Colorectal Cancer Screening: Confusion Reigns

In 2008, it is estimated that 148,810 new cases of colorectal cancer (CRC) will be diagnosed in the United States and that 49,960 people will die from the disease (1); lung cancer is the only cancer that kills more people each year in developed countries than CRC. According to almost universal consensus based largely on indirect evidence from a decade or more ago, effective implementation of screening would substantially reduce CRC morbidity and mortality (Table 1).

The population is divided into people at average (∼75%) and increased (∼25%) risk for CRC. People at average risk for CRC are the topic of the current discussion. The lengthening list of screening test options for average-risk subjects includes guaiac-based fecal occult blood testing (gFOBT), fecal immunochemical testing (FIT), stool DNA, double-contrast barium enema (DCBE), flexible sigmoidoscopy (FSIG), optical colonoscopy (OC), and computed tomographic colonography (CTC), which is colloquially referred to as virtual colonoscopy. The lengthening list of options has just been further complicated by suggested new joint guidelines from the American Cancer Society, the U.S. Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology (Table 2; ref. 2).

The U.S. Preventive Services Task Force assigns its strongest grade A category of recommendation “that clinicians screen men and women 50 years of age or older for colorectal cancer”¹

(accessed 7/13/2008). The U.S. Preventive Services Task Force approves the use of FOBT, FSIG, OC, and DCBE but not CTC or stool DNA. Until the latest American Cancer Society revisions, the American Cancer Society and U.S. Preventive Services Task Force guidelines were broadly similar (3). What is the current status of CRC screening in this setting of multiple alternative tests and changing recommendations?

In the general population of the United States, FOBT use is in decline and OC is supplanting FSIG (4). According to an analysis of data from the National Health Interview Survey for 2005, CRC screening rates exceeded 50% only in high-income men and women (5). Rates in low-income men and women were 32 and 35%, respectively. Will the new guidelines, summarized in Table 2, aid efforts to increase the uptake of CRC screening and to what extent will the new additions to the list of approved tests increase screening accuracy?

The decision to stratify tests into those that “detect adenomatous polyps and cancer” and “tests that primarily detect cancer” is baffling if an important purpose is to promulgate simple instructions for the health care community and general public that will increase CRC screening rates. Advanced colorectal adenomas (CRA), comprising CRAs with a diameter of ≥10 mm or CRAs of any size with villous histology or high-grade dysplasia (6), are those most likely to progress to CRC. The public health message should unequivocally be that preferred screening methods are those most likely to lead to the diagnosis of advanced CRAs as well as CRCs; CRCs and advanced CRAs are combined in the category of advanced colorectal neoplasms. A strong case can be made for reducing the number of recommended tests.

The sensitivity for advanced colorectal neoplasms of a “one-time” (i.e., three test cards and sampling of three bowel movements) gFOBT is only 24% (7). A more sensitive gFOBT and FITs have been developed in attempts to improve on gFOBT performance. The yield of advanced CRAs with the newer tests is not substantially improved but FIT sensitivity for asymptomatic CRCs approaches 66% (8, 9) compared with 26% with traditional gFOBT (10). Reluctance to jettison gFOBT is puzzling given the superior sensitivity of FIT for CRC.

The allure of molecular diagnostics has fueled considerable interest in stool DNA detection as a novel method for CRC screening; the concept is that neoplasm-specific DNA from CRA or CRC cells shed into the fecal lumen can be detected in stool. However, the methodology for stool DNA analysis is a work in progress. According to a web announcement on June 5, 2008,²

LabCorp has stopped offering what was at the time the only commercially available stool DNA test (PreGen-Plus), which required collection and shipping within 24 hours of collection an entire refrigerated stool specimen. In a recent experimental laboratory study comparing two generations of stool DNA tests with gFOBT, the latest stool DNA test detected 46% of advanced CRAs with a diameter of ≥10 mm compared with 17% with a sensitive gFOBT (11). There are no published studies comparing the sensitivity and specificity of stool DNA and FIT for asymptomatic CRC in a screening setting. Given the evolving nature of the technology and the lack of widespread evaluation of stool DNA testing in a population setting, not to mention the current lack of availability of any commercial stool DNA test, it is surprising, to say the least, that this technology has been recommended by the American Cancer Society and other professional organizations for CRC screening in the general population.

What are the relative merits of invasive (FSIG, OC, and DCBE) and noninvasive (CTC) methods of colorectal structural evaluation for CRC screening compared with stool-based technologies? Many would argue that DCBE is obsolete as a screening tool. It is rarely used for this purpose in clinical practice and the test is rapidly disappearing from radiology practice. In contrast to the United Kingdom and other countries, screening FSIG is rapidly disappearing from the practice scene in the United States. The relative merits of OC and CTC as screening tools are coming into clearer focus. In a recent comparison, the prevalence of advanced neoplasia in patients undergoing primary CRC screening by CTC or OC was 3.2% and 3.4%, respectively (12). Although there is now broad agreement that primary screening OC and CTC have equivalent sensitivity and specificity for detecting advanced neoplasms with a diameter of ≥10 mm, several other controversies concerning the relative merits of OC and CTC are unresolved.

Some argue that inferior sensitivity for lesions with a diameter of <10 mm is a drawback to CTC. Others have drawn attention to the prevalence of nonpolypoid (“flat”) CRAs, which might be more easily missed at CTC than OC (13), and reported a greater association for nonpolypoid than polypoid CRAs with carcinoma (14). What is beyond dispute is that CTC is greatly superior to gFOBT and FIT as a screening test for advanced neoplasia and CRAs of any size. Furthermore, devotees of OC tend to overlook the variable quality of screening OC in community practice. The length of time spent inspecting the colon during withdrawal of the instrument directly influences the sensitivity of OC, regardless of the colonoscopists' experience (15). Colonoscopists with mean withdrawal times of ≥8 minutes have higher rates of detection of any neoplasia and of advanced neoplasia compared with those with mean withdrawal times of <8 minutes (16).

Could the latest joint guidelines for CRC screening issued by the American Cancer Society, the U.S. Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology be simplified with public health benefit? I suggest that the focus should be on detecting advanced colorectal neoplasms without the test stratification shown in Table 2. A robust case can be made that FIT should supersede gFOBT. The case that stool DNA testing should be introduced into usual practice is extraordinarily weak and should be rejected until reputable tests are available. CTC has been thoroughly validated as a reliable method for CRC screening. OC remains the benchmark and the underlying premise of screening is that all average-risk patients who test positive by a primary method other than OC will be referred for diagnostic/therapeutic OC. A streamlined alternative to the new joint guidelines is proposed:

• OC if available.

• CTC if screening OC is declined or is unavailable.

• FIT if OC and CTC are unavailable.

Dissemination of guidelines for CRC screening is intended to increase the number of at-risk people who undergo screening. The purposes of screening are to identify and treat patients with premalignant CRAs and to maximize the proportion of patients whose CRC is diagnosed and treated at the localized stage while prognosis is excellent. In the recently issued revised CRC screening guidelines, the complexity of recommendations, which were already less than emphatic, is increased without good evidence that their implementation will increase screening uptake or the yield of screen-detected early stage disease. There is sufficient evidence to propose dispensing with several screening technologies, specifically gFOBT, stool DNA detection (until the technology is greatly improved), and DCBE. There is a pressing need for careful studies of alternative, simpler protocols, such as the OC or CTC or FIT strategies proposed above. The goal of these studies would be to optimize the costs and detection rates of CRC screening.