Abstract
Pain is a major clinical outcome in chronic inflammatory disease, yet molecular correlates of within-person pain change remain unclear and must be distinguished from inflammatory change. We used longitudinal whole-blood multi-omics data from the RA-MAP/TACERA early rheumatoid arthritis cohort to prioritize microRNA (miRNA) candidates whose 6-month changes co-varied with change in visual analogue scale pain (ΔVAS). Among 151 participants and 1,881 miRNA features, we applied a stability-oriented delta-omics workflow combining subsampling elastic-net selection frequency, bootstrap directional filtering, and post-selection rank-residual partial correlation adjusted for ΔCRP and ΔESR. This identified 29 directionally stable miRNA candidates; after ΔCRP/ΔESR adjustment, 16 retained sign-consistent support at post-selection BH-FDR < 0.05 within these 29 candidates. Matched miRNA–mRNA delta analysis found no individual miRNA–mRNA pair surviving pair-level BH-FDR correction. At the primary nominal pair-selection threshold of p < 0.05, 123 inverse pairs involving 117 target genes yielded exploratory Gene Ontology Biological Process enrichment involving synaptic and neuronal organization; however, no GO term remained BH-significant when the pair-selection threshold was tightened to p < 0.01. The exploratory plasma chronic neck pain comparison yielded 22 of 39 sign-consistent assay matches (56.4%; exact 95% CI, 39.6%–72.2%), which was not statistically distinguishable from the 50% chance expectation, and no match survived BH-FDR correction. These findings outline a stability-oriented, inflammation-adjusted approach for prioritizing blood miRNA candidates linked to pain change. The 16-miRNA subset should be interpreted as a hypothesis-generating candidate pool rather than a validated biomarker panel and provides a focused basis for future longitudinal validation and functional studies.