Background/Purpose: The immune system recognizes, controls, and eliminates tumors through a process of immunosurveillance. In RA, where there is an increased incidence of lymphoma and lung cancer, there has been much concern regarding the potential for biologic therapies to increase cancer incidence. While epidemiologic reports have been somewhat conflicting, several suggest an increased risk of skin cancer with anti-TNF therapy. Newer non-TNF biologic and small molecule therapies have been subject to limited investigation. We examined the associations of biologic and small-molecule therapies with incident cancer in RA.

Methods: We studied RA patients without prevalent cancer participating in the National Data Bank for Rheumatic Diseases (NDB), a longitudinal US-wide observational study that includes semiannual patient questionnaires from 1998 to 2015. Malignancy was determined per a standard protocol that included patient report, patient/family interview, review of medical records, and linkage with the National Death Index. We calculated cancer incidence rates using Poisson regression stratified by medications and the adjusted risk of incident cancer in multivariable Cox proportional hazards models. In Cox models, we analyzed medications individually (remains on drug indefinitely after initiation), grouped by mechanism of action (e.g. anti-TNF), and in a time-varying hierarchical model (MTX monotherapy, other DMARDs, anti-TNF, non-TNF biologics, and tofacitinib [TOF]).

Results: Among 11,582 patients, 6,262 biologic/TOF initiations (mean age 58 years, 80% female, and 48% current/former smoker; n=1,621 non-TNF and n=222 TOF) and 5,320  DMARD initiations (mean age 59 years, 79% female, and 49% current/former smoker) were identified with 1,456 incident cancers (812 excluding non-melanoma skin) occurring during 47,243 patient-years of follow-up. Cancer incidence rates/1,000 PY are shown in Table 1. Risk of incident cancer by medications in a hierarchical drug model and for individual therapies in separate models are shown in Table 2.

Conclusion: We observed increased rates of solid tumors with TOF and lymphoproliferative malignancies with non-TNF biologics, though both had imprecise estimates. After adjustment for several confounding variables, we did not find an increased risk of cancer with biologic or TOF relative to MTX monotherapy. Combination and non-MTX DMARD therapy had the lowest risk of incident cancer, and several therapies individually were associated with increased cancer risk modeled against MTX monotherapy. These findings highlight the complexity of determining cancer risk with therapies, and show how reasonable changes to methodological analysis can provide disparate results.