Is the Effectiveness of Colonoscopy "Good Enough" for Population-Based Screening?

In 2008, the American College of Gastroenterology published recommendations stating “colonoscopy every 10 years, beginning at age 50, remains the preferred colorectal cancer … screening strategy” ( 1 ) for average-risk individuals because of the potential effectiveness of colonoscopy to reduce the incidence and mortality from colorectal cancer. Rex et al. ( 1 ) acknowledged that this recommendation placed “greater emphasis on effectiveness than on risk.” Given the rapid rise in the use of colonoscopy for colorectal cancer screening in North America, this philosophy appears to be widely held ( 2 , 3 ). However, the rate of serious adverse complications ( 4–6 ) including death associated with colonoscopy, although low, is unprecedented for a screening maneuver recommended for use in the general population. Although colorectal cancer is a common cancer, the lifetime risk for an individual person is actually quite low, 5% ( 7 ). Because of this low risk, the vast majority of average-risk individuals screened for colorectal cancer with colonoscopy will be exposed to the potential complications of this procedure but will not benefit from it. Given the realities of screening, to justify the risks, colonoscopy in the population must be highly effective. Additionally, the effectiveness of colonoscopy must be substantially greater than lower-risk screening strategies, such as flexible sigmoidoscopy followed by colonoscopy in those with clinically significant findings. How effective is screening colonoscopy for reducing the burden of colorectal cancer in the population? The definitive answer to this question, best addressed by a randomized trial, is of course unknown. Although a trial designed to tackle this issue, the Northern European Initiative on Colorectal Cancer ( 8 ) is under way, final data collection for assessment of the primary outcomes (colorectal cancer incidence and mortality) in this study is not anticipated until 2026. Until then, observational evidence must inform our understanding of the effectiveness of colonoscopy and the study by Brenner et al. ( 9 ) in this issue of the Journal is an important contribution. In this cross-sectional survey, the authors determined the self-reported history of previous colonoscopy for 3287 patients undergoing screening colonoscopy during a 19-month period in the state of Saarland, Germany. Advanced neoplasia at screening colonoscopy was substantially more common among patients with no history of colonoscopy within the previous 10 years (308 [11%] advanced neoplasias among the 2701 patients in that group) than among patients with a history of colonoscopy in this time frame (36 [6%] in the 586 patients in that group), and importantly, detection of colorectal cancer was far more common among patients with no history of recent colonoscopy (41 [1.5%] in the 2701 patients) than among the group with a previous colonoscopy (one [0.2%] in the 586 patients). The rate of advanced neoplasia detection in patients with no previous colonoscopy in this study was quite similar to rates in the literature ( 10 , 11 ). The rate of advanced neoplasia detection in those with previous colonoscopy was higher than rates reported for patients with a history of negative colonoscopy ( 12 ) but was lower than rates reported for patients with a history of previously identified advanced neoplasia ( 13 ), indicating a blend of higher- and lower-risk individuals in this group. The association between previous colonoscopy and detection of advanced neoplasia at screening in the study, as measured by the adjusted prevalence ratio, indicates a substantial protective effect of colonoscopy (prevalence ratio = 0.52, 95% confidence interval [CI] = 0.37 to 0.73). However, the association varied with site of neoplasia; previous colonoscopy was strongly associated with detection of left-sided advanced neoplasia (prevalence ratio = 0.33, 95% CI = 0.21 to 0.53) but not right-sided advanced neoplasia (prevalence ratio = 1.05, 95% CI = 0.63 to 1.76). Overall, the rate of right-sided advanced neoplasia was low (3% of those undergoing screening colonoscopy in both groups); however, as a proportion of all advanced neoplastic lesions, right-sided lesions were overrepresented in the group having previous colonoscopy (50% of advanced neoplastic lesions were right-sided) as compared with those with no previous colonoscopy (23% of advanced neoplastic lesions were right-sided). This study has clear limitations, and the estimates of association produced by this study should not be considered precise. Because of the cross-sectional design, patients developing interval symptomatic cancers after colonoscopy were excluded and thus this study may overestimate the protective effect of colonoscopy. However, the previous colonoscopy group appeared to be at higher baseline risk of colorectal cancer and proximal neoplasia in particular ( 14 , 15 ); they were older and more likely to have a first-degree relative with colorectal cancer than the colonoscopy-naive group. Although the researchers adjusted for these variables, the risk of residual confounding is real and may have resulted in an underestimate of the association of colonoscopy with detection of advanced neoplasia, right-sided neoplasia in particular. Nevertheless, the results are remarkably consistent with a number of recently published studies, all of which demonstrate the overall effectiveness of colonoscopy for reduction of colorectal cancer incidence and mortality but with a marked variance in effectiveness for proximal and distal cancers ( 16–20 ). Although there is compelling evidence that colonoscopy is a less effective tool in the proximal colon than distal colon, the underlying reasons for the differential performance are unclear. Like previous studies ( 16–20 ), Brenner et al. (9) did not evaluate the quality of initial colonoscopy in the previously exposed group. Although the standards for screening colonoscopy in this region are high, these standards were set in 2002 after many of the colonoscopies of interest in this study were conducted. Certainly, inadequate performance of colonoscopy will reduce effectiveness and quality of colonoscopy may be more important for detection and treatment of neoplasia in the proximal colon than distal colon. However, there may be biological differences that limit the potential effectiveness of colonoscopy in the proximal colon. Right-sided colonic adenomas tend to be flatter than left-sided lesions ( 21 ) and are, therefore, harder to identify and remove. These differences may result from known variations in the molecular features of right-sided cancers ( 22 , 23 ). The predominant genetic pathways of carcinogenesis may differ between right-sided and left-sided cancers ( 23–26 ) in a fashion that influences the potential effectiveness of any currently available screening tool in the proximal colon but is not amenable to quality improvement strategies. The potential limitations of colonoscopy for prevention of incidence and mortality from colorectal cancer that develops in the proximal colon, as identified by this study ( 9 ) and others, raise very important questions: Is there an incremental benefit of colonoscopy over flexible sigmoidoscopy for colorectal cancer screening? If so, is the incremental benefit of sufficient magnitude to justify the additional risks and costs of colonoscopy for screening in the population? Simply put, is the effectiveness of colonoscopy “good enough” for population-based screening? As more observational evidence accumulates, the answer to this question becomes less certain. Cancer Care Ontario Chair in Health Services Research (to N.N.B.).

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.