The IARC Perspective on Colorectal Cancer Screening

Colorectal cancer is the third most common cancer in men and the second most common in women, and represents almost 10% of the annual global cancer incidence ¹. Incidence rates of colorectal cancer show a strong positive gradient with increasing level of economic development ²; the net five-year survival decreases with income, reaching 60% in high-income countries, yet falling to 30% or less in low-income settings ³.

Established risk factors for colorectal cancer include consumption of processed meat, consumption of alcoholic beverages ⁴, tobacco smoking ⁴, and excess body fatness ⁵, whereas consumption of dietary fiber and dairy products, and physical activity decrease colorectal cancer risk ^6,7. In addition, certain subgroups of the population are at increased risk, due to genetic predisposition (e.g. Lynch syndrome), family or personal history of colorectal neoplasia, or medical conditions (e.g. inflammatory bowel disease) that predispose to colorectal cancer.

Colorectal cancer can be classified based on location within the large bowel, histology, and molecular features. Advanced adenomas, in particular advanced adenomas of > 10 mm, are the most well-known precursor lesions of colorectal carcinoma ⁸. Screening aims to reduce colorectal cancer mortality through early detection of cancer and morbidity associated with detection of cancer at a later stage; it also aims to reduce incidence – and mortality – through detection and removal of precancerous lesions. Colorectal cancer screening is available in many high- and upper-middle-income countries worldwide, delivered by organized programs or through opportunistic screening; participation rates are highly variable between countries and settings ⁹ but have typically been below 40%. Insurance status and access to primary care are the main determinants of participation; costs, logistical challenges, lack of provider involvement, language barriers, cultural beliefs, and lack of awareness of colorectal cancer screening represent important additional barriers ^10,11.

There are several available methods for colorectal cancer screening. Stool-based tests to detect blood include the guaiac fecal occult blood test (gFOBT) and the more sensitive fecal immunochemical test (FIT) ¹². Endoscopic methods, which use optical approaches to directly examine the rectum and colon, include sigmoidoscopy and colonoscopy ¹³. Colonoscopy is used both as a primary screening tool and as follow-up for individuals testing positive with other screening methods. In addition, computed tomographic (CT) colonography, an imaging method based on scanning technology, has been developed as a less invasive visualization technique for colorectal cancer screening ¹⁴. Newer techniques, which have recently emerged but have not been widely tested yet, are based on either visual inspection (e.g. video capsule endoscopy) or analysis of biomarkers in stool (e.g. multitarget-stool DNA), blood (e.g. methylated Septin9 DNA), or breath, such as volatile organic compounds and various protein, RNA, and DNA markers.

We reviewed the published evidence from randomized controlled trials, observational studies, and modeling studies, for stool-based, endoscopic, and CT colonography-based screening methods and evaluated outcomes with respect to preventive effects, adverse effects, and the balance of benefits and harms in average-risk populations, for men and women combined (Table 1)[Please see the Supplemental appendix for members of the IARC Working group and how the review was conducted]. Where randomized trials reporting on colorectal cancer mortality and incidence were missing, evidence from the most similar screening test for which a reduction in colorectal cancer mortality and/or incidence has been shown (i.e. FIT vs gFOBT; colonoscopy vs sigmoidoscopy) and/or from comparative studies on test performance (i.e. CT colonography vs colonoscopy) was also considered. Evidence on newer techniques was reviewed but was deemed insufficient to enable any conclusions to be reached.

Table 1.

Evaluations of colorectal cancer screening with stool-based blood tests, endoscopic methods, and CTC

Screening technique	Strength of evidence 1
	Reduction in CRC incidence	Reduction in CRC mortality	Benefit–harm ratio1
Stool-based blood tests
Biennial screening with gFOBT without rehydration	ESLE	S	S
Annual or biennial screening with gFOBT of higher sensitivity	L	S	S
Biennial screening with FIT	L	S	S2
Endoscopic techniques
Single screening with sigmoidoscopy	S	S	S
Single screening with colonoscopy	S	S	S3
CTC
Single screening with CTC	L3,4		I

CRC, colorectal cancer; CTC, computed tomographic colonography; ESLE, evidence suggesting a lack of effect; FIT, fecal immunochemical test; gFOBT, guaiac-based fecal occult blood test; I, inadequate evidence; L, limited evidence; S, sufficient evidence.

¹Sufficient evidence applies only with the assumption that screening can be delivered with high quality and follow-up ensured.

²A variety of qualitative and quantitative FIT tests are available, with wide ranges of sensitivity and specificity. The net balance of benefits and harms will depend on the cut-off level for positivity.

³A minority of the Working Group favored a different evaluation (see text).

⁴The evaluation of limited evidence applies to the reduction in incidence and/or mortality (one single evaluation).

This article briefly summarizes the evaluation of the scientific evidence reviewed at the meeting (Table 1). The full report will be published as Volume 17 of the IARC Handbooks of Cancer Prevention. It is noteworthy that the majority of studies reviewed have been conducted: in middle- to high-income settings, where colorectal cancer incidence is generally high; in asymptomatic, average-risk populations typically aged 50–70 years; and under conditions in which colorectal cancer screening – including subsequent follow-up and treatment – can be delivered with high quality. Extrapolation of the conclusions to different settings (e.g. in level of health system development) needs to take into account these and other context-related specificities.

Stool-based blood tests

Beneficial effects of gFOBT

Five randomized studies of annual or biennial gFOBT screening conducted in North America or Western Europe ^15–19, and 10 observational studies of gFOBT conducted in screening settings in different geographical regions ^20–29, reported on colorectal cancer mortality and/or incidence. The studies used either gFOBT without rehydration, or a rehydrated gFOBT test, a test with higher sensitivity than the non-rehydrated test (see Table 1).

Based on two randomized clinical trials, two large cohort studies with up to 11 screening rounds, and one case–control study ^{15,18,22,23,25}, sufficient evidence supports that biennial screening with gFOBT without rehydration reduces colorectal cancer mortality. Sufficient evidence supports that annual or biennial screening with gFOBT of higher sensitivity reduces colorectal cancer mortality ^16,19,24. The mortality reduction reported in the randomized trials ranged from 9% to 14% with gFOBT without rehydration, and from 16% to 32% with gFOBT of higher sensitivity (coupled with colonoscopy in those with a positive test), and was significant in all trials compared to no screening.

Regarding screening with gFOBT without rehydration, there is evidence suggesting a lack of effect of biennial screening in reducing colorectal cancer incidence, based on three randomized trials and one cohort study after 11 screening rounds ^15,17,18,23, and limited evidence that annual or biennial screening with gFOBT of higher sensitivity reduces colorectal cancer incidence, based on one randomized trial with 18 years of follow-up ³⁰.

Beneficial effects of FIT

No randomized clinical trials with colorectal cancer incidence or mortality outcomes have been conducted with FIT, but the findings from observational studies in screening settings were highly consistent. Three cohort studies, including one incidence-based mortality study, showed a reduction of 10% to 40% in colorectal cancer mortality ^31–33. One ecological study in Italy that compared areas with early (2002–2004) versus late (2008–2009) implementation of an organized screening program with biennial screening with FIT also showed a rapid reduction in colorectal cancer mortality after introduction of screening in the area where screening was introduced first ³⁴ Overall, sufficient evidence supports that biennial screening with FIT reduces colorectal cancer mortality. In reaching this evaluation, the following evidence was also taken into account: (i) evidence from randomized trials of gFOBT screening; given the close similarity between the two tests, FIT should be at least as good as gFOBT in reducing colorectal cancer mortality; and (ii) evidence from randomized trials of the increased performance of FIT compared with gFOBT for the detection of advanced adenoma and colorectal cancer.

The evidence was deemed limited in relation to reduction of colorectal cancer incidence. Small to moderate reductions in cumulative incidence were observed in two cohort studies after three rounds of biennial FIT ^32,33, and in one ecological study conducted in Italy ³⁵.

Potential harms and benefit–harm ratio for stool-based blood tests

Potential harms of stool-based blood tests are related to psychological harms of screening per se and of a positive test; these were reported to be mild and transitory ^36,37. In addition, unnecessary referrals, and medical harms linked to follow-up colonoscopy and surveillance after a positive test can occur (see below).

In modeling studies, all stool-based blood tests provided gains in quality-adjusted life years compared with no screening, especially FIT and gFOBT of higher sensitivity ³⁸.

Overall, sufficient evidence supports that the benefits outweigh the harms for colorectal cancer screening with any type of stool-based blood test.

Endoscopic methods

Four large randomized clinical trials of sigmoidoscopy screening have been performed, three in Europe and one in the USA ^39–42. Statistically significant reductions of 18% to 26% were observed consistently in all studies for colorectal cancer incidence, and in all but one study ⁴² for colorectal cancer mortality (22–31% risk reduction). An extended follow-up of one trial up to 17 years showed a persistent statistically significant reduction of 26% in colorectal cancer incidence and of 30% in colorectal cancer mortality, in intention-to-treat analyses ⁴³. Four randomized trials of colonoscopy are currently in progress, but data on the effect on colorectal cancer incidence or mortality are not yet available.

A large number of observational studies were available for review; only those performed in a screening setting were included for evaluation, conducted mainly in the USA. Two cohort studies ^44,45 provided estimates on colorectal cancer incidence and/or mortality for sigmoidoscopy, and five cohort studies ^45–49 for colonoscopy. In addition, nine and five case–control studies, including several studies of more than 2000 persons, provided risk estimates for sigmoidoscopy and colonoscopy, respectively. Statistically significant risk reductions in both colorectal cancer incidence and mortality were reported in most cohort and case–control studies, although the relative risks varied greatly between studies. The most recent meta-analysis of observational studies estimated risk reductions in both incidence and mortality of almost 70% for colonoscopy and almost 50% for sigmoidoscopy. The effect was consistently stronger in the distal compared with the proximal colon ⁵⁰.

Sufficient evidence supports that a single screening with sigmoidoscopy or colonoscopy reduces colorectal cancer incidence and mortality (Table 1). In reaching the evaluation for colonoscopy, in addition to the consistent results from the observational studies with colonoscopy, evidence from randomized trials of sigmoidoscopy screening was taken into account: given that a full colonoscopy, by definition, includes a sigmoidoscopy and assuming comparable false negative rates for both procedures, colonoscopy will be at least as effective as sigmoidoscopy at detecting advanced adenomas and colorectal cancer. Insufficient evidence is currently available to assess the benefit of subsequent rounds of endoscopic screening.

Similar to stool-based blood tests, endoscopic screening may generate psychological harms, and unnecessary referrals after positive sigmoidoscopy. In addition, endoscopy may provoke serious medical harms, of which bleeding and perforations are the most frequent, although such harms remain uncommon (each occurs in 0.01–0.05% of colonoscopy procedures) ⁵¹. The proportion of overdiagnosis of cancer from endoscopic screening is uncertain.

In modeling studies, sigmoidoscopy and colonoscopy both provide gains in quality-adjusted life years compared with no screening ⁵².

Overall, sufficient evidence supports that the benefits of a single screening with sigmoidoscopy outweigh the harms.

Regarding colonoscopy, the consensus was that sufficient evidence supports that the benefits of a single screening with colonoscopy outweigh the harms, when screening can be delivered with high quality. A minority of the expert panel considered that the evidence is limited, due to the variability and the related limited accuracy of the effect estimates, the harms associated with colonoscopy, and the inherent limitations in extrapolating from data of screening with sigmoidoscopy.

Computed tomographic colonography

No published randomized clinical trials have assessed the effect of CT colonography screening on colorectal cancer incidence or mortality. One randomized trial ⁵³ and four tandem studies ^54–57 with consecutive or parallel screening of asymptomatic individuals compared adenoma detection rates with CT colonography versus colonoscopy, and were considered informative for the evaluation. In the tandem studies (a comparison study where the same person is screened sequentially with two methods), the detection rates of advanced neoplasia (advanced adenoma and cancer) were similar with both techniques; in the randomized trial, detection rates with CT colonography compared with colonoscopy were similar for colorectal cancer, but were lower for advanced adenomas (5.6% vs 8.2%) and for advanced adenomas ≥ 10 mm (5.4% vs 6.3%); this difference disappeared when adjusting for participation rate.

Harms associated with CT colonography include radiation-induced effects, the downstream effects from detection of extracolonic findings ⁵⁸, and the potential harms of follow-up colonoscopy. Based on these data, limited evidence supports that a single screening with CT colonography reduces colorectal cancer incidence and/or mortality. A minority of the expert panel considered that the evidence is inadequate, due to the lack of randomized trials or observational studies with incidence and mortality as end-points, the lack of studies with repeated CT colonography screening, the fact that only test performance and adenoma detection rates were available, and the wide extrapolation needed from the known detection rates of lesions to an expected reduction in colorectal cancer incidence and mortality in a screening setting.

Finally, inadequate evidence supports that the benefits of a single round of screening with CT colonography outweigh the harms.

Comparative effectiveness of main colorectal cancer screening techniques

Comparisons of colorectal cancer incidence and mortality reduction with stool-based versus endoscopic methods were available from network meta-analyses (using indirect comparisons of studies of screening versus no screening). One meta-analysis of 9 randomized clinical trials ⁵⁹ concluded that sigmoidoscopy performed better than gFOBT in reducing colorectal cancer incidence, but not mortality. Another meta-analysis that included randomized trials and observational studies in screening settings ⁶⁰ reported colonoscopy to be more effective than sigmoidoscopy and gFOBT in reducing colorectal cancer mortality, although the quality of the evidence as low because of the heterogeneity in study designs and inherent biases in such comparisons. In addition, when comparing the performance of a single screening round, endoscopic techniques, especially sigmoidoscopy, generally yielded higher detection rates of advanced neoplasia than one-time stool-based blood tests ^61–64. Recent data, however, suggest that detection rates of advanced neoplasia are similar in a program of biennial FIT over five consecutive screening rounds when compared with one-time colonoscopy ⁶⁵. Taken together, the evidence was considered insufficient to draw a conclusion.

In conclusion, sufficient evidence supports that screening for colorectal cancer with currently established stool-based blood tests (gFOBT and FIT), and lower endoscopy (sigmoidoscopy and colonoscopy) decreases colorectal cancer-specific mortality, and that the benefits outweigh the harms. Evidence from comparative effectiveness data to evaluate one test over another was inconclusive.