U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Lin JS, Piper MA, Perdue LA, et al. Screening for Colorectal Cancer: A Systematic Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2016 Jun. (Evidence Syntheses, No. 135.)

Cover of Screening for Colorectal Cancer

Screening for Colorectal Cancer: A Systematic Review for the U.S. Preventive Services Task Force [Internet].

Show details



The U.S. Preventive Services Task Force (USPSTF) will use this report to update its 2008 recommendation on screening for colorectal cancer.

Condition Background

Condition Definition

Colorectal cancer (CRC) or colorectal adenocarcinoma is a malignant tumor arising within the walls of the large intestine, which comprises the following segments: the cecum, ascending colon, transverse colon, descending colon, sigmoid, and rectum. CRC does not include tumors in the tissues of the anus or the small intestine. Adenomas are benign epithelial tumors or polyps that can progress to adenocarcinomas (Table 1). Adenomas or adenomatous polyps can be pedunculated (polypoid) or sessile (flat). Adenomas can have different degrees of dysplasia or different histologic characteristics (i.e., tubular, tubulovillous, and villous). Advanced adenomas are benign tumors with an increased likelihood to progress to CRC. The term advanced neoplasia, on the other hand, refers to a composite outcome of advanced adenomas and all stages of CRC (Table 1). Although there is some variation in the exact definition of advanced adenomas, they generally refer to adenomas 1 cm or larger, with villous components (tubulovillous or villous), or with high-grade or severe dysplasia.

Table 1. Definitions of Terms Describing Colorectal Cancer and Its Precursor Lesions.

Table 1

Definitions of Terms Describing Colorectal Cancer and Its Precursor Lesions.

Prevalence and Burden of Disease

CRC causes significant morbidity and mortality in the United States. Although CRC incidence rates have been declining for the past 20 years, among all cancers, CRC is third in incidence and cause of cancer death for both men and women.1 In 1999, the National Program of Cancer registries estimated the age-adjusted incidence rate of invasive CRC to be 56.5 cases per 100,000 persons. By 2011, the estimate had fallen to 39.9 cases per 100,000 persons.2 The National Cancer Institute (NCI) estimates that more than 50,000 persons will die in the United States from CRC in 2014.3 Data from the NCI’s Surveillance, Epidemiology, and End Results (SEER) Program from 2007–2011 indicate that the annual incidence of CRC in the United States is 43.7 cases per 100,000 persons, with approximately 95 percent of diagnoses occurring in adults older than age 45 years.3 The lifetime risk of acquiring CRC in the United States is about 5 percent, with an age-adjusted death rate of 15.9 deaths per 100,000 persons. Survival largely depends on the stage of cancer at the time of diagnosis. Patients with localized disease at diagnosis have a 5-year survival rate of 90 percent. Five-year survival rates drop to 70 percent, however, for those diagnosed with regionalized disease (cancer spread to regional lymph nodes). These rates drop to 12 percent for those with distantly metastasized disease.3

Increasing age, male sex, and black race are all associated with an increased incidence of CRC (Table 2). The median age at diagnosis is 68 years, and nearly half of all new cases are diagnosed in persons ages 65 to 84 years.3 Black men and women have the highest incidence of CRC compared to other racial/ethnic subgroups. This is troubling given that black men and women also have a disproportionately high mortality from CRC.4,5 This disparity has increased in the past 20 years, illustrated by the fact that CRC mortality rates have decreased more among whites than blacks.6 While the overall annual CRC-related death rate is 19.1 deaths per 100,000 men and 13.5 deaths per 100,000 women, it is 27.7 deaths per 100,000 black men and 18.5 deaths per 100,000 black women, which is nearly double the mortality for Hispanics and Asians or Pacific Islanders.3

Table 2. Age-Specific Colorectal Cancer Incidence Rates per 100,000 by Race/Ethnicity, United States, 1999–2011.

Table 2

Age-Specific Colorectal Cancer Incidence Rates per 100,000 by Race/Ethnicity, United States, 1999–2011.

Natural History

CRC usually develops over a period of several years, with the cancer beginning as a precancerous lesion.7,8 Experts estimate that at least 95 percent of cases of CRC arise from preexisting adenomas.9,10 This hypothesis that CRC arises from an adenoma-carcinoma sequence initially came from observations of a greatly elevated CRC risk status in patients with hereditary polyposis syndromes1113 and from observational studies showing a reduction in CRC incidence after polypectomy during colonoscopy or flexible sigmoidoscopy (FS).1421

Colorectal adenomas are very common. Based on a review of 14 studies (n=13,618), for example, the prevalence of adenomas in average-risk screening populations ranged from 22 to 58 percent.22 While adenomas can develop into cancers, most do not. Each adenoma’s tendency toward net growth or regression, however, may vary by polyp size and histology, as well as by other characteristics such as patient age, tumor location, and number of lesions.23,24 In general, larger adenomas and those with greater dysplasia are more likely to progress to cancer.25 Sessile serrated adenomas, as opposed to other adenomas, may not have dysplasia but do have malignant potential.26 These lesions are the major precursor lesion of serrated pathway cancers and are thought to represent 20 to 35 percent of CRC cases.26 Overall, the rate of progression of adenoma to cancer is variable and unknown, such that some lesions grow quickly and others very slowly. Better understanding of both the natural history of smaller adenomas and differences in the natural history of proximal versus distal lesions has implications for screening, as certain modalities may be better suited toward identifying smaller or proximal lesions.

Small Polyps or Adenomas (6–9 mm)

While there is general agreement that the risk of in situ cancer, or progression to cancer, for polyps 10 mm or larger is sufficiently high as to require immediate removal, the necessity and benefit of removing small polyps (<10 mm) is not clear.27,28 This stems from the fact that the natural history of smaller adenomas, particularly those 6 to 9 mm, remains uncertain. Greater understanding of the natural history of small adenomas will influence choice and implementation of screening test as well as definitions of test positivity (e.g., referral, polypectomy, or surveillance criteria for endoscopy and computed tomographic colonography [CTC]). In addition, unnecessarily removing smaller polyps can increase the risk of harms, including bleeding and perforation. Although promising, in vivo polyp discrimination methods are not yet (widely) used in clinical practice to distinguish neoplastic from nonneoplastic lesions.29,30

Studies using colonoscopy registries report the prevalence of advanced histology or CRC in polyps of various sizes. A limited number of studies have been conducted in screening cohorts. A systematic review by Hassan and colleagues, for example, assessed the distribution of advanced adenomas in average-risk screening populations according to polyp size and reported that the overall prevalence of advanced adenomas was 5.6 percent (95% CI, 5.3 to 5.9) in four studies (n=20,562). Polyps <10 mm were very common in this sample. The prevalence of diminutive polyps (≤5 mm) was 27 percent, prevalence of small polyps (6–9 mm) was 9 percent, and prevalence of large polyps (≥10 mm) was 6 percent. Diminutive polyps (≤5 mm) as the largest lesions accounted for 4.6 percent (95% confidence interval [CI], 3.4 to 5.8) of patients with advanced adenomas. Small polyps (6–9 mm) accounted for 7.9 percent (95% CI, 6.3 to 9.4) of patients with advanced adenomas. In contrast, large lesions (≥10 mm) accounted for 87.5 percent (95% CI, 86.0 to 89.4) of advanced adenomas.31 The largest screening study included in this review31 was a prospective cohort derived from the Clinical Outcomes Research Initiative (CORI) database by Lieberman and colleagues.32 In this study, polyps 6–9 mm were detected in 9.1 percent (1,275/13,992) of patients. The proportion of advanced histology was 6.6 percent in those with polyps 6–9 mm. Only two of these patients had CRC (0.2%).

Until very recently, only small, pilot-sized studies conducted in nonscreening populations have followed the natural history of smaller (<10 mm) lesions. These were observed in situ by serial endoscopy, suggesting that many remain dormant or regress during a 2- to 3-year period.23,33 More recently, however, a large cohort (n=22,006) of asymptomatic adults undergoing routine CRC screening with CTC at two U.S. medical centers has been published. In this study, the volumes and linear sizes of polyps in vivo were measured with CTC at baseline and surveillance (mean surveillance interval, 2.3 years).34 Nine percent (1,982/22,006) of adults had small polyps (6–9 mm) at baseline. Of the 306 small polyps in 243 adults who were followed with CTC surveillance, 22 percent (68/306) progressed (≥20% growth), 50 percent (153/306) were stable, and 28 percent (85/306) regressed (≥20% reduction). Histology was established in 43 percent of polyps (131/306) after final CTC. Ninety-one percent (21/23) of proven advanced adenomas compared to 37 percent (31/84) of proven nonadvanced adenomas progressed.

Proximal Versus Distal Lesions

The distal large intestine can be defined as distal to the splenic flexure (including the descending colon, sigmoid colon, and rectum). Some definitions are more limited and include only the sigmoid colon and rectum, or exclude rectal cancers (for a distinction between the distal large intestine vs. the distal colon). The proximal large intestine or colon is generally defined as proximal to the splenic flexure (including the cecum, ascending and transverse colon) (Figure 1).

Figure 1 illustrates the locations in the large intestine: proximal colon (cecum, ascending, hepatic flexure, and transverse colon) and distal colon (splenic flexure, descending, sigmoid colon, and rectum).

Figure 1

Locations in the Large Intestine: Proximal Colon (Cecum, Ascending, Hepatic Flexure, and Transverse Colon), Distal Colon (Splenic Flexure, Descending, Sigmoid Colon, and Rectum). Source: http://cisnet.cancer.gov/projections/colorectal/screening.php

While overall CRC incidence and mortality is decreasing over time, this trend is more apparent in distal than proximal cancers.35,36 Data from the NCI’s SEER Program, for example, demonstrate a proximal migration of CRC in the past two decades, which is attributed to a decrease in incidence of distal CRC (i.e., screening for primary prevention of cancer) and an aging population in which proximal lesions are more common.37 A growing body of evidence also suggests that colonoscopy is less effective in reducing proximal compared to distal CRC incidence and mortality.3842 The reason for this finding remains unclear, however, and we do not know if this discrepancy is due to inadequate quality/implementation of colonoscopy (e.g., failure to reach the cecum, poor bowel preparation) and/or to biologic differences in the types of lesions and natural history of lesions in the proximal versus distal large intestine. It is well established that there are physiological differences between the proximal and distal large intestine as well as differences in proximal and distal CRC.43 Cancers in the proximal and distal colon appear to arise from different molecular pathways (e.g., microsatellite instability and BRAF mutations in proximal cancers).43,44 Molecular differences may explain differences in morphology (e.g., higher proportion of flat polyps in the proximal colon) and natural history (e.g., hypothesized more rapid progression of adenoma to cancer).45

Based on data from the NCI’s SEER Program and the North American Association of Central Cancer Registries from 2006–2010, the age-adjusted incidence of cancer is 22.6 cases per 100,000 persons in the distal colon/rectum and 18.9 cases per 100,000 persons in the proximal colon. The proportion of proximal to total cancers is 42 percent.46 CRC prognosis and mortality also differ by tumor location in the colon. Analyses of SEER data have shown a higher late- to early-stage incidence for proximal compared to distal colon/rectum cancer.47 Proximal cancers have lower 5-year survival and greater mortality and SEER data show differences in stage at presentation.

Adenomas also appear to be more common in the distal colon/rectum than in the proximal colon. In the National Polyp Study, for example, the proportion of proximal to total adenomas was 36 percent.21 In more recent screening colonoscopy or CTC cohorts, the proportion of proximal to total adenomas ranges from 27 to 52 percent.4852 Data suggest that there is a higher rate of invasive cancer in adenomas in the rectum versus the colon; however, it is still unclear if there is a significant difference in cancer rates in adenomas in the proximal versus distal colon.53 One large retrospective cross-sectional analysis suggests that proximal polyps with advanced neoplasia are smaller than distal polyps (7.6 vs. 11.1 mm, respectively).54

The distribution of CRC (and adenomas) differs by age, sex, and race/ethnicity. The incidence of proximal cancers as well as the proportion of proximal cancers (to total cancers) is higher with advancing age.46 Again, based on data from the NCI’s SEER Program and the North American Association of Central Cancer Registries from 2006–2010, proximal cancers are also more common in women than in men; the proportion of proximal to total cancers is 46 versus 38 percent, respectively.46 Despite this difference, however, men have higher rates of CRC (distal and proximal) incidence and mortality.46

Based on SEER data, black men and women appear to have a higher proportion of proximal cancers than other racial/ethnic groups. In addition, 5-year survival rates for proximal cancers are worse for blacks (best for Asians and Pacific Islanders), and these survival disparities persist after adjusting for age, sex, stage of presentation, and therapy received.55 Although poverty is a confounder for CRC incidence and survival, recent data suggest that socioeconomic status plays a more prominent role for distal colon and rectal cancers than proximal cancers in whites, blacks, and Asians and Pacific Islanders.47

There is some evidence from separate analyses conducted from screening colonoscopy cohorts derived from the CORI database on the difference of prevalence and distribution of polyps among different racial/ethnic subgroups. However, the clinical importance of some of these differences is still unclear. These studies found that blacks (both men and women) had higher prevalence of large adenomas and proximal lesions (adenomas and advanced neoplasia).5659 Based on analogous data from CORI, there does not appear to be a difference in the distribution of large adenomas in Hispanics compared to whites, although Hispanics appear to have a lower age-adjusted prevalence of large adenomas than whites.59,60

Risk Factors

Most cases of CRC are sporadic, with 75 percent developing in average-risk persons, versus about 20 percent developing in persons with some type of family history. The remainder of cases develop in persons who have predisposing inflammatory bowel disease or a known inherited familial syndrome (defined by mutations in known high-risk cancer susceptibility genes), including familial adenomatous polyposis and Lynch syndrome (previously known as hereditary nonpolyposis colorectal cancer).6164 Family history of CRC that is not attributable to any known inherited syndromes is a well-established risk factor, with an average 2- to 4-fold increase in risk of CRC compared to those with no family history. Despite this finding, however, there is great heterogeneity in the published literature in how family history is defined (e.g., the age, number, and relationship to relative[s] with CRC).6567 As a result, the risk of developing CRC varies approximately 20-fold between persons in the lowest quartile (average lifetime risk, 1.25%) and the highest quartile (average lifetime risk, 25% in persons with an inherited familial syndrome).68

Some lifestyle factors have also been linked to risk of developing CRC, including lack of exercise, long-term smoking, heavy alcohol use, being overweight or obese, and having type 2 diabetes.1 Despite the large range in risk and known risk factors for CRC, risk prediction and use of risk prediction models for CRC is suboptimal.69

CRC Screening

Rationale and Current Clinical Practice

Because CRC has precursor lesions and survival largely depends on the stage at the time of diagnosis, screening can affect both primary prevention (finding precancerous lesions that could later become malignant) and secondary prevention (detecting early cancers that can be more effectively treated).

Large, well-conducted randomized, controlled trials (RCTs) have demonstrated that screening for CRC can reduce disease incidence and disease-specific mortality. The decrease in CRC incidence and mortality in the past two decades in the United States corresponds to an increase in self-reported screening rates from less than 25 percent in the 1980s to about 52 percent in 2002 and about 65 percent in 2012.70 Despite increases in CRC screening over time, screening rates remain well below optimal, as evidenced by the fact that approximately 28 percent of U.S. adults eligible for screening have never been screened for CRC.70 There is also evidence of racial/ethnic and socioeconomic disparities in CRC screening, with lower rates of CRC screening in nonwhite and Hispanic populations and less educated adults.71 Multiple patient, clinician, and health care delivery factors have been found to negatively influence CRC screening, including low socioeconomic or educational status, lack of physician recommendation, and lack of insurance or limited access to health care.72

Screening Tests

Multiple tests are available to screen for CRC, including stool-based tests (e.g., guaiac-based fecal occult blood test [gFOBT], fecal immunochemical test [FIT], fecal DNA tests), endoscopy (e.g., FS or colonoscopy), and imaging tests (e.g., double-contrast barium enema [DCBE], CTC, magnetic resonance colonography [MRC], capsule endoscopy). Screening tests currently used in the United States that have evidence to support their use include high- sensitivity gFOBT (Hemoccult SENSA®; Beckman Coulter, Brea, CA), FIT, FS, and colonoscopy.73

Despite being designated under a single test type, FITs are not a homogeneous class of stool testing. In fact, various types of FITs are available from multiple manufacturers (and therefore different proprietary names), with differing test methods and performance characteristics. Of the FITs available in the United States, some have been reviewed by the U.S. Food and Drug Administration (FDA) and cleared as test kits via 510(k) review, while many more have been granted waived status by the FDA.74 Waived status may be granted under the Clinical Laboratory Improvements Amendments of 1988 if the device is simple to use, is demonstrated at intended use sites to be accurate, and poses an insignificant risk of erroneous results. In contrast to FITs, high-sensitivity gFOBT is produced in the United States by one primary manufacturer (Hemoccult SENSA, Beckman Coulter). Stool testing is generally performed on spontaneously voided stool samples, as opposed to in-office stool samples obtained by digital rectal examination, because of the less sensitive or unclear test performance of the latter.75,76

Since 2001, when the Centers for Medicare & Medicaid Services (CMS) started covering screening colonoscopy, colonoscopy utilization for screening has increased and the use of FS has decreased.77,78 Despite lack of RCT evidence demonstrating a reduction in CRC mortality from a program of screening with colonoscopy, and some studies suggesting screening colonoscopy is not as effective in reducing incidence of or mortality from proximal compared to distal CRC,40,41, 7981 colonoscopy remains the most commonly used screening modality in the United States.78,82 In 2012, for example, 62 percent of persons who were screened had colonoscopy compared to 10.4 percent who were screened with stool testing and only 0.7 percent who were screened with FS in combination with stool testing.70 Public and clinician perceptions of accuracy of colonoscopy versus FS, given the reach of endoscopy, also play an important role in this issue.83 Newer technologies, specifically CTC and stool DNA testing, have a growing evidence base, and may play an important role in CRC screening. In 2013, the FDA Medical Advisory Panel agreed that the benefits of using CTC to screen for CRC outweigh the risks (e.g., radiation exposure and identification of extracolonic findings).84 Only one stool DNA test, a multitarget stool DNA (mtsDNA) test incorporating FIT testing, is currently available and approved by the FDA for use for CRC screening. One new blood test to detect circulating methylated septin 9 gene DNA (mSEPT9) is currently available.

While other screening tests are available for CRC, they are no longer widely used. The original gFOBT (i.e., Hemoccult I or II), for example, has largely been replaced by stool testing with higher sensitivity (i.e., Hemoccult SENSA or selected FITs). DCBE is also largely no longer used because of its suboptimal performance compared to other screening tests.73 Two newer technologies, MRC and capsule endoscopy (PillCam®; Given Imaging, Yokne’am Illit, Israel), are primarily used as diagnostic tools and are not currently used as screening tests. MRC, similar to CTC, can image the lumen of the colon but without the radiation exposure. Capsule endoscopy has the advantage of being noninvasive and requiring no sedation. Thus far, however, the efficacy of MRC and capsule endoscopy in screening populations have only had limited evaluation in small studies.85,86

Current Screening Recommendations

Most organizations agree that any CRC screening is better than no screening. Existing guidelines recommend that the age to begin screening in adults at average risk for CRC is 50 years. However, the optimal age to start screening may vary by sex or race/ethnicity based on differences in onset and incidence of CRC. The optimal time to stop screening in average-risk adults is uncertain, such that screening from ages 76 to 85 years should be individualized based on the patient’s comorbid conditions and prior screening results.

Currently, most U.S. guideline organizations, including the USPSTF, agree that the recommended options in screening for CRC include: colonoscopy every 10 years, annual high-sensitivity gFOBT or FIT, and FS every 5 years with stool blood testing (FOBT or FIT).87,88 There remains a number of important areas of disagreement about these options, however, as reflected by the variation in screening recommendations across professional societies in the United States and internationally (Appendix A Table 1).

The largest difference in recommendations exists between the USPSTF’s 2008 recommendation and the American Cancer Society (ACS), U.S. Multi-Society Task Force (MSTF), and American College of Radiology (ACR) 2008 joint recommendations (Appendix A Table 1).73,87,88 While the USPSTF recommendations stated that any number of options (listed above) are suitable for CRC screening, the ACS-MSTF-ACR joint recommendations supported newer technologies (i.e., stool DNA testing and CTC) and gave preference to “structural exams,” including colonoscopy and CTC as a means of preventing CRC.

Previous USPSTF Recommendation

In 2008, the USPSTF issued the following recommendations about screening for CRC:

  • The USPSTF recommends screening for CRC using FOBT, sigmoidoscopy, or colonoscopy in adults, beginning at age 50 years and continuing until age 75 years (A recommendation).
  • The USPSTF recommends against routine screening for CRC in adults ages 76 to 85 years (C recommendation). There may be considerations that support CRC screening in an individual patient.
  • The USPSTF recommends against screening for CRC in adults older than age 85 years (D recommendation).
  • The USPSTF concludes that the evidence is insufficient to assess the benefits and harms of CTC and stool DNA testing as screening modalities for CRC (I statement).

The USPSTF determined that for all screening modalities, starting screening at age 50 years resulted in a balance between life-years gained and colonoscopy risks that was more favorable than commencing screening earlier. Despite the increasing incidence of colorectal adenomas with age, for individuals previously screened, the gain in life-years associated with extending screening from age 75 to 85 years was small in comparison to the risks of screening persons in this age group. For adults who have not previously been screened, decisions about first-time screening in this age group should be made in the context of the individual’s health status and competing risks, given that the benefit of screening is not seen in trials until at least 7 years later. For persons older than age 85 years, competing causes of mortality preclude a mortality benefit that outweighs the harms.

The USPSTF concluded that there was insufficient evidence to assess the sensitivity and specificity of stool DNA testing for colorectal neoplasia; therefore, the balance of benefits and harms could not be determined for this test. The USPSTF concluded that the evidence for CTC to assess the harms related to extracolonic findings was insufficient, and, as a result, could not determine the balance of benefits and harms. It did state, however, that the option of CTC could help reduce CRC mortality in the population if patients who would otherwise refuse screening found it to be an acceptable alternative.


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.7M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...