A mechanistic quantitative systems pharmacology framework enriched by multiscale data enables model-informed...

Lupus nephritis (LN) is a complex autoimmune disease and mesenchymal stem cells (MSCs), given their unique immune-modulatory potential, have emerged as a promising route for treating LN. Understanding the multiscale dynamic LN-MSC interactions and selecting the optimal clinical design for MSCs is critical to enhance their therapeutic potential and achieve meaningful benefit-risk differentiation in patients. To guide MSC clinical development, we constructed the first mechanistic quantitative systems pharmacology model of LN pathophysiology that integrated multimodal preclinical-clinical data, with detailed illustration of standard-of-care immunosuppressant pharmacology and physiologically-based biodistribution and therapeutic mechanisms of MSCs. We created model-based virtual LN patients that well captured clinical interpatient heterogeneity and were validated by numerous sets of clinical efficacy readouts, including first-hand results from our investigator-initiated trial testing MSCs in LN patients. We used this computational framework and identified that a spaced two-infusion MSC dosing strategy would yield optimal clinical response in LN patients, which directly informed the phase 2 design of a novel MSC therapy. Modeling analyses further identified that CXCL10 in combination with proteinuria are potential predictive biomarkers that could boost clinical renal response to MSCs. These results demonstrated the exceptional importance of our QSP framework in driving MSC clinical development for LN patients.