Meta-analysis: colorectal and small bowel cancer risk in patients with Crohn s disease

Transcription

1 Alimentary Pharmacology & Therapeutics Meta-analysis: colorectal and small bowel cancer risk in patients with Crohn s disease C.CANAVAN*,K.R.ABRAMS & J. MAYBERRY* *Digestive Diseases Centre, University Hospitals of Leicester, Leicester General Hospital, Leicester, UK; Centre for Biostatistics and Genetic Epidemiology, Department of Heath Sciences, University of Leicester, Leicester, UK Correspondence to: Dr C. Canavan, Digestive Diseases Centre, University Hospitals of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UK. cc71@le.ac.uk Publication data Submitted 28 September 2005 First decision 29 October 2005 Resubmitted 21 January 2006 Second decision 22 January 2006 Resubmitted 23 January 2006 Accepted 24 January 2006 SUMMARY Background Crohn s disease is associated with small bowel cancer whilst risk of colorectal cancer is less clear. Aim To ascertain the combined estimates of relative risk of these cancers in Crohn s disease. Methods MEDLINE was searched to identify relevant papers. Exploding references identified additional publications. When two papers reviewed the same cohort, the later study was used. Results Meta-analysis showed overall colorectal cancer relative risk in Crohn s disease as 2.5 ( ), 4.5 ( ) for patients with colonic disease and 1.1 ( ) in ileal disease. Meta-regression showed reduction in relative risk over the past 30 years. Subgroup analysis showed Scandinavia had significantly lower colorectal cancer relative risk than the UK and North America. Cumulative risk analysis showed 10 years following diagnosis of Crohn s disease relative risk of colorectal cancer is 2.9% (1.5% 5.3%). Meta-analysis showed small bowel cancer relative risk in Crohn s disease is 33.2 ( ). Small bowel cancer relative risk has not significantly reduced over the last 30 years. Conclusion Relative risk of colorectal and small bowel cancers are significantly raised in Crohn s disease. Cumulative risk of colorectal cancer of 2.9% at 10 years suggests a potential benefit from routine screening. However, the value of screening requires rigorous appraisal. Aliment Pharmacol Ther 23, ª 2006 The Authors 1097 doi: /j x

2 1098 C. CANAVAN et al. INTRODUCTION Prevalence and incidence rates of Crohn s disease have been increasing in both the US 1 3 and Europe. 4 6 A recent study in the UK using primary care data reported a prevalence as high as 145 cases per Prevalence of Crohn s disease in North America is currently estimated between 26 and 199 cases per , which means there are as many as patients with Crohn s disease in North America alone. 8 An association between colorectal cancer and ulcerative colitis was first reported in Many studies have looked at the link between these diseases as well as risk factors that might increase the likelihood of an individual with ulcerative colitis developing cancer. Meta-analysis of these papers found that the risk of colorectal cancer for a patient with ulcerative colitis is 2% at 10 years, 12% at 20 years and 18% at 30 years disease duration. 10 In contrast, there have been fewer studies on the link between Crohn s disease and cancer, with most focusing on small bowel and colorectal cancers, the first report being that of an adenocarcinoma of the ascending colon in There is a strong association between Crohn s disease and developing cancer of the small bowel, with relative risks (RRs) reported of between and compared to the general population. Small bowel cancer is rare, accounting for only 1 5% of all gastrointestinal malignancies, 14 so the absolute risk remains small. It is, however, less clear if there is an increase in the RR of developing colorectal cancer in patients with Crohn s disease, reports vary from to This systematic review and meta-analysis of published data aims to provide an accurate overview of the current risk of colorectal and small bowel cancer in Crohn s disease. An accurate analysis of these data will help establish the likely risk of developing cancer for patients with Crohn s disease and so enable patients to make informed choices with regard to treatment and inform the debate regarding surveillance and screening for colorectal cancer in patients with Crohn s disease. METHODS Search strategy Published reports calculating the risk of cancer in Crohn s disease were identified through a literature search of MEDLINE using the following key words, both in free text and as MESH headings: Crohn s disease, inflammatory bowel disease, cancer, neoplasm. A comprehensive search of reference lists of all review articles and original studies retrieved by this method was performed to identify additional reports. Through this approach 544 papers were identified that had been published between 1972 and Inclusion and exclusion criteria Only English language journal papers were included where Crohn s disease patients had been studied and statistics regarding risk of cancer calculated that included either RR (incidence risk ratio compared to the general population) or reported observed and expected cancers. Studies were excluded if they only reported occurrence of cancer as case studies, with regard to surgical technique, histology, diagnostic techniques, possible biochemical pathways or as part of a randomized control trial for a drug. When two or more publications from one institution appeared to review the same patients only the most recent study results were included. This left 14 original papers for the analyses, 12 reporting colorectal cancer risk and eight reporting small bowel cancer risk (Figure 1). Data extraction Each paper was critically reviewed and the following data were extracted: 1 country of origin; 2 type of centre where study was conducted; 3 duration of study; 4 mean follow up of patients (patient years exposure); 5 number of patients in the study; 6 mean duration of disease in study participants; 7 the RR of colorectal cancer or small bowel cancer; 8 the confidence interval for RRs or observed and expected numbers of colorectal and small bowel cancers; 9 median year of diagnosis of patients. Study details are summarized in Table 1. Community-based studies are ones in which a population of patients with Crohn s disease lived or were diagnosed in a defined geographic area, whilst hospital or specialist centre-based studies include patients referred to the institute from an undefined area. All studies are

3 META-ANALYSIS: COLORECTAL AND SMALL BOWEL CANCER RISK IN CROHN S 1099 Potentially relevant papers identified on MEDLINE and through explosion of references (n = 544) Papers excluded as they were not written in English (n = 100) or were not published in a journal (n = 13) Figure 1. The Quality of Reporting of Meta-analyses (QUOROM) statement flow diagram 17 indicating the inclusion and exclusion criteria for papers into this meta-analysis. Abstracts of papers analysed for relevance (n = 435) Potentially appropriate papers to be included in meta-analysis (n = 19) Papers used in meta-analysis (n = 14) Papers excluded because they did not study Crohn s disease (n = 47), did not report cancer incidence or appropriate statistics (n = 80), were only case studies (n = 35), studied relatives (n = 1), were review articles only (n = 13), were drug trials (n = 7), or focused on: biochemical pathways (n = 125), histology or diagnostic techniques (n = 35), surgical methodology (n = 34), risk factor analysis (n = 7), treatment strategies (n = 27) Papers excluded because the same population was used again in a later study (n = 5) Table 1. Details of study and study population Paper (including country of origin and year) Median year of diagnosis with Crohn s Duration of study (years) Mean duration of follow up (years) Centre where study was conducted Study population size Fielding UK Hospital 295 Weedon US Hospital 449 Gyde UK Hospital 513 Greenstein US Hospital 579 Kvist Denmark Specialist referral centre 473 Gollop US Community based 103 Fireman Israel Community based 365 Ekbom Sweden Hospital 1655 Gillen UK Hospital 281 Persson Sweden No appropriate Community based 1251 data recorded Mellemkjaer Denmark Hospital 2645 Bernstein Canada Community based 2857 Jess Denmark Specialist referral centre 374 limited by bias 27 but this can be reduced if six standards are met. 27 These include appropriate and accurate case definition, high level of patient follow up, clear statement of outcomes measured, number of years over which the study was conducted, the centre where it was conducted and number of patients included. Studies conducted over a long period with a large population and long follow up with few patients lost represent better quality and the results would be more reliable. All papers included within this analysis reported on a population studied for at least 10 years and followed patients for a mean duration of at least 8.5 years. Statistical analysis All analyses were performed using STATA statistical software (College Station, TX, USA). The overall pooled estimate and 95% confidence interval of RR of colorectal and small bowel cancers were obtained separately for patients with Crohn s disease using either a fixed or random effects meta-analysis model on a log RR scale as appropriate depending on a test for heterogeneity using a 10% significance level. 28 Heterogeneity in the log RR over time was explored using random effects meta-regression for both cancer types. The weight given to studies is inversely proportional to the

4 1100 C. CANAVAN et al. variance associated with the RR reported in each. For the studies where the confidence interval was not reported, standard error was calculated using the observed cancers and expected cancers reported in the papers. The data used in both meta-analyses are summarized in Table 2. Heterogeneity was explored using meta-regression techniques to assess any change in RR of cancer with time, and subgroup analysis was used to explore differences due to site of initial Crohn s disease, country and type of centre in which study was undertaken, these data are included in Table 1. Sensitivity analyses were undertaken to explore the influence on the overall results of individual studies. The pooled estimate for RR of colorectal cancer in patients with Crohn s disease was then used in a cumulative probability analysis, using the national annual incidence in the UK reported by the National Institute for Clinical Excellence, 29 to calculate the risk of developing colorectal cancer over 30 years diagnosed with Crohn s disease, assuming a constant Table 2. Study data used in the meta-analyses of colorectal cancer and small bowel cancer risk in Crohn s disease Author RR Upper CI Lower CI log RR SE [log RR] Colorectal cancer Weedon et al Gyde et al Greenstein et al Kvist et al. 20 * Gollop et al. 21 * Fireman et al. 22 * ) Ekbom et al Gillen et al Persson et al ) Mellemkjaer et al Bernstein et al Jess et al Small bowel cancer Fielding et al Greenstein et al Ekbom et al Gillen et al Persson et al Mellemkjaer et al Bernstein et al Jess et al * Confidence intervals calculated using observed and expected data, rather than being directly extracted. RR, relative risk. annual incidence ratio. This was then compared to the cumulative risk in the general population 29 and in patients with ulcerative colitis. 30 RESULTS Colorectal cancer Two of the 12 papers in this study reported a RR less than , 22 and four further papers reported a confidence interval that included , 21, 24, 26 The risk reported in the remaining papers ranged from to A chi-squared test for heterogeneity based on these 12 studies yielded v 2 ¼ , P < and therefore a random effects model was used to produce an overall pooled estimate for the colorectal cancer RR of 2.5 with a 95% confidence interval , which was statistically significant (P ¼ 0.004; Figure 2). Subgroup analysis was carried out for colorectal cancer specific to the site of the Crohn s disease, colonic disease or ileal disease. Five papers specified colorectal cancer incidence in patients with colonic disease, 12, 15, 19, 23, 26 the overall pooled estimate for RR of colorectal cancer in these patients is 4.5 with a 95% confidence interval , which was statistically significant (P < 0.015; Figure 3). Only three papers specify the incidence of colorectal cancer in patients with ileal disease, 12, 15, 26 the combined RR for these patients is 1.1 with 95% confidence interval , this risk is not statistically significantly greater than the risk of the general population (P ¼ 0.7). Subgroup analysis was also carried out for the geographical Weedon 1973 Gyde 1980 Greenste in 1981 Kvist 1986 Gollop 1988 Fireman 1989 Ekbom 1990 Gillen 1994 Persson 1994 Mellemkjaer 2000 Bemstein 2001 Jess 2004 Combined Relative risk (log scale) Figure 2. Forest plot of studies reporting relative risk (RR) of colorectal cancer (log scale) for Crohn s patients.

5 META-ANALYSIS: COLORECTAL AND SMALL BOWEL CANCER RISK IN CROHN S % Gyde Ekbom Gillen Persson Jess Combined SIR (log scale) 10 Cumulative probability of CRC 10% 5% Figure 3. Forest plot of studies reporting relative risk (RR) of colorectal cancer (log scale) for patients with Crohn s disease located in the colon. 0% Time (years) region in which the study was conducted. This showed RR of colorectal cancer in Scandinavia is not significantly higher than seen in the general population, 1.4 ( ); whilst in the UK and North America risk it is significantly higher, with RR 3.9 ( ) and 8.5 ( ) respectively. Subgroup analysis of the type of centre where the study was conducted showed no significant difference in RR reported from primary, secondary or tertiary centres. When change in RR with time, assessed by taking the median year of diagnosis plus the average follow-up time, was assessed using a mixed effects meta-regression model there was a decrease in RR of 0.09 each year (95% confidence interval )0.12 to )0.06) this was statistically significant (P ¼ 0.016). The cumulative risk analysis for all patients with Crohn s disease at any site showed that the incidence of colorectal cancer in patients diagnosed with Crohn s disease for 10 years is 2.9% ( %), 5.6% ( %) for patients who have been diagnosed for 20 years and 8.3% ( %) for those diagnosed for 30 years. Cumulative risk of developing colorectal cancer once diagnosed with Crohn s disease is compared to risk in ulcerative colitis 30 and the general population 29 in Figure 4 and this shows that the risk of colorectal cancer for patients with Crohn s disease is statistically significantly greater than in the general population and not statistically significantly different to the risk for patients with ulcerative colitis. Sensitivity analysis shows the study by Weedon et al. 16 has the single greatest effect on the combined RR. However, omitting this study reduces the combined Figure 4. Cumulative risk of colorectal cancer in patients diagnosed with Crohn s disease (solid black line with dark grey 95% confidence interval) compared to that seen in ulcerative colitis, based on unstratified data (black and white dot and dashed line with light grey 95% confidence interval) and the cumulative risk in the general population (dashed line, no confidence interval). RR estimate to 2.1 (95% CI ), which remains within the 95% confidence interval of the overall estimate for all studies. Excluding Weedon et al. 16 from the meta-regression of change in colorectal cancer risk with time results in the decrease being no longer statistically significant, )0.029 (95% CI )0.074 to 0.017, P ¼ 0.21). If Weedon et al. 16 is removed from the sub-group analysis of country, then RR for North America is reduced to 3.8 (95% CI ). Small bowel cancer Only eight papers were found that reported RR for small bowel cancer in Crohn s disease. All report the risk as increased from that in the general population, ranging from to and none report confidence intervals that include 1.0. A chi-squared test for heterogeneity based on these studies yielded v 2 ¼ , P < and therefore a random effects model was used to produce an overall pooled estimate for small bowel cancer RR of 31.2 (95% CI ), which is highly statistically significant (P < ; Figure 5). Subgroup analysis was carried out for the geographic region in which the study was conducted. This showed that the RR of small bowel cancer in

6 1102 C. CANAVAN et al. Fielding, 1972 Greenstein, 1981 Ekbom, 1990 Gillen, 1994 Persson,1994 Mellemkjaer, 2000 Bemstein, 2001 Jess, 2004 Combined Scandinavia was again lower than in the UK and US, but on this occasion the difference was not significant and was still much higher than in the general population. RR by geographic region was: 19.4 ( ), 63.3 ( ) and 44.2 ( ) for Scandinavia, the UK and North America respectively. Subgroup analysis of the type of centre where the study was conducted, again showed no significant difference in RR reported from primary, secondary or tertiary centres. When change in RR with time, assessed by median year of diagnosis plus the average follow-up time, was assessed using a mixed effects meta-regression model there was no significant change. DISCUSSION RR (log scale) Figure 5. Forest plot of studies reporting relative risk (RR) of small bowel cancer (log scale) for Crohn s patients. This study shows the overall combined risk of colorectal cancer in Crohn s disease is more than two and a half times that of the general population. For patients with colonic disease this risk is increased to almost four and a half times that of the general population. From the data available, the risk of colorectal cancer in patients with isolated ileal disease is not statistically significantly different to the general population, although data on colorectal cancer risk in patients with ileal Crohn s disease is very limited. The study by Weedon et al. 16 has the single greatest effect on the combined RR, reporting an especially high RR. However, only patients diagnosed before the age of 21 were included, a patient group known to be at greater risk of colorectal cancer. 12, 23 The other studies have less individual effect on the combined RR, showing the result of the meta-analysis is robust. The percentage weight of each study in the meta-analysis was around 10% except the studies by Gollop 21 and Fireman 22 which contributed 5.5% each. Patients with Crohn s disease in the UK have a significantly higher RR of colorectal cancer than those in Scandinavia. The UK is less pro-active in its approach to surgery than Scandinavia 31 and this could explain the difference. A meta-analysis of international mortality for patients with Crohn s disease 32 shows that mortality for patients with Crohn s disease is not significantly different in these countries, so it would seem the difference in colorectal cancer incidence does not affect overall mortality. In Scandinavia patients with Crohn s disease are managed in specialist referral centres whilst in the UK care is mainly given within the community with occasional hospital support. 33 This difference in approach does not appear to be important, however, as subgroup analysis showed no significant difference between RR of colorectal cancer reported from different types of healthcare setting. Meta-regression analysis of change in RR of colorectal cancer over time shows risk has decreased from 1940 to 1990, but if the study by Weedon et al. 16 is excluded this decrease is no longer statistically significant. Whilst the introduction of steroids and azathioprine may have lowered mortality 32 they have not reduced the risk of developing colorectal cancer, suggesting that these drugs modify some aspects of Crohn s disease but do not affect the pathology that underlies development of colorectal cancer. Some recent studies have reported that whilst acute inflammation counteracts cancer development, chronic inflammation, as in Crohn s disease, promotes cancer in three main ways. 1 Tumour necrosis factor binding to the NFKappaB receptor promotes survival signalling to the cell, which is pro-oncogenic IL-6 counteracts apoptosis of cells COX-2 is elevated in both chronic inflammation and carcinogenesis. Raised expression is associated with poor prognosis in colorectal adenocarcinoma when inhibited tumour growth is suppressed. 35 Infliximab is an anti-tumour necrosis factor antibody, so it may modify the disease in a way that reduces colorectal cancer risk. In the US 20% of colorectal cancer is linked to smoking 36 and a high proportion of Crohn s patients smoke, 37, 38 which represents an additional risk factor. Patients with ulcerative colitis are known to have an increased RR of developing colorectal cancer with a

7 META-ANALYSIS: COLORECTAL AND SMALL BOWEL CANCER RISK IN CROHN S 1103 cumulative probability of 3% at 10 years. 10 It is recommended that such patients are screened for dysplasia of the colonic mucosa by colonoscopy 10 years 39, 40 after diagnosis and repeated every 2 years. The cumulative risk of colorectal cancer 10 years after diagnosis of Crohn s disease at any site is 2.9% and Figure 4 shows no significant difference in risk of colorectal cancer between patients with ulcerative colitis and Crohn s disease, implying that patients with Crohn s disease at any site should be offered the same screening opportunities as those with ulcerative colitis, the results of the subgroup analysis would suggest that this is even more important in patients with colonic involvement. There are many studies of colorectal cancer risk in ulcerative colitis that stratify data by duration of disease. This has allowed more accurate assessment of risk over time, with 18% cumulative probability after 30 years. Similar analysis is not possible with current papers in Crohn s disease because there are too few that stratify risk of developing colorectal cancer at different durations of disease, but it is possible that cumulative risk will increase in a similar way. The benefit of screening has yet to be established. Indeed it has been suggested that rather than increasing survival following diagnosis, screening only increases the lead time 39 and a recent paper on cost efficacy reports that for biannual screening to be cost effective in ulcerative colitis the risk must be greater than 27%. 41 Even after 30 years the risk of colorectal cancer in ulcerative colitis is only 18%, so it would seem guidelines for screening for colorectal cancer in inflammatory bowel disease need revision. The pooled estimate from the meta-analysis of small bowel cancer risk showed an overall RR of 31.2 compared to the general population. Whilst this is very high, the real risk is small because small bowel cancer is rare and accounts for less than 5% of all gastrointestinal cancers. 14 There are currently no effective methods to screen for small bowel cancer and the low incidence, even in Crohn s disease, means there is no possibility of screening. This may change, however, with the increasing use of capsule endoscopy. There is no statistically significant difference in risk of small bowel cancer between different countries or study centres, neither is there any statistically significant change in risk of small bowel cancer in Crohn s disease over 45 years, using median year of diagnosis with Crohn s disease as the marker for passage of time. This would further suggest that steroids and azathioprine have not modified the pathological process that increases the risk of intestinal cancer. There are few papers reporting RR of colorectal or small bowel cancer in Crohn s disease, far less than there are for ulcerative colitis and this limits this meta-analysis. Whilst the papers reporting RR of small bowel cancer report similar findings, the results of studies into RR of colorectal cancer vary greatly, thus there is a considerable uncertainty for the true value of RR. It is unlikely that the true risk of colorectal or small bowel cancer in Crohn s disease has been reported as a capture recapture technique was not used in any study and no studies had an average follow up exceeding 20 years, with most less than 15 years. Also as there is no screening programme for small bowel cancer some asymptomatic patients would not have been identified. Further investigation via an individual patient detail meta-analysis is required to fully elucidate the colorectal and small bowel cancer risks experienced by patients with Crohn s disease and utilize individual patient covariates. ACKNOWLEDGEMENT The study was sponsored by the University of Leicester NHS Trust. REFERENCES 1 Calkins BM, Lilienfeld AM, Garland CF, Mendeloff AI. Trends in incidence rates of ulcerative colitis and Crohn s disease. Dig Dis Sci 1984; 29: Sedlack RE, Whisnant J, Elveback LR, Kurland LT. Incidence of Crohn s disease in Olmsted County, Minnesota, Am J Epidemiol 1980; 112: Nunes GC, Ahlquist RE Jr. Increasing incidence of Crohn s disease. Am J Surg 1983; 145: Trallori G, Palli D, Saieva C, et al. A population-based study of inflammatory bowel disease in Florence over 15 years ( ). Scand J Gastroenterol 1996; 31: Thomas GA, Millar-Jones D, Rhodes J, Roberts GM, Williams GT, Mayberry JF. Incidence of Crohn s disease in Cardiff over 60 years: an update. Eur J Gastroenterol Hepatol 1995; 7: Barton JR, Gillon S, Ferguson A. Incidence of inflammatory bowel disease in Scottish children between 1968 and 1983; marginal fall in ulcerative colitis, three-fold rise in Crohn s disease. Gut 1989; 30:

8 1104 C. CANAVAN et al. 7 Rubin GP, Hungin AP, Kelly PJ, Ling J. Inflammatory bowel disease: epidemiology and management in an English general practice population. Aliment Pharmacol Ther 2000; 14: Loftus EF Jr, Schoenfeld P, Sandborn WJ. The epidemiology and natural history of Crohn s disease in population-based patient cohorts from North America: a systematic review. Aliment Pharmacol Ther 2002; 16: Crohn B, Rosenburg H. The simoidoscopic picture of ulcerative colitis (nonspecific). Am J Med Sci 1925; 170: Eaden JA, Abrams KR, Mayberry JF. The risk of colorectal cancer in ulcerative colitis: a meta-analysis. Gut 2001; 48: Warren S, Sommers SC. Cicatrizing enteritis (regional ileitis) as a pathologic entity. Am J Pathol 1948; 24: Ekbom A, Helmick C, Zack M, Adami HO. Increased risk of large-bowel cancer in Crohn s disease with colonic involvement. Lancet 1990; 336: Greenstein AJ, Sachar DB, Smith H, Janowitz HD, Aufses AH. A comparison of cancer risk in Crohn s disease and ulcerative colitis. Cancer 1981; 48: Torres M, Matta E, Chinea B, et al. Malignant tumors of the small intestine. J Clin Gastroenterol 2003; 37: Persson PG, Karlen P, Bernell O, et al. Crohn s disease and cancer: a population-based cohort study. Gastroenterology 1994; 107: Weedon DD, Shorter RG, Ilstrup DM, Huizenga KA, Taylor WF. Crohn s disease and cancer. N Engl J Med 1973; 289: Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUO- ROM statement. Quality of Reporting of Meta-analyses. Lancet 1999; 354: Fielding JF, Prior P, Waterhouse JA, Cooke WT. Malignancy in Crohn s disease. Scand J Gastroenterol 1972; 7: Gyde SN, Prior P, Macartney JC, Thompson H, Waterhouse JA, Allan RN. Malignancy in Crohn s disease. Gut 1980; 21: Kvist N, Jacobsen P, Nordgaard H, et al. Malignancy in Crohn s disease. Scand J Gastroenterol 1986; 21: Gollop JH, Phillips SF, Melton LJ, Zinsmeister AR. Epidemiological aspects of Crohn s disease: a population based study in Olmsted County, Minnesota, Gut 1988; 29: Fireman Z, Grossman A, Lilos P, et al. Intestinal cancer in patients with Crohn s disease. Scand J Gastroenterol 1989; 24: Gillen CD, Andrews HA, Prior P, Allan RN. Crohn s disease and colorectal cancer. Gut 1994; 35: Mellemkjaer L, Johansen C, Grindley G, Linet M, Kruger Kjear S, Olsen JH. Crohn s disease and cancer risk (Denmark). Cancer Causes Control 2000; 11: Bernstein CN, Blanchard JF, Kliewer E, Wajda A. Cancer risk in patients with inflammatory bowel disease. A population based study. Cancer 2001; 91: Jess T, Winther KV, Munkholm P, Lanholz E, Binder V. Intestinal and extra-intestinal cancer in Crohn s disease: follow-up of a population-based cohort in Copenhagen County, Denmark. Aliment Pharmacol Ther 2004; 19: Sackett DL. Bias in analytical research. J Chronic Dis 1979; 32: Sutton AJ, Abrams KR, Jones DR. An illustrated guide to the methods of meta-analysis. J Eval Clin Pract 2001; 7: National Institute of Clinical Excellence. Improving Outcomes in Colorectal Cancers, Available at: nice.org.uk (Last accessed 20 June 2005). 30 Eaden JA. A Meta-analysis Determining the Risk of Colorectal Cancer in Ulcerative Colitis. Colorectal Cancer in Patients with Ulcerative Colitis. MD University of Leicester, 2000: Jess T, Winther VK, Munkholm P, Langholz E, Binder V. Mortality and causes of death in Crohn s disease: follow up of a population based cohort in Copenhagen County, Denmark. Gastroenterology 2002; 122: Canavan C, Abrams K, Mayberry J. Mortality in Crohn s disease: no change in 34 years. Gut 2005; 54 (Suppl. 11): Hungin P. Working together across international boundaries: improving the primary-secondary care network. Eur J Gastroenterol Hepatol 2001; 13 (Suppl. 2): S Philip M, Rowley DA, Schreiber H. Inflammation as a tumor promoter in cancer induction. Semin Cancer Biol 2004; 14: Ristimaki A. Cyclooxygenase 2: from inflammation to carcinogenesis. Novartis Found Symp 2004; 256: Giovannucci E. An updated review of the epidemiological evidence that cigarette smoking increases risk of colorectal cancer. Cancer Epidemiol Biomarkers Prev 2001; 10: Casellas F, Lopez-Vivancos J, Badia X, Vilaseca J, Malagelada JR. Influence of inflammatory bowel disease on different dimensions of quality of life. Eur J Gastroenterol Hepatol 2001; 13: Card T, Hubbard R, Cogan RF. Mortality in inflammatory bowel disease: a population based cohort study. Gastroenterology 2003; 125: Carter MJ, Lobo AJ, Travis SPL, on behalf of the IBD Section of the British Society of Gastroenterology. Guidelines for the management of inflammatory bowel disease in adults. Gut 2004; 53 (Suppl. V): v Itzkowitz SH, Harpaz N. Diagnosis and management of dysplasia in patients with inflammatory bowel diseases. Gastroenterology 2004; 126: Inadomi JM. Cost-effectiveness of colorectal cancer surveillance in ulcerative colitis. Scand J Gastroenterol 2003; 237 (Suppl.):