Serum Soluble Mediator Signatures of Lupus Nephritis: Histologic Features and Response to Treatment

Objective: Lupus nephritis (LN) management remains challenging, and novel noninvasive biomarkers are needed. This study quantified serum soluble mediators in the Accelerating Medicines Partnership (AMP) LN cohort to identify biomarkers of histologic features and treatment response. Methods: Patients with systemic lupus erythematosus (SLE) (n = 268) undergoing clinically indicated kidney biopsies (urine protein/creatinine ratio [UPCR] ≥ 0.5) were recruited through the AMP Rheumatoid Arthritis and SLE Network. Serum was collected at biopsy and 3-, 6-, and 12-month postbiopsy, alongside samples from 22 healthy controls. Concentrations of 66 immune mediators were quantified using xMAP multiplex assays, and (TACE) measured by enzyme-linked immunosorbent assay. Seven mediators with >95% values below detection limits were excluded from analyses. Bootstrapped least absolute shrinkage and selection operator (LASSO) regression identified proliferative LN (class III/IV ± V) predictors from baseline mediators. Associations with 12-month treatment response (complete/partial vs no response) were tested using three-month changes in LASSO-selected mediators and UPCR via logistic regression. Molecular clustering of mediator profiles was performed to identify LN subgroups. Results: Proliferative patients with LN (class [III or IV] ± V; n = 160) displayed a distinct mediator profile compared with nonproliferative LN (class I, II, or V; n = 96). LASSO regression identified 20 mediators predictive of proliferative LN (areas under the curve, 0.82; 95% confidence interval [CI], 0.81–0.91), including elevated syndecan-1, tumor necrosis factor receptor type I, tumor necrosis factor receptor type II, and vascular cell adhesion molecule 1 (VCAM-1), as well as decreased CCL3//macrophage inflammatory protein 1α, CD40 ligand, and interleukin-5 levels. Among proliferative patients with LN, 3-month reductions in syndecan-1 and VCAM-1, mediators associated with intrarenal LN activity and/or chronicity, predicted the 12-month treatment response. A model incorporated these reductions and a decline in UPCR-predicted treatment response in proliferative LN (0.90; 95% CI, 0.82–0.98). Molecular clustering revealed four distinct LN subgroups with unique soluble mediator signatures and clinical features not captured by histology alone. Conclusion: Serum soluble mediators, particularly syndecan-1 and VCAM-1, reflect LN histologic activity, and early decreases predict treatment response, supporting their potential use as noninvasive longitudinal biomarkers. The substantial heterogeneity within LN highlights the potential for biomarker-guided reclassification to advance precision medicine approaches.