BACKGROUND: Choosing optimal biologic and targeted synthetic DMARD (b/tsDMARD) regimens for individual RA patients can be challenging. This study used electronic health records and linked genotyping data from two large biobanks to investigate genetic factors associated with RA treatment patterns.
METHODS: We identified RA patients initiating b/tsDMARDs and with available genotyping data in the Million Veteran Program (MVP) and Mass General Brigham (MGB) Biobank. We used mixture Markov modeling to identify patterns of b/tsDMARD use. We calculated inflammatory arthritis genetic risk scores (GRS) for RA, spondyloarthritis, psoriatic arthritis, lupus, and gout and examined the association of the GRS with b/tsDMARD treatment pattern using logistic regression.
RESULTS: Among 2,454 RA patients in the MVP and 461 patients in the MGB Biobank, we identified common patterns of b/tsDMARD use: TNFi persisters, TNFi to abatacept, and multi-bDMARD cyclers. In the meta-analysis, a higher RA GRS was associated with higher odds of belonging to the TNFi to abatacept group (OR 1.129 95%CI 1.008-1.264 per GRS standard deviation) and multi-b/tsDMARD cycling (1.137 95%CI 1.056-1.223) compared to TNFi-persisters. The spondyloarthritis GRS was associated with a lower odds belonging to multi-b/tsDMARD (OR 0.824 95%CI 0.762-0.892) or TNFi to abatacept (OR 0.883 95%CI 0.786-0.992) patterns compared to TNFi persisters.
CONCLUSIONS: Among patients with RA, we observed common patterns of b/tsDMARD use in two separate healthcare systems. Moreover, these patterns had different underlying genetic associations. Higher RA GRS was associated with lower odds of TNFi persistence, while a higher SpA GRS was associated with higher odds of TNFi persistence.