Serum protein profiles in lupus nephritis associated with initial-onset systemic lupus erythematosus: Characterization through PEA immunoassay and preliminary development of predictive model.

[OBJECTIVE] Given the constraints inherent in current biomarkers and renal biopsy techniques for lupus nephritis (LN), this exploratory investigation was conducted to identify novel serum protein expression profiles through proximity extension immunoassay (PEA, Olink) in patients experiencing initial-onset LN for early identification of LN.

[METHODS] The PEA immunoassay was utilized to quantify serum concentrations of ninety-two inflammation-associated proteins among individuals with initial-onset LN (diagnosed concurrently with systemic lupus erythematosus (SLE), n = 23), patients with initial-onset non-LN manifestations (limited to cutaneous and musculoskeletal symptoms, specifically lupus arthritis (LA), n = 21), and age-matched healthy controls (HC, n = 22). Simultaneously, baseline clinical data were collected from the study population. Subsequently, machine-learning-based statistical modeling was employed to uncover prospective biomarkers indicative of initial-onset LN.

[RESULTS] A sum of 36 serum proteins was found to be differentially expressed in initial-onset LN relative to HC, comprising 25 upregulated and 11 downregulated proteins. Notably, transforming growth factor beta-1 proprotein, C-X-C motif chemokine 5, interleukin-15 receptor subunit alpha (IL-15RA), and C-X-C motif chemokine 11 were included in the final predictive model, achieving a sensitivity of 91.67 % and a specificity of 95.24 %. Compared with LA patients, 12 proteins were differentially regulated in the LN group, including 11 downregulated and 1 upregulated protein. Among these, Oncostatin-M, CC motif chemokine 28, Fibroblast growth factor 19, IL-15RA, leukemia inhibitory factor, neurotrophin-3, and TNF-like weak inducer of apoptosis were incorporated into a model yielding a sensitivity of 74.07 % and specificity of 82.35 %.

[CONCLUSION] The application of highly sensitive PEA profiling facilitated the identification of distinct serum protein signatures across SLE, initial-onset LN, and LA, uncovering multiple novel protein candidates. These promising biomarkers, whose precise role and predictive potential within the context of SLE merit further validation.