Abstract
Screening colonoscopy was first suggested as a potential future colorectal cancer (CRC) screening tool for average risk adults in 1988 in the absence of efficacy data, but with the hope that solid evidence in its support would accumulate in the future [Neugut and Forde, 1988]. While indirect evidence was published in the ensuing years (e.g. the ability of colonoscopy to detect and remove adenomas), until recently definitive proof of a mortality benefit has been elusive; indeed, whether it is superior to flexible sigmoidoscopy has been questioned [Neugut and Lebwohl, 2010]. Nonetheless, it has become the primary CRC screening modality in the United States [Joseph et al. 2012].
It is important to recognize that two criteria are the fundamental cornerstones by which a screening test is evaluated from a public policy perspective [Brawley and Kramer, 2005]. The first is whether the test reduces disease-specific mortality. The second criterion is one that is omnipresent in the evaluation of any intervention in medicine, i.e. that the risk–benefit ratio is worthwhile. Other endpoints are frequently discussed and used as surrogates or alternatives and, when this is done, the evaluation becomes confused. Examples in the case of CRC screening include adenoma detection, yield of CRC found, or the stage distribution of cancers detected. These may all be positive outcomes, and may all suggest that a potential screening test may have promise as a future addition to the menu of options. However, they are not the true criteria by which to establish the efficacy and role of a screening test: the reduction of CRC mortality. Even the reduction of CRC incidence [Brenner et al. 2011], which may seem like the equivalent of CRC mortality, is not a substitute for this endpoint.
The tests that are acceptable for CRC screening, and the preferences among them, have been in evolution. The United States Preventive Services Task Force and a guidance statement from the American College of Physicians list three acceptable tests: fecal occult blood testing (FOBT), sigmoidoscopy, and colonoscopy [U.S. Preventive Services Task Force, 2008; Qaseem et al. 2012]. Two of these tests have randomized trial-based evidence for improved mortality: FOBT and sigmoidoscopy. But colonoscopy, which is usually recommended as the premiere screening tool and is the most widely used CRC screening modality in the United States, has had only limited evidence of efficacy.
Over the last 3 years, there has been a plethora of new data regarding the efficacy of colonoscopy in the reduction of CRC mortality. While none are randomized trials (and such trials have only recently been initiated), they shed light on the shortcomings, as well as the potential, of screening colonoscopy as an effective public health intervention.
Colonoscopy earned its popularity in the absence of data showing a reduction of mortality or cancer incidence, and was primarily justified on the basis of a logical argument based on the proven efficacy of sigmoidoscopy; colonoscopy must be superior because it also explores the right side of the colon. Thus, the CRC screening community was taken aback by the first two observational studies of CRC mortality with colonoscopy in 2009 and 2010. While they showed a reduction in overall CRC mortality, this reduction was similar to that of sigmoidoscopy and colonoscopy did not show any benefit on the right side of the colon (see Table 1) [Baxter et al. 2009; Singh et al. 2010]. These two studies, performed in Canada rather than the US, were criticized since the bulk of exams were not performed by gastroenterologists, which could have accounted for the lack of efficacy on the right side of the colon. Nevertheless, these results led to the hypothesis, troubling to gastroenterologists as well as the general public which has embraced colonoscopy, that colonoscopy may not be effective in preventing death from proximal colon cancer.
Studies testing for an association between colonoscopy and mortality from colorectal cancer (CRC).
In parallel with these studies challenging the efficacy of colonoscopy in the proximal colon, came three randomized trials of sigmoidoscopy [Atkin et al. 2010; Segnan et al. 2011; Schoen et al. 2012]. Two of these three studies showed a reduction in mortality of 26% and 31% for CRC mortality overall, mostly limited to the left colon, while the third showed a reduction in incidence, thus reinforcing sigmoidsocopy’s efficacy as a screening tool [Inadomi, 2012]. Thus, we reached the conclusion that the available evidence was insufficient to support colonoscopy’s primacy as a CRC screening tool, and we raised the possibility that flexible sigmoidoscopy could rival or even supplant colonoscopy in an era of cost-conscious healthcare [Neugut and Lebwohl, 2010].
However, now the first population-based study of colonoscopy and CRC mortality in the US has been published [Baxter et al. 2012]. This game-changing case-control study of Medicare recipients by Baxter and colleagues demonstrates a 60% reduction in CRC mortality for colonoscopy, clearly superior to that of sigmoidoscopy. More importantly, while the study does not show the same reduction in mortality on the right and left sides of the colon, it does show good efficacy on both sides (76% reduction in mortality on the left, 42% on the right) (see Table 1). This efficacy on the right, not seen in the Canadian studies, is likely due in large part to the fact that 65% of the colonoscopies were performed by gastroenterologists versus 31% in one of the Canadian studies. Another study, limited to patients who underwent both colonoscopy and polypectomy, suggested a similar overall mortality reduction [Zauber et al. 2012].
While we do not yet have the results of randomized trials of colonoscopy, which will be available in 15 years, the study by Baxter and colleagues provides high-quality observational evidence that colonoscopy has a clear protective effect on CRC mortality. On the other hand, the evidence for sigmoidoscopy is firmly established on the basis of three randomized trials, the gold standard for efficacy. Extending the use of endoscopic screening to the proximal colon comes at the cost of substantial added invasiveness, risk, logistical effort on the part of the patient, and significant added economic cost. These issues will need to be weighed in conjunction with the putative gain in efficacy.
Cancer screening tests arouse more controversy than almost any area of clinical decision-making. To a large degree, this arises because the scientific evidence often does not support the clinical intuition that leads to the initial widespread adoption of a test before it is proven. Screening colonoscopy has persisted for decades in a similar limbo state, without definitive mortality evidence in its support, and with recent studies questioning its use. The study by Baxter and colleagues [Baxter et al. 2012] offers the first definitive evidence of superior efficacy, with mortality reduction throughout the colon, making ongoing efforts to promote colonoscopy appropriate.
Footnotes
Acknowledgements
Dr Neugut and Dr Lebwohl jointly conceived, drafted, and revised this manuscript.
Funding
Dr Lebwohl is supported by the National Center for Research Resources, a component of the National Institutes of Health (KL2 RR024157).
Conflict of interest statement
The authors declare no conflicts of interest in preparing this article.