Biomarker-Based Prediction of Tyrosine Kinase Inhibitor Cardiotoxicity using Human iPSC-Derived Cardiomyocytes [SPS 2019]
Presented by Jessica Palmer at the Safety Pharmacology Society Meeting in Barcelona, Spain (September, 2019)
- Tyrosine kinase inhibitors (TKI) have greatly improved the treatment and prognosis for a wide range of cancers. Unfortunately, numerous TKIs produce cardiotoxic effects, which were not well predicted during preclinical studies.
- We developed an in vitro assay, Cardio quickPredict, for predicting cardiovascular liability based on changes in human iPSC-derived cardiomyocytes (iPSC-CM) metabolism and cell viability, which identifies both functional and structural cardiotoxicants. The assay’s prediction model (PM) is based on the response of four metabolites: lactic acid (LAC), arachidonic acid (ARA), thymidine (THY), and 2’-deoxycytidine (2DC)) and predicts the concentration at which a drug exhibits cardiotoxicity potential. The PM classified 81 drugs with known cardiotoxicity outcomes (54 cardiotoxic, 29 noncardiotoxic) with 86% balanced accuracy, 83% sensitivity, and 90% specificity.
- The current study evaluated the utility of this assay for evaluating the cardiotoxicity potential of TKIs. We tested 10 TKIs that induce a variety of cardiotoxic effects, including eight drugs clinically associated with cardiotoxicity (crizotinib, dasatinib, imatinib, lapatinib, nilotinib, sorafenib, sunitinib, and vandetanib) and two drugs considered to be relatively cardiac-safe (axitinib and erlotinib) to compare changes in metabolism of the PM ratios. Human iPSC-CMs were exposed to eight concentrations of each drug for 72 hours and cell viability and metabolites in the spent media were analyzed.
- Every drug altered at least one metabolite independent of a change in cell viability. Crizotinib, imatinib, sorafenib, sunitinib, and vandetanib elicited a response in all four metabolites indicative of cardiotoxicity; however, a difference was observed in which metabolite was impacted at the lowest concentration. For example, crizotinib altered LAC at significantly lower concentrations (?5-fold) than where a response was observed in ARA and THY. In contrast, sorafenib elicited a response in ARA prior to LAC, THY, and 2DC.