Use of a Molecular Signature Response Classifier to Inform Treatment Selection Improves Clinical Disease Activity Among Patients with Rheumatoid Arthritis Initiating a Biologic or Targeted Synthetic Disease-Modifying Antirheumatic Drug.

Abstract

We assessed the effectiveness of PrismRA to improve clinical outcomes among patients with rheumatoid arthritis (RA) initiating treatment with a biologic or targeted synthetic disease-modifying antirheumatic drug (b/tsDMARD). PrismRA incorporated 19 gene expression features and four clinical features to assess a patient's likelihood of inadequate response to tumor necrosis factor inhibitor (TNFi). PrismRA was assessed in a prospective, interventional cohort study of patients initiating treatment with a b/tsDMARD. PrismRA results were provided to treating rheumatologists and incorporated into the selection of TNFi vs non-TNFi for study treatment. External comparator patients were identified in a rheumatology provider electronic health records system and matched to PrismRA patients using propensity scores. All patients had moderate-high disease activity at baseline. The primary study outcome was achievement of minimal important difference (MID) in clinical disease activity index (CDAI) at 24 weeks. Last observation carried forward was used to impute missing data. There were 330 PrismRA cohort patients and 990 matched comparator patients. Key baseline patient characteristics were all well-balanced between cohorts. Study treatment selection was consistent with PrismRA results in 82% of PrismRA cohort patients. CDAI MID at 24 weeks was achieved by 63.0% of PrismRA patients and 42.4% of comparator patients (odds ratio 2.31, confidence interval 1.79-2.99). PrismRA results informing selection of TNFi vs non-TNFi treatment was associated with better CDAI outcomes compared to matched external comparator patients. PrismRA testing helps fill the need for a precision medicine approach to more rapidly identify the most effective therapy for individual patients with RA.