html xmlns=”http://www.w3.org/1999/xhtml”>
18
Colorectal cancer: Population screening and surveillance
Rules of evidence and feasibility of evidence
The rules of evidence for evaluating studies of treatment interventions are well developed. The randomized controlled trial (RCT) is the scientific gold standard [1]. The prominence of RCT evidence is underscored in a comprehensive guideline that outlined five phases of biomarker development for the early detection of cancer, which described the final phase as “standard parallel-arm randomized clinical trial … with one arm consisting of subjects undergoing the screening protocol and the other arm consisting of unscreened subjects” [2]. However, large-scale RCTs are expensive to conduct and their results may not be available for many years. This is certainly the case for RCTs of colorectal cancer (CRC) screening if CRC mortality is the primary outcome. A good example is the UK Flexible Sigmoidoscopy Trial which compared flexible sigmoidoscopy as an intervention for CRC screening versus usual care, with CRC mortality as the primary outcome. After several years of work, during which the trial was designed, the protocol was developed, funding was obtained, patients were enrolled during 1996-1999. The baseline findings were published in 2002 [3]. The results for the primary outcome, CRC mortality, are not yet available in early 2010. Since the original conception of this trial, CRC screening technology has evolved considerably.
What to do when ideal evidence is lacking
The first US Multi-Society Task Force (USMSTF) guideline states that:
The authors believe that it might be appropriate in the future to substitute a newer test for currently recommended ones if there is convincing evidence that the new test has: (1) comparable performance (e.g., sensitivity and specificity) in detecting cancers or adenomatous polyps at comparable stages; (2) is equally acceptable to patients; and (3) has comparable or lower complication rates and costs. Provided this is the case, the authors feel that it would not be necessary to submit each new technology to the original standard of proof, i.e. a RCT with death from CRC as the outcome measure [4].
A recently published framework for evaluating diagnostic tests refines this approach by clarifying that the cases detected by the new technology should represent a similar spectrum of disease, so that the evidence of effectiveness applies to these new cases as well as the cases detected by the older technology [5]. It also goes one step further, indicating a role for decision analytic modeling when the new test has both favorable and unfavorable attributes, such as improved sensitivity but reduced specificity. A good example of the use of decision analysis is the most recent US Preventive Services Task Force (USPSTF) recommendations for CRC screening [6], which used microsimulation modeling to help inform the recommendations. As noted by the authors “well-validated microsimulation models may be used to highlight the tradeoff between clinical benefit and resource utilization from different screening policies and inform decision making with standardized comparisons of net benefits and risks” [7]. As new technologies arise that offer trade-offs in performance, acceptability to patients and costs, we have to acknowledge that RCTs with CRC mortality as the primary outcome will not be feasible, and we will need to agree on what constitutes adequate evidence.
Colorectal cancer: an ideal target for prevention and early detection through screening
Colorectal cancer is a common condition, with a pre-malignant lesion (the tubular adenoma) with a long dwell time, making it an ideal target for early detection and cancer prevention through screening. CRC is the second leading cause of cancer death in the USA and the third leading cause of cancer death among both men and women [8]. Over 148,000 people are diagnosed with colorectal cancer in the USA every year, and over 49,000 people die from it [8]. It is generally accepted that detection and removal of polyps at colonoscopy decreases colorectal cancer incidence. B The magnitude of that benefit is estimated to be between 76–90% [9]. There are data that show a gradual decline in CRC incidence in the USA since 1985 [10]. This incidence decline is greater for distal CRC than for proximal CRC. Possible explanations for this include: (1) earlier uptake of flexible sigmoidoscopy than colonoscopy for screening; (2) higher effectiveness of polypectomy for cancer prevention in the left than in the right colon due to colonoscopy technique or different polyp morphology; (3) biologic differences between distal and proximal cancers [11].
Screening for colorectal cancer is now widely recommended and screening programs are in place in many countries [12]. A number of randomized trials are also underway that evaluate flexible sigmoidoscopy compared with no screening [3, 13]. These data, coupled with evidence from the randomized trials of fecal occult blood testing [14–18], should provide ample evidence on the overall benefit of screening. As policy makers, clinicians and researchers launch their screening programs, they have to choose a test or strategy to implement, based on the trade-offs of operational demands, patient acceptance, capacity constraints and overall program cost. These decisions are often difficult and are often driven by political or economic issues.
Organized vs opportunistic screening
The International Agency for Research on Cancer (IARC) defines an organized screening program as one that has the following features: (1) an explicit policy with specified age categories, method and interval for screening; (2) a defined target population; ( 3) a management team responsible for implementation; (4) a health care team for decisions and care; (5) a quality assurance structure; and (6) a method for identifying cancer occurrence in the population [19]. In contrast, opportunistic screening is done outside of an organized screening program, often delivered through fee-for-service reimbursement of physicians. Compared with opportunistic screening, organized screening focuses much greater attention on the quality of the screening process including follow-up of participants [20]. Thus, a key advantage of organized screening is that it provides greater protection against the harms of screening, including over-screening, poor quality and complications of screening, and poor follow-up of those who test positive [20].
A critical appraisal has assessed the evidence for organized cancer screening programs by systematically evaluating the published literature from 1966 to 2002 [21]. The authors reported that although there is a substantial body of literature on organized cancer screening programs, most studies are descriptive, and of those that are evaluative, the focus is on components of the programs, rather than the organized screening programs as a whole. For the relatively few studies that evaluated the effectiveness of organized versus opportunistic screening, most are from the Scandinavian countries and focus on cervix cancer screening. The authors concluded that there is limited evidence to directly support the effectiveness of organized cervix cancer screening and somewhat weaker evidence for other cancers. The promise of organized cancer screening programs is that they achieve better accessibility, quality, accountability and outcomes. In this way, benefits are maximized and harms are reduced. We need further research that compares organized versus opportunistic CRC screening to determine whether organized screening delivers on this promise.
Fecal occult blood testing
Fecal occult blood testing (FOBT) is the only colorectal cancer screening approach demonstrated to be effective in randomized controlled trials. Depending on whether the tests were done biennially or annually, and whether the tests were or were not rehydrated, fecal occult blood testing is associated with a 15–33% reduction in colorectal cancer mortality [14–16, 18], and a 17–20% reduction in colorectal cancer incidence [17]. The randomized trials that provided this evidence were all conducted with the originally available, standard guaiac FOBT (gFOBT), which is still widely used throughout the world. Fecal blood testing, offers the advantage of being non-invasive and convenient for patients. Tests can be sent through the mail directly to patients, samples are collected in the privacy of their homes and can be returned by mail to a central processing laboratory. All of these are advantages for population screening.
Since the publication of those landmark studies, many newer tests for fecal blood have been approved for clinical use. There is evidence demonstrating that fecal immuno-chemical tests (FIT) are more sensitive then standard guaiac tests (GT), such as Hemoccult [22]. There is also evidence from a randomized trial that they are associated with improved adherence to screening, perhaps because they have improved collection devices, and require fewer samples and no dietary restrictions [23]. There is also mounting evidence that FITs have better sensitivity than high sensitivity GTs such as Hemoccult II SENSA, with preserved specificity [24, 25]. Selected FITs offer the option of automated test reading, with resulting improved precision and reliability of the interpretation, and also the possibility to report a quantitative result, but with resultant trade-offs in sensitivity and specificity [26].
Expected sensitivity of FITs
The standard hemoglobin concentration in most studies of FITs has been 100 ng of hemoglobin per ml of stool. At this concentration, results have varied somewhat between settings and depend on the number of samples and the gold standard used to determine the true prevalence of cancer. All studies have evaluated a single application of the FIT, leaving to modeling and speculation about the benefits of screening that accrue to patients who continue to screen in an annual or biennial program of screening. In a preventive health appraisal population from Kaiser Permanente, Allison used the three samples of the HemeSelect FIT and reported that the sensitivity for cancer was 68.8% and specificity for cancer was 94.4% [22]; this study used two-year follow-up for clinical cancer incidence as the gold standard for the presence of cancer. In a large screening colonoscopy cohort from Japan, Morikawa et al. reported a sensitivity of 65.8% and specificity of 95.5%, using a single application of the Magstream 1000 test [27]. The highest cancer sensitivity reported to date was 81.8%, with a specificity of 98.1%, using the three samples of the FlexSure OBT FIT and a definition of cancer that was a based on a combination of flexible sigmoidoscopy evaluation for determination of the presence of left-sided colorectal cancers and two years of clinical follow-up for the presence of proximal colorectal cancers [24].
An explicit evaluation of varying the numbers of samples and the target concentration of hemoglobin was performed by Levi et al. in a high risk colonoscopy population [26]. Increasing the number of specimens increases the sensitivity for cancer detection from 64.7 to 82.4% to 88.2%, for one, two or three specimens. Specificity declines with the addition of each sample, from 94.3% to 91.9% to 89.7%. Lowering the hemoglobin cutoff from 100 to 75, increases sensitivity by 0-6%, depending on the number of specimens and decreases specificity by 2% [26].
Studies that have directly compared the sensitive gFOBT (Hemoccult SENSA) to a FIT, in a population large enough to obtain precise point estimates of sensitivity and specificity have been difficult to evaluate. Each study used a different reference standard and a different population. Since the same criterion was applied to both tests in each study, they do provide some evidence of relative sensitivity and specificity. In the one study in which a standard GT (Hemoccult II) was used, both the sensitive gFOBT and the FIT showed improved sensitivity (37.1% for Hemoccult II vs 68.8% for the FIT vs 79.4% for SENSA). The major drawback with sensitive gFOBT has been its much lower specificity (94.4% for the FIT vs 86.7% for the gFOBT) [22].
At the current time, due to heterogeneity in the evidence, it is difficult to say that one FIT is clearly superior [28]. Sensitivity and specificity point estimates have varied across studies due to differences in populations and the criterion standard used to determine the true incidence of cancer. In most recent studies, however, the FIT has performed better than either standard GT or the increased sensitivity GT. In population screening practical considerations such as patient acceptance and reliability of results reporting, become nearly as important as pure test operating characteristics. This was recently demonstrated in Holland, where higher acceptance of FIT resulted in higher neoplasia detection [23]. In mass screening, the automated test processing available with the FIT offers an important advantage of improved test reliability and less risk of repetitive strain injury. The non-invasive, easily scaled option of the FIT is an important option for CRC screening applicable for most population screening programs.
Stool DNA
Tests for deoxyribonucleic acid (DNA) mutations in stool were first described in 1992 [29]. There are two pathways providing targets for fecal screening. Genomic instability is caused by oncogene and tumor suppressor gene mutations, believed to occur in a specific sequence as colonic epithelium progresses through the adenoma-carcinoma sequence [30]. Microsatellite instability may be inherited as mutations in specific genes or acquired through epigenetic DNA methylation, providing an alternative pathway to tumorigenesis, and another target for fecal screening [31]. The initial approach to fecal DNA testing was in the form of a multi-target assay panel, combining detection of mutations in chromosomal instability markers k-ras, p53 and APC with BAT-26 (a marker of microsatellite instability). A DNA integrity assay (long DNA) was added as a generalized marker of escape from apoptosis. The initial report demonstrated sensitivity of 93% for cancer and 87% for adenomas, with a specificity of 93% [32]. Other promising results were reported in several other small studies [33–36].
As evaluation of stool DNA moved from small, single center studies in high risk patients to large, multi-site clinical trials in average risk, asymptomatic screening subjects, sensitivity decreased. The sensitivity of the initial version of the stool DNA test, using 21 mutations in a range of markers (Version 1.0 or SDT-1) has been reported in larger studies to be between 25–52% for cancer and 18–20% for cancers plus advanced adenomas [37, 38]. This decrease in performance occurred for several reasons. The initially selected DNA mutations were not as commonly found among sporadic colorectal cancers and adenomas in a screening population compared to the high risk populations in the earlier studies. In addition, bacterial enzymes in stool digested the DNA during transport to a central testing site. The 21 marker panel is also very expensive to run, leading to a test that is too expensive to be clinically viable.
The testing panel has evolved to a more streamlined set of markers, primarily exploiting the hypermethylation of vimentin as a screening target, and a buffering system to preserve the long DNA marker during transport. In smaller, high risk patient populations, the newer assay (also known as Version 2.0) has demonstrated 86–88% sensitivity and 82–83% specificity for cancer [39, 40]. In the NCI funded screening study, the addition of a methylated vimentin based assay to a k-ras mutation assay combined with a scanning of cluster regions in APC, without the use of the buffer (SDT-2), was reported to be 58% sensitive for cancer alone, and 40% sensitive for cancer plus advanced adenomas [37]. Due to US FDA concerns, neither version 2.0 nor SDT-2 is available. A single marker test, based on hypermethylation of vimentin alone, is available from LabCorp. The performance of the current version of this test for screening patients is not known, and it has not been widely used.
Flexible sigmoidoscopy
Flexible sigmoidoscopy is an endoscopic procedure that examines approximately the lower half of the colon. It may be performed with a variety of endoscopic instruments, including a 60 to 70 cm version of a standard colonoscope, an upper endoscope, a standard pediatric colonoscope or a short (70cm) pediatric colonoscope. It is typically performed without sedation and a more limited bowel preparation than is used for standard colonoscopy. Since sedation is not required, it can be performed in office based settings and by diverse examiners, including specially trained nurses or physician assistants [41].
Flexible sigmoidoscopy use in the USA has been decreasing in the recent decade. Approximately 2.8 million flexible sigmoidoscopies and 14.2 million colonoscopies were estimated to have been performed in 2002 [42]. Low reimbursement and a shortage of adequately trained examiners are two barriers to flexible sigmoidoscopy availability [43, 44]. In settings where reimbursement has not been a concern, and where nurse endoscopists have been employed, high rates of flexible sigmoidoscopy utilization have been achieved and it continues to be performed [45, 46].
The use of flexible sigmoidoscopy for colorectal cancer screening is supported by high quality case-control and cohort studies. B2, 3 In general, flexible sigmoidoscopy appears to be associated with a 60% reduction in colorectal cancer mortality for the area of the colon within its reach, and this protective effect appears to persist for 10 years or more [47]. A small randomized trial demonstrated decreased colorectal cancer incidence in the sigmoidoscopy screened group compared to a non-screened control group, but not an overall mortality benefit from screening [48]. There was higher all-cause mortality in the screening group, perhaps due to unbalanced distribution of cardiac risk factors in the two groups, or due to changes in lifestyle behavior after screening [49]. There are four randomized controlled trials ongoing in the USA and Europe [3, 50–52], and results should be reported soon.
The evidence for the diagnostic accuracy of flexible sigmoidoscopy is primarily derived from colonoscopy studies. Flexible sigmoidoscopy is 60–70% as sensitive as colonoso-copy for detection of advanced adenomas and cancers in the colon [53, 54]. However, this figure varies according to age, since proximal neoplasia becomes more common after age 65 [55]. Flexible sigmoidoscopy may also be less sensitive in women than in men [56], but the overall prevalence of advanced colonic neoplasia is lower in women than in men [57].
The key limitation in the evidence is the lack of any longitudinal head- to-head comparison of flexible sigmoidoscopy screening with other screening techniques, such as colonoscopy or fecal blood testing. The key question for screening policy is the incremental benefit of colonoscopy over flexible sigmoidoscopy, given the higher direct medical and indirect (patient) costs of colonoscopy and the higher risk of complications with colonoscopy. There is also a lack of evidence on how race, ethnicity, gender or other cultural factors affect patients’ perceptions of the optimal screening test.
Several lines of evidence support the idea that the incremental benefit of colonoscopy is less than simply the difference in sensitivity for advanced adenomas between colonoscopy and flexible sigmoidoscopy. A key issue is the rate at which these advanced adenomas progress to invasive and life-threatening colorectal cancers. Proximal and distal colorectal cancers are, generally speaking, biologically different [11], and these biologic differences are known to affect the rate of progression through the adenoma-carcinoma sequence. Colonoscopy follow-up studies, including the National Polyp Study and the recent Manitoba Colonoscopy Study, have found that colorectal cancers that occur shortly after a colonoscopy with or without polypectomy are found disproportionately in the proximal colon [9, 58]. Therefore, published evidence suggests that most of the benefit of colonoscopy for cancer prevention through polyp detection and removal occurs in the distal colon, a region of the colon that is examined by flexible sigmoidoscopy. This feature decreases the risk-benefit advantage of colonoscopy over flexible sigmoidoscopy.
The chief advantage of flexible sigmoidoscopy is that it can be performed without sedation, by a variety of examiners in diverse settings. The absence of sedation is perceived by some patients as an advantage and by others as a disadvantage. The chief limitation of flexible sigmoidoscopy is that it does not examine the entire colon, but only the rectum, sigmoid and descending colon. The complications of flexible sigmoidoscopy include colonic perforation, even if no biopsy or polypectomy is performed, but this complication occurs in fewer than one in 20,000 examinations [3, 59].
The available published evidence supports performing flexible sigmoidoscopy screening at least every 10 years. More frequent examinations may be justified based on patient or provider preference, but there is no evidence to suggest that more frequent examinations will result in improved patient outcomes. Recent colorectal cancer screening guidelines have recommended a five-year interval between normal flexible sigmoidoscopies, while recommending a 10-year interval between colonoscopies [6, 60, 61]. The shorter interval is recommended for flexible sigmoidoscopy out of concern that it may be less sensitive than colonoscopy even in the area examined because of the differences in bowel preparation, the experience of the examiners performing the procedure and the effect that patient discomfort may have on depth of sigmoidoscope insertion and adequacy of mucosal inspection. C5 The five-year interval also led to improved outcomes at acceptable cost in the USPSTF simulation modeling study [7]. In examinations where an experienced examiner feels comfortable that a well cleansed colon has been thoroughly examined, a 10-year interval between exams may be sufficient.
There may be considerable variation in adenoma detection at flexible sigmoidoscopy between different examiners [62, 63], and this variability may reduce the effectiveness of flexible sigmoidoscopy for colorectal cancer screening. Quality assurance is an important issue for flexible sigmoidoscopists, and has been reviewed in detail elsewhere [41]. Providers should be well trained, and should consider exceeding the published ASGE standards for a minimum number of training examinations prior to performing sigmoidoscopy without supervision.
An incomplete flexible sigmoidoscopy, defined as a depth of insertion less than 40 cm, is associated with an increased risk for interval colorectal cancers in the time after a flexible sigmoidoscopy [55]. Incomplete examinations should be followed by alternative screening using a different method (annual fecal blood tests, double contrast barium enema or colonoscopy) if the examination was limited by angulation or patient discomfort, or if the colonic preparation was inadequate. C5
Ideally, providers performing flexible sigmoidoscopy should be skilled and comfortable taking biopsies, to allow the determination of whether small polyps are adenomas. The decision to follow up a flexible sigmoidoscopy with colonoscopy in patients with 1–2 adenomas smaller than 1cm should be individualized based on provider and patient preferences, and concern about the risk of complications. In the absence of biopsies, referring all patients with polyps larger than 5 mm for colonoscopy is a reasonably effective management strategy [64]. C5
Flexible sigmoidoscopy is an effective, affordable and safe option for colorectal cancer screening. A shortage of adequately trained examiners and low reimbursement has limited its use. For providers who have been able to implement large-scale flexible sigmoidoscopy programs, it has been a feasible and safe test for screening. There are four randomized trials of this intervention currently ongoing, and we should learn significantly more about the performance of this test when the results of these are available.
Radiologic screening
Double contrast barium enema
Double contrast barium enema (DCBE) is endorsed for CRC screening in the current joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer and the American College of Radiology (ACS/USMSTF/ACR) [60, 61], but not in the current USPSTF guideline [6].
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
