Abstract 15110: Atherospectrum Identified Novel Foaming Programs and Monocyte Predictor Genes for Residual Cardiovascular Risk

Abstract

Despite lipid-control therapies, substantial risk of major cardiovascular event remains. New approaches are needed to define and mitigate residual risks in a patient specific manner. Monocytes, critical player in CVD, transcriptionally respond to metabolic and inflammatory cues in the circulation and contribute to atherogenesis. However, previous studies of monocytes only found moderate associations to CVD events and with limited predictive power. Here we developed a novel computational approach, named AtheroSpectrum, which effectively depicted heterogeneity of foam cells. Our analysis revealed two distinct programs of plaque macrophages: the homeostatic-foaming and the inflammatory pathogenic-foaming, which we found to be associated with severity of atherosclerosis in multiple studies. Next, we screened the genes enriched in the pathogenic-foaming program using our original self-optimizing progressive machine learning algorithms, and identified 29 signature genes (including SOD2, SLC39A8, etc) that were strongly associated with cardiovascular events in 936 participants from the Multi-Ethnic Study of Atherosclerosis (MESA) whose LDL levels were below 129 mg/dl, and were further validated in additional human data (gtexportal.org, GSE43292), while traditional risk factors (BMI, lipid profiles, smoking, etc) showed no significant association. We further developed a novel residual CVD risk score system that incorporated the newly identified 29 pathogenic foaming signature genes, gender, age, and the inflammation index (MPI) of AtheroSpectrum. Our scores presented strong prediction power for depicting residual cardiovascular risks (AUC 0.805) compared with the JAMA 2001 (AUC 0.620) and Framingham risk scores (AUC 0.650).In summary, we successfully identified pathogenic foam cell subpopulations with active inflammatory and lipid metabolic programs. Characterization of these subsets revealed novel feature-based key genes directly contributing to CVD residual risks and endowed creation of a CVD prediction score with high accuracy. Further optimizing this score system will facilitate both mechanistic investigations and development of therapeutic and prognosis strategies of CVD residual risk.