Identifying Biomarkers of Cardiotoxicity with a Genome-Scale Model of Metabolism Constrained with 'Omics Data

With recent improvements in the detection and treatment of cancer, the adverse side effects of chemotherapeutics, particularly cardiotoxicity, have become more apparent. Many chemotherapeutics are now associated with adverse cardiovascular events, however, there are no clinical measures to detect, limit, or prevent cardiotoxicity. Although research work has demonstrated that metabolites are good biomarkers of early cardiotoxicity, further work is needed. Here, we utilize genome-scale metabolic network reconstructions (GENREs) to identify new biomarkers of cardiotoxicity. First, we built a heart-specific GENRE and used the model with a novel approach, the Tasks Inferred from Differential Expression (TIDEs) approach, to identify shifts in metabolic functions in heart failure. Next, we collected paired transcriptomics and metabolomics data in primary rat neonatal cardiomyocytes exposed to three compounds (5-fluoruracil, acetaminophen, and doxorubicin) to characterize in vitro cardiotoxicity. Finally, we integrated our collected data with a model of heart metabolism to identify shifts in metabolic functions, unique metabolic reactions, and shifts in metabolic reactions that are unique to cardiotoxicity. For each compound, we identified unique shifts in metabolism, confirming mechanisms of toxicity for doxorubicin and proposing new hypotheses for mechanisms of toxicity for 5-fluorouracil and acetaminophen. Given that our experiments are done in rats, future work is needed to address translatability in humans. To this end, finally, we highlight the utility of data-driven and mechanistic modeling approaches in making cross-species comparisons.