The CiPA paradigm addresses many of the concerns related to the traditional ICH S7B and ICH E14 approaches. Importantly, CiPA focuses on the assessment of potential ventricular proarrhythmia risk via a mechanistically robust set of input data rather than employment of the surrogates of I _Kr reduction and QTc prolongation. Drugs with increased proarrhythmic risk would be assessed for their potential to enhance vulnerability to disrupt repolarization rather than simply to delay repolarization. Moving the mainstay of the assessment of proarrhythmic risk away from later (clinical) drug development to earlier stages of drug discovery, hence permitting its deployment in candidate selection, will likely prevent the current common scenario that a small QTc prolongation signal seen in drug development constitutes uncertain risk that is to be avoided, resulting in the premature discontinuation of a compound that may have been therapeutically beneficial and had an acceptable benefit–risk balance.

A second advantage is that the CiPA paradigm is not dependent on dichotomous categorization, i.e., prolongation vs. no prolongation of QTc. A graduated risk scale that would permit improved benefit–risk assessments is envisioned. Some degree of proarrhythmic risk may be considered acceptable for some drugs addressing unmet medical needs, e.g., serious oncologic diseases and chronic, debilitating diseases, but not for other drugs such as those for allergic rhinitis. The vision for CiPA is a nonbinary output in which novel compounds are given a proarrhythmic risk score predicated on a continuous scale that has been calibrated against a test set of clinical drugs spanning the range of proarrhythmic risk.

This paradigm also offers the facilitation of more informative package inserts at the time of a new drug’s marketing approval and also the relabeling of those compounds that currently have QTc warnings and precautions but for which a true low proarrhythmic risk can now be identified.

Numerous overlapping ionic currents contribute to the determination of the morphology and duration of the ventricular action potential: those discussed in this chapter are summarized in Table 9.1.

While there are multiple repolarizing potassium currents, I _Kr has attracted particular attention in two contexts: the inherited channelopathy LQT2 (Bunch and Ackerman 2007; McBride et al. 2013; Smith et al. 2013) (recall discussions in Chap. 3) and drug-induced (acquired) QTc prolongation. In LQT2, an abnormal variant of the hERG gene produces hERG channels with decreased expression or function, leading to a decrease in repolarizing I _Kr. In the case of drug-induced hERG channel blockade, a small-molecule drug typically binds to the inner walls of the central pore of the channel and impedes I _Kr. Many noncardiac drugs block the hERG channel in preference to other cardiac potassium channels as a result of differences in its intracellular channel vestibule and pore structure, characteristics that have led to the channel being described as “an unusually promiscuous target among potassium channels” (Lagrutta et al. 2008). While the mechanism of action is different (inherited vs. acquired), hERG blockade therefore results in the same cascade of occurrences as does LQT2, i.e., a reduction of repolarizing I _Kr and delayed ventricular repolarization manifesting as QT prolongation. Given this commonality, drug-induced QT prolongation, like LQT2, is also of considerable clinical concern.

All other influences being kept equal, a drug-induced decrease in I _Kr will lead to a reduction in overall (net) repolarizing current. However, the relationships between the amount of hERG block, the extent of delayed repolarization/QT prolongation, and the proclivity and incidence of proarrhythmia/torsades are uncertain and likely dependent on the kinetics and extent of I _Kr block, block of other cardiac channels during repolarization (termed multi-channel block), and the underlying electrophysiologic substrate (which may be altered in the setting of cardiac disease). Two possibilities arise. The same drug molecule can affect one or more of the multiple other cardiac ion channels, thereby affecting multiple ionic currents to affect net outward current. Depending on the channels affected, as well as kinetics and extent of block, multi-channel block may effectively minimize or cancel out the decrease in repolarizing influence due to hERG channel block, leading to minimal (if any) proarrhythmic effects. Alternatively, decreases in both I _Kr and I _Ks could have additive/synergistic effects not appreciated by evaluating hERG channel block alone. This notion is embodied in the term repolarization reserve, which recognizes that both I _Kr and I _Ks define ventricular repolarization: with one current (e.g., I _Ks) already reduced, block of the second (I _Kr) will have greater effects on repolarization and potentially convey more proarrhythmic liability in the presence of reduced baseline repolarizing current. Malik (2016) commented on the term repolarization reserve as follows:

It suggests that the interplay between different ion channels that maintain myocardial repolarization is to some extent redundant and that this redundancy offers mechanisms protecting against externally induced abnormalities including drug-induced anomalies. The susceptibility to arrhythmic consequences of drug-induced repolarization changes markedly increases when other pathological processes reduce this built-in protection.

Advances in the capabilities and adoption of higher-throughput automated voltage clamp patch platforms will facilitate more efficient characterization of drug effects on multiple cardiac currents (Dale et al. 2007; Castle et al. 2009; Ma et al. 2011; Farre and Fertig 2012; Di Veroli et al. 2013). Employing higher-throughput automated patch techniques will also provide sufficient sample size and statistical power to facilitate parameterization of subsequent in silico reconstruction efforts and to determine IC₅₀ and other characteristics of block as deemed necessary to provide reliable, reproducible characterization of integrated electrophysiological effects.

As previously noted, the relationships between the degree of hERG block, the extent of delayed repolarization/QT prolongation, and the proclivity and incidence of proarrhythmia/torsades are uncertain. Not all drugs that prolong QTc are proarrhythmic; examples include ranolazine, phenobarbital, and tolterodine. Verapamil is a potent hERG current blocker, but it does not cause QTc prolongation (except possibly at very high intravenous exposures), likely as a result of its concomitant blockade of calcium current (Zhang et al. 1999). Amiodarone is an example of a drug that causes marked QTc prolongation (not infrequently to lengths >550 msec) and yet only very rarely causes torsades.

It is likely that drug effects on multiple cardiac currents, especially reductions of inward (depolarizing) calcium and sodium currents during the action potential plateau, provide protection from proarrhythmia when coupled with a decrease of I _Kr. The concept that block of non-hERG currents may mitigate proarrhythmic effects of hERG current block is not new: it has been known for some years that combining block of repolarizing potassium current with either sodium or calcium channel block may reduce or reverse early after-depolarization (EAD) formation (Bril et al. 1996; Martin et al. 2004), a topic discussed in the following section. A review of the potency of I _Kr block (relative to clinical exposures) and TQT study results for 39 drugs demonstrated the need for additional nonclinical assays addressing drug effects on other currents to assess more comprehensively the risk of QTc prolongation (Gintant 2011). More recently, a logistic-regression approach involving assessment of drug effects on three cardiac channels, Kv11.1 or hERG (I _Kr), fast sodium Nav1.5 (I _{Na fast}), and Cav1.2 (I _CaL), showed a significant reduction in false-positive and false-negative classifications for 55 drugs from multiple classes (32 torsadogenic and 23 non-torsadogenic drugs) as compared with predictions based on I _Kr block alone (Kramer et al. 2013).

These observations reinforce the need to consider drug effects on multiple cardiac currents when assessing proarrhythmic liabilities. Some companies are already screening multiple cardiac ion channels in drug discovery (Davies et al. 2012). The ionic current studies component of the CiPA paradigm would provide standardized, best practice assays to ensure valuable data sets that inform decisions regarding cardiac safety early in drug discovery, provide guidance on first-in-human studies, and generate valuable information for regulatory considerations.

The core in vitro strategy focuses on the evaluation of drug effects on multiple isolated human cardiac currents via heterologous expression systems assessed electrically using voltage/patch clamp techniques. As we have seen, multiple currents define the cardiac action potential, and knowledge gleaned from inherited LQTS and drug-induced proarrhythmia convincingly demonstrate that repolarizing and depolarizing currents must both be considered to understand proarrhythmia. I _Kr represents only one of multiple potassium and sodium currents that, when differing from normal, are associated with long QT and proarrhythmia. Studies of acquired LQT syndromes and proarrhythmia demonstrate that fast inward sodium current and enhanced inward current (I _Na-Late or reactivation of calcium current during the action potential plateau) are also involved in proarrhythmia. Thus, a more comprehensive in vitro set of ion current assays could conceivably explore I _Kr, I _Ks, and I _K1, as well as I _Na-Fast, I _Na-Late, and I _Ca-L for drug effects. The specific currents to be evaluated to generate a sufficiently comprehensive and predictive data set are currently under discussion by various workstreams.

The use of voltage clamp studies for unbiased and standardized decision-making in arrhythmia evaluation will necessitate the development of consensus on best practices and/or standardization of protocols, positive/negative controls, and experimental conditions. This effort will reduce variability, allow comparisons across assays and laboratories, and generate movement towards more uniform data quality for purposes of decision-making, both by sponsors internally and by regulatory agencies. As one example, multiple studies demonstrate that the potency of I _Kr blockade is, at least for some drugs, affected by the experimental temperature, i.e., room temperature vs. physiologic temperature (37^oC). It is becoming clear that the potency of drugs that demonstrate prominent temperature-dependent effects is uncertain, necessitating an evaluation of hERG blocking potency at physiologic temperature.

Within CiPA, the potency of current block (based on IC₅₀ values) relative to free drug plasma concentration will be a key component in evaluating a drug’s proarrhythmic liability. Further characterization of the kinetics of block, e.g., including voltage, time, and concentration dependence, might be critical for some currents, likely including I _Kr: comparison of in silico studies incorporating conductance block models with those incorporating kinetics of drug block and unblock will guide future discussions (Di Veroli et al. 2013).

In silico models of cellular human ventricular activity are employed to integrate drug effects on multiple cardiac currents, providing reconstructions of cellular electrical activity. Electrophysiological models have been used since the pioneering work of Hodgkin and Huxley to reconstruct neuronal excitability of squid giant axons based on contributions of overlapping voltage- and time-dependent sodium and potassium currents (Hodgkin and Huxley 1952; Krouchev et al. 2015). In the CiPA paradigm, voltage clamp data describing a drug’s effects on multiple ionic currents, based on the O’Hara–Rudy model (O’Hara et al. 2011), will describe effects on ventricular repolarization that are not easily understood from effects on any individual cardiac current.

It is envisioned that in silico reconstructions will provide information on two fronts: drug effects related to the ability to elicit EADs during Phase 3 repolarization based on measures of net current during repolarization and evaluation of the robustness of repolarization based on the ability of depolarizing currents applied during the action potential plateau to amplify delayed repolarization and EAD activity. A scoring system may be necessary to rank order proarrhythmic risk based on measures of repolarization instability calibrated against a training set of compounds affecting multiple cardiac ion channels that have been ranked according to clinical torsades risk. This continuous scoring system could then be used to rank order drug candidates’ risk of torsades proarrhythmia in the context of therapeutic margins, e.g., therapeutic concentration and plasma protein binding.

In support of the use of integrative in silico models, a study by Mirams and colleagues (Mirams et al. 2011) employing in silico modeling to measure action potential prolongation demonstrated that evaluation of drug effects across three human ion channels (Kv11.1, Nav1.5, and Cav1.2) improved prediction of torsadogenic risk compared with evaluations based solely on hERG channel block. Numerous studies have described the general utility of in silico ventricular reconstructions in evaluating overall delayed repolarization liabilities and/or proarrhythmic risk (Valentin and Hammond 2008; Fletcher et al. 2011; Mirams et al. 2011, 2012; Kramer et al. 2013; Beattie et al. 2013).

The best in silico cellular model(s) for reconstructions will have to be selected and then made available to users in a standardized format to provide meaningful ranking of proarrhythmia across different laboratories or, alternatively, be made widely available on a centralized cloud-based resource.

CiPA’s third component is the evaluation of drug effects on the electrical activity of hiPSC-CMs. This approach yields a cell-based integrated electrophysiological drug response, providing a check on the adequacy of the voltage clamp data and in silico reconstructions of ventricular electrical activity: it is critical to ensure that drug effects not detected in voltage clamp-based ionic current assays and in silico models are detected and evaluated. Isolation and propagation of human-induced pluripotent stem cells and hiPSC-CMs has provided a useful source of cells for applications in drug discovery and cardiotoxicity screening (Mordwinkin et al. 2013).

Voltage clamp studies of hiPSC-CMs have demonstrated the presence of currents expected in adult ventricular myocytes (Ma et al. 2011; Hoekstra et al. 2012) and effects on repolarization consistent with human responses (Hoekstra et al. 2012; Peng et al. 2010). However, studies have shown that hiPSC-CMs demonstrate a relatively immature phenotype compared with adult human myocytes (Jonsson et al. 2012; Qu et al. 2013a). Despite these limitations, numerous studies have demonstrated their ability to detect responses consistent with clinical findings. Thus, while they represent a model system, human ventricular myocytes presently do not fully recapitulate the native adult ventricular myocyte in all functional aspects. Efforts are ongoing in multiple laboratories to provide more representative, or mature, ventricular myocytes. Once successful, fully mature hiPSC-CMs (expressing all the ionic currents with the same densities and characteristics as adult human ventricular myocytes) should replace most in vitro proarrhythmia testing approaches. Furthermore, it should also be possible to test evolving diseased ventricular myocyte models in vitro to evaluate drug effects on at-risk populations (the so-called disease in a dish studies).

A critical assessment of present practices and data yielded from hiPSC-CMs will be necessary in defining the most appropriate experimental methodologies and their limitations. As hiPSC-CMs represent a relatively new and rapidly evolving area of investigation, it is necessary to characterize these preparations more fully and build consensus on their ability to provide consistent data across laboratories and methods. As is the case for all new in vitro preparations, the selection of hiPSC-CMs and the experimental conditions in which they are employed will need to be rigorously defined to facilitate subsequent standardization for use in CiPA.

Updates from each working group are provided in turn.