AI-Driven Prediction of Cardio-Oncology Biomarkers Through Protein Corona Analysis.

Protein corona, a layer predominantly composed of proteins and other biomolecules that forms on nanoparticle surfaces upon interaction with biological fluids, has recently been extensively utilized to enhance the depth of plasma proteomics and biomarker discovery. In this study, we integrate protein corona profiling with mass spectrometry (MS)-based bottom-up proteomics (BUP), machine learning, and causality analysis to identify potential biomarkers in the field of cardio-oncology. We selected prostate cancer (PC) and atherosclerosis as model cardio-oncology diseases, given that PC is the most prevalent cancer among men in the United States and frequently coexists with atherosclerotic cardiovascular disease (ASCVD), which contributes to the progression of metastatic PC (mPC). Protein corona profiles were generated from 35 plasma samples categorized into four groups: mPC with ASCVD, nonmetastatic PC (nmPC) with ASCVD, mPC without ASCVD, and nmPC without ASCVD. MS-based BUP analysis identified 887 unique proteins within the protein corona. Gene Ontology (GO) analysis of the 260 proteins common to all samples revealed key plasma proteomic pathways significantly associated with ASCVD and mPC. Using Least Absolute Shrinkage and Selection Operator (LASSO) regularization, we isolated 22 proteins strongly associated with ASCVD or mPC, including chaperonin containing TCP1 subunit 7 (CCT7), which was common to both conditions. Automated formal reasoning and causality analysis of these 22 proteins identified thromboxane-A synthase 1 (TBXAS1) as a primary causal factor linked to both ASCVD and mPC. TBXAS1 plays a critical role in promoting platelet aggregation, vascular smooth muscle cell proliferation, endothelial dysfunction, and thrombosis. In this proof-of-concept study, CCT7 and TBXAS1 emerged as potential biomarkers for both ASCVD and mPC, suggesting their utility as dual biomarkers for early detection and targeted therapeutic interventions. By combining nanomedicine with advanced analytical methods, our integrated approach provides a robust framework for uncovering causal relationships between biomarkers and disease states.