Validation of a guinea pig Langendorff heart model for assessing potential cardiovascular liability of drug candidates

Abstract

Introduction: Cardiac liabilities represent a major cause of recent withdrawal of marketed drugs and also for the high attrition rate evidenced during late stage drug development. To identify molecules with potential cardiovascular risks early in drug development, a screening model of ex vivo Langendorff hearts has been validated with 26 reference compounds of various chemical and therapeutic classes.

Methods: The hearts of adult guinea pigs were maintained by retrograde perfusion in Langendorff mode, beating spontaneously at sinus rhythm or paced via the right atrium at 200 and 300 beats per minute. Multiple parameters consisting of hemodynamic function (coronary and left ventricle pressure), cardiac electrophysiology (electrocardiogram and monophasic action potential) and indices of arrhythmogenesis (triangulation, reverse-use dependence, repolarization dispersion and beat-to-beat instability), together with overt arrhythmia were evaluated simultaneously. Ascending concentrations up to either 100-fold of the determined hERG IC(50) or nominally 100 microM were routinely tested utilizing 4-6 hearts per compound.

Results: Each compound exhibited a unique cardiovascular profile: (i) the majority displayed concentration and heart rate-dependent mixed-ion channel or multiple-target effects that frequently resulted in bradycardia, atrioventricular block, negative inotropy, coronary vasodilatation, and QRS widening. (ii) Compounds associated with high arrhythmogenic risk in the clinic exhibited more "positive signals" at concentrations within 30-fold of their maximal therapeutic free plasma concentration than those with less arrhythmogenic potential. (iii) For several potent torsadogens, proarrhythmic indices other than the prolongation of QT/QTc and MAP duration appeared more sensitive in representing proarrhythmic liability. (iv) A scoring system was developed to assist in the rank-ordering of potential cardiotoxicants.

Discussion: The cardiovascular action of reference compounds profiled by this isolated heart model was generally consistent with their known mechanisms and, except for the sinus heart rate, correlated well with that observed in the clinic. Further, the overall cardiac liability estimated by the scoring system matched the clinical documentation, suggesting this model could serve as a valuable tool in early cardiovascular drug safety assessment.