Statistical modeling of extracellular vesicle cargo to predict clinical trial outcomes for hypoplastic left heart syndrome

Abstract

Summary: Cardiac-derived c-kit+ progenitor cells (CPCs) are under investigation in the CHILD phase I clinical trial (NCT03406884) for the treatment of hypoplastic left heart syndrome (HLHS). The therapeutic efficacy of CPCs can be attributed to the release of extracellular vesicles (EVs). To understand sources of cell therapy variability we took a machine learning approach: combining bulk CPC-derived EV (CPC-EV) RNA sequencing and cardiac-relevant in vitro experiments to build a predictive model. We isolated CPCs from cardiac biopsies of patients with congenital heart disease (n = 29) and the lead-in patients with HLHS in the CHILD trial (n = 5). We sequenced CPC-EVs, and measured EV inflammatory, fibrotic, angiogeneic, and migratory responses. Overall, CPC-EV RNAs involved in pro-reparative outcomes had a significant fit to cardiac development and signaling pathways. Using a model trained on previously collected CPC-EVs, we predicted in vitro outcomes for the CHILD clinical samples. Finally, CPC-EV angiogenic performance correlated to clinical improvements in right ventricle performance.

Keywords

• Medicine
• Drug delivery system
• Statistical physics