Does Flexible Sigmoidoscopy Reduce Colorectal Cancer Incidence and Mortality?

Article Outline

• Comment
• Copyright

Hoff G, Grotmol T, Skovlund E, et al. (Cancer Registry of Norway, Oslo, Norway). Risk of colorectal cancer seven years after flexible sigmoidoscopy screening: randomized controlled trial. BMJ 2009;338:b1846.

Colorectal cancer (CRC) is the second leading cause of cancer death in the United States (CA Cancer J Clin 2008;58:71–96) and Europe (Ann Oncol 2007;18:581–592). Aside from a small trial of screening flexible sigmoidoscopy (Scan J Gastroenterol 1999;34:414–420), prospective, controlled studies of CRC screening are only available for fecal occult blood testing (FOBT), where improvement in both CRC incidence and mortality has been demonstrated (Am J Gastroenterol 2008;103:1541–1549). However, several large-scale, prospective, randomized screening flexible sigmoidoscopy trials are now underway in the United Kingdom, the United States, and Europe (Lancet 2002;359:1291–1300; J Natl Cancer Inst 2005;97:989–997; J Natl Cancer Inst 2002;94:1763–1772).

In this study, Hoff et al report the first interim results from the NORwegian Colorectal Cancer Prevention (NORCCAP) trial 1, a population-based, randomized, controlled trial comparing 1-time flexible sigmoidoscopy with usual care. The primary study outcome was incidence of CRC on an intention-to-screen basis. Secondary outcomes included incidence and mortality from rectosigmoid cancer for those who attended screening (per protocol analysis). The study sample consisted of all residents aged 55–64 years living in the city of Oslo and in Telemark County, Norway, in November, 1998. Of these 55,736 residents, 13,823 were randomly selected and invited to undergo a 1-time screening flexible sigmoidoscopy. Half of these individuals were also invited to complete 3 FOBT samples. The other 41,913 residents served as the control population and were offered no screening and simply were followed for study outcomes via national cancer and death registries.

Overall, 8846 of those patients invited to undergo flexible sigmoidoscopy attended screening (64.8%). Flexible sigmoidoscopy was performed after a single sorbitol enema using a 140-cm colonoscope (average insertion depth 44.0 for women and 48.9 cm for men) and all identified lesions were sampled at the time of the examination. Neoplasia was found in 19% of those screened, and 5% had high-risk adenomas or cancer. All participants found to have adenomas of any size, polyps ≥10 mm in diameter, carcinoma, or positive FOBT were referred for colonoscopy; this occurred in 21% of individuals.

In the intention-to-screen analysis at a median follow-up of 7 (range, 6–8) years, the incidence of CRC was the same for the screening and control groups (134.5 vs 131.9 cases per 100,000 person-years). Of the 123 CRC in the screening group, 33 were detected as a direct result of the screening process (prevalent lesions). In the per protocol analysis, the cumulative incidence of rectosigmoid cancer among those attending screening versus controls was 58 versus 79 per 100,000 person-years (P = .103). Overall CRC mortality was nonsignificantly reduced in the intention-to-screen analysis (hazard ratio [HR], 0.73; 95% confidence interval (CI), 0.47–1.13), with similar results for rectosigmoid cancer mortality (HR, 0.63; 95% CI, 0.34–1.18). In the per protocol analysis, however, total CRC mortality was reduced by 59% (HR, 0.41; 95% CI, 0.21–0.82) and rectosigmoid cancer mortality was reduced by 76% (HR, 0.24; 95% CI, 0.08–0.76).

The authors conclude that the effect of screening flexible sigmoidoscopy on reducing CRC incidence may be lower and occur later than expected. However, the results suggest a reduction in distal CRC with a 76% reduction in CRC mortality for those who attended screening, although they acknowledge that this analysis is prone to selection bias.

Back to Article Outline

Comment 

Current CRC screening guidelines from the United States Multi-Society Task Force, the American Cancer Society, and the American College of Radiology recommend a panel of options including FOBT, flexible sigmoidoscopy, colonoscopy, CT colonography, and fecal DNA tests (Gastroenterology 2008;134:1570–1595). In fact, only FOBT is supported by large scale, prospective evidence (Am J Gastroenterol 2008;103:1541–1549), whereas the remainder are recommended based on observational studies and inferential evidence (Gastroenterology 2008;134:1570–1595). However, recent studies have raised concerns regarding the effectiveness of colonoscopic screening (Ann Intern Med 2009;150:1–8; Gastroenterology 2008;134:A-111–112; Gastrointest Endosc 2005;61:385–391). It may well be that our assumptions about many recommended CRC screening tests may be overly optimistic. Therefore, there is a great need for comparative effectiveness research in the form of large-scale, prospective, controlled trials to determine the true effectiveness of CRC screening methods (N Engl J Med 2009;361:325–327).

The NORCCAP investigators have performed one such important study. Although approximately 60% of Americans have undergone recent CRC screening, with colonoscopic screening becoming increasingly common (MMWR Morb Mortal Wkly Rep. 2008;57:253–258), no organized CRC screening program utilizing any modality exists in Norway. Therefore, these investigators had the opportunity to compare flexible sigmoidoscopy with a control group that often went unscreened. Given the enthusiasm in the United States for colonoscopy-based screening, one might wonder why there is interest in studying flexible sigmoidoscopy. This study is important because flexible sigmoidoscopy does have advantages over colonoscopy, including lower provider training requirements, lack of need for sedation, fewer complications, and lower cost. Furthermore, the effectiveness of colonoscopy in reducing death from right-sided CRC has been questioned (Ann Intern Med 2009;150:1–8). Therefore, if flexible sigmoidoscopy is effective, it may be more readily implemented in areas where capacity for screening colonoscopy does not exist.

Although the finding of no significant reduction in CRC mortality in the intention-to-screen analysis might disappoint CRC advocates, there are reasons to not be disheartened. First, adherence to screening may be an issue because only 8,846 of 13,823 randomized to flexible sigmoidoscopy actually attended screening. Because only 1 invitation to attend screening was offered, additional invitations or other interventions may have improved compliance. The per protocol results strongly suggest that benefits can be achieved for those that complete screening. Second, given what we know from FOBT trials, perhaps this first interim result is not surprising (additional outcome assessment at 10 and 15 years is also planned). Recall that in the large FOBT trial by Mandel et al, the reduction in CRC mortality was demonstrated after 13 years of follow-up (N Engl J Med 1993;328:1365–1371) and CRC incidence reduction was shown after 18 years of follow-up (N Engl J Med 2000;343:1603–1607). Nevertheless, a clinically significant (albeit not statistically significant) 27% reduction in CRC mortality was shown by the NORCCAP study. Finally, Hoff et al chose important but challenging end points (ie, cumulative CRC incidence and mortality). Because it is not possible to exclude prevalent cancers from the control group, all cancers found at screening must be counted in the screening group, making it difficult to demonstrate improved cumulative incidence of CRC. This factor should be less of an issue during assessments at later time points. Furthermore, by including both proximal and distal CRC in their analysis, they are holding flexible sigmoidoscopy, an examination of only the distal colon, to a high standard. In the landmark case-control study of flexible sigmoidoscopy by Selby et al, all the benefits were limited to the rectosigmoid (N Engl J Med 1992;326:653–657).

Given all the above, it would be a mistake to interpret this study as evidence that flexible sigmoidoscopy is not an effective method to screen for CRC. The significant benefits seen in both rectosigmoid cancer mortality and overall CRC mortality in screening attendees is certainly encouraging. Among those screened, there were only 3 rectosigmoid cancer deaths, where >12 deaths would have been expected based on findings in the control population (BMJ 2009;338:b2531). Nearly half of all CRC in those who attended was detected through this screening, and these cancers tended to be more localized and had a lower case fatality rate than in the control group. Also, the incidence curve for rectosigmoid cancers was relatively flat compared with the control group, suggesting that screening may ultimately reduce cancer incidence, potentially through the removal of precancerous lesions. However, per protocol analyses are subject to selection bias; those who attend screening may be different from those who do not, such as in diet, exercise, and other important health behaviors. In fact, compared with controls, those attending screening had higher rates of smoking, but were also more physically active and reported more adherence to general dietary recommendations (Eur J Cancer Prev 2006;15:10–19). Interpretation of the study results is further complicated by the fact that half of those randomized to screening were offered both flexible sigmoidoscopy and FOBT. Three of the 13 cancers detected by screening were found among participants with a positive FOBT but a negative flexible sigmoidoscopy. Ultimately, to specifically understand the effectiveness of flexible sigmoidoscopy alone, these cases may need to be removed from the analysis.

In summary, the NORCCAP study has added considerably to our understanding of the role of flexible sigmoidoscopy in CRC screening, but some important questions remain. First, the study emphasizes the importance of adherence to the success of any screening program and, therefore, raises questions about how adherence can be optimized in clinical practice. Recent cost-effectiveness studies that informed the United States Preventive Services Task Force guidelines regarding CRC screening have come to this conclusion as well (Ann Intern Med 2008;149:659–669). Second, the study also suggests that flexible sigmoidoscopy does lead to the detection of earlier and more treatable cancers, though perhaps not to the extent that was anticipated by the investigators. Although the per protocol analysis results are encouraging, more information is needed to definitively understand the overall benefits of this screening modality with longer follow-up. It is interesting to note that no benefits in overall mortality were apparent when comparing the screened versus the control group (HR, 1.02; 95% CI, 0.98–1.07; P = .28). An earlier Norwegian study reported increased all-cause mortality in patients randomized to flexible sigmoidoscopy screening (Gastroenterology 2001;120:A14) and also pooled results of large FOBT trials showed a reduction in CRC specific mortality yet a simultaneous small but significant increase in non-CRC mortality (Am J Gastroenterol 2006;101:380–384). Long-term data from this study combined with results from other ongoing large flexible sigmoidoscopy trials examining both cancer specific and overall mortality are required to fully explore this issue. Finally, this study highlights the need for large-scale, randomized, controlled trials of other CRC screening methods, including colonoscopy and CT colonography, because we cannot fully understand the effectiveness of these screening interventions without such important comparative effectiveness research studies.