Abstract 4143250: Individualized Interactomes from Pulmonary Arterial Hypertension Biopsies Predict Therapeutic Response: an <i>In Silico</i> Clinical Study

Introduction: Pulmonary arterial hypertension (PAH) is a highly morbid cardiopulmonary disease characterized by substantial pathobiological and clinical heterogeneity, which is believed to underlie variable treatment-response observed in clinical trials and at point-of-care during real-world practice. Utilizing patient-specific pathogenetic information to inform drug selection for use in clinical practice is expected to improve clinical efficacy of PAH drugs, but such strategies are currently lacking. Methods: We conducted a retrospective analysis of 25 PAH patients (32 samples, receiving PAH therapies) undergoing right heart catheterization (RHC) for the routine clinical assessment of dyspnea at Rhode Island Hospital (Warren Alpert Medical School of Brown University) , along with 3 healthy controls. A minimally invasive "cell biopsy" approach was used to collect Pulmonary Arterial Endothelial Cells (PAECs) from the tips of pulmonary artery catheters. Transcriptomic RNA-seq was performed on mRNA isolated from cell passage 3 or 4 for all samples. We deployed an innovative network medicine method to construct sample-specific interactomes from the transcriptomic data and human protein-protein interactome. The individualized interactomes captured patient-specific molecular features, which could then be used to predict therapeutic response in PAH patients. Results: The individualized interactomes have a range of 5276 [4900, 7263] nodes and 7204 [6606, 12992] edges. We observed that an increase in 6MWD from baseline that was sustained at 12 months following treatment initiation was associated with higher fraction of drug targets in the individualized interactomes. By contrast, such a positive association was not observed in comparator drugs. Similarly, we observed a transition in BNP levels from elevated at baseline to within the normal range at 12 months that corresponded to an absolute decrease of ~250 pg/mL. Higher fraction of drug targets in the individualized interactomes was also associated with a significant improvement in REVEAL 2.0 Risk score at 6 months (p=0.004) and 12 months (p&lt; 0.001). Conclusions: "Cell biopsies" taken at point-of-care for clinically indicated RHCs may be leveraged to predict therapeutic response and inform precision-based initiatives in pulmonary vascular disease through transcriptome-derived sample-specific interactomes and systems pharmacology tools.