Discussion
This set of 15 hERG blocking drugs reported as causing TdP includes 9 of the most potent hERG blockers of the Kramer list (Kramer et al., 2013). All these torsadogenic drugs induced hemodynamic effects activating in turn autonomic reflex mechanisms. These hemodynamic effects were different depending on drug and dose levels. In most of cases, they were more or less compensated by reflex mechanisms making them difficult to detect directly from blood pressure changes only. In parallel, combining QTc prolongation and HFQT oscillations analysis allowed detection of all tested torsadogenic drugs since visible from at least one of these two biomarkers. In the absence of QTc prolongation, enhancement of HFQT oscillations also constitutes a sensitive surrogate for suspicion of possible concealed QTc prolongation at low doses of torsadogenic hERG blockers. This paradigm is due to specific off-targets which conceal the hERG blockade induced QTc prolongation by activating the sympathetic system during HF cycles. This phenomenon likely involves the recruitment of the IKs repolarisation reserve by β-adrenoceptors mediated sympathetic activation (Volders et al., 2003). The α1 adrenoceptors blocking property is by far the most common off-target among torsadogenic drugs. This is not the only one since drugs like pimozide and cisapride also induced autonomic coactivation while they were not reported as exhibiting this α1 adrenoceptors blocking property. The milrinone profile shows that phosphodiesterase inhibition could also constitute an off-target leading to an autonomic coactivation and increase in HFQT oscillations. This example is important because it demonstrates that increase in HFQT oscillations do not require necessarily any mechanism lengthening the ventricular repolarisation since milrinone is devoid of hERG blocking properties and even causes QTc shortening. First of all, this work demonstrates that HFQT oscillations magnitude is a sensitive biomarker of the sympathetic activity on autonomic beat to beat variability of ventricular repolarisation. This biomarker is particularly useful when the parasympathetic system activity is predominant as with dofetilide at a low dose or in LQT patients with concealed QT prolongation. This dataset of hERG blockers also shows that this biomarker is more sensitive for detection of drug torsadogenic profile than short term QT variability that rather reflects instability of ventricular repolarisation in its principle.
Specificity for TdP risk assessment appears promising too. Indeed, increases in HFQT oscillations were found with torsadogenic hERG blockers only whereas non torsadogenic hERG blockers did not increase HFQT oscillations. However, two non arrhythmic drugs, prazosin and milrinone, were found sharing this ability to enhance this biomarker. Investigating available data from literature about sudden cardiac death risk reveals that milrinone (Packer et al, 1991), prazosin and other α1 adrenoceptors antagonists (O’ Connell et al., 2014) were reported to unexplainably worsen mortality in patients with heart failure. Despite its hERG blocking properties (Thomas et al., 2004), we failed to find clinical reports of Tdp induced by prazosin. Conversely, milrinone was recently cited as involved in TdP events in humans (Chiba et al., 2020) although devoid of hERG blocking properties (Yunomae et al., 2007) and considered as a non torsadogenic drug. Milrinone was also reported as causing cardiotoxicity resulting from adrenergic hyperactivity (Chiba et al., 2020). Their propensity to cause excess mortality in the context of heart failure is certainly not a coincidence. Indeed, ventricular repolarization is lengthened in heart failure due to down regulation of potassium current (Näbauer & Kääb, 1998) and enhancement of the late sodium current (Maltsev et al., 2007) creating as well favourable conditions for the triggering of ventricular arrhythmias by sympathetic activation as in LQT syndromes. These two examples of molecules open new perspectives for sudden cardiac death risk assessment out the area of TdP induced by arrhythmic drugs.
One of the most important finding in this work is the central role played by the stroke volume and cardiac output changes in reflex mechanisms leading to autonomic coactivation and increase in beat to beat ventricular repolarisation variability. This autonomic coactivation corresponds to an enhancement of oscillatory reflex mechanisms aimed to maintain stable the mean diastolic arterial pressure during HF parasympathetic cycles elicited by mild increases or decreases in stroke volume. Transient drops in diastolic arterial pressure following large RR pauses during HF cycles support involvement of the baroreflex in this autonomic coactivation phenomenon. However, mechanoreceptors sensitive to ventricular distension and volume could also be involved. They were shown to trigger large sympathetic burst following transient drop in arterial pressure in particular following premature ventricular contraction where the systolic pressure peak required to trigger arterial baroreceptors to terminate sympathetic outflow is absent (Zamir et al., 2012). Lowering in stroke volume induced by α1 adrenoceptors blocking drugs and other peripheral vasodilator drugs is considered as due to blood pooling in peripheral venous compartments below the diaphragm causing in turn a reduction in venous return and cardiac output (Brignole et al., 2018). Conversely, outcomes of dofetilide at a high dose under β-adrenoceptors blockade support that QT prolongation may result in a mild increase in stroke volume. Accordingly, positive inotropic effects correlated to their effects on ventricular repolarisation duration were reported for dofetilide and E4031 in ventricular muscle preparations (Tande et al., 1990; Abrahamsson et al., 1993).
This work establishes for the first time complex relationships between hemodynamic effects leading to sympathetic compensatory reflexes with arrhythmic electrophysiological mechanisms related to ventricular repolarisation prolongation. These relationships very well fit with the concept of the Coumel’s triangle and allow to dramatically refining and update (Figure 8) the model based on the Coumel’s triangle proposed earlier (Champéroux et al., 2015). According to this concept, TdP triggering involves 1/ a modulator: the autonomic nervous system: its contribution is featured by an enhancement of HF oscillations through a sympathetic coactivation in response to hemodynamic off-targets in case of vaso-active torsadogenic and/or due to ventricular repolarisation prolongation itself for drugs acting selectively on ventricular repolarization and in LQT syndromes, 2/ a substrate: the lengthening in ventricular repolarisation due to genetic mutations in LQT syndromes or related to intrinsic electrophysiological (off-)targets in case of torsadogenic drugs, 3/ a trigger: the sympathetic system that increases the probability of rate dependent arrhythmias under conditions of ventricular repolarization prolongation (Shimizu & Antzlevitch, 1999) during acceleration phases of HF cycles or following large RR pause, reminding that 80% of TdP published in the clinical literature are pause dependent (Viskin et al., 2000).
Finally, two specific further features common to torsadogenic hERG blocking drugs, LQT1 and LQT2 syndromes should be emphasized. First, the magnitude of HFQT oscillations is increased in LQT1 and LQT2 syndromes as with most of torsadogenic hERG blockers causing QTc prolongation. In both cases, this enhancement of HFQT oscillations reflects an increase in sympathetic activity. In LQT syndromes, TdPs are mainly triggered in humans by sympathetic activation achieved in various situations such as physical exercice, swimming, fear reaction or arousal events (Schwartz et al., 2001, Kim et al., 2010, ). Like in LQT syndromes, these compensatory sympathetic reflexes caused torsadogenic drugs also appear as playing a key role for their arrhythmic profile. Besides, the proposed sympathetic reflex mechanisms provide a robust rational support to current therapeutic strategies applied in LQT syndromes based on use of β-blockers or sympathetic left stellate ganglia ablation (Priori et al., 2015). Secondly, drug induced QTc prolongation potency can be concealed as in LQT1 and LQT2 syndromes. This latter feature should have strong implications for preclinical safety pharmacology and drug safety. Indeed, the current preclinical strategy based on risk assessment of QT prolongation and hERG blockade showed a poor sensitivity in the 1x-10x exposure multiple range (Park et al., 2018). Combining QTc prolongation potency assessment, hemodynamic and autonomic modelling and HFQT oscillations analysis offers real opportunity for better estimation of drug safety margins and improving preclinical assessment of TdP risk and sudden cardiac death.