomical location, size (as recorded on clinical endoscopy reports), and histology of removed polyps and masses. Dates of diagnosis, TNM clinical stages, and TNM pathological stages were collected for subjects with invasive colorectal cancer.
Categories were defined for the scoring of each screen-detected colorectal growth. Advanced adenomas included villous or tubulovillous adenomas, large (≥1.0 cm) adenomas, and severe or high-grade dysplasias. Carcinoma in situ was categorized with severe dysplasia for the purposes of our analysis. In the PLCO trial, the colorectal cancer endpoint included carcinoma staged according to the American Joint Committee on Cancer (AJCC) criteria (5th edition) and carcinoid tumors involving the appendix, colon, or rectum but excluded cancers involving the anus or anal canal. To estimate diagnostic consequences and yields, we tabulated all lower endoscopic procedures and all diagnoses of colorectal cancer or adenoma that occurred within 365 days of a positive screening FSG examination by PLCO investigators.
We used binomial regression (implemented in PROC GENMOD, SAS System for Windows Release 9.2, Cary, NC) to compare FSG outcomes according to sex and age group. Models were adjusted for sex, enrollment age (four 5-year categories: ages 55–59, 60–64, 65–69, and 70–74 years), enrollment year (nine single-year categories between 1993 and 2001), and screening center (10 centers). Models used the identity link function to evaluate the difference and the log link function to evaluate the ratio between two proportions (ie, proportion of subjects with at least one FSG, proportion of subjects with inadequate FSG, and proportion of subjects with advanced colorectal neoplasia) (23). In SAS, a monotonic differentiable link function specifies the relationship between the expected value of the response variable and the linear predictor. Results reflect the status of PLCO data as of January 31, 2010.
The PLCO screening protocol offered a screening FSG on study entry and a second screening 3 or 5 years later. The 77447 persons aged 55–74 years whom we randomly assigned to screening (as opposed to usual care) included 39105 (50.5%) women, 38342 (49.5%) men, and 27814 (35.9%) persons older than 65 years (Table 1). In our study population, there were 8702 (11.2%) participants who were black, Hispanic, or other nonwhite race (66874 white, non-Hispanic; 3883 black, non-Hispanic; 1421 Hispanic; 2793 Asian; 605 Pacific Islander or American Indian; and 1871 missing), and 26660 (34.4%) participants who were college graduates (20). Among the 77447 participants, 67073 (86.6%) were considered to be adherent subjects because they received at least one sigmoidoscopy. This group included 64655 (96.4%) subjects who were first examined at study entry, 718 (1.1%) who were first examined in the third year after enrollment, and 1700 (2.5%) who were first examined in the fifth year after enrollment. Of the original 77447 subjects, 39443 (50.9%) received both an initial and a repeat sigmoidoscopy; these included 11449 (29.0%) subjects who were reexamined in the third year and 27994 (71.0%) subjects who were reexamined in the fifth year. Age-specific rates of inadequate FSG (ie, persons for whom all examinations were inadequate) were much higher in women than in men (at 55–59 years, 11.0% vs 3.6%; at 60–64 years, 11.2% vs 3.9%; at 65–69 years, 13.2% vs 4.6%; at 70–74 years, 16.1% vs 5.8%). Overall, 22083 (32.9%) of 67073 of adherent subjects had at least one mass or polyp detected (ie, were screen positive). Slightly greater than one-half of subjects who received a diagnostic intervention had a cancer or adenoma (positive predictivity). We tabulated these screening and diagnostic outcomes according to sex and age (Table 1).
Among the 15839 individuals who were screen positive at the first screen, 11867 (74.9%) received a diagnostic intervention within 1 year of the first screen. Among the 9133 individuals who were screen positive at the repeat screen, 7184 (78.7%) received a diagnostic intervention within 1 year of the repeat screen (Table 2). Medical records confirmed colonoscopy as part of the follow-up for 11159 (94.0%) of the 11867 and for 6940 (96.6%) of the 7184 subjects with diagnostic intervention after a first and repeat screen, respectively. Follow-up for subjects who did not receive colonoscopy, as documented by medical records, took the form of sigmoidoscopy, endoscopy (not otherwise specified), and/or surgical biopsy. Cumulatively, nearly one-quarter (17772; 22.9%) of all enrolled subjects had diagnostic intervention, 11867 (15.3%) for a positive first screen, and 7184 (9.3%) for a positive repeat screen. Repeat screening increased the proportion of subjects who received a diagnostic intervention by 50% (from 11867 to 17772). Among enrolled subjects, the cumulative probability of diagnostic intervention was substantially higher in men (28.0%) than in women (18.0%).
Combined, first screening and repeat screening detected advanced colorectal neoplasia (ie, colorectal cancer or advanced adenoma) in 49.8 of every 1000 persons screened. Initial screening detected advanced neoplasia in 37.8 of every 1000 persons screened (Table 3). Therefore, repeat screening increased the cumulative yield of detected advanced neoplasia by 32%. When we analyzed screening yields by sex, the first FSG detected advanced colorectal neoplasia in 25.9 of every 1000 women and 49.1 of every 1000 men screened (relative risk [RR] of advanced colorectal neoplasia in men vs women, adjusted for age at enrollment, year of enrollment, and screening center = 1.95, 95% confidence interval [CI] = 1.80 to 2.12); the second FSG detected advanced colorectal neoplasia in 13.0 of every 1000 women and 27.1 of every 1000 men screened (adjusted RR = 2.19, 95% CI = 1.88 to 2.56). At a program level, the increase in yield produced by the second FSG was 26% in women and 34% in men. When these data were adjusted for age at first or repeat FSG, instead of age at enrollment, there was no effect on the adjusted differences in FSG yield between men and women (data not shown).
Several factors contributed to sex-specific differences in yield (Supplementary Table 1, available online). More men (89.4%) than women (83.8%) had at least one sigmoidoscopy (adjusted difference = 3.9%, 95% CI = 3.4% to 4.3%). More men (39.4%) than women (26.1%) screened positive for the presence of a polyp or mass; of those, slightly more women (82.1%) than men (79.5%) had a diagnostic intervention. The presence of a polyp or mass was confirmed upon diagnostic intervention in more men than women, so that the positive predictive value of screening was higher in men: Among those who had diagnostic interventions for positive FSGs, 55.6% of men, compared with 42.6% of women, had a cancer or any adenoma; 21.1% of men, compared with 15.3% of women, had cancer or an advanced adenoma; and 1.6% of men, compared with 1.1% of women, had colorectal cancer. Overall, diagnostic yields were higher in men: 17.4% of men who had a screening FSG had colorectal cancer or adenoma, whereas 9.1% of adherent women had colorectal cancer or adenoma (Table 3).
We also examined age-specific yields (Supplementary Table 2, available online). Cancer or advanced adenoma yields were higher in persons aged 60–64 years (50.7 per 1000 people screened) than in those aged 55–59 years (41.3 per 1000 people screened; adjusted RR = 1.30, 95% CI = 1.19 to 1.42) and higher in persons aged 65–69 years (57.0 per 1000 people screened) than in those aged 60–64 years (adjusted RR = 1.14, 95% CI = 1.04 to 1.24). However, cancer or advanced adenoma yields were similar in persons aged 70–74 years (57.1 per 1000 people screened) and 65–69 years (adjusted RR = 1.02, 95% CI = 0.92 to 1.14).
The subjects in our study who were screened twice by FSG fell into five subgroups defined by the results of the first FSG screening: 29699 (75.3%) had received a negative first screen, 3558 (9.0%) had received an inadequate first screen, 2005 (5.1%) had received no diagnostic intervention after a positive first screen; in 2934 (7.4%), no adenoma or cancer was found upon diagnostic intervention after a positive first screen, and in 1247 (3.2%), an adenoma or cancer had been found upon diagnostic intervention after a positive first screen (Table 4). Among those subjects who returned for screening after a negative or inadequate first screen, 18.8% had a positive result on the repeat FSG screen (18.8% after a negative first screen and 18.4% after an inadequate first screen; Table 4). In comparison, 15839 (23.6%) of 67073 subjects tested positive at the time of their first screening FSG (Table 2). In persons with negative and inadequate first FSG reexamined by repeat FSG, yields of advanced neoplasia were 17.8 and 16.0 per 1000 subjects examined by repeat FSG, respectively. In all persons with first FSG reexamined by repeat FSG, the yield of advanced neoplasia was 20.8 per 1000 subjects examined by repeat FSG (Table 4), whereas all first time screens yielded 37.8 subjects with advanced neoplasia per 1000 subjects screened (Table 3).
As one might expect, among subjects who had a positive first FSG but did not receive any diagnostic intervention after the positive first FSG, a repeat FSG frequently found a polyp or mass (1197 [59.7%] of 2005) and often led to the diagnosis of advanced neoplasia (in 64.3 subjects per 1000 rescreened; Table 4). The remainder of subjects in this category (ie, those 808 [40.3%] of 2005 who had a positive first FSG, no diagnostic intervention, and no detectable polyp or mass on repeat FSG) included persons with an inadequate repeat FSG (6.3%), only small (<0.5 cm) polyps on first FSG (27.9%), no more than medium-sized (0.5–0.9 cm) polyps on first FSG (4.8%), one or more large (≥1 cm) polyps on first FSG (1.0%), and polyps of unknown size on first FSG (0.2%) (data not shown). Among the 1247 subjects who had a positive first FSG, and diagnosis of an adenoma after the first screen, there were 467 (37.4%) for whom a repeat FSG detected a polyp or mass, which led to the diagnosis of advanced neoplasia in 28.9 persons per 1000 rescreened. Repeat FSG in this setting, although discouraged by protocol, contributed to the surveillance that is normally offered to persons with a history of colorectal adenoma. Among the 3338 subjects with advanced neoplasia detected by the PLCO FSG program, the first screen detected 2535 (75.9%) and the repeat screen detected the remaining 803 (24.1%) (data not shown). Seventeen subjects had advanced adenoma detected at both a first and repeat FSG.
Excluding 20 subjects who had carcinoid tumors, there were 223 subjects who were diagnosed with colorectal carcinomas within 1 year of a positive first or repeat FSG: their tumor characteristics were recorded (Table 5). Diagnosis occurred within 3 months for 183 (82.1%) of this population and within 6 months for 207 (92.8%). Overall, 144 (64.6%) of these tumors were stage I and 39 (17.5%) were stage II, so that 183 (82.1%) were early stage (I or II). Approximately four (80.3%) out of five tumors were located in the splenic flexure or more distally. Distal location was more frequent after initial screening than repeat screening (150 [84.7%] of 177 tumors after initial FSG vs 29 [63.0%] of 46 tumors after repeat FSG, P = .001). Although colorectal carcinomas that were detected after the initial and repeat screenings had similar stage distributions, the fraction of individuals detected with cancers (including carcinoid tumors) or advanced adenomas, when expressed as a percentage of all individuals detected with cancers or adenomas, was higher at the initial FSG than at the repeat FSG (2535 [41.0%] of 6185 vs 820 [27.8%] of 2951, P < .001).
As a supplement, we have provided additional data, which may be useful to calibrate future models (Supplementary Tables 3, and 4, available online). Future models might compare the net benefits afforded by competing screening modalities, periodic as opposed to one-time screening, longer as opposed to shorter intervals between periodic screens, and different age thresholds to start or stop screening.
With 77447 subjects enrolled, 67073 screened at least once, and 39443 screened twice, repeating the FSG increased the yield of cancers and advanced adenomas detected in the PLCO trial by 32%. A high percentage (22.9%) of enrollees had at least one abnormal FSG, leading to diagnostic intervention. About one in five (3338 [18.8%; 1.4% + 17.4%] of 17772; Table 1) subjects who received diagnostic intervention because of a positive FSG received a diagnosis of advanced neoplasia. Among subjects with a diagnosis of advanced neoplasia, 75.9% were diagnosed as a result of the initial FSG and 24.1% were diagnosed as a result of the repeat FSG. There were 223 subjects diagnosed with colorectal carcinoma within 1 year of a positive FSG: 64.6% at stage I and 82.1% at stage I or II.
Repeated testing or relaxation of the threshold for referral to colonoscopy will affect the frequency of the colonoscopic evaluation of the proximal colon and the associated detection of premalignant adenomas and early colorectal cancers. The rate of colonoscopy under the PLCO FSG screening program approached 22% (Table 2). By using two FSG cycles, the investigators of the PLCO trial increased the percentage of all enrollees who received a diagnostic intervention from 15.3% after the first screen to 22.9% after both screens (Table 2). The European single-FSG trials reported colonoscopy rates of 5.0% [United Kingdom (18)], 7.8% [Italy (19)], and up to 21% [Norway (16,17)], and the 5% colonoscopy rate in the United Kingdom had no measurable effect on proximal cancer incidence (15). Assuming benefit from colonoscopy, higher rates of colonoscopy such as those in the PLCO trial, which used two rounds of screening FSGs, could produce greater reductions in proximal colorectal cancer incidence or mortality. Subsequent analyses that compare the PLCO intervention group and control group (usual care) will determine whether screening works to reduce incidence or mortality, from colorectal cancer overall and from cancer in the rectum, distal colon, and proximal colon specifically.
There are also reasons why repeating an FSG screening might not be beneficial. Although repeated FSG increases screening yields, it also increases the costs of screening, the need for diagnostic intervention, and the risks of complication. In addition, participants may become fatigued or disenchanted by requests for more frequent screenings or by unnecessary diagnostic interventions generated by false-positive screens. Both the frequency of repeat screening and the number of primary sites screened increase the cumulative incidence of unnecessary diagnostic intervention (24). Also, analysis of PLCO trial data regarding complications identified three instances when the bowel was perforated during one of 107236 FSG screening examinations (2.8 perforations per 100000 FSGs). In persons without detected colorectal cancer, there were 19 instances when the bowel was perforated during one of 17672 diagnostic colonoscopy examinations completed for evaluation of a positive FSG screening (1.1 perforations per 1000 diagnostic colonoscopies). Benefits produced by the PLCO FSG program must be weighted against costs and harms, including bowel perforation, which occurred at rates similar to those described in the literature (25–29), typically one perforation for every 10000 FSGs and one perforation for every 1000 colonoscopies. The current analysis shows equivalent, if not higher, rates of diagnostic intervention after a positive repeat FSG compared with a positive first FSG (78.7% vs 74.9%, Table 2). These results suggest that PLCO screening and intervention procedures, instead of producing participant fatigue, sustained participants’ willingness to undergo screening and diagnostic intervention for FSG-detected abnormalities.
We observed some sex-specific differences in the results from PLCO FSG screening. Inadequate FSGs occurred more frequently in women than men, a difference consistent with our 2005 report (20) and the reports of others (30–32). FSGs are sometimes said to be more difficult in women than in men because of anatomical differences, including those caused by hysterectomy (32), which may make the procedure more uncomfortable (33). Also, consistent with other studies (34–36), adenoma was more frequent in men than women. Consistent with the higher colorectal cancer risk generally observed in men relative to women, the diagnosis of colorectal cancer or adenoma after FSG was two times more common in men than women, whether evaluated after first or after repeat FSG (Table 3).
We also observed some age-specific differences in the results from PLCO FSG screening. As reported in our 2005 analysis of data from the entry FSG (20), cancer or advanced adenoma yields from the PLCO FSG program increased with age, at least through age 65–69 years (Supplementary Table 2, available online). Cancer or advanced adenoma yields per 1000 persons screened were 30% higher for subjects who enrolled at 60–64 years of age than for subjects who enrolled at 55–59 years of age. Yields were 14% higher in subjects who enrolled at age 65–69 years than in subjects who enrolled at age 60–64 years. Perhaps affected by a small decline in repeat FSG among subjects in the oldest age group (data not shown), yields were not statistically greater in subjects who enrolled at age 70–74 years compared with subjects who enrolled at age 65–69 years. The value of screening the elderly for colorectal cancer is uncertain. These observations serve to remind us that screen detectable adenoma, though common in the elderly, may stop increasing in prevalence after a certain age. Any age-related increase in yield from screening must be balanced against age-related risks from diagnostic or therapeutic intervention (37,38). The elderly (ie, persons aged ≥80 years) can plausibly expect to gain less life expectancy from screening colonoscopy than younger persons (39). Though guidelines suggest exceptions for persons not previously screened, the US Preventive Services Task Force now discourages routine colorectal cancer screening in persons aged 76–85 years (40).
There are limitations to this study. The efficacy endpoints for PLCO FSG are colorectal cancer mortality and incidence, for which data are not yet available. Because the PLCO FSG program included two rounds of screening, any differences in colorectal cancer mortality and incidence attributed to the PLCO FSG program will include contributions from both rounds. Despite our ability to measure the yields from first and repeat FSG, it may not be possible to distinguish specific mortality benefits derived from first as opposed to repeat FSG.
There are also several factors to consider with regard to the generalizability of our study. PLCO used 10 geographically representative centers to enroll a large study group, completed screening FSG using dedicated examiners, and relied on community resources for diagnostic intervention. These factors make our results especially relevant to any discussion of results one should expect from efforts to apply FSG to the US population. However, conclusions about outcomes from the PLCO FSG program must take the following four factors into account. First, the yields observed from programmatic screening partially reflected the repeat FSG completion rates achieved in PLCO. Although 86.6% of participants overall completed at least one FSG, only 58.8% of those screened at least once also completed the repeat FSG. Unlike our detailed 2002 analysis of factors associated with repeat FSG adherence (41), these estimates of FSG completion should not be accepted as measures of participant adherence to PLCO FSG protocols. Most notably, our calculations of FSG completion do not take into account those persons ineligible or unavailable for repeat FSG by virtue of death, interval colorectal cancer, or adenoma detection at first FSG. Second, yields from repeat FSG may not be able to be generalized to repeat colonoscopy because adenoma recurrence risks are higher in persons who are found to harbor proximal as opposed to distal adenoma (42,43). Third, practitioners of sigmoidoscopy (44,45) and colonoscopy (46,47) vary considerably with respect to detecting adenoma and, to a lesser extent, advanced adenoma (46). The accumulated experience of individual PLCO FSG examiners or temporal changes in the panel of PLCO FSG examiners could have contributed to the different outcomes observed at repeat FSG, relative to first FSG. Finally, outcomes specifically related to repeat FSG reflect not only aging subjects or different examiners but also the characteristics of subjects who returned for repeat FSG.
Midway through the trial, PLCO extended the interval between first and repeat FSG from 3 to 5 years. Ostensibly, this change permits examination of the effect of time interval on incident colorectal neoplasia risk. Results from such a comparison in PLCO could add to the evidence base that supports the currently recommended 5-year interval between FSG screenings (48). However, in addition to temporal change in examiner and subject composition, other factors distinguish the 3- and 5-year repeat FSG study populations. Most importantly, diagnostic intervention for screen-positive repeat FSG occurred more often in the 5-year than the 3-year group (data not shown). For these reasons, comparison of clinical outcomes at 3-year as opposed to 5-year repeat FSG demands careful analysis best reserved for a separate report.
In conclusion, repeat FSG in the PLCO Cancer Screening Trial increased colorectal cancer or advanced adenoma detection in women by one-fourth and in men by one-third. Final study results will examine the effects of the PLCO FSG program on colorectal cancer incidence and mortality.
This work was supported by individual contracts from the Division of Cancer Prevention, National Cancer Institute, NIH, DHHS, to each of the 10 screening centers and to the coordinating center (N01-CN-25404 to University of California Los Angeles Immunogenetics Ctr., N01-CN-25476 to Westat, Inc, N01-CN-25511 to University of Pittsburgh Cancer Institute, N01-CN-25512 to Henry Ford Health System, N01-CN-25513 to University of Minnesota School of Public Health, N01-CN-25514 to University of Colorado, N01-CN-25515 to Pacific Health Research & Education Institute, N01-CN-25516 to Washington University, N01-CN-25518 to Marshfield Clinic Research Foundation, N01-CN-25522 to Georgetown University Medical Center, N01-CN-25524 to University of Utah, and N01-CN-75022 to University of Alabama at Birmingham).
The National Cancer Institute (NCI) participated in the trial design and provided oversight during the conduct of the study. The authors thank the Screening Center investigators and staff of the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial, Mr Tom Riley and staff, Information Management Services, Inc, Ms Barbara O’Brien and staff, Westat, Inc. Most importantly, we acknowledge the study participants for their contributions to making this study possible.