Comment: Cardiotoxic side effects of anti-cancer drugs have gained much attention recently. New therapies effectively treat various cancers and patients often survive for many years; thus, the significance of potentially irreversible adverse effects of anti-cancer therapies is substantial1. This is also the case with novel, more molecularly targeted therapies.
Protein kinase inhibitors (KIs) are anti-cancer drugs that are rapidly expanding with more than 40 approved KIs in clinical use and several hundred in development2. For many KIs cardiovascular adverse effects such as hypertension, thromboembolic events, QT prolongation, or heart failure have been described1.
In the current study, the authors approached cardiotoxicity testing by combining known adverse event profiles derived from the FDA adverse event reporting (FAER) database with transcriptomic signatures from human cardiomyocyte cell lines treated with 26 FDA-approved KIs. Via biostatistical computation, they develop and validate models with good predictive performance. I.e. early gene expression patterns in treated cardiomyocytes may predict observed adverse events derived from clinical databases. They further assess the association of molecular structures (e.g. similarity) and known KI target profiles (promiscuity of KI binding) with predicted cardiotoxicity risk and thereby assess the importance of chemical structure and selectivity.
The authors' approach is innovative and integrates unbiased transcriptomic profiling in human cardiomyocytes with observed adverse events in a large ‘real life’ cohort. Good performance of risk prediction may help to prioritise drug candidates in development based on predicted risk. Further, the approach is a step toward more personalised treatments as one may conceive using patient-derived cells (such as IPSc derived cardiomyocytes) for risk prediction in the future3.
However, there are some caveats. First, the authors use dedifferentiated cardiomyocytes from heart healthy donors. While we know that there are cardiomyocyte-intrinsic effects of KIs4, adverse events may also be due to non-myocyte effects, such as vascular effects like hypertension or thromboembolic events1, the interaction of different cell types5, or electrophysiological alterations1. With certain KIs that approach may be problematic. For example, ponatinib that has high relative observed odd ratios and predicted risk scores often exhibits predominantly vascular-mediated cardiac adverse events in clinical cohorts and less primary heart failure6.
The clinical cardiac composite outcome that integrates myocyte functional, electrophysiological, and vascular effects on potentially pre-diseased hearts with tuned up signalling pathways may thus differ.
Still, the current study advances the field with novel and sophisticated approaches towards more tailored drug development and personalised treatments.