T vector velocity distribution: A new biomarker for identifying drug effects on cardiac myocyte ion channels

Drug effects on ion channels affecting the cardiac ventricle's repolarization are well understood on the cellular level. On the ECG level, QT interval prolongation is a biomarker for a drug's potential for proarrhythmic risk, but the QT interval alone doesn't allow further differentiation of a drug's effects on various ion channel types. Besides other ECG biomarkers, the J-Tpeak interval has been reported as the best biomarker for differentiating QTc prolonging drugs with predominant potassium channel block from drugs with multichannel block. We present a new class of ECG biomarkers that are based on the time required to reach specific quantiles of the cumulative distribution of the spatial heart vector velocity in the T loop, and we assess their performance using the published ECGs from two studies funded by the FDA's Critical Path Initiative. We demonstrate that double-delta values of the 40% quantile of the T loop vector velocity distribution separate pure potassium channel blocking drugs from multichannel blocking drugs with an area under the ROC curve (AUC) value of 0.90, which is significantly larger than the published performance based on JTpeak (0.83) using the same ECG data. The performance could be further improved by combining the 20%, 40%, 60%, 80% and 100% quantiles in a logistic regression model, resulting in an AUC value of 0.93. We further present a new effect profile concept that allows qualitative and quantitative comparison of drug effects on the repolarization process. We demonstrate that the effect profile based on the T vector velocity quantiles indicates increasingly delayed electrical activity over most of the repolarization process for pure potassium efflux ion channel blocking drugs. For drugs solely blocking the early phase influx ion channels (mexiletine, lidocaine), the effect profiles indicate accelerated electrical activity in the initial repolarization phase. For drugs (ranolazine) or drug combinations blocking multiple ion channels (dofetilide + mexiletine, dofetilide + lidocaine), the overall effect profiles indicate a superposition of the individual effect profiles. In conclusion, the proposed T vector velocity based parameters greatly enhance characterization of multichannel drug effects on cardiac repolarization and may help in bridging the gap between ion channel effects and the overall cardiac electrophysiology reflected in the body surface ECG.