The Big Picture

Colonoscopy for average-risk colorectal cancer screening has transformed the practice of gastrointestinal medicine in the United States. However, although the dominant screening strategy, its use is not supported by randomized controlled trials. Observational data do support a protective effect of colonoscopy and polypectomy on colorectal cancer incidence and mortality, but the level of protection in the proximal colon is variable and operator-dependent. Colonoscopy by high-level detectors remains highly effective, and ongoing quality improvement initiatives should consider regulatory factors that motivate changes in physician behavior.

Key points

•

Colonoscopy is the dominant colorectal screening strategy in the United States, yet its use is not supported by randomized controlled trials.
•

Observational data do support a protective effect of colonoscopy and polypectomy on colorectal cancer incidence and mortality, but the level of protection in the proximal colon is variable and operator-dependent.
•

Reducing operator dependence and developing new technical improvements in colonoscopy are and will remain priorities in colorectal cancer prevention.
•

Ongoing quality improvement initiatives should consider regulatory factors that motivate changes in physician behavior.

Colonoscopy was first endorsed in the United States as a screening strategy for colorectal cancer in 1997. This followed the publication of several observational studies of screening colonoscopy, highlighting the prevalence of adenomas in asymptomatic volunteers. Screening recommendations prior to this centered on structural examination of the distal colon and rectum with flexible sigmoidoscopy and noninvasive stool tests. At the time, colonoscopy was the logical extension to sigmoidoscopic screening, as it allowed complete and direct visualization of the whole colon and combined diagnosis and therapy in a single session, compared with noninvasive, two-stage screening strategies.

Two multi-center studies in 2000 definitively established the safety and feasibility of screening colonoscopy, while highlighting the prevalence of proximal neoplasia in patients without any findings in the distal colon and rectum. These results, combined with academic and media attention, provided support for lobbying efforts by professional societies with the US Congress, which passed legislation mandating coverage for screening colonoscopy every 10 years in Medicare beneficiaries aged 50 years and older from July 1, 2001.

This law heralded the transformation of gastroenterology practice in the United States. Colonoscopy was adopted by endoscopists and patients, and in the ensuing years, it has become the dominant colorectal cancer screening strategy in the United States. Screening colonoscopy is now recommended by 95% of primary care physicians, and it is estimated that approximately 12 million colonoscopies are performed each year in the United States.

However, the position of colonoscopy as a colorectal cancer screening strategy has not been evaluated in randomized controlled trials. Detractors of screening colonoscopy highlight this lack of clinical trial evidence, together with concerns about lack of cost-effectiveness. In contrast, programmatic screening with flexible sigmoidoscopy is supported by high-level evidence. Four randomized controlled trials have now been performed of flexible sigmoidoscopy, all showing a reduction in distal colorectal cancer incidence and mortality. The challenge for colonoscopy is whether it offers any additional benefit in screening the proximal colon.

There is ample indirect evidence of a protective effect for colonoscopy on colorectal cancer. For example, population-level cancer statistics show that coincident with the establishment of screening colonoscopy in the United States, colorectal cancer incidence and mortality have progressively fallen over recent decades. A report from the US Centers for Disease Control and Prevention described an overall 30% reduction in colorectal cancer incidence between 2001 and 2010 (3.4% per year), with the greatest impact in the screening-eligible age group (3.9% per year). Colorectal cancer mortality has been falling in the United States since 1975, and this has been attributed to improvements in treatment (12%) and risk factor profiles (32%) and increases in screening (53%).

Other indirect evidence comes from the original study of the fecal occult blood test (FOBT). The Minnesota randomized controlled trial of FOBT was the first to establish a role for screening for colorectal cancer. Although not a trial of screening colonoscopy, colonoscopy was performed in patients with positive tests, and polyps were removed. Subsequent long-term follow-up showed 17% to 20% reductions in colorectal cancer incidence compared with those who were not screened, and 22% to 32% reductions on colorectal cancer mortality at 30 years.

The proximal colon

However, the incremental benefit of colonoscopy over flexible sigmoidoscopy depends on extending the protection against colorectal cancer to the proximal colon. In the early days of screening colonoscopy, it was assumed that proximal colon protection would naturally extend from a structural examination of the whole colon. Here are reviewed the data on colorectal cancer protection from colonoscopy, focusing on protection in the proximal colon.

By definition, flexible sigmoidoscopy is a screening test of the distal colon and rectum, so any impact on the incidence of proximal cancer in sigmoidoscopy trials is attributable to colonoscopy, performed following the detection of sentinel lesions on initial index sigmoidoscopy. For example, in the US study (Prostate, Lung Colorectal and Ovarian Cancer Screening Trial), a 14% reduction in the incidence of proximal colorectal cancer was seen. This reduction is consistent with exposure to colonoscopy in 21.9% of participants, either from referral for colonoscopy for sigmoidoscopy findings, or colonoscopy outside the study protocol. In contrast, the UK flexible sigmoidoscopy trial showed no impact on proximal cancer risk. However, these studies differed in participant recruitment and in criteria for referral for colonoscopy, with only 5% of participants in the UK trial referred for colonoscopy (compared with 21.9% in the US trial ).

During the initial lobbying for introduction of screening colonoscopy in the United States, observational data on the protective effect of colonoscopy on colorectal cancer incidence were derived from studies in adenoma-bearing cohorts. The first and most widely quoted of these studies was the US National Polyp Study, in which patients were initially randomized to different colonoscopic surveillance intervals. Findings showed a 76% to 90% reduction in colorectal cancer incidence compared with reference cohorts. A related Italian study with comparable methodology also demonstrated a 76% reduction in cancer incidence after colonoscopic polypectomy.

These studies, showing high levels of cancer protection, were widely cited as evidence of the effectiveness of colonoscopy. Colonoscopy was regarded as preventing approximately 80% of colorectal cancer, and the National Polyp Study was extensively utilized in lobbying of the US Congress for coverage of screening colonoscopy. Long-term follow-up from the National Polyp Study was published in 2012, showing a sustained reduction in colorectal cancer mortality of 53% at a mean of 15.8 years in the adenoma-bearing cohort.

Other adenoma cohort studies have not shown such high levels of colorectal cancer protection. Several studies were not able to demonstrate any reduction in colorectal cancer incidence compared with the general population. Methodologic variation makes direct comparison of these cohort studies difficult, although an analysis of the cancers detected in these other studies indicated the occurrence of early interval (often proximal) cancers may explain variable levels of cancer protection after colonoscopy.

The incidence of colorectal cancer has also been studied in long-term follow-up of patients who had colonoscopy performed in the original screening studies. Colorectal cancer incidence reductions of up to 67% were documented. However, further analysis of the interval cancers found during follow-up, which represent a failure of colonoscopic protection, showed that most were located in the proximal colon and found within 3 to 5 years of the index colonoscopy.

The highest protective effect of colonoscopy comes when the index baseline examination is negative. In 2 prospective cohort US studies of average-risk screening patients with a negative baseline colonoscopy, no cancers were found at 5 years in either group. Two retrospective Canadian studies of large cohorts of patients with negative, complete colonoscopies (n = 32,203 and n = 111,402 ) in which colorectal cancer incidence was compared with the general population, showed incidence ratios of 0.66 to 0.80 at 1 to 2 years respectively, and 0.25 to 0.28 at 10 to 14 years. However, cancers that occurred during follow-up were more likely proximal, and the reduction in risk for proximal cancers was less.

More recently, analysis of large US cohorts from the Nurses Health Study and the Health Professionals Follow-up Study (n = 88,902) showed a reduction in risk of colorectal cancer after colonoscopy. Specifically, adjusted hazard ratios were 0.57 (95% confidence interval [CI] 0.45–0.72) after colonoscopy with adenoma removal and 0.44 (95% CI 0.38–0.52) after negative colonoscopy. Again, risk reductions were greater for distal than proximal cancers, and colonoscopy with polypectomy was not associated with a significant reduction in the incidence of proximal colon cancer. The proportion of incident cases prevented with colonoscopy was 40% overall, 22% for proximal cancers, and 61% of distal colorectal cancer.

Multiple case–control studies have examined the association between colonoscopy exposure and colorectal cancer occurrence. Initial studies from Canada in nonscreening populations failed to show any protection against colorectal cancer in the proximal colon. Complete colonoscopy was associated with a 47% to 67% reduced risk of death from left-sided colorectal cancer but no reduction in right-sided colorectal cancer mortality. However, these data are influenced by the Canadian context, where colonoscopy is typically performed by nongastroenterologists, who are known to be less effective than gastroenterologists in preventing colorectal cancer.

In contrast, subsequent data from Germany and the United States have shown a beneficial impact from colonoscopy on proximal colon cancer incidence and mortality. In 2011, a German population-based study showed a 77% reduction in overall colorectal cancer risk, with reductions in both the right (odds ratio [OR] 0.44, 95% CI 0.35–0.55) and left colon (0.16, 95% CI 0.12–0.20). In the United States, a case–control study of screening colonoscopy showed a reduced risk of colorectal cancer overall of 70%, with reductions in both left-sided (OR 0.26, 95%CI 0.06–0.11) and right-sided colorectal cancers (0.37, 0.16–0.82). In the most recent German population-based study, there was a substantial reduction in cancer risk after screening colonoscopy (OR 0.09, 95% CI 0.07–0.13), with a 78% incidence reduction in the right colon (0.22, 95% CI 0.14–0.33).

In summary, despite a lack of randomized trials, substantial observational data indicate that colonoscopy offers protection against colorectal cancer incidence and death. Colonoscopy does offer an incremental benefit over flexible sigmoidoscopy by extending the protection to the whole colon, even though its impact on proximal cancer incidence and mortality can be lower. Factors contributing to this inconsistent protection in the proximal colon are now considered.