Background In patients who have failed 1 or more TNF-α inhibitors (TNFi), there is little data to guide clinical decision making in terms of changing mechanism of action or prescribing another TNFi.

Objectives To compare the effectiveness of rituximab (RTX) versus a subsequent TNFi among RA patients with prior TNFi exposure using data from a multi-center observational registry within the United States (the Consortium of Rheumatology Researchers of North America: CORRONA).

Methods We identified RA patients from 3/1/06 to present who had discontinued at least 1 TNFi and initiated either RTX or a subsequent TNFi, had moderate or high disease activity based on the Clinical Disease Activity Index (CDAI) at the time of initiation and had follow-up at 12 months (3 month window). A propensity score (PS) for TNFi vs. RTX was estimated using patient demographic information, disease characteristics (severity, duration, activity), comorbidities, past medication history and concurrent medications. We trimmed the PS distributions, excluding patients who fell outside the region of common support (n=2). Our primary outcome was achievement of low disease activity (CDAI ≤10). Multivariable logistic regression models with adjustment for fixed and random effects (patient and provider) were performed. Covariates with a standardized difference of <0.1 as well as 4 factors chosen a priori (baseline CDAI, steroid use, background methotrexate use, and number of prior TNFi’s) were included in the models. As a sensitivity analysis, we reran the analyses using PS matched TNFi and RTX initiators.

Results 266 RTX users and 744 TNFi users who met inclusion criteria were included in the analyses. Baseline characteristics are described in Table 1. As compared to the TNFi users, RTX users were older, had a greater duration of RA, had prior exposure to a greater number of prior biologics and were more likely to be receiving prednisone. Achievement of low disease activity occurred in 35% of the RTX users and 28% of the TNFi users. RTX was associated with a greater likelihood of achievement of low disease activity (1.74, 95% CI 1.22-2.47) after adjustment for age, race, insurance status, comorbidity, and RA characteristics (baseline disease activity, severity and medication use [current and prior]). Results from PS matched multivariable analyses were similar. Overall reported rates in RTX users were 0.02 events per person-year (PPY) (95% CI 0.01-0.04) for cardiovascular events, 0.02 PPY (95% CI 0.01-0.05) for serious infections and 0.02 PPY (95% CI 0.01-0.04) for malignancies. The rates in TNFi users were 0.02 PPY (95% CI 0.01-0.03), 0.03 PPY (95% 0.02-0.05) and 0.02 PPY (95% 0.01-0.03), respectively.

Conclusions In RA patients who failed 1 or more TNFi agents, use of RTX was associated with a higher likelihood of achieving CDAI low disease activity than use of a subsequent TNFi.

Disclosure of Interest: L. Harrold Grant/research support from: National Institute of Health - K23AR053856, Consultant for: CORRONA, G. Reed Consultant for: University of Massachusetts Medical School, Employee of: CORRONA, R. Magner Employee of: University of Massachusetts Medical School, A. Shewade Employee of: Genentech, Inc., A. John Employee of: Genentech, Inc., W. Reiss Employee of: Genentech, Inc., J. Greenberg Shareholder of: CORRONA, Consultant for: AstraZeneca, CORRONA, Novartis and Pfizer, J. Kremer Shareholder of: CORRONA, Employee of: CORRONA