Background Herpes zoster (HZ) is an opportunistic infection with a high incidence in Rheumatoid Arthritis (RA) patients. There are contradictory data regarding the risk of HZ associated with treatment with biologic and non biologic disease modifying agents (DMARD).
Objectives To compared the risk for HZ among new users of TNF inhibitors (TNFi), non-TNF inhibitor biologics (nTNFi) and non biologic DMARD (nBIO) in RA patients participating in the Consortium of Rheumatology Researchers of North America (CORRONA) registry.
Methods The CORRONA registry is a network of rheumatology practices across the U.S. with patient and physician derived data on more than 27,000 RA patients. Patients with or without background methotrexate were categorized to three treatment cohorts (TNFi vs nTNFi vs nBIO). Follow-up time began with initiation of a TNFi; a nTNFi; or Leflunomide, sulfasalazine, or hydroxychloroquine (nBIO cohort). Exposure was considered “as treated” and analyses were censored if patients switched treatment groups and after the 1st HZ event. HZ events were based on physician report. Patients with a history of HZ were excluded (analysis without their exclusion yielded similar results). Propensity scores (PS) were created using multinomial logistic regression, and the PS included age, gender, prednisone use, mHAQ, CDAI, duration of RA, Rheumatoid factor positivity (RF+), history of diabetes mellitus (DM) cardiovascular disease (CVD) and cancer (Ca). Stratification on propensity score quintile was conducted and hazard ratios (HR) were estimated using Cox regression to examine the risk of HZ associated with each treatment. Any factors in imbalance within stratified quintiles were examined as additional confounders. In a separate analysis, three way propensity score matching of nBIO, TNFi, and nTNFi biologic (1:2:1) was performed with a propensity score using age, gender, prednisone treatment (tx), mHAQ, duration of RA, and RF+.
Results The propensity stratified analysis included 2,809 in the TNFi cohort, 1,182 in the nTNFi cohort and 1,070 in the nBIO 842 cohort. At baseline: the nTNFi cohort included more females, patients with longer RA duration, more frequent RF+ and steroid tx, higher mHAQ, and CDAI. The TNF cohort included younger patients with fewer comorbidities (DM, Ca, CVD). The overall rate of HZ was 8/1000 person yrs [95%CI: 7, 11]. Using the TNFi cohort as the referent group the hazard ratio for nBIO was 0.97 (95%CI:0.54-1.77, p=0.93) and for nTNFi was 0.81 (95%CI:0.44-1.48, p=0.49). Adjustment of additional confounders after stratification by propensity quintiles did not change results. Propensity matching resulted in 1,658 TNFI, 829 nBio and 829 nTNFi. The HZ rate in the matched cohorts was 9/1000 person yrs [95% CI: 6, 11]. Results are presented in Table 1 and are similar between the two analytic approaches.
Conclusions The rate of HZ was comparable among RA patients initiating TNFi, nTNFi or nBIO DMARDs agents in the CORRONA registry.
Disclosure of Interest D. Pappas: None Declared, M. Hooper Shareholder of: Amgen, Employee of: Amgen, G. Reed Grant/Research support from: CORRONA, Consultant for: CORRONA, Employee of: University of Massachusetts Medical School, Paid Instructor for: Harvard Medical School, Y. Shan: None Declared, D. Wenkert Shareholder of: Amgen, Employee of: Amgen, J. Zhang Grant/Research support from: Amgen, J. Greenberg Shareholder of: CORRONA, Consultant for: AstraZeneca, Novartis, Pfizer, CORRONA, J. Curtis Grant/Research support from: NIH (AR053351), AHRQ (R01HS018517), Amgen, UCB, CORRONA, Genentech, Janssen, Pfizer, Crescendo Bioscience, Consultant for: Amgen, UCB, CORRONA, Genentech, Janssen, Pfizer, Crescendo Bioscience