
Novel Predictive Model May Help Identify Patients at Risk for Difficult-to-Treat RA
Key Takeaways
- A predictive model identifies RA patients at risk of D2T-RA, using machine learning and real-world data for early intervention.
- Patient-reported outcomes, such as pain and fatigue, are stronger predictors of D2T-RA than traditional biomarkers.
The study highlighted pain, fatigue, and functional status as key early warning signs.
A group of researchers has developed and externally validated the first-ever predictive model to help clinicians identify patients with
The study, published in
“Overall, our predictive model demonstrated moderate discrimination and calibration in derivation and validation cohorts,” wrote the researchers, noting that while the model’s performance was modest, they view it as an important proof of concept.
While biologic and targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs) have transformed care for RA, about 20% of patients cycle through multiple therapies without achieving remission or low disease activity and are categorized as difficult-to-treat. Previous
In 2021, the European League Against Rheumatism established formal criteria for defining D2T-RA, requiring 3 elements: multiple DMARD failures, ongoing disease activity, and management perceived as problematic by the patient or physician. Yet, no clinical tool has been able to identify patients before they reach this state.
To address the gap, the research team analyzed data from 2 large longitudinal registries—the Brigham and Women’s Arthritis Sequential Study (BRASS) and the CorEvitas CERTAIN cohort—comprising over 2700 RA patients treated with biologic or targeted synthetic DMARDs.
Using a random survival forest model, the investigators assessed 23 clinical and patient-reported variables, including pain, fatigue, global disease activity, functional status, and laboratory markers such as C-reactive protein (CRP). The model was trained on BRASS participants and externally validated in CERTAIN.
Over a median follow-up of 40 months in BRASS, 16% of patients progressed to D2T-RA, compared to 28% in the CERTAIN validation cohort. The model achieved a C-index of 0.64 in the derivation dataset and 0.62 in validation, indicating moderate predictive accuracy. The researchers noted that integrating biomarkers, genetic profiles, or imaging data could enhance predictive power.
Surprisingly, patient-reported outcomes (PROs) emerged as the strongest predictors of progression, surpassing traditional biomarkers and physician assessments. Specifically, functional impairment, pain, fatigue, and patient global assessment (PtGA) scores were most strongly linked to future D2T-RA development, suggesting these data should be included in decisions about treatment.
“Our finding that PROs of pain, fatigue, functioning, and PtGA score were top predictors of progression to D2T-RA has not been previously reported. However, a connection between persistent unresolved pain in patients with RA treated with b/tsDMARDs has been reported in the literature,” explained the researchers, noting that the previous research was not able to fully determine the trajectory of pain and non-inflammatory predictors.
Other predictors included higher disease activity, obesity, and ongoing prednisone use, while concurrent methotrexate therapy appeared protective against progression.
The team plans to refine the tool and evaluate its utility in clinical practice to support shared decision-making and personalized care.
References
1. Paudel ML, Shadick N, Weinblatt, et al. Development and external validation of a multivariable predictive model for progression to difficult-to-treat rheumatoid arthritis in biologic-experienced patients. Arthritis Care Res. Published online September 22, 2025. doi:10.1002/acr.25654
2. Hofman ZLM, Roodenrrijs NMT, Nikiphorou E, et al. Difficult-to-treat rheumatoid arthritis: what have we learned and what do we still need to learn? Rheumatology. 64(1):65-73. doi:10.1093/rheumatology/keae544
Newsletter
Stay ahead of policy, cost, and value—subscribe to AJMC for expert insights at the intersection of clinical care and health economics.