Categorizing TdP risk with qNet: Effects of protocol design on HERG models and TMS outcomes

The Comprehensive in vitro Proarrhythmia Assay (CiPA) requires assessing drug effects on multiple ionic currents to measure pro-arrhythmic risk and a detailed model of drug binding to HERG channels. Manual patch clamp recordings of hERG current are performed using the Milnes protocol. The CiPA procedure has strong predictive ability, but extracting HERG model parameters is challenging at the experimental and analytical levels. We examined a two-voltage pulse (2P) protocol to categorize TdP risk using the qNet metric and Torsade Metric Score (TMS) in simulated and experimental data using improved perfusion techniques. We also explored whether these shorter and simpler experimental protocols could produce qNet/TMS results similar to those of the Milnes protocol. We tested the simulated predictions for dofetilide block of hERG current expressed in HEK cells (room and physiological temperatures) using the FDA-CiPAORdv1.0 drug-binding model. Our CYBERQ-qNET analysis software calculated the qNet and TMS values and compared the TdP risk category to those from the Milnes protocol. There were no significant differences in the qNET, TMS values, and TdP risk categories when using simulated/experimental data from either the two voltage pulses or the Milnes protocol. Using simulated data, the TMS/TdP risk values for the 2P protocol (using fractional block across the inactivation pulse) vs. Milnes protocol: dofetilide (0.0473/2 vs. 0.0524/2), bepridil (0.0453/2 vs. 0.0452/2), terfenadine (0.0635/1 vs. 0.0605/1) and diltiazem (0.0888/0 vs. 0.092/0), respectively. The experimental dofetilide data resulted in correct TdP categorization across the activation and inactivation pulses at room temperature (0.0481/2 and 0.0305/2) and physiological temperature (0.0309/2 and − 0.0189/2). However, the physiological temperature data displayed significantly more block than the room temperature and published Milnes data, which were comparable. Combining both pulses produced a miscategorization for the room temperature data (0.0625/1) but not the physiological temperature data (0.0440/2). Simulated data had consistent TMS scores. In contrast, experimental data showed more variable TMS scores depending on the pulse analyzed. These results suggest that model development can be coupled to simpler and less demanding experimental protocols for assessing arrhythmogenic potential using CiPA.