Genetic and Epigenetic Predispositions, Shared Mechanisms, and Common Biomarkers Between Cancer and CVD—Machine Learning-Based Insights

Abstract

People affected by cancer are at an elevated risk of cardiovascular disease (CVD)-associated morbidity and mortality; the reverse association is also substantially evident. This bidirectional association can be characterized by their genetic susceptibility, common modifiable and non-modifiable risk factors, shared molecular mechanisms, and diagnostic biomarkers. The root of this intricate relationship between these two disempathetic phenotypes is not clearly understood yet, though their mechanistic correlations and underlying shared pathways are consistently investigated. Along with bioinformatics methods, machine learning capabilities are evolving and may enhance the understanding of this complex relationship through genetic variant and epigenetic modification detection, biological mechanism elucidation, common biomarker discovery, and predictive model development for cancer-related morbidity and mortality in CVD patients and CVD-related events in cancer patients or survivors. In this review, we critically analyze the advancements, contributions, potential, and challenges of machine learning applications. As a relatively new approach, we identify the necessity of optimized computational methods for important feature selection, efficient data imputation algorithms for handling missing data, effective data harmonization pipelines for multi-modal data, computational frameworks for the simultaneous cancer-CVD risk stratification, and explainable algorithms for comprehensive clinical translation. Finally, we recommend machine learning-based causal inference analysis of their interconnections in order to understand their mutual causal effects.