Background: Targeted treatments for rheumatoid arthritis (RA) have significantly improved disease management by controlling symptoms and slowing progression. Despite these advances, a substantial proportion of patients exhibit suboptimal responses to specific biologic and targeted synthetic DMARDs (b/ts DMARDs), often cycling through multiple treatments before achieving disease control. This variability in therapeutic response underscores a critical gap in treatment selection and the pressing need for a precision medicine approach. A predictive test to identify responders to specific b/ts DMARDs could support more informed and effective treatment decisions, improving patient outcomes and cost-effectiveness.

Objectives: This study aims to develop and validate a novel blood-based assay leveraging cell-free DNA derived from regulatory-active chromatin (cfDNAac) to predict response to bDMARDs and tsDMARDs using data from participants enrolled in the PRIMA-102 study, an ongoing prospective, multi-center, observational trial (NCT05936970).

Methods: Baseline plasma samples from 188 RA participants in the PRIMA-102 study were collected prior to therapy switch to TNFi (n=102), JAKi (n=44), or anti-T-cell therapies (n=42). Participants were classified as responders (n=106) or non-responders (n=82) based on Clinical Disease Activity Index (CDAI) scores at baseline and three months post-therapy. Based on published definitions for the minimal clinically important difference in the CDAI, response was defined as a CDAI improvement of ≥12 units for patients starting in high disease activity (CDAI ≥22) and improvement ≥6 units for patients starting in moderate disease activity (CDAI ≥10). Samples were processed using a novel cfDNAac capture assay to access tissue- and disease-specific molecular signatures. Of 161 samples, 70 were used to develop classifiers for each mechanism of action by identifying features discriminating responders from non-responders. Machine-learning (ML) models were trained with 25 bootstrap iterations and tested on the remaining 91 samples. The performance of the three classifiers was further validated using an independent, blinded hold-out cohort of 27 randomly selected participants, balanced across b/ts DMARD classes and response categories.

Results: Participants had a mean age of 57.8 years, with 85.6% female and 76.1% white. Mean baseline CDAI was 29.5, and 52.1% of participants were seropositive. The cohort included 18.1% b/tsDMARD-naive participants, with remaining participants having exposure to various numbers of prior b/tsDMARD: 38.8% with exactly 1, 24.5% with exactly 2, 5.9% with exactly 3, and 12.8% with ≥4 (Table 1). The classifiers demonstrated strong discriminatory performance across all drug classes, achieving an overall area under the curve (AUC) of ≥ 0.82. Sensitivity at 80% specificity ranged from 70.6% to 76.4%, depending on mechanism of action (Table 2). Validation in the hold-out cohort provided evidence of the classifiers’ robustness. Specifically, the TNFi classifier demonstrated 80% sensitivity at 100% specificity (n=16), the JAKi classifier showed 100% sensitivity at 100% specificity (n=6), and the anti-T-cell classifier exhibited 67% sensitivity at 100% specificity (n=5).

Conclusion: This interim analysis demonstrates the potential of novel blood-based cfDNAac classifiers to predict therapeutic response to TNFi, JAKi, and anti-T-cell therapies in RA patients. While the small training and validation cohorts to date limit the scope of these findings, recruitment is ongoing, and these findings provide a critical first step in evaluating the utility of this approach. The larger cohort in the ongoing PRIMA-102 study will be pivotal in validating these classifiers for future clinical use and advancing personalized treatment strategies for RA.

REFERENCES: NIL.

Acknowledgements: NIL.

Disclosure of Interests: Peter C. Taylor AbbVie, Aqtual, Inc., Biogen, Fresenius, Alfasigma, Gilead, Roche, Takeda, Lilly, Nordic Pharma, Pfizer, Sanofi, UCB, Acelyrin Inc., Immunovant, Sanofi and Moonlake, Alfasigma, Kevin Lai Aqtual Inc., Signe Fransen I am an employee of Aqtual Inc., Gordon Lam Abbvie, Amgen, AstraZeneca, GlaxoSmithKline, Johnson&Johnson, Pfizer, Sanofi, Abbvie, Amgen, Aqtual, AstraZeneca, Bristol Myers Squibb, GlaxoSmithKline, Johnson&Johnson, Pfizer, Sanofi, Setpoint, UCB, David Chernoff SetPoint Medical, SetPoint Medical, Aqtual, Inc. and Reflexion Pharma, Diana Abdueva Aqtual Inc., Aqtual Inc., Jason Carlson Aqtual Inc., Jeffrey R Curtis AbbVie, Amgen, Aqtual, BMS, GSK, Janssen, Lilly, Moderna, Novartis, Pfizer, Sanofi, Scipher, Setpoint, TNacity Blue Ocean, UCB, AbbVie, Amgen, Aqtual, BMS, CorEvitas, GSK, Janssen, Lilly, Moderna, Novartis, Pfizer, Sanofi, Scipher, Setpoint, UCB.

© The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.