ObjectiveAlthough pain affects the assessment of disease activity in patients with rheumatoid arthritis (RA), pain is not always directly related to peripheral joint inflammation. Peripheral and central nervous system regulatory mechanisms also affect pain perception. We used regression tree methodology to identify mechanisms most predictive of disease activity after disease-modifying antirheumatic drug (DMARD) treatment.MethodsDisease activity was evaluated using the Disease Activity Score in 28 joints (DAS28) in 176 patients with RA, before and after starting a DMARD. Quantitative sensory testing (QST), including pressure pain thresholds (PPTs), temporal summation, and conditioned pain modulation (CPM), were used to assess pain mechanisms. Regression tree methodology was used to determine the QST modalities most predictive of DAS28 after DMARD treatment.ResultsThis analysis identified 4 groups defined by baseline DAS28 category and either knee PPT (a combined measure of peripheral and central nervous system dysregulation) or CPM (a measure of descending pain inhibition). Among patients starting with low/moderate disease activity, lower knee PPT (PPT ≤ 4.65 kgf) most strongly predicted higher posttreatment disease activity (group 1 mean DAS28 2.8 [SD 1.0] vs group 2 mean DAS28 3.5 [SD 1.0]). Among patients starting with high baseline disease activity, less efficient descending pain modulation (CPM ≤ 1.55) most strongly predicted higher posttreatment disease activity (group 3 mean DAS28 3.4 [SD 1.4] vs group 4 mean DAS28 4.6 [SD 1.1]).ConclusionThese results highlight the importance of identifying and treating aberrant peripheral and central pain regulation in patients with RA starting or switching DMARD therapy.