Preclinical assessment of drug-induced cardiotoxicity is still a challenge for pharmaceutical industries, which require new tools for its accurate and early identification. In-silico trials using multiscale biophysically-detailed human-based models have proven to be powerful as a new tool for pro-arrhythmic risk prediction. Our goal is to evaluate the consistency between in-silico simulations using a novel human Purkinje model with experimental data obtained in rabbit Purkinje fibres, a preparation often used for preclinical drug testing.
Action potential (AP) recordings were acquired from isolated cardiac Purkinje fibres (n=3) from rabbit hearts (N=3). The effects of 4 drugs with varied ion block profiles (Diltiazem, Dofetilide, Risperidone, and Verapamil) were evaluated at 4 concentrations, through 8 AP biomarkers, at 1 and 0.2 Hz. Simulations were conducted using a recently published computational model for human Purkinje electrophysiology and the simple pore-block model to reproduce drug-induced effects on the ionic channels, using IC50s and Hill coefficients. Simulated APs quantitatively reproduced the drug-induced effects experimentally observed for both Dofetilide and Risperidone at 1 Hz. Simulations of Diltiazem and Verapamil correctly reproduced drug-induced effects at the early repolarisation, though, they showed AP prolongation instead of shortening during the late repolarisation. The interplay between Ca2+ and K+ channels inhibition was identified as the leading cause of the discrepancy. In order to reproduce the experimental AP response, additional simulations were performed for Diltiazem and Verapamil. The experimentally observed AP shortening was quantitively matched, only when not considering IC50s for K+ currents. At 0.2 Hz, Dofetilide and Risperidone developed EADs, in line with their known cardiotoxicity risk, whereas safe drugs as Diltiazem and Verapamil did not.
In-silico drug trials are a promising tool for drug safety and efficacy, and caution needs to be taken to the IC50s values that describe the drug behaviour, since they critically affect the in-silico model response.