- Email: [email protected]
Flexible sigmoidoscopy to prevent colorectal cancer
Correspondence
870
AJB is a scientific advisory board member and founder for NuMedii and Genstruct; is a scientific advisory board member for Johnson and Johnson; has received consultancy fees from Lilly, NuMedii, Johnson and Johnson, Genstruct, Tercica, and Prevendia and honoraria from Lilly and Siemens; and holds stock in NuMedii and Genstruct. The other authors declare that they have no conflicts of interest.
Alexander A Morgan, Rong Chen, Atul J Butte, *Euan A Ashley [email protected] Department of Pediatrics (AAM, RC, AJB) and Center for Inherited Cardiovascular Disease (EAA), Stanford University School of Medicine, Stanford, CA 94305, USA 1
Morgan AA, Chen R, Butte AJ. Likelihood ratios for genome medicine. Genome Med 2010; 2: 30. Ioannidis JP. Why most published research findings are false. PLoS Med 2005; 2: e124. Chen Y, Zhu J, Lum PY, et al. Variations in DNA elucidate molecular networks that cause disease. Nature 2008; 452: 429–35.
2 3
Flexible sigmoidoscopy to prevent colorectal cancer Wendy Atkin and colleagues (May 8, p 1624)1 report that, 11·2 years after once-only sigmoidoscopy screening, colorectal cancer incidence was reduced by 23% and mortality by 31% in intention-to-screen analyses and by 33% and 43%, respectively, in perprotocol analyses. At first sight, these reductions are substantially higher than for guaiacbased faecal occult blood tests (gFOBT), which have been shown to reduce 50
colorectal cancer mortality by about 16%,2 and colorectal cancer incidence by about 20% 18 years later.3 However, Atkin and colleagues’ results were obtained in a selected population from which 55% of the invited participants were excluded because they did not respond or responded “no” to the invitation.1 Consequently, the participation rate for sigmoidoscopy was high at 71%. If the people who did not respond and those who responded “no”, had been included, participation would have dropped from 71% to about 32%. This latter proportion is in line with most other trials. On the basis of data provided by Atkin and colleagues, I have plotted participation in sigmoidoscopy screening against the reduction in colorectal cancer mortality and incidence (figure). From this figure, one can deduce that, at 40% participation, the reported reduction in mortality for once-only sigmoidoscopy screening is identical to that for gFOBT programmes (ie, 16%).2 Immunochemical FOBT (iFOBT) screening has been shown to identify more large (>1 cm) advanced adenomas than sigmoidoscopy and to be associated with a higher participation rate (60% vs 30%).4,5 Programmes based on iFOBT are thus very competitive with sigmoidoscopy screening, and are much easier and
Mortality Incidence
45 40 35 Reduction (%)
variants reported to contribute to disease risk in decreasing order of the number of published studies and the number of patients within each study. Our aim was to allow the individual specialist to include or exclude variants on the basis of a personal threshold of their choice. For example, a common standard is to accept as valid variants replicated in at least one additional study. Applying this standard to figure 4 would lead to the inclusion of 15 variants in the risk assessment. Some might hold a more rigorous standard, some less. Pierce and Ahsan suggest a personal threshold of around 20 variants for this analysis. Such information is highly dynamic. At the time of publication, our database held 10 201 associations between genetic variants and phenotype derived from primary research publications. This number has now grown to 18 781 associations. And perhaps because odds ratios have become a reporting standard for genome-wide association studies, many researchers do not include genotype-specific case– control frequencies in their reports. Although we encourage full reporting, limiting reporting limits the application of the likelihood ratio approach.1 We are also constrained by a publication record that is substantially biased towards the reporting of positive results and larger effect sizes—an inherent problem of any meta-analysis.2 The paper reports our best attempt to present, today, in an unbiased manner, the potential of genomics for individual patients. We hope that in the future, the medical community will create more refined models for diseases with a genetic risk component, not only with improved consensus on which variants to include but also on the synthesis of their interaction. For most diseases, the underlying biology is likely to be best represented by a complex network of interactions with many variants contributing differing risk under different environmental conditions.3
30 25 20 15 10 5 0 0
10
20
30
40
50 60 Participation (%)
70
80
90
100
Figure: Relation between participation in sigmoidoscopy screening and reduction in colorectal cancer mortality and incidence, deduced from data provided in reference 1
www.thelancet.com Vol 376 September 11, 2010
Correspondence
cheaper to do. I therefore conclude that well designed screening programmes based on iFOBT will be more suitable than sigmoidoscopy for populationbased colorectal cancer screening. I declare that I have no conflicts of interest.
Jan B M J Jansen [email protected] Elkerliek Ziekenhuis, 5700 AB Helmond, Netherlands 1
2
3
4
5
Atkin WS, Edwards R, Kralj-Hans I, et al. Onceonly flexible sigmoidoscopy in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010; 375: 1624–33. Hewitson P, Glasziou P, Watson E, Towler B, Irwig L. Cochrane systematic review of colorectal cancer screening using fecal occult blood test (hemoccult): an update. Am J Gastroenterol 2008; 103: 1541–49. Atkin W. Options for screening for colorectal cancer. Scand J Gastroenterol 2003; 237 (suppl): 13–16. Hol L, van Leerdam ME, van Ballegooijen M, et al. Screening for colorectal cancer: randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut 2010; 59: 62–68. Van Rossum LG, van Rijn AF, Laheij RJ, et al. Random comparison of guaiac and immunochemical fecal occult blood tests for colorectal cancer in a screening population. Gastroenterology 2008; 135: 82–90.
In the UK trial of once-only sigmoidoscopy screening,1 Wendy Atkin and colleagues estimated a 50% reduction in the incidence of distal colorectal cancer in the screened group after adjusting for non-compliance according to the method of Cuzick et al.2 However, this estimate included prevalent cases of cancer discovered at sigmoidoscopy, cases which by definition could not have been prevented by that sigmoidoscopy. We propose an alternative measure of efficacy based on a modification of Cuzick et al’s method. Cases
From numbers presented in the paper,1 we calculated incidence rates of distal colorectal cancer excluding prevalent tumours at baseline (table). To do this we assumed that a similar proportion of prevalent cases as was detected in the screening group (126 of 57 099 or 0·22%) would have been detected in the control group, had screening been offered (based on an assumed equivalence of the randomised groups). We further assumed that essentially all these prevalent cases in the control group would have been clinically detected during the follow-up period of the trial (ie, little overdiagnosis and sufficient followup to assure that all prevalent cases would become symptomatic and hence detected). With these rates, the Cuzick method estimates a greater than 70% reduction in the incidence of distal colorectal cancer (noncompliance adjusted rate ratio of 0·29). Subject to the limitations of the above assumptions, this efficacy estimate might provide a better measure of the ability of endoscopy to reduce the future incidence of colorectal cancer. We declare that we have no conflicts of interest.
*V Paul Doria-Rose, Chris Vinden, David F Ransohoff, Philip C Prorok [email protected] *Biometry Research Group, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA (VPDR, PCP); Cancer Prevention Fellowship Program, Center for Cancer Training, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA (VPDR); Department of Surgery, University of Western Ontario, London, ON, Canada (CV); and Departments of Medicine and
Person-years
Incidence (per 100 000 person-years)
Intervention group Screened
89*
445 488
20·0
Unscreened
171
172 565
99·1
943†
1 220 175
77·3
Control group
*215 total cases minus 126 prevalent cases. †1192 total cases minus 249 cases assumed to be prevalent among those who would have been screened by sigmoidoscopy if offered.
Table: Incidence of distal colorectal cancer by screening group and compliance, after exclusion of prevalent cases
www.thelancet.com Vol 376 September 11, 2010
Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA (DFR) 1
2
Atkin WS, Edwards R, Kralj-Hans I, et al. Onceonly flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010; 375: 1624–33. Cuzick J, Edwards R, Segnan N. Adjusting for non-compliance and contamination in randomized clinical trials. Stat Med 1997; 16: 1017–29.
Authors’ reply Jan Jansen makes a strong case for faecal occult blood tests (FOBT) rather than sigmoidoscopy in terms of efficacy and uptake rate. A systematic review1 showed that the guaiac FOBT (gFOBT) reduces colorectal cancer mortality by 25% among test users, which compares with a 43% reduction after a single sigmoidoscopy. However, as Jansen illustrates, the effectiveness of a screening programme is dependent on uptake rates. Although our data could not be used to estimate population uptake, a study in an ethnically diverse population in London, UK, found that attendance for sigmoidoscopy was as high as participation for gFOBT in the same geographic area.2 There are many unknowns for immunochemical FOBT (iFOBT). A Dutch trial found that the yield of advanced neoplasia per 100 invitees was lower for iFOBT than for sigmoidoscopy (1·5 vs 2·4) despite higher uptake rates (61·5% vs 32·4%).3 How frequently iFOBT needs to be repeated or what the cumulative uptake or positivity rate would be over a programme of screening is not known. The positivity rate for a single test was 5% in the Dutch trial. If the test were offered biennially between 55 and 74 years of age, the cumulative requirement for colonoscopy could be substantial. iFOBT shows promise, but we cannot conclude that it would be more cost effective or acceptable in a population-based programme than once-only sigmoidoscopy. Paul Doria-Rose and colleagues suggest that exclusion of prevalent cancers detected at screening gives a better estimate of the efficacy of 871
870
AJB is a scientific advisory board member and founder for NuMedii and Genstruct; is a scientific advisory board member for Johnson and Johnson; has received consultancy fees from Lilly, NuMedii, Johnson and Johnson, Genstruct, Tercica, and Prevendia and honoraria from Lilly and Siemens; and holds stock in NuMedii and Genstruct. The other authors declare that they have no conflicts of interest.
Alexander A Morgan, Rong Chen, Atul J Butte, *Euan A Ashley [email protected] Department of Pediatrics (AAM, RC, AJB) and Center for Inherited Cardiovascular Disease (EAA), Stanford University School of Medicine, Stanford, CA 94305, USA 1
Morgan AA, Chen R, Butte AJ. Likelihood ratios for genome medicine. Genome Med 2010; 2: 30. Ioannidis JP. Why most published research findings are false. PLoS Med 2005; 2: e124. Chen Y, Zhu J, Lum PY, et al. Variations in DNA elucidate molecular networks that cause disease. Nature 2008; 452: 429–35.
2 3
Flexible sigmoidoscopy to prevent colorectal cancer Wendy Atkin and colleagues (May 8, p 1624)1 report that, 11·2 years after once-only sigmoidoscopy screening, colorectal cancer incidence was reduced by 23% and mortality by 31% in intention-to-screen analyses and by 33% and 43%, respectively, in perprotocol analyses. At first sight, these reductions are substantially higher than for guaiacbased faecal occult blood tests (gFOBT), which have been shown to reduce 50
colorectal cancer mortality by about 16%,2 and colorectal cancer incidence by about 20% 18 years later.3 However, Atkin and colleagues’ results were obtained in a selected population from which 55% of the invited participants were excluded because they did not respond or responded “no” to the invitation.1 Consequently, the participation rate for sigmoidoscopy was high at 71%. If the people who did not respond and those who responded “no”, had been included, participation would have dropped from 71% to about 32%. This latter proportion is in line with most other trials. On the basis of data provided by Atkin and colleagues, I have plotted participation in sigmoidoscopy screening against the reduction in colorectal cancer mortality and incidence (figure). From this figure, one can deduce that, at 40% participation, the reported reduction in mortality for once-only sigmoidoscopy screening is identical to that for gFOBT programmes (ie, 16%).2 Immunochemical FOBT (iFOBT) screening has been shown to identify more large (>1 cm) advanced adenomas than sigmoidoscopy and to be associated with a higher participation rate (60% vs 30%).4,5 Programmes based on iFOBT are thus very competitive with sigmoidoscopy screening, and are much easier and
Mortality Incidence
45 40 35 Reduction (%)
variants reported to contribute to disease risk in decreasing order of the number of published studies and the number of patients within each study. Our aim was to allow the individual specialist to include or exclude variants on the basis of a personal threshold of their choice. For example, a common standard is to accept as valid variants replicated in at least one additional study. Applying this standard to figure 4 would lead to the inclusion of 15 variants in the risk assessment. Some might hold a more rigorous standard, some less. Pierce and Ahsan suggest a personal threshold of around 20 variants for this analysis. Such information is highly dynamic. At the time of publication, our database held 10 201 associations between genetic variants and phenotype derived from primary research publications. This number has now grown to 18 781 associations. And perhaps because odds ratios have become a reporting standard for genome-wide association studies, many researchers do not include genotype-specific case– control frequencies in their reports. Although we encourage full reporting, limiting reporting limits the application of the likelihood ratio approach.1 We are also constrained by a publication record that is substantially biased towards the reporting of positive results and larger effect sizes—an inherent problem of any meta-analysis.2 The paper reports our best attempt to present, today, in an unbiased manner, the potential of genomics for individual patients. We hope that in the future, the medical community will create more refined models for diseases with a genetic risk component, not only with improved consensus on which variants to include but also on the synthesis of their interaction. For most diseases, the underlying biology is likely to be best represented by a complex network of interactions with many variants contributing differing risk under different environmental conditions.3
30 25 20 15 10 5 0 0
10
20
30
40
50 60 Participation (%)
70
80
90
100
Figure: Relation between participation in sigmoidoscopy screening and reduction in colorectal cancer mortality and incidence, deduced from data provided in reference 1
www.thelancet.com Vol 376 September 11, 2010
Correspondence
cheaper to do. I therefore conclude that well designed screening programmes based on iFOBT will be more suitable than sigmoidoscopy for populationbased colorectal cancer screening. I declare that I have no conflicts of interest.
Jan B M J Jansen [email protected] Elkerliek Ziekenhuis, 5700 AB Helmond, Netherlands 1
2
3
4
5
Atkin WS, Edwards R, Kralj-Hans I, et al. Onceonly flexible sigmoidoscopy in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010; 375: 1624–33. Hewitson P, Glasziou P, Watson E, Towler B, Irwig L. Cochrane systematic review of colorectal cancer screening using fecal occult blood test (hemoccult): an update. Am J Gastroenterol 2008; 103: 1541–49. Atkin W. Options for screening for colorectal cancer. Scand J Gastroenterol 2003; 237 (suppl): 13–16. Hol L, van Leerdam ME, van Ballegooijen M, et al. Screening for colorectal cancer: randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut 2010; 59: 62–68. Van Rossum LG, van Rijn AF, Laheij RJ, et al. Random comparison of guaiac and immunochemical fecal occult blood tests for colorectal cancer in a screening population. Gastroenterology 2008; 135: 82–90.
In the UK trial of once-only sigmoidoscopy screening,1 Wendy Atkin and colleagues estimated a 50% reduction in the incidence of distal colorectal cancer in the screened group after adjusting for non-compliance according to the method of Cuzick et al.2 However, this estimate included prevalent cases of cancer discovered at sigmoidoscopy, cases which by definition could not have been prevented by that sigmoidoscopy. We propose an alternative measure of efficacy based on a modification of Cuzick et al’s method. Cases
From numbers presented in the paper,1 we calculated incidence rates of distal colorectal cancer excluding prevalent tumours at baseline (table). To do this we assumed that a similar proportion of prevalent cases as was detected in the screening group (126 of 57 099 or 0·22%) would have been detected in the control group, had screening been offered (based on an assumed equivalence of the randomised groups). We further assumed that essentially all these prevalent cases in the control group would have been clinically detected during the follow-up period of the trial (ie, little overdiagnosis and sufficient followup to assure that all prevalent cases would become symptomatic and hence detected). With these rates, the Cuzick method estimates a greater than 70% reduction in the incidence of distal colorectal cancer (noncompliance adjusted rate ratio of 0·29). Subject to the limitations of the above assumptions, this efficacy estimate might provide a better measure of the ability of endoscopy to reduce the future incidence of colorectal cancer. We declare that we have no conflicts of interest.
*V Paul Doria-Rose, Chris Vinden, David F Ransohoff, Philip C Prorok [email protected] *Biometry Research Group, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA (VPDR, PCP); Cancer Prevention Fellowship Program, Center for Cancer Training, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA (VPDR); Department of Surgery, University of Western Ontario, London, ON, Canada (CV); and Departments of Medicine and
Person-years
Incidence (per 100 000 person-years)
Intervention group Screened
89*
445 488
20·0
Unscreened
171
172 565
99·1
943†
1 220 175
77·3
Control group
*215 total cases minus 126 prevalent cases. †1192 total cases minus 249 cases assumed to be prevalent among those who would have been screened by sigmoidoscopy if offered.
Table: Incidence of distal colorectal cancer by screening group and compliance, after exclusion of prevalent cases
www.thelancet.com Vol 376 September 11, 2010
Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA (DFR) 1
2
Atkin WS, Edwards R, Kralj-Hans I, et al. Onceonly flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010; 375: 1624–33. Cuzick J, Edwards R, Segnan N. Adjusting for non-compliance and contamination in randomized clinical trials. Stat Med 1997; 16: 1017–29.
Authors’ reply Jan Jansen makes a strong case for faecal occult blood tests (FOBT) rather than sigmoidoscopy in terms of efficacy and uptake rate. A systematic review1 showed that the guaiac FOBT (gFOBT) reduces colorectal cancer mortality by 25% among test users, which compares with a 43% reduction after a single sigmoidoscopy. However, as Jansen illustrates, the effectiveness of a screening programme is dependent on uptake rates. Although our data could not be used to estimate population uptake, a study in an ethnically diverse population in London, UK, found that attendance for sigmoidoscopy was as high as participation for gFOBT in the same geographic area.2 There are many unknowns for immunochemical FOBT (iFOBT). A Dutch trial found that the yield of advanced neoplasia per 100 invitees was lower for iFOBT than for sigmoidoscopy (1·5 vs 2·4) despite higher uptake rates (61·5% vs 32·4%).3 How frequently iFOBT needs to be repeated or what the cumulative uptake or positivity rate would be over a programme of screening is not known. The positivity rate for a single test was 5% in the Dutch trial. If the test were offered biennially between 55 and 74 years of age, the cumulative requirement for colonoscopy could be substantial. iFOBT shows promise, but we cannot conclude that it would be more cost effective or acceptable in a population-based programme than once-only sigmoidoscopy. Paul Doria-Rose and colleagues suggest that exclusion of prevalent cancers detected at screening gives a better estimate of the efficacy of 871