Plasma proteomics for risk prediction and identification of novel drug targets in systemic lupus erythematosus.

Objectives: The relationship between proteomic profiles and incident systemic lupus erythematosus (SLE) remains unclear. We aimed to identify candidate plasma proteins for SLE risk in women, discover potential treatment targets for SLE, and develop and validate a protein-based prediction model for SLE risk.

Methods: 28 220 women from the UK Biobank were randomly split into training (70%) and testing (30%) sets.

Results: During a median follow-up of 13.3 years, 90 and 38 SLE events occurred in the training and testing set, with an average onset age of 63 years. In the training set, a protein risk score for new-onset SLE was constructed using 22 out of 2,911 proteins. Of the 22 proteins, ITGAV, TNFRSF4, CXCL10 and CD207 were causally associated with SLE, and TNFRSF4 and CXCL10 have been developed as drug targets for cancer and autoimmune diabetes. The 22 proteins enriched in cytokine-cytokine receptor interaction, inflammatory, and immune response. TNF, CD80, and LGALS9 were key proteins within the protein network. In the testing set, the SLE protein risk score has a better predictive capability for new-onset SLE (C-index, 0.801; 95%CI, 0.715, 0.887) than clinical risk factors of SLE (C-index, 0.706; 95%CI, 0.612, 0.800) and polygenic risk score for SLE (C-index, 0.638; 95%CI, 0.555, 0.721). Adding SLE protein risk score to clinical risk factors of SLE (C-index from 0.706-0.827) improved the predictive performance.

Conclusion: The SLE protein risk score had a good predictive performance for new-onset SLE in women. ITGAV, TNFRSF4, CXCL10 and CD207 were potential treatment targets for SLE.