Background Rheumatoid arthritis (RA) is a significant global health issue. Early diagnosis remains clinically challenging, particularly in patients with inflammatory arthritis who do not yet fulfill 2010 ACR/EULAR classification criteria (undifferentiated arthritis, UA). Methods This retrospective study enrolled 83 treatment-experienced RA (TE-RA), 49 treatment-naïve RA (TN-RA), and 51 seropositive UA patients, as well as 60 healthy controls. Peripheral lymphocyte subsets and serum cytokines were profiled using flow cytometry and bead arrays. Least absolute shrinkage and selection operator (LASSO) regression was utilized to develop an immune-based derivation-stage classifier for distinguishing TN-RA from UA, with internal validation by bootstrap resampling (B = 1000). Results Immunophenotypic analysis identified distinct immune profiles between TN-RA and UA at initial presentation. TN-RA was characterized by IL-2 signaling exhaustion (elevated sIL-2R, decreased IL-2), systemic inflammation (elevated IL-6, IFN-γ, and TNF-α), and compensatory memory Treg expansion (increased CD45RO+ Tregs). In contrast, UA exhibited a Th17/Treg imbalance with relatively preserved Th2 and Th17 responses. An eight-feature immune signature (sIL-2R, IL-6, CD45RO+ Tregs, CD45RO+ Treg%, IFN-γ, Th17/Treg ratio, Th2, Th17%) discriminated TN-RA from UA with an optimism-corrected AUC of 0.959 (95% CI: 0.923–0.995), adjusted for age, sex, BMI, and disease duration. sIL-2R and IL-6 were the strongest contributors, consistent with their central roles in RA pathophysiology. Conclusions In the present study, an immune-based derivation-stage classifier showed potential for distinguishing TN-RA from UA at initial presentation. The IL-2-Treg axis perturbation represents a potential pathophysiological distinction, with sIL-2R as a candidate biomarker. These findings suggest that objective immune profiling may inform clinical decision-making when conventional criteria are inconclusive.
Zi et al. (Tue,) studied this question.