Document resource
Objectives To model SLE clinical manifestations using pathway-level patterns derived from plasma proteomics, with the aim of identifying molecular mechanisms that explain clinical outcomes and autoantibody profiles.Methods We analysed plasma proteomic profiles from 100 patients with SLE, recruited cross-sectionally at Karolinska University Hospital, who fulfilled the revised 1982 ACR classification criteria. The patients were selected to represent the four autoantibody-defined SLE subgroups previously described (PMID:34658170). Proteins (n = 2,141) were quantified using liquid chromatography–mass spectrometry (LC-MS/MS), and pathway-level activity scores were computed using Gene Set Variation Analysis (GSVA), which summarizes protein expression into biologically meaningful pathways.We modelled 29 binary clinical manifestations and 17 autoantibody profiles separately using elastic net logistic regression, allowing the selection of pathways that best explain each clinical outcome. Model performance was evaluated using five-fold cross-validation to ensure robustness.To validate our findings, we applied the same pathway definitions to other 99 patients with SLE from the same cohort, profiled using the Olink Explore cardiometabolic panel (n=1,436 proteins), 19 of which were also present in the MS dataset. Predictive accuracy was assessed using receiver operating characteristic (ROC) curves and the area under the curve (AUC), which reflect how well the selected pathways distinguish between patients with and without each feature.Results Ten pathway-based models achieved moderate to good discrimination (AUC 0.65–0.80) for predicting antibody positivity or clinical manifestations. The best performance was observed for combined Sm/RNP antibody positivity (AUC = 0.799). Pathways associated with Sm, RNP68, RNPA, and anti-nucleosome positivity consistently included upregulation of interferon alpha, interferon gamma and cholesterol homeostasis.For clinical manifestations, serositis was best defined by two pathways, lupus nephritis by 19 pathways, ischemic cerebrovascular disease (ICVD) by seven, and lymphopenia by two (figure 1A–C).Abstract PO:08:218 Figure 1Conclusions Pathway-level proteomic data effectively captures molecular patterns associated with clinical manifestations in SLE. This approach provides a framework for molecular characterisation and mechanistic interpretation of disease manifestations, opening possibilities for tailored drug repurposing strategies to target the dysregulated pathways in SLE.