Arthritis Res Ther. 2025 Oct 27;27(1):197. doi: 10.1186/s13075-025-03660-2.
ABSTRACT
BACKGROUND: The aim of this study was to use clinical data to develop a prediction model for treatment response, comparing tofacitinib to methotrexate or etanercept in individual patients with Psoriatic Arthritis.
METHODS: Data from the development cohort (n = 80) of the TOFA-PREDICT trial were used. The cohort included PsA patients naïve to disease-modifying antirheumatic drugs (DMARDs) randomised to tofacitinib (n = 20) or methotrexate (n = 20), and patients failing conventional synthetic DMARD (csDMARD) treatment randomised to add-on tofacitinib (n = 20) or etanercept (n = 20). Treatment response was defined as achievement of minimal disease activity at week 16. Elastic net regression was used to select relevant baseline predictors in the complete cohort and in treatment subgroups. Using Ridge regression with different modelling strategies, three prediction models were developed and compared. Based on performance, a final model was selected.
RESULTS: Increased Health Assessment Questionnaire score, increased tender joint count, increased Leeds Enthesitis Index score, decreased VAS global assessment by physician and previous Tumour Necrosis Factor alpha inhibitor treatment were selected as predictors for non-response. The final cross-validated model had an AUC-ROC of 0.76 and predicted clinically relevant differences in response to the compared treatments. The predicted probability of response was higher for methotrexate compared to tofacitinib in 85% of DMARD naïve patients. The predicted probability of response was higher for etanercept compared to tofacitinib in all patients failing csDMARD treatment.
CONCLUSION: Our results support the use of baseline clinical data for prediction of response to different treatments. We intend to validate this prediction model and to assess the additional predictive value of imaging and multi-omics biomarkers in future analyses.
TRIAL REGISTRATION: EudraCT Trial registration number 2017-003900-28, registration date January 25th 2018.
PMID:41146314 | DOI:10.1186/s13075-025-03660-2
