Systematic review and network meta-analysis of interventions for articular involvement in patients with systemic lupus erythematosus.

Several immunosuppressive therapies beyond corticosteroids have demonstrated efficacy in controlling symptoms and reducing inflammation in lupus-related arthritis. The current standard of care (SOC) remains the combination of antimalarials and low-dose steroids, with the addition of other immunosuppressants in refractory cases. This study aimed to compare the efficacy of available biological and non-biological immunosuppressive therapies in controlling articular activity in patients with systemic lupus erythematosus (SLE). A systematic review and network meta-analysis were conducted following the PRISMA-NMA checklist (PROSPERO registration: CRD42024562392). Randomized controlled trials up to December 2023 evaluating biological and non-biological immunosuppressive treatments in lupus with articular involvement were included. Searches were performed in MEDLINE, EMBASE, LILACS, and SCIELO, including gray literature. Two reviewers independently selected studies, extracted data, and assessed risk of bias (RoB2). A fixed-effects frequentist model was applied using netmeta, estimating relative risks (RR), heterogeneity (I2, τ2), and efficacy ranking using P-scores. A total of 1635 records were identified, and five randomized controlled trials met inclusion criteria, comprising 1476 patients, 93.3% of whom were women, with a mean age of 40 years. Evaluated interventions included methotrexate, anifrolumab, and baricitinib, all compared to placebo. Three studies showed low risk of bias, while two presented some concerns in specific domains. The network meta-analysis revealed no significant heterogeneity (Q = 2.44; P = .29) or inconsistency between designs, though global inconsistency was detected by netsplitting analysis. No statistically significant differences in efficacy were observed among the evaluated treatments. In the exploratory ranking analysis, P-scores indicated that methotrexate (P-score = 1.0), baricitinib (P-score = .62), and anifrolumab (P-score = .37) were numerically positioned at the top of the ranking. Nevertheless, these estimates represent probabilistic rather than inferential measures and should therefore be interpreted with caution. Meta-regression did not identify a significant influence of risk of bias on the results (P = .42), although substantial residual heterogeneity was observed (I2 = 98.8%), which limits the comparative interpretation of treatment efficacy.