Virtual patients inspired by multiomics predict the efficacy of an anti-IFNα mAb in cutaneous lupus

Lupus Erythematosus is a heterogeneous autoimmune disease that requires treatments tailored to specific patient subsets. To evaluate in silico the efficacy of the anti-IFNα S95021 monoclonal antibody, we created a Quantitative Systems Pharmacology model of cutaneous lupus and a virtual patient population, with attributes matching the diversity of actual patients. To this aim, we performed a multiomics profiling analysis of 337 lupus patients from the PRECISESADs cohort, thereby identifying four patient clusters with distinct immune dysregulation patterns, including various levels of type I IFN pathway upregulation. Simulation of S95021 treatment in the virtual patient cohort (n=241) predicted distinct clinical responses in patient clusters, with machine learning analysis further revealing biomarkers that distinguish predicted responders from non-responders. Combining multiomics profiling of actual patients with mechanistic mathematical modeling supports precision medicine by predicting drug responses based upon patient characteristics in a complex heterogeneous disease. [Display omitted] •QSP modeling of lupus can be used to predict the efficacy of drug candidates.•A cohort of virtual lupus patients was created from profiling data of actual patients.•Virtual patient simulations predicted distinct anti IFN treatment responses.•Machine learning found biomarkers to differentiate responders from non-responders.