The research, published in Frontiers in Musculoskeletal Disorders, evaluated a molecular signal response classifier, or MSRC, in Hispanic patients with rheumatoid arthritis recruited from a private rheumatology practice in El Paso, Texas, where 82.9 percent of residents are Hispanic and 78.2 percent identify as Mexican-American. The test detects gene expression patterns in peripheral blood that predict an inadequate response to TNFi therapy. It combines RNA sequencing data from 19 genes with three clinical characteristics, namely sex, body mass index, and patient global assessment of disease activity, along with antibody status for anticitrullinated peptide antibodies. A numeric score of at least 10.6 on the test’s 1-to-25 scale signals a signal of inadequate response, or SIR, to TNFis, a threshold that carried a positive predictive value of 87.7 percent in earlier studies conducted predominantly in non-Hispanic white patients.

Whether such a classifier performs equally well in Hispanic populations has been an open question. Hispanics with rheumatoid arthritis exhibit different clinical features and worse patient-reported outcomes than non-Hispanic whites, including higher pain scores and poorer global assessments of disease activity, and unique genetic, socioeconomic, and environmental factors may influence both disease activity and treatment response. Because predictive biomarkers are generally tested and validated in non-Hispanic whites, their generalizability to diverse groups has remained uncertain, a gap the new study set out to close.

The investigators compared 108 patients who underwent MSRC testing with 206 untested controls in an unmatched case-control design. All participants had at least moderate disease activity despite conventional synthetic disease-modifying antirheumatic drugs and were planning to initiate or change biologic or targeted synthetic therapy through shared decision-making with their rheumatologist. Patients were reevaluated roughly 4.5 months after treatment change, and outcomes included low disease activity on the clinical disease activity index, the minimum clinically important difference in that index, low disease activity on the patient-completed Routine Assessment of Patient Index Data-3, low patient global assessment, and minimal pain, all defined by validated thresholds.

The headline finding was striking: 64.8 percent of tested Hispanic patients, 70 of 108, exhibited an SIR to TNFis, meaning the majority of patients with significant ongoing disease activity were not poised to respond to the preferred first-line biologic class. This figure aligns closely with the 56.3 percent rate reported in unselected non-Hispanic white populations, indicating that the classifier operates similarly across ethnic groups. Tested patients with an SIR received TNFis far less frequently than untested controls, at 14.3 percent versus 57.8 percent, an adjusted odds ratio of 0.12. The effect was even more pronounced among patients who had never received a biologic, where only 9.1 percent of SIR-positive patients were prescribed a TNFi compared with 76.9 percent of bDMARD-naive controls, an adjusted odds ratio of 0.03.

Notably, the test results influenced not only physician prescribing but also patient agreement. Because treatment decisions were made through shared decision-making, Hispanic patients themselves appeared to heed the MSRC results when weighing therapeutic options. Overall, 90.7 percent of all tested patients and 95.1 percent of biologic-naive tested patients received therapies aligned with the MSRC result, rates comparable to the 75.6 percent alignment reported in majority white samples. One sociodemographic pattern emerged: patients with an SIR who nonetheless received a TNFi had lower area-level high school graduation rates, a proxy for socioeconomic status, suggesting that health literacy, cost, insurance coverage, or communication barriers may have shaped some treatment choices.

The clinical payoff was substantial. Patients whose therapy was aligned with the MSRC result were dramatically more likely to achieve low disease activity on the clinical disease activity index, with an adjusted odds ratio of 9.79, and on the patient-reported RAPID3, with an adjusted odds ratio of 11.55, compared with those whose treatment was misaligned. They also showed greater likelihood of achieving the minimum clinically important difference, low patient global assessment, and minimal pain, with adjusted odds ratios ranging from 6.49 to 9.74. In linear models, aligned patients had greater reductions in disease activity and in the pain and overall health components of the RAPID3 questionnaire, though not in physical function.

When the investigators compared MSRC-aligned patients with untested controls receiving standard care, the differences were narrower but still meaningful. Aligned patients were more likely to report low patient global assessment, with an adjusted odds ratio of 1.76, and minimal pain, with an adjusted odds ratio of 1.94, though no differences emerged on physician-reported measures. Among biologic-naive patients, however, the advantages widened considerably, with aligned patients showing greater likelihood of low disease activity, clinically important improvement, low patient global assessment, and minimal pain compared with controls. The authors suggest that patient-reported outcomes, which distinguish active treatment from placebo in randomized trials as well as or better than physician measures, may be particularly sensitive to the benefits of empiric rather than trial-and-error therapy selection, especially in a population known to report worse pain and function than non-Hispanic whites.

The study also examined the classifier’s diagnostic performance in the 31 tested patients who actually received a TNFi. Among those with an SIR, 80 percent failed to achieve low disease activity, yielding a positive predictive value of 80.0 percent, a sensitivity of 50 percent, a specificity of 86.7 percent, and an area under the curve of 0.68. These figures are comparable to those reported in white patients using a different response criterion, providing additional evidence that the test operates equally well across ethnic groups. Patients with an SIR who received a TNFi were significantly less likely to achieve low disease activity than those without an SIR, with an odds ratio of 0.15.

The findings carry important implications for how rheumatoid arthritis is treated in Hispanic communities and beyond. Because Hispanics remain underrepresented in clinical trials, there has long been concern about the generalizability of therapeutic evidence to this population, and the new study provides some of the first prospective data showing that a molecular classifier validated in non-Hispanic whites performs comparably in a predominantly Mexican-American cohort. The authors caution that the study was not randomized, that the small number of misaligned patients produced wide confidence intervals, and that disclosure of test results during shared decision-making may have introduced reporting bias into patient self-reported outcomes. Still, they conclude that broader use of the MSRC test may optimize outcomes and alter treatment paradigms, potentially sparing two-thirds of tested patients from a biologic class they are unlikely to benefit from and accelerating the path to effective therapy.

Subject of Research: Health

Article Title: Use of a molecular signal response classifier informs treatment selection and improves outcomes in Hispanic patients with rheumatoid arthritis

Article References: Karpouzas, G. A., Rodriguez, M., Bui, V. L., Strand, V., & Ormseth, S. R. (2026). Use of a molecular signal response classifier informs treatment selection and improves outcomes in Hispanic patients with rheumatoid arthritis. Frontiers in Musculoskeletal Disorders, 3, Article 1728996. https://doi.org/10.3389/fmscd.2025.1728996

Image Credits: AI Generated

DOI: 10.3389/fmscd.2025.1728996

Keywords: rheumatoid arthritis, personalized medicine, biomarkers, TNF inhibitors, gene expression, health disparities, molecular, signal, response, classifier, informs, treatment