Familial And Acquired Long QT Syndrome And The Cardiac Rapid Delayed Rectifier Potassium Current

Authors

  • Harry J Witchel,

    1. Department of Physiology and Cardiovascular Research Laboratories, School of Medical Sciences, University of Bristol, Bristol, United Kingdom
    Search for more papers by this author
  • Jules C Hancox

    1. Department of Physiology and Cardiovascular Research Laboratories, School of Medical Sciences, University of Bristol, Bristol, United Kingdom
    Search for more papers by this author

Harry J Witchel, Department of Physiology and Cardiovascular Research Laboratories, School of Medical Sciences, University of Bristol, University Walk, Bristol, BS8 1TD, United Kingdom. Email: harry.witchel@bristol.ac.uk

SUMMARY

1. Long QT syndrome (LQTS) is a cardiac disorder characterized by syncope, seizures and sudden death; it can be congenital, idiopathic, or iatrogenic.

2. Long QT syndrome is so-named because of the connection observed between the distinctive polymorphic ventricular tachycardia torsade de pointes and prolongation of the QT interval of the electrocardiogram, reflecting abnormally slowed ventricular action potential (AP) repolarization. Acquired LQTS has many similar clinical features to congenital LQTS, but typically affects older individuals and is often associated with specific pharmacological agents.

3. A growing number of drugs associated with QT prolongation and its concomitant risks of arrhythmia and sudden death have been shown to block the ‘rapid’ cardiac delayed rectifier potassium current (IKr) or cloned channels encoded by the human ether-a-go-go-related gene (HERG; the gene believed to encode native IKr). Because IKr plays an important role in ventricular AP repolarization, its inhibition would be expected to result in prolongation of both the AP and QT interval of the electrocardiogram.

4. The drugs that produce acquired LQTS are structurally heterogeneous, including anti-arrhythmics, such as quinidine, non-sedating antihistamines, such as terfenadine, and psychiatric drugs, such as haloperidol. In addition to heterogeneity in their structure, the electrophysiological characteristics of HERG/IKr inhibition differ between agents.

5. Here, clinical observations are associated with cellular data to correlate acquired LQTS with the IKr/HERG potassium (K+) channel. One strategy for developing improved compounds in those drug classes that are currently associated with LQTS could be to design drug structures that preserve clinical efficacy but are modified to avoid pharmacological interactions with IKr. Until such time, awareness of the QT-prolongation risk of particular agents is important for the clinician.

INTRODUCTION

The term ‘long QT syndrome’ (LQTS) refers to a disorder, diagnosed by an abnormality of the electrocardiogram (ECG), that predisposes individuals to potentially fatal ventricular arrhythmias. Long QT syndrome is of particular interest not simply due to the severity of the syndrome, but because it now reflects one of the better understood disorders of cardiac rhythm.

Therefore, in the present article we aim to provide an overview of LQTS and the impact made on the understanding of the syndrome by advances in molecular genetics and in the electrophysiology of cardiac muscle. The disorder exists in both ‘familial’ (congenital) and ‘acquired’ forms. The acquired form is associated with a range of clinically used pharmacological agents and recent increases in understanding the basis of acquired LQTS have followed the elucidation of the genetics of the familial disorder. In the present review article, the history and basis of the familial disorder will be considered, together with an explanation of the cellular consequences of ventricular action potential (AP) prolongation. Consideration will then be given to pharmacological agents associated with the acquired syndrome.

QT INTERVAL REPRESENTS THE DURATION OF THE VENTRICULAR AP

Electrical surface recordings made during the ECG correspond to the electrophysiological events occurring during impulse generation and conduction in the heart. Each heartbeat starts as electrical excitation, is generated in the sinoatrial node and is rapidly conducted throughout the atria. On surface ECG measurements, the P wave represents the combined electrical activity of atrial depolarization. Impulse conduction to the ventricles occurs following conduction through the atrioventricular node and excitation is transmitted rapidly across both ventricles via the His–Purkinje system and by virtue of the tight electrical coupling between ventricular cells. The QRS complex of the ECG corresponds to the depolarization of the ventricles (and masks the electrical activity associated with AP repolarization in the atria); the T wave is associated with ventricular repolarization. Thus, the QT interval represents the duration of the ventricular AP. This is shown diagrammatically in Fig. 1, in which action potential ‘A’ corresponds to QT(A). It follows, therefore, that prolongation of the QT interval corresponds to prolongation of the ventricular AP (in Fig. 1, AP ‘B’ corresponds to QT(B)).

Figure 1.

This is a schematic diagram showing the relationship between the ventricular action potentials (AP) and the electrocardiogram (ECG). The QRS complex corresponds to the AP upstroke and the T wave corresponds to ventricular repolarization for the AP. (A) Normal AP lasting 200–300 msec; (B) delayed repolarization. This correlates with the prolongation of the QT interval, QT(A)→ QT(B).

The ventricular AP is typically 200–300 msec in duration. Prolongation of the heart rate-corrected (QTc) interval in excess of approximately 440 msec is associated with an increased cardiovascular risk. 1 Such a prolongation predisposes the patient to torsade de pointes (a polymorphic ventricular tachycardia). The degeneration of distinct (although prolonged) ECG complexes into torsade de pointes is characteristic of the condition. Figure 2 shows a sample of an ECG from a patient with LQTS. It shows the characteristic prolonged QT duration and the short–long–short RR durations preceding the initiation of torsade de pointes with a ‘twisting’ or sinusoidal waveform.

Figure 2.

Part of a continuous single-channel Holter recording from a patient with long QT syndrome. In this record, the characteristic prolonged QT interval (A), followed by giant late-repolarization ‘T-wave humps’ (B), led to premature beats with a bigeminal pattern with short-long-short sequences of R–R intervals (C) before onset of torsade de pointes (D). The episode progressed into ventricular fibrillation before spontaneously resolving into sinus rhythm; the patient was later treated successfully with pacing and beta-blockers (Reproduced with permission from Benhorin and Medina. 159)

IONIC CURRENTS INVOLVED IN VENTRICULAR AP REPOLARIZATION

Control of the duration of the cardiac ventricular AP is mediated by the equilibrium between inward and outward currents across the cell membrane. 2 Under physiological conditions, the AP progresses through the phases of depolarization, plateau phase and repolarization over the course of 200–300 msec but, in LQTS, repolarization is delayed and the AP duration at times will be able to exceed 440–460 msec. Electrophysiologically, this delay in repolarization implies a deviation from the usual balance of currents across the myocyte membrane responsible for normal repolarization. Thus, prolongation of the AP could, theoretically, arise from an increase in inward (depolarizing) current or, alternatively, from a decrease in outward (repolarizing) current carried by potassium ions. Both of these conditions have been found to exist in different LQTS patients.

Potassium channels play prominent roles in the initiation and completion of AP repolarization, as well as during the plateau phase. The rate of net K+ efflux and, by extension, the rate of repolarization is determined by the density and gating properties of different K+ channels. The K+ channels mediating an inwardly rectifying current (IK1) are important in maintaining the normal resting potential and during the final stage of AP repolarization. 2 A rapidly activating and inactivating transient outward current (Ito) contributes to early AP repolarization and underlies the initial ‘notch’ before the plateau phase. Of particular relevance to the AP plateau is the delayed rectifier current (IK), comprised of rapid (IKr) and slow (IKs) components mediated by distinct channel subtypes with distinct kinetic properties. 3,4 The IK develops gradually during the plateau phase, opposing the inward currents underlying sustained depolarization. As the net balance of current alters and net outward current exceeds inward current, repolarization occurs. The IK and IK1 can be considered to regulate ventricular AP repolarization over the plateau range and final rapid repolarization phase, respectively. 4

POSSIBLE ORIGINS OF DYSRHYTHMIA ASSOCIATED WITH LQTS

Familial LQTS is characterized by an early onset of symptoms (a mean age of 24 years in the international registry 5), frequent, recurrent, non-sustained tachyarrhythmias and a labile, prolonged QT interval on the ECG. In a large prospective study, Moss et al.6 reported that patients diagnosed as having LQTS had a higher frequency of pre-enrolment syncope (fainting/collapse) or cardiac arrest with resuscitation (80%), a resting heart rate less than 60 b.p.m. (31%), a history of ventricular tachyarrhythmia (47%) and a higher rate of congenital deafness (7%) than other unaffected family members. The arrhythmogenic syncope often can be associated with acute physical, emotional or auditory arousal and these syncopal episodes are due to torsade de pointes, which can degenerate into ventricular fibrillation, ultimately leading to sudden death in most untreated patients with the disorder. For some time there was no consensus as to the cause of torsade de pointes in LQTS, which was thought to be possibly either neural or cardiac in origin.

Neural

In a neural scheme, the primary abnormality would be lower than normal right sympathetic activity compared with left sympathetic activity, leading to a left–right imbalance in sympathetic activity and the QT prolongation, per se, would not be responsible for arrhythmogenesis. The use of β-adrenoceptor blockade as a treatment for LQTS would be consistent with this hypothesis. Furthermore, stimulation of the left stellate ganglion causes QT prolongation and ablation causes QT abbreviation, while those procedures applied to the right stellate ganglion have the opposite effects. In some cases, left stellate ganglion block or ablation has been used in the treatment of LQTS. 7 In addition, secondary torsade de pointes is produced by various drugs and by intracranial disease, such as subarachnoid haemorrhage. Evidence for a role of sympathetic imbalance comes from single photon emission computed tomography comparing healthy volunteers with LQTS patients. 8 Although consistent with sympathetic imbalance, such evidence does not exclude the possibility of a concurrent cardiac cause for LQTS.

Cardiac

At a cellular level, pharmacologically induced slowing of AP repolarization can lead to spontaneous depolarizations generated on the falling phase of the AP plateau ( Fig. 3). Unlike the AP itself, these events (termed ‘early afterdepolarizations’ or ‘EAD’) are not synchronized with an excitatory stimulus (experimentally, an applied injection of current; in vivo, an incoming wave of excitation) and, so, can give rise to asynchronous tissue excitation (and thereby arrhythmogenesis). Early afterdepolarizations and triggered firing were shown electrophysiologically to be the mechanism for initiation of tachyarrhythmias in the LQTS model system of the dog heart in which caesium is used to block potassium currents and prolong the QT interval. 9 Becuase many of the pharmacological agents that cause acquired LQTS also block potassium currents, these agents may share the mechanism for caesium-induced tachyarrhythmia, EAD and triggered firing. 10

Figure 3.

Early afterdepolarizations (EAD) in a sheep isolated cardiac Purkinje fibre that has been induced by increasing dosages of the L-type calcium channel agonist Bay K 8644. The upper trace shows intracellular recordings of action potentials and the lower trace shows measurements of cellular tension. With increasing concentrations of Bay K 8644, the EAD become more pronounced. The cell stimulation rate was 0.2 Hz. (Reproduced with permission from January et al.160)

Events discernible from the ECG may correspond to EAD at the cellular level. For example, the increase in the late component of the slow waves following the QRS complex in LQTS (often called TU waves to describe the apparent merging of T and U waves) may correlate with EAD 11 given that, by using cardiac pacing at varying rates interspersed with pauses of various duration, it was demonstrated that the amplitude of the terminal portion of the TU wave is directly proportional to the immediately preceding RR intervals.

The mediating step between a prolonged AP and EAD is believed to involve L-type Ca2+ channels, 12 such that EAD may result from depolarizing current due to the reopening of a small percentage of L-type Ca2+ channels while the membrane potential passes through their ‘window’ voltage range. 13 The ‘window’ is a range of voltages over which steady state voltage-dependent activation and inactivation curves for L-type ICa overlap, permitting steady state Ca2+ entry. If the membrane potential remains above the ‘window’, the L-type calcium channels remain inactivated and non-conducting, while stepping the membrane potential below the window causes the channels to change from the inactivated to closed states and these closed channels remain non-conducting until the membrane potential depolarizes.

It has been proposed that two phases exist for the induction of EAD: (i) an initial conditioning phase, in which the membrane potential must remain depolarized long enough to permit calcium channel inactivation; and (ii) a recovery phase, in which the membrane potential is within the ‘window’ and calcium channels can make the transition from the inactivated to closed states, followed by a reopening of a small proportion of channels. At prolonged AP durations, more channels can complete the conditioning phase, facilitating the generation of EAD (see Makielski and January 13 for a review). In the cultured rabbit ventricular myocyte model, in which cells are predisposed to prolonged AP durations and EAD, adenovirus-mediated overexpression of the human ether-a-go-go-related gene (HERG) channel was sufficient to shorten AP duration and reduce EAD by four-fold. 14

In addition, the well-known upregulation of L-type ICa by β-adrenoceptor stimulation would be expected to further increase Ca2+ entry over window voltages, explaining the increased likelihood of arrhythmogenesis under conditions of exertion or extreme emotion. The arrhythmias in the familial LQTS are often precipitated by sympathetic excess and treated by β-adrenoceptor blockade. 5

ION CHANNEL GENES AND FAMILIAL LQTS

Molecular genetic studies have played important roles not only in establishing links between particular channelopathies and LQTS, but also in producing important insights into the function of individual ion channel types. Therefore, before considering the pharmacological modulation of ion channel currents involved in acquired LQTS, it is first useful to consider the basis for congenital LQTS.

Thus far, five genes have been identified as sites of mutation for the congenital disorder. The genetic approaches show that all five genes are responsible for ion channel proteins that are present in the myocardium, providing insight into the electrophysiological causes of LQTS. A K+ channel protein named KvLQT1 15 (locus 11p15.5) is the α-subunit (the main body of the ion channel) responsible for IKs. Mutations in a Na+ channel gene (SCN5A;16 locus 3p21-p23) can interfere with channel inactivation, causing the channel to conduct a sustained inward current during the plateau phase of the AP, thereby delaying repolarization. The gene KCNE1 (21q22.1-p22) encodes a K+ channel β-subunit (an accessory subunit) called minK that is associated with KvLQT1 to form functional channels mediating IKs. 17,18 Another gene encoding a K+ channel β-subunit, MiRP1 (21q11.1), has recently been shown to be associated with clarithromycin-induced arrhythmias and mutations in this gene are thought to be clinically silent until combined with additional stressors. 19 The first gene identified as responsible for LQTS was HERG20,21 (7q35-q36); this gene encodes a K+ channel that mediates IKr. 22 When HERG was first demonstrated to be a potassium ion channel responsible for chromosome 7-associated familial LQTS, 21 it was also proposed that pharmacological inhibition of the HERG channel was a possible mechanism for acquired LQTS. 22

The putative structure of the HERG channel, as determined by analysis of the derived amino acid sequence of the cloned transcripts, shows HERG to be similar in many ways to other members of the shaker-type voltage-gated potassium channel families. 20 These potassium channels are made up of four subunits, each of which has six α-helical transmembrane domains and a looping ‘pore region’ ( Fig. 4a; see also Sanguinetti and Keating 4). The transmembrane domains are functionally organized such that S5 and S6 and the looping pore region contribute to the pore and the S4 region includes regularly spaced charged amino acids that function as the voltage sensor. 23 The HERG channel is 1159 amino acids long and has extended N- and C-termini that project intracellularly. The N-terminus of HERG is functionally related to deactivation of the channel 24 and X-ray crystallography has recently shown the structure of the N-terminus to be related to a PAS regulatory domain. 25 Many different mutations in the channel have been found and these provide clinical correlates for the functional domains of the channel (see Table 1 and Roden and Balser 26).

Figure 4.

Figure 4.

(a) Diagrammatic representation of a single human ether-a-go-go-related gene (HERG) potassium channel protein α-subunit. There are six transmembrane domains and a looping ‘pore’ domain between S5 and S6. The charged S4 domain is the ‘voltage sensor’ that responds to changes in membrane potential. The N- and C-termini are marked. (b) Schematic representation of the profile of rapidly activating delayed rectifier K+ current (IKr) during the ventricular action potential (based on data in Hancox et al.31). The upper panel shows the action potential profile, while the lower panel shows the corresponding current.

Figure 4.

Figure 4.

(a) Diagrammatic representation of a single human ether-a-go-go-related gene (HERG) potassium channel protein α-subunit. There are six transmembrane domains and a looping ‘pore’ domain between S5 and S6. The charged S4 domain is the ‘voltage sensor’ that responds to changes in membrane potential. The N- and C-termini are marked. (b) Schematic representation of the profile of rapidly activating delayed rectifier K+ current (IKr) during the ventricular action potential (based on data in Hancox et al.31). The upper panel shows the action potential profile, while the lower panel shows the corresponding current.

Table 1.  Mutations in HERG associated with congenital long QT syndrome
MutationLocalizationFunctional effectReferences
  1. The mutations are named by the amino acid substitution (or deletion) predicted from the genetic variation. Also shown are the regions of the K+ channel protein predicted to be affected, as well as the reference for the mutation. A question mark (?) indicates that the functional effect of the mutation is not yet known.

F 29 LPAS domainRapid deactivation111
N 33 TPAS domainRapid deactivation111
G 53 RPAS domainRapid deactivation111
R 56 QPAS domainRapid deactivation111
C 66 GPAS domainRapid deactivation111
H 70 RPAS domainRapid deactivation111
A 78 PPAS domainRapid deactivation111
L 86 RPAS domainRapid deactivation111
Δb.p. 1261 (frameshift)S1Decreased expression21, 112, 113
N 470 DS2Decreased expression, altered deactivation21, 113
T 474 IS2/S3?114
ΔI 500–F 508S3No expression21, 113
R 534 CS4Altered activation115
A 558 PS5?116
A 561 VS5?21, 113
A 561 TS5?117
R 582 CS5-Pore?116
I 593 RS5-Pore?118
G 601 SS5-Pore?119
G 604 SS5-Pore?116
Y 611 HS5-Pore?114
V 612 LS5-Pore?120
T 613 MS5-Pore?116
A 614 VS5-Pore?114, 120
G 628 SPoreNo expression21, 113
N 629 DPore?120
N 629 SPore?120
V 630 LPore?114
N 633 SPore?120
F 640 LPore?116
Splice site alterationcNBDNo expression113, 122
Splice site alterationcNBD?21, 113
S 818 LcNBDSensitive to hypokalaemia123
V 822 McNBDSensitive to hypokalaemia121, 123

When measurements were made from the expressed HERG product, a unique current profile was observed. Measurements of the native current from heart cells had shown inward rectification of the channel at positive voltages. 3 Similar observations were made for HERG 22,27 and were later found to result from unusually rapid, C-type inactivation at positive potentials. 28,29 This has important implications for the physiological profile of IKr/HERG during the AP. At the AP peak, the current activates quickly, but relatively little current flows at the peak due to rapid inactivation. As inactivation is removed during the plateau, outward current flow gradually increases, peaking just before the final rapid repolarization phase ( Fig. 4b). This current profile has been confirmed by measurements of both HERG and native IKr made using the ‘AP clamp’ technique 30,31 and is consistent with a critical role for IKr in regulating the duration of the plateau (and, therefore, AP duration).

There are phenotypic correlations between the triggers for cardiac events and different genotypes in LQTS, raising the possibility of gene-specific therapy. In LQTS related to mutations in IKs/KvLQT1, exercise-related events dominate the clinical course, 32 whereas patients with loss of inactivation of SCN5A experience events at rest or during sleep and are able to shorten their QT interval during exercise. 33 Long QT syndrome patients with mutations in IKr/HERG have arrhythmias associated with rest and exercise, 33 but such patients can have events associated with auditory stimuli, which is not the case with the IKs/KvLQT1 patients. The HERG can be inhibited by low external K+22 and there is some evidence that raising serum K+ may be useful in chromosome 7-associated LQTS. 34

ACQUIRED LQTS AND PHARMACOLOGICAL INHIBITION OF IKr

An increased understanding of congenital LQTS has helped shed light on the acquired syndrome. A variety of pharmaceutical agents in clinical use can be linked to LQTS and its associated pro-arrhythmic risk. The original approach for identification of the compounds producing acquired LQTS was clinical and empirical, using ECG or symptomology in animal models or in clinical subjects. With increased understanding of the electrophysiological basis of LQTS, pharmacological agents could be tested on cardiac myocytes, suggesting a mechanism for the generation of arrhythmia. Also, the increased knowledge regarding ion channels involved in LQTS has led to direct electrophysiological testing of agents on ion channels expressed in model cell systems (heterologous expression of the gene in mammalian cell lines and in Xenopus oocytes). A range of cardiac and non-cardiac agents can precipitate LQTS and have been demonstrated to inhibit either HERG or IKr (summarized in Table 2). We will consider here a cross-section of such agents.

Table 2.  Important pharmaceutical agents associated with prolonged QT intervals and the associated risks of arrhythmia
 References
Pharmacological agentClinicalIKr/HERG*
  1. Citations of relevant clinical findings and relevant studies at the cellular level, observing pharmaceutical inhibition of the native rapidly activating delayed rectifier K+ current (IKr) or of heterologously expressed human ether-a-go-go-related gene (HERG) current (the latter indicated by an asterisk) are given.

  2. TCA, tricyclic antidepressants; GI, gastrointestinal.

Class III anti-arrhythmics
Amiodarone12462, 65*
Dofetilide12537, 45*
D-Sotalol 5149
Bretylium 126
Almokalant127128
Sematilide129130
Ibutilide131132
Tedisamil133134
Azimilide13561*
Class I anti-arrhythmics
Quinidine13636 (review), 73*, 137
Propafenone13879, 139
Procainamide140
Disopyramide14176
Bepridil8381*
Prenylamine142
Terodiline9091
Non-cardiac drugs
Antihistamines
Terfenadine14398*
Astemizole14399*
Psychiatric drugs
Haloperidol144103*
TCA145107,* 146
Chlorpromazine147
Thioridazine148149
Sertindole150104*
Antimicrobial and antimalarial
Erythromycin151109, 152
Pentamidine153
Quinine154
Chloroquine155
Halofantrine156
GI motility
Cisapride157158*

Anti-arrhythmic agents and LQTS

While anti-arrhythmic drugs in common use are, to varying extents, effective against particular rhythm disturbances, some are also associated with the risk of pro-arrhythmia. The traditional classification of anti-arrhythmic agents, the Singh–Vaughan-Williams (S-V-W) classification, categorizes agents in an effort to describe ways in which abnormal rhythms can be corrected or prevented, rather than by their chemical structures or physical properties. 35

1. Class I drugs act primarily by reducing inward sodium current at concentrations that do not reduce the resting membrane potential.

2. Class II drugs are sympatholytic drugs, including beta-blockers.

3. Class III drugs prolong the AP duration, typically by blocking potassium channels. 4

4. Class IV drugs include calcium channel blockers.

Theoretically, a drug that blocks only one ion channel or receptor is likely to have a more specific action; however, many of the drugs in clinical use exhibit actions from more than one S-V-W category.

Action potential prolongation (a Class III action) can be both anti-arrhythmic and pro-arrhythmic. Moderate AP prolongation enhances the refractory period, thereby helping limit ventricular rate and re-entry. Excessive prolongation, as explained earlier, can facilitate production of EAD and lead to resultant ventricular tachycardia. Acquired LQTS with anti-arrhythmic therapy is well illustrated by the association of ‘quinidine syncope’ with paroxysmal ventricular tachycardias. 36 Although quinidine is a ‘Class I’ agent, it also possesses strong ‘Class III’ actions (as will be discussed later), which explains this clinical observation.

The ideal Class III anti-arrhythmic effect would be a prolongation of the AP duration that was augmented as the AP rate itself increased. Thus, at normal rates, there would be comparatively little effect of the agent, while during tachycardia, drug-induced AP prolongation would become more prominent. A drug with this effect would be described as exhibiting positive ‘use dependence’. Unfortunately, many drugs with Class III effects show exactly the opposite behaviour (reverse use dependence): AP prolongation is greatest at slower rates. 37 This is undesirable, because it can lead to EAD and resultant ventricular rhythm disturbance at slow rates, while it would lead to reduced AP prolongation at faster rates.

Use dependence of drug action

Therefore, the issue of ‘use-dependence’ merits some consideration. It is important to be aware that a distinction exists between a global drug effect (e.g. use dependence at the level of the ventricular AP) and an effect at the level of individual ion channel types. For a single ion channel substrate, ‘use-dependent block’ occurs on repetitive stimulation of the channel, may increase with stimulation frequency and is most likely to occur with agents that bind to ion channels in either the ‘open’ or ‘inactivated’ states. It is possible, however, for use dependence of drug action at the level of AP to differ from that at the level of individual channel types. This is highlighted by the reverse use-dependent effects of the Class III drug dofetilide. In 1993, Jurkiewicz and Sanguinetti showed that the IKr blocker dofetilide produced reverse use-dependent AP prolongation in guinea-pig ventricular myocytes, while the drug effect on IKr was found to be rate independent. 37 However, they observed that at faster rates IKs deactivated incompletely between successive pulses, an effect consistent with an increasing role for IKs in repolarization at faster heart rates. This property of IKs was considered to partially offset the rate-independent effects of dofetilide on IKr. 37

While dofetilide is highly selective as a channel blocker for IKr over IKs and Jurkiewicz and Sanguinetti did not observe effects of the drug on IKs, 37 a subsequent report suggested that it may delay deactivation of IKs. 38 Although such an effect may itself actually enhance the role of IKs during repolarization and although a later study 39 did not support a similar role for IKs in canine myocytes, the hypothesis that IKs may assume a more prominent role in AP repolarization at faster heart rates does offer an attractive explanation for reverse use-dependent effects on the AP of drugs that inhibit IKr. In this scheme, selective IKr inhibition will result in a greater prolongation of the AP at slow rates (at which IKr predominates) than at faster ones (at which IKs plays a greater role).

Long QT syndrome and Class III anti-arrhythmic agents

Following studies in which certain Class I anti-arrhythmic agents increased mortality in high-risk postinfarction patients, much anti-arrhythmic research has focused on agents with Class III actions. 40 As discussed above, a Class III action mediated by selective IKr blockade may not be without risk of pro-arrhythmia. Therefore, it is valuable to consider in more detail LQTS associated with Class III S-V-W agents. Despite their use in the prophylaxis of arrhythmia, a number of Class III anti-arrhythmic agents carry a risk of pro-arrhythmia. It is likely that such effects are associated with selective IKr (over IKs) blockade.

The Class III agent E-4031 produces dose-related QTc prolongation in humans and, at some doses, seems well tolerated. 41 In addition to blocking native IKr, 3 E-4031 is also associated with reverse use-dependent effects 42 and, in experimental models, the drug can induce EAD. 43 E-4031 was the first agent to be demonstrated at the molecular level to mediate its effects by blocking the HERG K+ channel. 27 Other methanesulphonanilides shown to block HERG include MK-499 and dofetilide, 44,45 both of which were proposed to block the open state of the channel. The half-maximal inhibitory concentrations (IC50) for these compounds in the heterologous Xenopus oocyte system are 151 and 12 nmol/L, 44,45 respectively, and comparisons to their block of other K+ currents suggest that IKs and IK1 are affected by these compounds only at doses 100-fold greater than HERG. The open state channel block was also observed in single channel studies on heterologously expressed channels. 46 Because the block by these agents is more potent when the agents are applied intracellularly, it is likely that the site of action of the drugs is inside the membrane and that the drugs must diffuse across the plasma membrane to act. The block by dofetilide was further examined by making chimeric proteins from human HERG and the non-inactivating bovine homologue of HERG (BEAG), the latter being 100-fold less sensitive to dofetilide. 47 The results of this study suggested that the pore region of HERG contains important determinants of dofetilide binding and that a specific serine residue (S620) in the S5-S6 linker region ( Fig. 4a) appears critical for high-affinity dofetilide binding.

One Class III agent of particular interest with regard to acquired LQTS is sotalol. As a racemic ( D, L) mixture, sotalol exhibits both β-adrenoceptor blocking and potassium channel blocking activities. 48,49 The ability of sotalol to suppress ventricular ectopics is similar to that of Class I agents and is better than that of standard beta-blockers. 50 The isomer D-sotalol is a pure Class III agent without β-adrenoceptor blocking effects. The Survival With ORal D-Sotalol (SWORD) trial of D-sotalol was stopped due to drug-associated mortality. 51 Although sudden death precipitated by torsade de pointes was initially presumed to be responsible for the results, it was found that none of the torsade-associated risk factors (i.e. time from beginning of therapy, QTc, serum potassium, renal function, dose of D-sotalol), except for female gender, were associated with mortality. 52 Even so, there are reports of torsade de pointes associated with high doses of D-sotalol 53 and D-sotalol can increase dispersion of ventricular repolarization (a predisposing factor for torsade de pointes; see Verduyn et al.54).

Unlike D-sotalol, the racemic D, L mixture was not associated with increased mortality rate in a postinfarction trial. The D, L mixture has shown superior efficacy for prevention of recurrent ventricular tachycardia and ventricular fibrillation, making racemic sotalol an increasingly important anti-arrhythmic. 50 The more desirable profile of racemic sotalol than that of D-sotalol is likely to be attributable to β-adrenoceptor blocking properties, suggesting that the combination of β-adrenoceptor blockade and IKr block may be superior to IKr blockade alone as an anti-arrhythmic strategy.

In rabbit perfused isolated hearts, clofilium, another Class III anti-arrhythmic, was shown to be less rate dependent than dofetilide. 55 It does, however, produce QT prolongation and (in association with hypokalaemia) arrhythmias with the characteristics of torsade de pointes.56 Clofilium has been shown to block IKr in a voltage- dependent fashion. 57 Both clofilium and its tertiary analogue LY97241 were demonstrated to block HERG with a positive voltage and a positive use-dependent mechanism 58 suggestive of a block that is open state dependent. Experiments using heterologous expression of a mutated version of the channel demonstrated the block of LY97241 to be associated with channel inactivation. This was determined using a mutant of HERG called ‘S631A’ (for which the voltage dependence of channel inactivation is shifted to substantially more positive voltages 59). The effect of LY97241 on this mutated channel was reduced seven- to 33-fold, thereby suggesting a role for the inactivated state in block. 58

Class III agents that inhibit both IKr and IKs may be valuable in that reverse use-dependent effects on the ventricular AP may be less likely. The drug azimilide blocks both IKr and IKs and, so, the drug may have less potential to cause arrhythmia. 60 There is evidence that HERG channel block by azimilide is reverse use dependent; 61 the block increases with subsequent stimuli, but the increase in block is more rapid when the stimulation frequency is lower. The block is not significantly different in the mutant S631A from the block in the wild-type channel, nor is the rate of inactivation increased by azimilide. 61 Busch et al.61 concluded that the mechanism of block by azimilide is distinct from that of E-4031 and dofetilide. While clinical trials of azimilide are not complete, emerging information does suggest that the drug has a good safety profile. 60 Thus, the additional drug effects on IKs may counteract pro-arrhythmic potential, which may otherwise arise due to selective IKr blockade.

Another agent that affects both IKr and IKs is amiodarone, 62 which is a well-established Class III drug not associated with polymorphic ventricular tachyarrhythmias. Amiodarone also blocks sodium and calcium currents, as well as being a non-competitive α- and β-adrenoceptor blocker. 63 Thus, amiodarone exhibits actions from each of the four S-V-W classes. In guinea-pig ventricular myocytes, the efficacy of block of IKr was greater than that of IKs for identical concentrations. 64 Amiodarone has also been shown to block HERG in a heterologous system, 65 the block consisting of two main components: (i) a closed channel block that could not be reversed within the time of those experiments; and (ii) an open channel block with a slow unblock, having a recovery time constant of 73 s at –80 mV.

In clinical trials, the Class III anti-arrhythmic amiodarone has been found to show a more favourable profile than Class I anti-arrhythmics in terms of sudden cardiac death and total mortality. 66 It is usually classified as a Class III anti-arrhythmic because it lengthens the AP duration and the QT interval but, in clinical trials, amiodarone almost never causes torsade de pointes or sustained ventricular tachycardia. 40 This may be due to the wide range of ion channel-blocking actions of amiodarone. The drug is associated with a range of non-cardiac side effects but is very useful in the treatment of life-threatening arrhythmias that have not responded or cannot tolerate adequate doses of other available anti-arrhythmics. 66

In concluding the discussion of Class III agents, it is useful to consider data arising from selective IKs blockade and drug- combination experiments. Chromanol 293B is a specific blocker of IKs and it increases human and guinea-pig AP duration in ventricular myocytes to a similar fractional extent at all frequencies. 67 This may suggest that selective IKs blockade offers a useful anti-arrhythmic strategy, which may in some ways compare favourably with selective IKr blockade. For IKr blockers, drug combination may offer some additional benefit. In experiments on isolated guinea-pig papillary muscles, the reverse use-dependent effects on the AP duration of E-4031 could be abolished by coadministration with low levels of the calcium channel blockers verapamil, nitrendipine or ryanodine. 42 Moreover, the compound BRL-32872, which combines calcium channel and potassium channel blocking effects, 68 increases AP duration equally at basic cycle lengths of 300 and 1000 msec. 42

Long QT syndrome and Class I anti-arrhythmics

Anti-arrhythmic drugs not originally anticipated to affect cardiac K+ channels can also cause lengthening of the QT interval. One of the first medicinal agents associated with what is now known to be LQTS was quinidine, a Class I anti-arrhythmic drug that remains the most commonly prescribed agent for atrial fibrillation. 69 Syncope and sudden death associated with its use have been noted as potential complications for a considerable time. 70 It is now known that these dangerous complications are due to torsade de pointes36 and measurements show that quinidine is associated with LQTS at a minimum frequency of 1.5–8% per patient year of treatment. 71,72 Quinidine has been shown to block native cardiac IKr62 and has recently been shown to block HERG current. 73

Over time, it has become clear that while most Class I anti-arrhythmics primarily act by reducing sodium permeability and, thus, slowing conduction velocity, many Class I anti-arrhythmics can also block potassium currents. 74,75 In addition to quinidine (vide supra), disopyramide 76,77 and propafenone can also affect the QT interval and IKr. 78,79 For disopyramide, the QT prolongation may be isomer specific. 77

Calcium channel blockers, IKr/HERG inhibition and LQTS

Calcium channel antagonists are commonly prescribed for the management of hypertension, angina and supraventricular arrhythmias. They constitute a heterogeneous class of compounds, including the 1,4-dihydropyridines (e.g. nitrendipine and amlodipine, which fall outside the S-V-W classification), the phenylalkylamines (e.g. verapamil), the benzothiazepines (e.g. diltiazem) and others, such as bepridil. They vary in their ability to block IK and those that do block can be used in the suppression of certain ventricular arrhythmias (e.g. the verapamil sensitivity of certain ventricular tachycardias 80). Of HERG and KvLQT1/IsK, verapamil only blocks HERG current, 81,82 while diltiazem seems selective for ICa. 81,82 Bepridil prolongs the AP duration 83 and was shown, apart from being a potent anti-arrhythmic agent, to be associated with polymorphic ventricular arrhythmias. 84 Mibefradil (Ro 40-5967) was introduced as a calcium channel antagonist that could preferentially inhibit the T-type, dihydropyridine-insensitive, calcium channels at therapeutic concentrations 85 and that did not have the negative inotropic effects of diltiazem or verapamil. However, mibefradil has been demonstrated to block both HERG and KvLQT1/IsK currents in the range of 1–10 μmol/L, as does bepridil, 81 and the IC50 for mibefradil-induced T-type channel block versus delayed rectifier block are quite similar in human fusion-competent myoblasts. 86 The blockade exerted by bepridil or mibefradil did not seem to be either voltage dependent or accompanied by a significant change in the channel gating. 81 At the clinical level, mibefradil was not associated with QT prolongation but was associated with bradycardia and an increase in sinus node recovery time. 87,88 Another anti-anginal agent associated with QT prolongation, perhexiline, has been shown to inhibit heterologously expressed HERG channels. 89 The block by perhexiline is use and voltage dependent and a hyperpolarizing shift in the voltage dependence of steady state inactivation suggests that the agent may bind to the inactivated state of the channel.

Terodiline is a calcium channel antagonist originally used as an anti-anginal agent and later as an antispasmodic used to treat detrusor instability in urinary incontinence. However, it has been shown to be associated with torsade de pointes.90 When tested on isolated ventricular myocytes in the concentration range of 1–10 μmol/L, terodiline was shown to block IKr; the block was preferential to IKs at these concentrations. 91

An interesting Ca2+ channel-blocking agent, from the standpoint of the present article, is prenylamine. Although this agent is associated with torsade de pointes, 92 it appears to mediate its pro-arrhythmic effects via a calcium-dependent, rather than potassium-dependent, mechanism. 93 With rapid stimulation rates, prenylamine is associated with block of calcium channels but, at low stimulation rates, prenylamine acts as an agonist to calcium channels (possibly a stereoselective action mediated by the + isomer), which can lead to calcium channel reactivation and EAD. 92 Although unusual, this observation suggests that while acquired LQTS commonly correlates with IKr/HERG blocking actions, this may not always be the underlying basis for pro-arrhythmia.

Long QT syndrome and non-cardiac drugs

The preceding discussion demonstrates that many agents designed for use in cardiovascular pharmacology affect IKr/HERG and that this may be the most common mechanism underlying acquired LQTS. However, the syndrome is not restricted to this broad category of drugs. A number of non-cardiac drugs are also associated with acquired LQTS.

The non-sedating antihistamines, such as terfenadine and astemizole, carry a pro-arrhythmic risk in patients suffering from predisposing factors, such as ischaemic heart disease, congestive heart failure, congenital prolongation of the QT interval, electrolyte imbalance or reduced drug-metabolizing activity. 94,95 Also, they have been shown to affect the K+ currents associated with repolarization of the cardiac AP. 96,97 When tested on heterologously expressed HERG in Xenopus oocytes, terfenadine was found to block the current within the micromolar range, although there was disagreement as to whether the block was tonic or open state dependent. 98,99 Similar blocking effects were also obtained for astemizole 99 and ebastine. 100 However, the block of IK and its associated risk of arrhythmia may not be shared by the entire class of non-sedating antihistamines; other non-sedating antihistamines, such as loratidine 100 and cetirizine, 101 are not associated with arrhythmia or HERG inhibition when considered at the same concentrations as terfenadine and astemizole.

Overdose of antipsychotics, such as haloperidol, can be associated with ventricular arrhythmia 102 and heterologous HERG has been shown to be blocked by the antipsychotics haloperidol and sertindole when expressed in oocytes and cell lines, respectively. 103,104 Sertindole inhibits HERG with particular avidity, being more potent than dofetilide in the same experimental series and being 500-fold more potent than haloperidol. The tricyclic antidepressant imipramine has also been shown to be a blocker of IK105,106 and, recently, of HERG. 107 Such findings may be of particular relevance to cardiac side effects associated with tricyclic antidepressant overdose.

Pharmacological effects on HERG may also play a role in the toxicity of ketoconazole when used in conjunction with terfenadine. 108 Both drugs use the cytochrome P450 pathway for breakdown and combined drug use may result in the high concentrations associated with ketoconazole-induced direct block of the HERG channel. Another antimicrobial agent known to prolong the QT interval, the antibiotic erythromycin, has been shown to block IKr in M cells of isolated strips of canine cardiac muscle. 109

The HERG is sensitive to external K+22 and hypokalaemia is a risk factor for pro-arrhythmia. Thus, drugs that tend to produce hypokalaemia may carry some pro-arrhythmic risk, especially if used in combination with agents that themselves prolong the QT interval. This raises an important general issue, namely that of synergistic drug interactions. The diversity in drugs predisposing patients towards LQTS is marked and it is therefore important that combinations of drugs that share the potential to prolong the QT interval (e.g. patients receiving both erythromycin and terfenadine) be avoided wherever possible.

PHARMACEUTICAL DEVELOPMENT AND LQTS

Familial LQTS is thought to be more prevalent than measures suggest because it can go undiagnosed. It is to be expected that the increasing knowledge of the genetic basis for congenital LQTS variants 32 may help drive the development of both appropriate treatments and screening technology. The pharmaceutical agents causing acquired LQTS are more likely to be known because of rigorous screening. Even so, it is postulated that some agents will only cause complications in a subpopulation of sensitive patients 19 and recent examples 104 suggest that it remains difficult to detect absolutely all agents with potential problems before they reach the market. As the example with cetirizine shows, 101 it is possible to identify agents that do and do not affect HERG current and to correlate those data with clinical experience of arrhythmogenesis.

Given the pressures on drug development, the use of combinatorial chemistry and the use of high through-put screening, it seems that the use of whole animal tests on a large number of candidate drugs would be both time consuming and expensive. While it would be impossible to substitute a battery of in vitro cellular electrophysiological tests for whole animal tests on QT prolongation, investigating drug actions on either native IKr or HERG should be considered as important. Such a series of tests could be used as a rapid screen for channel sensitivity to the candidate compounds and this could be done earlier in the screening process because technological developments in electrophysiology make the process more amenable to high through-put drug screening. The diversity of the compounds that have been found to block IKr/HERG suggests that routine screening may be the only viable option at the present time, whether or not cellular electrophysiological experiments comprise the main line of screening or are combined with in vivo testing. 110

As more becomes known about the structure of the HERG-encoded channel and about the molecular nature of the channel block by the various drug classes, it may be feasible to implement improved rational drug design. One potentially attractive strategy would be to use compounds that are currently associated with LQTS as the basis of new chemical structures that preserve the clinical effects of the parent compound but are modified to avoid molecular interactions with the appropriate binding site on the channel for the parent compound.

NOTE ADDED IN PROOF

The authors regret that in a brief review it is not possible to cite all the work in this rapidly moving research area.

Acknowledgements

The authors acknowledge Terri Harding, Richard Helyer and Robert Meech (Bristol, UK) for critical help during the composition of this review. HJW acknowledges LTJ Bukem (London, UK) for helpful guidance. The authors also acknowledge the Massachusetts Medical Association and Lippincott, Williams & Wilkins for permission to use Figs 2 and 3. The authors gratefully acknowledge the British Heart Foundation (PG/98081) and the Wellcome Trust for funding and thank Dr Phil Dooley (La Trobe University, Victoria, Australia) for particularly useful comments on a draft of the manuscript.

Ancillary