Risk of lymphoma: inflammatory bowel disease and immunomodulators

We read with great interest the meta‐analysis by Kandiel et al (Gut 2005;54:1121–5), the purpose of which was to provide a more precise estimate of the relative risk of lymphoma among inflammatory bowel disease (IBD) patients treated with azathioprine or 6‐mercaptopurine (6‐MP). Based on the six studies included in their meta‐analysis, Kandiel et al determined that the risk of lymphoma among IBD patients who did not receive immunomodulators was similar to that of the general population while the lymphoma risk among IBD patients treated with azathioprine/6‐MP was increased approximately fourfold. It is reassuring that the largest study included in their meta‐analysis, which utilised the UK General Practice Research Database, failed to show a significantly higher risk of lymphoma among 16 996 IBD patients, regardless of whether or not patients received azathioprine/6‐MP.¹ While several population based and hospital based studies have similarly failed to identify a significantly increased risk of lymphoma in IBD patients who never received immunomodulators,² it is worth noting that two studies have recently reported an increased lymphoma risk in IBD patients independent of immunomodulator or biological therapy.

In 2000, an Italian study of 920 IBD patients reported a ninefold increased risk of Hodgkin's disease among ulcerative colitis patients but not among Crohn's disease patients.³ More recently, a large population based study from Manitoba, Canada, compared 2857 Crohn's disease and 2672 ulcerative colitis patients with age, sex, and geographic location matched controls and reported an almost fourfold increased risk of lymphoma among males with Crohn's disease.⁴ This suggests that disease activity and underlying disease related immune alterations in IBD may be associated with a small but significant inherent risk of lymphoma.

In support of this association, a Swedish nested case control study reported a strong linear association between disease activity in rheumatoid arthritis (RA) patients and risk of developing lymphoma, which was not only independent of immunomodulator treatment but of a larger magnitude than the risks linked to immunomodulator therapy reported in other RA studies.⁵ Compared with RA patients with low inflammatory activity, patients with high and medium inflammatory activity had odds ratios of 25.8 and 5.4, respectively, for the development of lymphoma. This led the authors to recommend “the use of potent immunosuppressive treatment to reduce disease activity, not only to prevent joint damage but possibly also to protect against lymphomas.”

The meta‐analysis by Kandiel et al reported a pooled relative risk of 4.18 for lymphoma risk among IBD patients treated with azathioprine/6‐ MP. As the authors acknowledge, previously published population based and hospital based cohort studies have presented conflicting results regarding the risk of lymphoma in IBD patients treated with azathioprine and 6‐MP. Reasons for these conflicting results may be related to inadequate sample size given the rarity of the outcome of interest as well as a wide range of estimates for the expected number of lymphomas. Discrepancies in the duration and dosage of treatment as well as limited length of follow up have also contributed to the conflicting results. Furthermore, it is well known that cohort studies such as ours⁶ and that reported by Kinlen⁷ which compared hospitalised or referral exposed cohorts to unexposed population based cohorts can result in artificially inflated relative risks as hospitalised or referral centre patients are more likely to have severe disease and multiple illnesses which are not necessarily related (that is, Berkson's bias).⁸ Ultimately, as IBD patients receiving immunomodulators by definition have more severe or refractory disease, it may be difficult to separate whether an increase in lymphoma risk is related to immunomodulator therapy or the severity of the underlying disease activity.

Unfortunately, as this meta‐analysis demonstrates, the actual risk of lymphoma in IBD patients treated remains unknown and our safety data with respect to azathioprine/6‐MP and lymphoma, albeit reassuring, remains largely based on observational studies. Thus it is not possible to fully exclude the possibility that an increased risk of lymphoma is associated with disease severity rather than immunomodulator therapy. Larger scale, prospective, population based studies are necessary to define the actual risk of lymphoma in IBD and adequately ascertain the influence of immunomodulators on that risk. Until then, we agree with Kandiel et al that the lymphoma risk associated with azathioprine and 6‐MP in IBD is likely to be of minimal clinical significance compared with the established and more frequent risks of myelosuppression and infection, and is far outweighed by the benefit of immunosuppression in IBD.