Lymphoma risk with azathioprine/6-MP therapy—read beyond the headlines

Abstract

Kandiel A, Fraser AG, Korelitz BI, Bresinger C, Lewis JD (University of Pennsylvania, Philadelphia, Pennsylvania). Increased risk of lymphoma among inflammatory bowel disease patients treated with azathioprine and 6-mercaptopurine. Gut 2005;54:1121–1125. Azathioprine and 6-mercaptopurine (6-MP) remain important agents for the induction and maintenance of remission in inflammatory bowel disease, and the prevention of postoperative Crohn’s disease (Gastroenterology 2004;127:723–729). Whether their use is associated with an increased risk of lymphoma has been a matter of controversy for many years as studies examining this risk have produced conflicting results. This is further complicated by the uncertainty over the overall risk of lymphoma in patients with IBD, and the significance of immunomodulator therapy as a marker of disease severity. At present, we lack prospective data measuring lymphoma risk in patients taking azathioprine/6-MP when compared with IBD patients taking other medications. Given the clinical benefits to patients seen in daily practice, the risk of lymphoma would need to be significant to outweigh the benefits of therapy with azathioprine/6-MP (Gastroenterology 2000;118:1018–1024). In an attempt to clarify this issue, Kandiel and colleagues recently performed a systematic review and meta-analysis of studies that examined the risk of lymphoma in IBD patients treated with azathioprine or 6-MP. They confined their meta-analysis to cohort studies that comprised IBD patients who received azathioprine or 6-MP, and were designed to measure cancer as an adverse outcome. The meta-analysis only included full articles published in the English language. In order to calculate a pooled estimate of the relative risk of lymphoma in each study, the authors extracted the primary data where available and compared observed numbers of lymphomas in patients with IBD on azathioprine/6-MP to age- and sex-specific rates for the general population in that region (eg, national rates). From this, they were able to calculate a standardized incidence ratio (SIR) for each study. In addition, where data was available, they compared lymphoma risk among patients receiving azathioprine/6-MP to those IBD patients who did not receive azathioprine/6-MP using Mantel-Haenszel methods. They also examined heterogeneity among the studies and the individual cases of lymphoma in each study. In total, 6 studies met their inclusion criteria and were reviewed in the paper. When the data from all 6 studies were pooled, the SIR for lymphoma (ratio of observed cases in IBD patients to expected cases for the general population) was 4.18 (95% CI, 2.07–7.51). However, there was significant heterogeneity between the studies because of major differences in results between 2 of the studies. In addition, the size of the study population was inversely related to the relative risk of lymphoma; the larger studies reported a lower estimated relative risk of lymphoma. When the authors examined various combinations of studies to determine the influence of individual studies on the overall results, the SIR ranged from 3.49 to 5.21. Three of the included studies had sufficient data to compare the relative risk of lymphoma in IBD patients on azathioprine/6-MP compared to those IBD patients who had not received these medications. A pooled analysis estimated the relative risk of lymphoma in IBD patients on these immunomodulators to be 2.95 (95% CI, 1.05–8.13). Finally, the authors looked at the risk of non-Hodgkin’s lymphoma in the 6 studies and calculated the SIR as 3.92, although again there was a wide variation between the individual studies as reflected by wide confidence intervals. The authors concluded that although there is an approximate 4-fold increased risk of lymphoma in IBD patients treated with azathioprine/6-MP, the benefit-to-risk ratio of these medications still favors therapy. How should these results be interpreted in practice? Like all meta-analyses, this study is only as good as the included studies, and the authors acknowledge this. The exclusion of non-English language papers, and all abstracts, leaves this meta-analysis open to publication bias (Lancet 1997;350:326–329). The authors did not include a funnel plot to examine this issue, but one would expect asymmetry given the heterogeneity of included studies. Even among full published papers, citation bias and multiple publication bias can increase the probability that only “positive” studies are found in systematic reviews (Control Clin Trials 1989;10:31–56). Thus, it is possible that trials not included in this analysis would have produced a different conclusion if added to the analysis. As is common in other studies of association between disease and outcome, the smaller trials published have much higher risks reported than the larger studies. All the studies included except one are based on patients treated in tertiary referral centers. These patients may have more severe inflammatory bowel disease than those in the community, and thus higher comorbidity. Indeed, measuring any outcome in tertiary referral patients introduces Berkson’s bias (eg, patients within the health care system are more likely to have additional diagnoses made than those in the general community) (Biometrics 1946;2:47–53). In fact, the only population-based study included in this publication found no significantly increased risk of lymphoma among patients receiving azathioprine/6-MP (Gastroenterology 2001;121:1080–1087), suggesting a disease-severity bias in the results (Gastroenterology. 2001;121:1239–1242). The actual meta-analysis of SIRs seems to be consistent across sensitivity analyses, which strengthens the confidence in the results. However, given the significant heterogeneity in the included studies, their combination in a meta-analysis may not be appropriate. A 4-fold increased risk of lymphoma compares observed cancers in IBD patients on azathioprine/6-MP to the expected cases in the general population. Therefore, these results do not exclude the possibility that either having IBD alone or having IBD of sufficient severity to warrant azathioprine/6-MP therapy is responsible for the increased risk. The risk of lymphoma in patients with IBD not receiving azathioprine/6-MP is a matter of debate, with conflicting results from population-based studies (Gut 2005;54:617–622, Gastroenterology 2001;121:1080–1087, Cancer 2001;91:854–862). In fact, population-based studies have shown conflicting results in the risk of many cancers in ulcerative colitis (Clin Gastroenterol Hepatol 2004;2:1088–1095, Am J Gastroenterol 1999;94:1047–1052) and Crohn’s disease (Aliment Pharmacol Ther 2004;19:287–293). For example, patients with ulcerative colitis who require immunomodulator therapy have, by definition, more severe disease, which is an established independent risk factor for colorectal cancer. Similarly, the representation of smokers, a risk for Hodgkin’s lymphoma, is not known in the studies included in this meta-analysis. In summary, the increased risk of lymphoma reported in this meta-analysis is unlikely to warrant changes in our current clinical practice. When contemplating starting IBD patients on azathioprine or 6-MP, it remains important to reassure them that while these drugs may carry a potentially small increased risk of lymphoma, the benefits of such therapy still outweigh the risks, and the alternatives carry significant risks of their own.