Background/Purpose: Whether anti- TNF agents increase the frequency of neoplasms remains debated.  The drug regulatory agencies still mandate black boxes on the related package inserts or drug advertisements. We hypothesized that one important reason for this continuing concern was that the majority of studies addressing this issue had methodological weaknesses.  Specifically, we had the impression that tabulating the number of cancers in the comparator groups (CGs), the expected number of cancers, was especially problematic. In observational studies when the CG is a national registry like SEER, the expected cancer incidence is given as new cancers/number of population rather than the number of patients with a new cancer which is the common method in assessing the observed incidence. Since multiple primary cancers, (around 1/6th) of all new cancers in the general population (1), are not separately tabulated as such in the national registries, a fair comparison of the cancer incidence between a patient cohort and a national registry is prone to important error. This potential error is further compounded by the fact patients with a previous history of cancer are not usually prescribed anti-TNF agents.  Thus we set out to formally test our impression that the issue of multiple primary cancers is not accounted for in the debate about whether anti-TNF agents increase the incidence of neoplasms.   

Methods: To test our hypothesis we first selected a recent comprehensive review (2) related to this debate which had concluded that anti-TNF agents did not increase cancer rate. All references to human studies were reanalyzed by 2 independent observers seeking whether the issue of multiple primary cancers was considered in studies which used cancer rates in the background general population as comparator. Any discrepancies between observers were to be settled by combined reassessment.

Results: Thirty-six references were selected for analyses. 24 were observational studies and 12 were meta-analyses or systematic reviews of controlled clinical trials and/or observational studies. 19 studies used national general population databases such as SEER as CGs. In none of these studies the issue of multiple primary cancers in the national database was addressed, causing a potential decrease in SIR for cancers in the anti-TNF using group, due to counting patients with multiple cancers in the general population more than once. There were no discrepancies in data assessment between the 2 observers.                                                                                                         

Conclusion: The important issue of multiple primary cancers have not been given due consideration in studies comparing the cancer incidence associated with anti-TNF use. We propose that all previous related studies which had used this problematic methodology should be reanalyzed and reinterpreted.

References

1. Wood ME, Vogel V, Ng A, Foxhall L, Goodwin P, Travis LB. Second malignant neoplasms: assessment and strategies for risk reduction. J Clin Oncol. 2012;30(30):3734-45.
2. Lebrec H, Ponce R, Preston BD, Iles J, Born TL, Hooper M. Tumor necrosis factor, tumor necrosis factor inhibition, and cancer risk. Curr Med Res Opin. 2015;31(3):557-74

« Back to 2015 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/assessing-the-possible-association-of-anti-tnf-use-with-new-neoplasms-an-important-methodological-consideration/