Should organised faecal occult blood test screening be established?

Abstract

Clinical trials conducted in the Nottingham (UK), Funen (Denmark) and Minnesota (USA) using faecal occult blood testing (FOBT) showed evidence that population-based screening with FOBT could reduce mortality from colorectal cancer (CRC) [1.Hardcastle J.D. Chamberlain J.O. Robinson M.H. et al.Randomised controlled trial of faecal-occult-blood screening for colorectal cancer.Lancet. 1996; 348: 1472-1477Abstract Full Text Full Text PDF PubMed Scopus (2406) Google Scholar, 2.Kronborg O. Fenger C. Olsen J. et al.Randomised study of screening for colorectal cancer with faecal-occult-blood test.Lancet. 1996; 348: 1467-1471Abstract Full Text Full Text PDF PubMed Scopus (2166) Google Scholar, 3.Mandel J.S. Church T.R. Ederer F. Bond J.H. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood.J Natl Cancer Inst. 1999; 91: 434-437Crossref PubMed Scopus (687) Google Scholar]. At first glance, these results could be interpreted as positive signals for implementing large-scale screening programmes using FOBT. Faecal occult blood testing was proposed for screening in the 1970s, when fibre-optic endoscopic techniques were in development, and CRC incidence was rising. Since then, endoscopic technologies have constantly improved and become much more widely available. Furthermore, significant advances have been made in the knowledge of the biology and in the medical management of CRC. Therefore, one should be cautious with results from FOBT trials, and examine carefully how these trials relate to the changing epidemiology of CRC, and the advent of new screening techniques. Hereafter, we list seven arguments that should be considered before the implementation of screening programmes based on FOBT. Reductions in CRC mortality achieved by the immense Nottingham, Funen and Minnesota trials were modest, in the order of 15% to 21%, and 10–18 years of follow-up were carried out before statistically significant differences were seen between intervention and control groups. This modest impact on CRC mortality is mainly due to the known low sensitivity of FOBT. Translated in absolute terms, these figures indicate that for 1000 persons invited for FOBT screening once every2 years for 10 years, one death due to CRC would be avoided [4.Gotzsche P. Screening for colorectal cancer.Lancet. 1997; 349: 356Abstract Full Text Full Text PDF PubMed Scopus (7) Google Scholar]. In FOBT trials, not all of the reduction in CRC mortality was attributable to the FOBT itself, but also to better medical attention given to those subjects who complied with the screening test. This statement comes from the observation that in the Nottingham and Funen trials [1.Hardcastle J.D. Chamberlain J.O. Robinson M.H. et al.Randomised controlled trial of faecal-occult-blood screening for colorectal cancer.Lancet. 1996; 348: 1472-1477Abstract Full Text Full Text PDF PubMed Scopus (2406) Google Scholar, 2.Kronborg O. Fenger C. Olsen J. et al.Randomised study of screening for colorectal cancer with faecal-occult-blood test.Lancet. 1996; 348: 1467-1471Abstract Full Text Full Text PDF PubMed Scopus (2166) Google Scholar], mortality due to interval CRC in intervention groups was lower than mortality due to CRC in control groups. Because of its low sensitivity, interval CRCs were more numerous than screen-detected CRC in subjects who accepted at least one FOB test, and interval CRCs were detected at an earlier stage than CRC diagnosed in the control group (Table 1). The latter observation is surprising, since one would not expect interval cancers to be detected earlier than in the absence of a screening test. In the Nottingham and Funen trials, control subjects received normal medical care. In contrast, physicians and perhaps many subjects in the intervention group were aware that they were part of a clinical study, and most physicians did not ignore the fact that they were trying to detect interval cancers with a test known for its low sensitivity. Therefore, it is highly probable that among screened subjects, more attention was paid to any symptom that might suggest the eventual presence of interval CRC, leading to more precocious diagnosis and lower mortality.Table 1Dukes’ A colorectal cancers in the Nottingham [1] and Funen [2] trialsOrigins of colorectal cancers in screening groupControlsTotalSubjects with ≥ 1 screening testNon-respondersaSubjects allocated to the screening group but who were never screened with the FOB test.Screen detectedIntervalNottingham trial [1.Hardcastle J.D. Chamberlain J.O. Robinson M.H. et al.Randomised controlled trial of faecal-occult-blood screening for colorectal cancer.Lancet. 1996; 348: 1472-1477Abstract Full Text Full Text PDF PubMed Scopus (2406) Google Scholar]Number of subjects75253448383041574998All CRCs885236249400836Number (%) Dukes’ A178 (20)97 (41)39 (16)42 (11)95 (11) Dukes’ B286 (32)71 (30)76 (31)139 (35)285 (33) Dukes’ C211 (24)51 (22)71 (29)89 (22)264 (31) Dukes’ D191 (22)13 (6)61 (22)117 (29)179 (21) Not known19 (2)4 (2)2 (1)13 (3)33 (4)P valuebχ2 test for percentage of Dukes’ A CRCs compared with controls.< 0.0001<0.00010.070.65–P valuecχ2 test for trend compared with Dukes’ distribution among controls.0.0001<0.00010.810.15–Funen trial [2.Kronborg O. Fenger C. Olsen J. et al.Randomised study of screening for colorectal cancer with faecal-occult-blood test.Lancet. 1996; 348: 1467-1471Abstract Full Text Full Text PDF PubMed Scopus (2166) Google Scholar]Number of subjects30967206721029530966All CRCs481138148195483Number (%) Dukes’ A105 (22)53 (38)31 (21)21 (11)54 (11) Dukes’ B164 (34)52 (38)46 (31)66 (34)177 (37) Dukes’ C90 (19)20 (14)35 (24)35 (18)111 (23) Dukes’ D98 (20)11 (8)27 (18)60 (31)114 (24) Not known24 (5)2 (1)9 (6)13 (7)27 (6)P valuebχ2 test for percentage of Dukes’ A CRCs compared with controls.<0.0001<0.00010.0020.88–P valuecχ2 test for trend compared with Dukes’ distribution among controls.0.001<0.00010.030.21–CRC, colorectal cancer.a Subjects allocated to the screening group but who were never screened with the FOB test.b χ2 test for percentage of Dukes’ A CRCs compared with controls.c χ2 test for trend compared with Dukes’ distribution among controls. Open table in a new tab CRC, colorectal cancer. In contrast, in the Minnesota trial, similar to subjects being allocated to the two intervention arms, control subjects were contacted annually about their vital status or about eventual diagnosis of CRC or colonic adenoma [5.Mandel J.S. Bond J.H. Church T.R. et al.Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study.N Engl J Med. 1993; 328: 1365-1371Crossref PubMed Scopus (2885) Google Scholar]. As a consequence of similar medical attention being paid to all subjects, survival was similar for interval CRCs and for CRCs diagnosed among control subjects [5.Mandel J.S. Bond J.H. Church T.R. et al.Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study.N Engl J Med. 1993; 328: 1365-1371Crossref PubMed Scopus (2885) Google Scholar]. One can calculate that the ‘disease awareness’ effect probably accounted for one-quarter of the impact of biennial FOBT on CRC mortality, and thus CRC mortality reductions achieved in the Nottingham and Funen trials were 11% to 14% rather than 15% to 18% (calculations available upon request). One could argue that disease awareness is part of the screening process and its eventual effects should thus be attributed to the screening procedure. This argument does not hold, since the evaluation of the efficacy of a screening test in a randomised trial should not be influenced by factors dependent on the trial design, such as the modalities of follow-up of subjects according to their allocation in the intervention or in the control group. Colorectal polyps rarely bleed, and thus FOBT is not very sensitive for detecting the presence of polyps. However, it can detect some of the large polyps that are more likely to bleed, and also to become cancerous. After 18 years, the Minnesota trial obtained a 17% reduction in CRC incidence [4]. FOBT is not well accepted by most people, and participation levels outside the context of randomised trials rarely exceeds 30%, with low adherence (i.e. subjects rarely attend more than one screening round) [7.Vernon S.W. Participation in colorectal cancer screening: a review.J Natl Cancer Inst. 1997; 89: 1406-1422Crossref PubMed Scopus (671) Google Scholar]. Hence, it is not necessarily the case that modest gains observed in FOBT trials could be replicated in areas where acceptance of FOBT is low. For obtaining participation rates around 50% to 60%, much effort must be devoted to motivation of the target population, with intense information and invitation–reinvitation procedures, or with strong commitment of general practitioners (GPs) [8.Hart A.R. Eaden J. Barnett S. et al.Colorectal cancer prevention. An approach to increasing compliance in a faecal occult blood test screening programme.J Epidemiol Community Health. 1998; 52: 818-820Crossref PubMed Scopus (14) Google Scholar, 9.Tazi M.A. Faivre J. Dassonville F. et al.Participation in faecal occult blood screening for colorectal cancer in a well defined French population: result of five screening rounds from 1988 to 1996.J Med Screening. 1997; 4: 147-151Crossref PubMed Scopus (86) Google Scholar]. It is doubtful that such efforts would be easy to replicate in other areas and maintained in the long term. Participation rates in organised screening programmes with pap smear test for cervical cancer and mammography for breast cancer usually exceed 60%, with false-negative results at 25% to 35%. International experience and recommendations suggest that higher numbers of false-negative results should prompt revision of screening programmes, or (in the case of cervical cancer) adoption of adjunct screening methods. In view of these requirements, why would one accept large-scale use of the FOB test, which is not well accepted and misses >50% of CRC? In addition, in the longer term, the numerous interval CRCs are likely to shake the confidence of the general public in FOBT screening. Decreasing trends in mortality from CRC have already taken place in many parts of the world. A general decline in CRC mortality is noticeable since the 1950s and 1960s in the UK, Germany, France, the USA and Canada [10.Canadian Cancer Society Canadian Cancer Statistiques, 1997. Toronto: Canadian Cancer.Society. 1997; Google Scholar, 11.Chu K.C. Tarone R.E. Chow W.H. et al.Temporal patterns in colorectal cancer incidence, survival and mortality from 1950 through 1990.J Natl Cancer Inst. 1994; 86: 997-1006Crossref PubMed Scopus (167) Google Scholar, 12.Coleman M.P. Esteve J. Damiecki P. et al.Trends in cancer incidence and mortality. International Agency for Research on Cancer.IARC Scientific Publications. 1993; : 1-806Google Scholar]. Possible reasons for the decline are multiple: changes in diet patterns, occasional removal of polyps, declining incidence, earlier diagnosis and improved management of CRC. It is difficult to know which of these factors accounts for the observed decline in mortality. Earlier detection encompasses various methods, from physicians and patients giving more attention to early symptoms of the eventual presence of CRC to sporadic screening tests in asymptomatic subjects using one or a combination of available screening methods. It is possible that the widespread use of FOBT would accelerate the decline, but it will be the use of more sensitive screening methods, able to detect the majority of cancerous lesions at an early stage, or lesions before they have evolved in cancer, that are likely to make the biggest difference and to lead to substantial reductions in mortality from CRC. Most countries experience considerable difficulties in deciding whether or not to organise or support nationwide screening programmes. Usually, these difficulties relate to the costs of such programmes, and to the human and material resources necessary for their proper functioning in the long term. If significant efforts are devoted to the organisation of a FOBT screening programme, then the arrival of a newer, more efficient screening method may well be hindered. As a consequence, populations where large-scale FOBT screening is instituted as a ‘gold standard’ will probably not benefit from reductions in CRC mortality as large as those expected if more sensitive methods are used. New techniques are now proposed or under investigation for early detection of CRC, including endoscopic techniques [13.Toribara N.W. Sleisenger M.H. Screening for colorectal cancer.N Engl J Med. 1995; 332: 861-867Crossref PubMed Scopus (168) Google Scholar] and detection in stools of molecular alterations that indicate the presence of malignant processes in the large bowel [14.Dong S.M. Traverso G. Johnson C. et al.Detecting colorectal cancer in stools with use of multiple genetic targets.J Natl Cancer Inst. 2001; 93: 858-865Crossref PubMed Scopus (320) Google Scholar]. Endoscopic techniques would be able to detect >90% of cancers and large polyps within reach of the endoscope [13.Toribara N.W. Sleisenger M.H. Screening for colorectal cancer.N Engl J Med. 1995; 332: 861-867Crossref PubMed Scopus (168) Google Scholar]. Colonoscopy would detect twice as many CRCs as sigmoidoscopy, as the entire colon would be explored [15.Imperiale T.F. Wagner D.R. Lin C.Y. et al.Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings.N Engl J Med. 2000; 343: 169-174Crossref PubMed Scopus (915) Google Scholar, 16.Podolsky D.K. Going the distance – the case for true colorectal cancer screening.N Engl J Med. 2000; 343: 207-208Crossref PubMed Scopus (144) Google Scholar]. Several prospective studies now provide evidence that colonoscopy combined with polypectomy may result in a 66% to 90% decrease in CRC incidence after 6–13 years [17.Winawer S.J. Zauber A.G. Ho M.N. et al.Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup.N Engl J Med. 1993; 329: 1977-1981Crossref PubMed Scopus (3830) Google Scholar, 18.Thiis-Evensen E. Hoff G.S. Sauar J. et al.Population-based surveillance by colonoscopy: effect on the incidence of colorectal cancer. Telemark Polyp Study I.Scand J Gastroenterol. 1999; 34: 414-420Crossref PubMed Scopus (406) Google Scholar, 19.Citarda F. Tomaselli G. Capocaccia R. et al.Efficacy in standard clinical practice of colonoscopic polypectomy in reducing colorectal cancer incidence.Gut. 2001; 48: 812-815Crossref PubMed Scopus (611) Google Scholar]. Although one must be cautious with results from observational studies aimed at assessing efficacy of screening technologies, data available so far indicate a 60% to 80% decrease in CRC mortality among those subjects who underwent a sigmoidoscopy [20.Selby J.V. Friedman G.D. Quesenberry Jr, C.P. Weiss N.S. A case–control study of screening sigmoidoscopy and mortality from colorectal cancer.N Engl J Med. 1992; 326: 653-657Crossref PubMed Scopus (1560) Google Scholar, 21.Newcomb P.A. Norfleet R.G. Storer B.E. et al.Screening sigmoidoscopy and colorectal cancer mortality.J Natl Cancer Inst. 1992; 84: 1572-1575Crossref PubMed Scopus (863) Google Scholar]. According to these results, an endoscopic examination of the bowel once every 10 years would be sufficient to achieve this reduction. One study using Surveillance, Epidemiology, and End Results (SEER) data in the USA suggested that colonoscopy and polypectomy were the major factor responsible for the declining CRC incidence and mortality of CRC observed in that country since 1986 [22.Nelson R.L. Persky V. Turyk M. Determination of factors responsible for the declining incidence of colorectal cancer.Dis Colon Rectum. 1999; 42: 741-752Crossref PubMed Scopus (61) Google Scholar]. One should be cautious, however, with the available efficacy data on colonoscopic screening, as these data were obtained from selected subsets of the general population, and not from true population-based studies. The Funen and the Nottingham FOBT trials were population-based randomised trials [1.Hardcastle J.D. Chamberlain J.O. Robinson M.H. et al.Randomised controlled trial of faecal-occult-blood screening for colorectal cancer.Lancet. 1996; 348: 1472-1477Abstract Full Text Full Text PDF PubMed Scopus (2406) Google Scholar, 2.Kronborg O. Fenger C. Olsen J. et al.Randomised study of screening for colorectal cancer with faecal-occult-blood test.Lancet. 1996; 348: 1467-1471Abstract Full Text Full Text PDF PubMed Scopus (2166) Google Scholar], while the Minnesota trial was conducted in volunteers [5.Mandel J.S. Bond J.H. Church T.R. et al.Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study.N Engl J Med. 1993; 328: 1365-1371Crossref PubMed Scopus (2885) Google Scholar]. Hence, data from FOBT trials and colonoscopic screening are not readily comparable. However, a reasonable approximation would be that with a 50% compliance rate to a single colonoscopic screening (that is the compliance to the initial FOB test observed in the Nottingham and Funen trials), compared with FOBT screening, one could expect at least a doubling in the reduction of CRC mortality in <10 years. Colonoscopy is often presented as an expensive technique that is not without risk. In expert hands, routine colonoscopy has an overall complication rate of 0.3% (e.g. perforation, side effects due to sedation, bleeding of removed polyp) [16.Podolsky D.K. Going the distance – the case for true colorectal cancer screening.N Engl J Med. 2000; 343: 207-208Crossref PubMed Scopus (144) Google Scholar]. These complication rates are likely to decrease as modern endoscopic techniques allow steadily easier, safer and cheaper colonoscopy [13.Toribara N.W. Sleisenger M.H. Screening for colorectal cancer.N Engl J Med. 1995; 332: 861-867Crossref PubMed Scopus (168) Google Scholar, 23.Bhattacharya I. Sack E.M. Screening colonoscopy: the cost of common sense.Lancet. 1996; 347: 1744-1745Crossref PubMed Scopus (34) Google Scholar, 24.Rogge J.D. Elmore M.F. Mahoney S.J. et al.Low-cost, office-based, screening colonoscopy.Am J Gastroenterol. 1994; 89: 1775-1780PubMed Google Scholar]. It is not impossible that the cost-effectiveness of colonoscopic screening would compare favourably with the cost-effectiveness of FOBT screening since, first, reductions in CRC mortality would most probably be much higher, and secondly, the cost of colonoscopies would be offset by savings from having to treat fewer patients with CRC, or treating patients with an earlier stage of CRC [13.Toribara N.W. Sleisenger M.H. Screening for colorectal cancer.N Engl J Med. 1995; 332: 861-867Crossref PubMed Scopus (168) Google Scholar]. Acceptability of colonoscopy remains low [7.Vernon S.W. Participation in colorectal cancer screening: a review.J Natl Cancer Inst. 1997; 89: 1406-1422Crossref PubMed Scopus (671) Google Scholar]. However, accepting a single colonoscopic screening at 50 or 60 years of age would likely reduce the probability of the patient suffering and/or dying from CRC more than accepting a single FOBT screening. The safety and acceptability of colonoscopy should be enhanced by virtual colonoscopy [25.Morrin M.M. Farrell R.J. Kruskal J.B. LaMont J.T. Virtual colonoscopy: a kinder, gentler, colorectal cancer screening test?.Lancet. 1999; 354: 1048-1049Abstract Full Text Full Text PDF PubMed Scopus (29) Google Scholar]. This novel non-invasive imaging technology consists of thin section, helical computed tomography (CT) of the colon. These sections are used to generate high resolution, two-dimensional axial images. Virtual colonoscopy holds the promise of enabling colonoscopy to be performed without an endoscope, and seems better accepted than conventional colonoscopy [26.Fenlon H.M. Nunes D.P. Schroy III, P. et al.A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps.N Engl J Med. 1999; 341: 1496-1503Crossref PubMed Scopus (666) Google Scholar]. In our opinion, the data available on CRC screening efficacy suggests that FOBT is not the appropriate method for screening for CRC. We do not see the relevance of mobilising millions of men and women for a screening procedure that will have little impact on their likelihood of dying of CRC. It is probable that its use (alone or in conjunction with flexible sigmoidoscopy) should be limited to subjects not willing to undergo colonoscopic screening.