Flexible sigmoidoscopy screening: is once enough?

www.thelancet.com Published online February 21, 2017 http://dx.doi.org/10.1016/S0140-6736(17)30542-1 1 Worldwide, colorectal cancer is the third most common incident cancer and the fourth leading cause of cancer death, with 1·4 million new cases and 700 000 deaths estimated in 2012—most of these from the developed world. Over the past three decades, several methods to screen for colorectal cancer, and to detect, remove, or both, colorectal cancer precursors— namely, adenomatous polyps—have been developed. Current guidelines in the European Union include recommendations for stool-based tests (fecal occult blood test [FOBT] and fecal immunochemical test [FIT]) and fl exible sigmoidoscopy, whereas most US guidelines include those tests as well as colonoscopy. In The Lancet, Wendy Atkin and colleagues report on the extended follow-up of the UK Flexible Sigmoidoscopy Screening Trial (UKFSST) through a median of 17 years of follow-up. This randomised controlled trial enrolled 170 432 eligible men and women to either an intervention group (off ered one-time fl exible sigmoidoscopy screening) or control group (not contacted), with primary outcomes of colorectal cancer incidence and mortality. The analysed cohort included 57 098 people in the intervention group, of whom 40 621 (71%) received screening, and 112 936 people in the control group. A previous report of 11 years of follow-up reported hazard ratios (HRs) in intention-to-treat (ITT) analyses of 0·77 (95% CI 0·70–0·84) for colorectal cancer incidence and 0·69 (0·59–0·82) for colorectal cancer mortality for the intervention versus control group. Rather remarkably, because the trial used only one fl exible sigmoidoscopy screen at baseline, these HRs were maintained through an additional 6 years, with ITT HRs now of 0·74 (95% CI 0·70–0·80) for incidence and 0·70 (0·62–0·79) for mortality. These new results are important for two reasons. First, and especially with regards to the reported reduction in incidence of distal colorectal cancer, the implications for our understanding of the natural history of adenomas and colorectal cancer are intriguing. Second, the reported reduction in incidence and mortality of overall colorectal cancer provides needed data from a public health perspective on the optimal design of a screening programme that incorporates fl exible sigmoidoscopy. Because of ethical concerns in leaving colon polyps intact for appreciable time periods, the natural history of adenomas, especially with regards to quantitative transition rates to colorectal cancer, is largely unknown. Estimates of transition rates and intervals largely derive from modelling studies, which generally rely on data covering only short periods of observation (5–10 years). The fi ndings of Atkin and colleagues’ extended followup should inform future modelling eff orts and lead to a better understanding of the natural history of adenoma. Based on the per-protocol analysis of Atkin and colleagues of the eff ectiveness in participants who received fl exible sigmoidoscopy screening, the reduction in incidence for distal colorectal cancer was actually slightly greater at 17 years of follow-up (HR 0·44 [95% CI 0·38–0·50]) than at 11 years (0·50 [95% CI 0·42–0·59]). After exclusion of prevalent cases, and with the assumption that the rate of such cases was the same in the control group as in those actually screened (126 [0·31%] of 40 621), a per-protocol analysis would show a striking 66% reduction in the incidence of distal colorectal cancer over 17 years. With the exception of the small subset referred for surveillance colonoscopy, this prevention was all due to a single fl exible sigmoidoscopy screen at baseline. These fi ndings suggest that the rate at which de-novo adenomas formed and transitioned to symptomatic colorectal cancer within 17 years was quite low. Current models of the natural history of adenoma should be reassessed to see if they are consistent with these long-term results. The trial protocol called for surveillance colonoscopy for patients found to be at high risk on the baseline screen. Of the 40 621 people actually screened, 1745 (4%) were reported to have undergone surveillance. Additional information about this surveillance cohort in terms of yield of adenomas and colorectal cancer over time would be useful to understand the various components of the sustained incidence reduction. From a public health perspective, the updated fi ndings of a reduction in the overall colorectal cancer incidence and mortality through to 17 years raises the issue of whether once-only fl exible sigmoidoscopy might be a reasonable screening strategy. Outside of the USA, where colonoscopy is the primary method of colorectal cancer screening, colorectal cancer screening guidelines generally recommend either Flexible sigmoidoscopy screening: is once enough?