Reply

Article Outline

• Copyright

We thank Peyrin-Biroulet and colleagues for their comments regarding our manuscript. They raise useful considerations; however, we believe that we have performed the best possible analysis with currently available data.

First, Peyrin-Biroulet et al suggest that we should have included studies with a follow-up duration as short as 12 weeks. We made the decision to include studies with longer follow-up a priori based on the reasoning that (1) it would be impossible to know if non-Hodgkin's lymphoma (NHL) diagnosed within 12 weeks of initiating anti-TNF therapy was pre-existing or biologically plausible, and (2) patients in short-term trials may have developed NHL after the study was completed, but it likely would not have been reported within that original publication. We feel that 48 weeks was the most appropriate time period to capture the cases of NHL that were most likely attributable to the treatment.

Second, the correspondents suggest that, in order to increase our statistical power, we should have included other anti-TNF agents that have not been shown to be effective in the treatment of Crohn's disease. We respectfully disagree. We do not see the value of including agents that are not and will not be in clinical use in our patients, and we would prefer to accept the possibility of a type II error rather than dilute clinical applicability.

We completely agree with the comment that we cannot make definitive conclusions regarding the risk of lymphoma and anti-TNF therapy. Meta-analyses are not intended to provide definitive evidence of causality; rather, they are intended to best summarize the available data in a systematic manner. Using this technique, we show that there is an association between combination therapy and a higher rate of NHL in patients with Crohn's disease when compared with the expected rate seen in Surveillance, Epidemiology, and End Results (SEER). The influence of anti-TNF therapy as sole therapy is simply unknown at this time, as there are not enough available data on patients receiving anti-TNF therapy who have never been exposed to immunomodulators.

It is clear that we need to learn more about how anti-TNF and immunomodulator therapies interact to influence the risk of NHL. Unfortunately, we did not have patient level data regarding concomitant therapy. As Peyrin-Biroulet et al suggest, such data would have provided further insight into this question. Nevertheless, we believe that the main point of our manuscript is that while combination therapy is probably associated with an increased risk of lymphoma, this risk is very small (about 4 extra lymphomas per 10,000 patient-years), and in most cases the benefits of treatment to the group of patients who need our most effective therapies will likely outweigh the risks.