7
Original article Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit Ingemar Jacobson, Leif Carlsson, Göran Duker AstraZeneca R&D Mölndal, Bioscience, Mölndal, Sweden abstract article info Article history: Received 4 March 2010 Accepted 23 April 2010 Keywords: Rabbit QT QT interval prolongation QT interval variability Torsades de pointes Introduction: Accumulating evidence suggest that drug-induced QT prolongation per se poorly predicts repolarisation-related proarrhythmia liability. We examined whether beat-by-beat variability of the QT interval may be a complementary proarrhythmia marker to QT prolongation. Methods: Anaesthetised rabbits sensitized towards developing torsades de pointes (TdP) were infused for 30 min maximum with explorative antiarrhythmic compounds characterised as mixed ion channel blockers. Based on the outcome in this model the compounds were classied as having a low (TdPlow; n = 5), intermediate (TdPintermedi- ate; n = 7) or high (TdPhigh; n = 10) proarrhythmic potential. Dofetilide (n =4) was included as a representative of a selective IKr-blocking antiarrhythmic with known high proarrhythmic potential. QT interval prolongation and beat-by-beat QT variability (quantied as the short-term variability, STV) were continuously assessed during the infusion or up to the point where ventricular proarrhythmias were induced. Results: All compounds signicantly prolonged the QT interval. For TdPlow and TdPhigh compounds the QT interval maximally increased from 169 ± 14 to 225 ± 28 ms (p b 0.05) and from 186 ± 21 to 268 ± 42 ms (p b 0.01), respectively. Likewise, in the dofetilide-infused rabbits the QT interval maximally increased from 177±11 to 243±25 ms (p b 0.01). In contrast, whereas the STV in rabbits administered the TdPhigh com- pounds or dofetilide signicantly increased prior to proarrhythmia induction (from 1.6 ± 0.4 to 10.5 ± 5.6 ms and from 1.6 ± 0.5 to 5.9 ± 1.8 ms, p b 0.01) it remained unaltered in the TdPlow group (1.3 ± 0.6 to 2.2 ± 0.9 ms). In the TdPintermediate group, rabbits experiencing TdP had a similar maximal QT prolongation as the non-susceptible rabbits whereas the change in the STV was signicantly different (from 0.9 ± 0.5 to 8.7 ± 7.3 ms vs 0.8 ± 0.3 to 2.5 ± 1.1 ms). Discussion: It is concluded from the present series of experiments in a sensitive rabbit model of TdP that increased beat-by-beat QT interval variability precedes drug-induced TdP. In addition, assessment of this potential proarrhythmia marker may be useful in discriminating highly proarrhythmic compounds from compounds with a low proarrhythmic potential. © 2010 Elsevier Inc. All rights reserved. 1. Introduction Accumulating evidence suggest that drug-induced prolongation of the QT interval per se poorly predicts a drug's propensity to induce repolarisation-related proarrhythmias such as torsades de pointes (TdP) (Carlsson, 2008; Shah & Hondeghem, 2005; Thomsen, Matz, Volders, & Vos, 2006). From a drug development perspective this implies that potentially benecial compounds may be discarded based on unrighteous grounds implying a need for alternative risk markers with improved predictive value. Repolarisation instability as mea- sured from e.g. beat-by-beat variability of the QT interval or the action potential duration has been put forward as such a marker (Hinterseer et al., 2008; Hondeghem, Carlsson, & Duker, 2001; Thomsen et al., 2004). However, additional nonclinical and clinical evidence in sup- port of the use of beat-by-beat variability of the QT interval to im- prove overall risk assessment of drug-induced TdP is required. The methoxamine-sensitized rabbit model of TdP is frequently used to assess the proarrhythmic potential of repolarisation-delaying agents (Carlsson, 2008). From several studies using this sensitive TdP model, it has been shown that mixed ion channel blockers such as H 345/52, BRL-32872, AZD7009, vernakalant, AZD1305 and ranolazine can sub- stantially prolong the QT interval without inducing TdP whereas compounds that selectively block the rapid delayed rectier potassi- um current (IKr) cause TdP at a high incidence (Amos et al., 2001; Bril et al., 1996; Carlsson, Andersson, Linhardt, & Löfberg, 2009; Orth et al., Journal of Pharmacological and Toxicological Methods 63 (2011) 4046 Corresponding author. AstraZeneca R&D, Bioscience, S-431 83 Mölndal, Sweden. Tel.: +46 31 7761000; fax: +46 31 7763766. E-mail address: [email protected] (G. Duker). 1056-8719/$ see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.vascn.2010.04.010 Contents lists available at ScienceDirect Journal of Pharmacological and Toxicological Methods journal homepage: www.elsevier.com/locate/jpharmtox

Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

Embed Size (px)

Citation preview

Page 1: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

Contents lists available at ScienceDirect

Journal of Pharmacological and Toxicological Methods

j ourna l homepage: www.e lsev ie r.com/ locate / jpharmtox

Original article

Beat-by-beat QT interval variability, but not QT prolongation per se, predictsdrug-induced torsades de pointes in the anaesthetisedmethoxamine-sensitized rabbit

Ingemar Jacobson, Leif Carlsson, Göran Duker ⁎AstraZeneca R&D Mölndal, Bioscience, Mölndal, Sweden

⁎ Corresponding author. AstraZeneca R&D, BiosciencTel.: +46 31 7761000; fax: +46 31 7763766.

E-mail address: [email protected] (G. D

1056-8719/$ – see front matter © 2010 Elsevier Inc. Aldoi:10.1016/j.vascn.2010.04.010

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 4 March 2010Accepted 23 April 2010

Keywords:RabbitQTQT interval prolongationQT interval variabilityTorsades de pointes

Introduction: Accumulating evidence suggest that drug-induced QT prolongation per se poorly predictsrepolarisation-related proarrhythmia liability. We examined whether beat-by-beat variability of the QTinterval may be a complementary proarrhythmia marker to QT prolongation. Methods: Anaesthetisedrabbits sensitized towards developing torsades de pointes (TdP) were infused for 30 min maximum withexplorative antiarrhythmic compounds characterised as mixed ion channel blockers. Based on the outcomein this model the compounds were classified as having a low (TdPlow; n=5), intermediate (TdPintermedi-ate; n=7) or high (TdPhigh; n=10) proarrhythmic potential. Dofetilide (n=4) was included as arepresentative of a selective IKr-blocking antiarrhythmic with known high proarrhythmic potential. QTinterval prolongation and beat-by-beat QT variability (quantified as the short-term variability, STV) were

continuously assessed during the infusion or up to the point where ventricular proarrhythmias wereinduced. Results: All compounds significantly prolonged the QT interval. For TdPlow and TdPhigh compoundstheQT intervalmaximally increased from169±14 to 225±28 ms (pb0.05) and from186±21 to 268±42 ms(pb0.01), respectively. Likewise, in the dofetilide-infused rabbits the QT interval maximally increased from177±11 to 243±25 ms (pb0.01). In contrast, whereas the STV in rabbits administered the TdPhigh com-pounds or dofetilide significantly increased prior to proarrhythmia induction (from 1.6±0.4 to 10.5±5.6 msand from 1.6±0.5 to 5.9±1.8 ms, pb0.01) it remained unaltered in the TdPlow group (1.3±0.6 to 2.2±0.9 ms). In the TdPintermediate group, rabbits experiencing TdP had a similar maximal QT prolongation as thenon-susceptible rabbits whereas the change in the STV was significantly different (from 0.9±0.5 to 8.7±7.3 ms vs 0.8±0.3 to 2.5±1.1 ms). Discussion: It is concluded from the present series of experiments in asensitive rabbitmodel of TdP that increased beat-by-beat QT interval variability precedes drug-induced TdP. Inaddition, assessment of this potential proarrhythmia marker may be useful in discriminating highlyproarrhythmic compounds from compounds with a low proarrhythmic potential.

© 2010 Elsevier Inc. All rights reserved.

1. Introduction

Accumulating evidence suggest that drug-induced prolongation ofthe QT interval per se poorly predicts a drug's propensity to inducerepolarisation-related proarrhythmias such as torsades de pointes(TdP) (Carlsson, 2008; Shah & Hondeghem, 2005; Thomsen, Matz,Volders, & Vos, 2006). From a drug development perspective thisimplies that potentially beneficial compoundsmay be discarded basedon unrighteous grounds implying a need for alternative risk markerswith improved predictive value. Repolarisation instability as mea-

e, S-431 83 Mölndal, Sweden.

uker).

l rights reserved.

sured from e.g. beat-by-beat variability of the QT interval or the actionpotential duration has been put forward as such a marker (Hinterseeret al., 2008; Hondeghem, Carlsson, & Duker, 2001; Thomsen et al.,2004). However, additional nonclinical and clinical evidence in sup-port of the use of beat-by-beat variability of the QT interval to im-prove overall risk assessment of drug-induced TdP is required. Themethoxamine-sensitized rabbit model of TdP is frequently used toassess the proarrhythmic potential of repolarisation-delaying agents(Carlsson, 2008). From several studies using this sensitive TdP model,it has been shown that mixed ion channel blockers such as H 345/52,BRL-32872, AZD7009, vernakalant, AZD1305 and ranolazine can sub-stantially prolong the QT interval without inducing TdP whereascompounds that selectively block the rapid delayed rectifier potassi-um current (IKr) cause TdP at a high incidence (Amos et al., 2001; Brilet al., 1996; Carlsson, Andersson, Linhardt, & Löfberg, 2009; Orth et al.,

Page 2: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

Table 1Sequential testing of drug-induced torsades de pointes liability in the anaesthetisedrabbit.

Sequential experiment no Accept if X≤ Reject if X≥

1 – –

2 – 23–6 – 37 – 48–10 0 4

X=accumulated number of rabbits experiencing torsades de pointes in the series.

41I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

2006; Wang, Robertson, Dhalla, & Belardinelli, 2008; Wu, Carlsson,Liu, Yan, & Kowey, 2005). The objective of the present series ofexperiments was to collect evidence that beat-by-beat QT variabilitywould be a superior risk marker for drug-induced proarrhythmialiability as compared to QT lengthening per se. Hence, QT prolongationand beat-by-beat QT interval variability for a number of investiga-tional compounds with known ion channel blocking characteristicsand repolarisation-delaying potency and varying degrees of proar-rhythmic potential in vivo were compared in the methoxamine-sensitized rabbit model of TdP.

2. Material and methods

All animal studies were approved by the ethical committee foranimal research at the University of Göteborg, Sweden and was con-ducted in accordance with Swedish animal care guidelines.

2.1. Estimation of proarrhythmic potential in the anaesthetisedmethoxamine-sensitized rabbit

To enable assessment and comparison of proarrhythmia liability ofthe investigational compounds in the methoxamine-sensitized rabbitat equipotent repolarisation-delaying doses, all compounds wereinitially studied in anaesthetised guinea pigs in which the potency interms of prolonging themonophasic action potential duration for eachcompound was determined as described elsewhere (Carlsson, 2008;Carlsson et al., 1997). Based on the repolarisation-delaying potencyobserved in the guinea pig, a dosing regimen (infusion rate and max-imal cumulative dose) to be used in the rabbits was defined. Theinfusion regimen adopted assumed a similar repolarisation-delayingpotency in guinea pigs and rabbits, a relationship that previously hasbeen demonstrated for a number of different compounds (Carlsson,2008).

2.2. Animal preparation

Male New Zealand White rabbits (2–3 kg, n=39) were includedin the study. The animals were anaesthetised with methohexitalsodium (5 mg/kg i.v.) and α-chloralose (90 mg/kg, infused through amarginal ear vein for 20 min). Subsequently the animals were pre-pared and instrumented as previously described in detail (Carlsson,Almgren, & Duker, 1990; Carlsson et al., 2009; Wu et al., 2005). Bloodgases in arterial blood were estimated by means of a blood gas andelectrolyte analyser (ABL505, Radiometer, Copenhagen, Denmark).Blood gases were kept within physiological limits for the species.

2.3. Experimental protocol

After baseline measurements, a continuous infusion of methox-amine (70 nmol/kg/min) was commenced. Ten minutes later the testcompound was administered as a continuous intravenous infusion for30 minmaximum. ECGswere continuouslymonitored and sampledona computer. The QT interval and the beat-by-beat QT interval vari-ability were measured using an in house interactive ECG analysisprogram. Starting approximately 2.5 min before the commencementof the infusion of the investigational drug and until appearance ofventricular extrasystoles and TdP rendering the ECG analysis unreli-able or for the entire drug infusion period (i.e., 30 min) if TdP was notinduced, the QT interval for every single beat was measured. Hence,in a typical experiment in which TdP was not induced, approximately8000 consecutive ECG complexes were analyzed.

2.4. Investigational compounds

The investigational compounds assessed in the present studyemanated from the AstraZeneca antiarrhythmic project and were all

structurally related to AZD7009, a drug developed for management ofatrial fibrillation and characterised as a mixed ion channel blocker(Geller et al., 2009; Persson, Carlsson, & Duker, 2005;Wu et al., 2005).In addition to blocking IKr, the investigational compounds also blockthe sodium current and the L-type calcium current at varying degrees.The potent and selective IKr blocker dofetilide, known to possess ahigh proarrhythmic potential in various experimental proarrhythmiamodels and in the clinical setting, was included as a proarrhythmicreference test compound (Thomsen et al., 2006; Wu et al., 2005).

Following assessment of the TdP inducibility the investigationalcompounds were separated into two different groups, one defined ashaving a low proarrhythmic potential (TdPlow, 5 compounds), one ashaving a high proarrhythmic potential (TdPhigh, 10 compounds) asjudged on the following criteria and outcome during the assessment.Some of the compounds were tested in more than one individualexperiment. Thus, the TdPlow group includes data from 8 experimentsusing 5 different compounds. The TdPhigh group includes 13 individualexperiments, using 10 different compounds. To categorise the testcompounds based on proarrhythmic potential, a sequential screeningprocedure was adopted. Thus, depending on the outcome of eachexperiment one of two decisions was taken: classify the compound as acompound having a low proarrhythmic potential (i.e., a true incidenceof TdP of less than 10%) or classify the compound as highlyproarrhythmic (i.e., a true TdP incidence exceeding 40%). The sequentialscreening planwas derived from a statistical simulation in the followingway (assuming each compound is associatedwith a certain risk of TdP):

(i) Consider sequential plans for which the probability of accept-ing a compound with a risk of developing TdP of 10% is at least95%, and the probability of rejecting a compound with a risk ofdeveloping TdP of 40% is at least 95%.

(ii) From all sequential plans fulfilling the above requirementschoose the one with the lowest expected number of experi-ments needed for reaching a decision about the compound.

Based on these assumptions and requirements the screening plandepicted in Table 1 was chosen.

In conclusion, compounds to be included in the group judged ashaving a low proarrhythmic potential were studied in at least8 sequential experiments without any rabbit experiencing TdP. Onthe contrary, compounds to be included in the group judged as havinga high proarrhythmic potential were selected from compounds thatinduced TdP in the first two sequentially studied rabbits. Furthermore,in the present study a third group (n=14) including compounds(n=7) with 50% TdP (TdPintermed) incidence as based on at leasttwo sequential experiments, one inwhich TdPwas induced and one inwhich TdP was not induced, was included in the analysis of QTprolongation and beat-by-beat QT interval variability.

2.5. Estimation of hERG-blocking potency and calcium and sodiumchannel binding affinity

Effects on hERG activity were determined by measuring rubidiumefflux fromHEK-292 cells stably expressinghERGchannels (Chenget al.,2002; Tang et al., 2001). Rubidium efflux was determined by atomic

Page 3: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

42 I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

absorption spectroscopy. Data were expressed as mean percentageinhibition andwere fittedwith sigmoidal curves to estimate IC50 values.

Estimates of affinity to Na channels (site 2) and L-type Ca channels(verapamil binding site) were obtained by using rat cerebral cortexmembranes and standard radioligand binding assays (Brown, 1986;Reynolds, Snowman, & Snyder, 1986). The specific ligand binding wasdefined as the difference between total binding and the nonspecificbinding determined in the presence of an excess of unlabelled ligand.For Na channels 3H-batrachotoxin-A-benzoate was used as the ligandand for L-type Ca channels (−)[3H]desmethoxyverapamil was used asthe ligand.

Data was expressed as percentage of control specific binding in thepresence of test compound and was fitted with sigmoidal curves toestimate IC50 values. Inhibition constants (Ki) were then calculatedusing the Cheng Prusoff equation, Ki=IC50/(1+L/KD) where L is theconcentration of radioligand in the assay and KD is the affinity of theradioligand to the receptor.

2.6. Measurement of the QT interval and the beat-by-beat QT intervalvariability

The QT interval was defined as the time between the first devia-tion from an isoelectric PR interval until the return of the ventricular

Fig. 1. (Upper panel) Representative examples of single beat QT intervals plotted as afunction of corresponding beat number in two methoxamine-sensitized anaesthetisedrabbit administered a compound not inducing torsades de pointes (TdPlow, blackcurve) or a compound inducing torsades de pointes (TdPhigh, red curve), respectively.Arrow indicates start of drug infusion. The QT interval was defined as the time betweenfirst deviation from an isoelectric PR interval until the return of the ventricularrepolarisation to the isoelectric TP baseline, and, as such, included a “U”wave if present.(Lower panel) QT interval before commencement of drug infusion (baseline) andduring infusion (on drug) with dofetilide (n=4) or investigational compounds withlow or high proarrhythmic potential (TdPlow; n=8 and TdPhigh; n=13), respectively,in the methoxamine-sensitized rabbit. Shown are mean±SD, *, **; pb0.05 and 0.01 vsbaseline, †; pb0.05, TdPhigh vs TdPlow. For further details, see Material and methods.

repolarisation to the isoelectric TP baseline and, if present, included aU wave. To assess the beat-by-beat QT interval variability, nextinterval or Poincaré plots were constructed by plotting the QT intervalagainst the preceding QT interval. The mean orthogonal distance (30beat moving average) from the diagonal to the points of the Poincaréplot was determined and referred to as short-term variability(STV=Σ|Dn+1−Dn|/[30⁎√2], where D represents the duration ofthe QT interval) (Thomsen et al., 2004). Baseline QT intervals and themaximal prolongation of the QT interval were determined from 30beat moving averages. The maximal drug-induced QT intervalprolongation and beat-by-beat QT interval variability change weredefined as the maximal effect observed before the occurrence of thefirst ventricular extrasystole or, as in the case of the TdPlow com-pounds, during the 30 min period of drug infusion. Only sinus beatsup to the appearance of the first ventricular extrasystole (if induced)were included in the analysis.

2.7. Statistics

Comparisons between groups were performed using a t-test or,when appropriate, a two-way ANOVA followed by a Bonferroni t-test.Values are given as mean±SD if not otherwise indicated.

Fig. 2. (Upper panel) Representative examples (same as in Fig. 1) of beat-by-beat QTinterval variability (STV) plotted as a function of corresponding beat number in twomethoxamine-sensitized anaesthetised rabbits administered a compound not inducingtorsades de pointes (TdPlow, black curve) or a compound inducing torsades de pointes(TdPhigh, red curve), respectively. Arrow indicates start of drug infusion. STV wasdefined as themean orthogonal distance (30 beatmoving average) from the diagonal tothe points of the Poincaré plot (STV=Σ|Dn+1−Dn|/[30⁎√2] where D represents theduration of the QT interval). (Lower panel) Beat-by-beat QT interval variability (STV)before commencement of drug infusion (baseline) and during infusion (on drug) withdofetilide or investigational compounds with low or high proarrhythmic potential(TdPlow; n=8 and TdPhigh; n=13), respectively, in the methoxamine-sensitizedrabbit. Shown are mean±SD, **; pb0.01 vs baseline, ††; pb0.01 TdPhigh vs TdPlow. Forfurther details, see Material and methods.

Page 4: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

43I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

3. Results

3.1. Effects of mixed ion channel blockers with high and lowproarrhythmic potential on QT interval prolongation and beat-by-beatQT interval variability

Following the start of the drug infusion, the QT interval markedlyincreased in rabbits administered proarrhythmic as well as nonproar-rhythmic compounds (Fig. 1). However, whereas the increase in theQT interval in the TdPlow group reached a plateau, no such flatteningwas observed in the TdPhigh group of rabbits. In contrast, the QTinterval steadily increased up to a point, usually within a few minutesof infusion, when ventricular extrasystoles and eventually TdP wereinduced (Fig. 1). During the drug infusion the QT interval maximallyincreased from 169±14 to 225±28 (n=8) ms and from 186±21 to268±42 ms (n=13) the TdPlow and TdPhigh groups, respectively(pb0.05 and 0.01 vs pre-drug, pb0.05 vs groups, Fig. 1).

Despite prominent increases in the QT interval, no marked changesin the beat-by-beat QT interval variability were evident during druginfusion in the TdPlow group (Fig. 2). Furthermore, overlay plots ofsingle beats aswell as Poincaré plots of the QT interval indicated a stableincrease of repolarisation delay (Fig. 3). In the TdPlow group (n=8) themaximal STV was 2.2±0.9 ms, a value not significantly different fromthe value obtained before commencement of drug infusion (1.3±0.6 ms, pN0.05, Fig. 2 see above). In contrast, in the TdPhigh group(n=13) the drug-induced QT prolongation was accompanied by a

Fig. 3. (Upper panels) Representative examples of overlay plots of single beat ECG traces obred traces). The upper left recordings stem from a rabbit infusedwith a compound not inducincompound inducing TdP (same examples as in Figs. 1 and 2). (Lower panels) Next interval orthe upper panel. The plots were constructed by plotting the QT interval n against the subsequthe compound not inducing TdP and the right panel the effects in the rabbit administered

marked increase in the beat-by-beat QT interval variability as evidentfrom the STV (from 1.6±0.4 ms to 10.5±5.6 ms, pb0.01 vs pre-drugand TdPlow) and the ECG overlay and Poincaré plots (Figs. 2 and 3).

As shown in Fig. 4, there was generally a rather poor correlationbetween the degree of the maximal change of the QT interval and themaximal change in STV. For TdPlow compounds STV did not exceed3.5 ms despite the fact that QT changed in the interval 30–110 ms.Likewise, changes in STV for TdPhigh compounds did not show astrong correlation to changes in QT, e.g. similar STV changes could beobserved both at moderate (60 ms) and pronounced (160 ms) QTprolongations.

3.2. Dofetilide-induced QT interval prolongation and beat-by-beat QTinterval variability

For comparative reasons the effects of a continuous infusion ofdofetilide (n=4) on the QT interval and the STV are also presented inFig. 1. The baseline (i.e., pre-drug) QT interval and STV in the group ofrabbits destined to dofetilide infusion were 177±11 ms and 1.6±0.5 ms, respectively. These values were very similar to the baselineQT interval and the STV of the low and the high TdP incidence groupsdestined to the test compound infusions (Fig. 1). Dofetilide maximallyprolonged the QT interval to 243±25 ms (pb0.005 vs baseline) andincreased the STV to 5.9±1.8 (pb0.005; Fig. 2) before inducing TdP inall rabbits.

tained during drug-free conditions (Pre-drug, black traces) and drug infusion (On drug,g torsades de pointes (TdP) and the upper right recordings from a rabbit administered aPoincaré plots obtained from analysis of single beat QT interval durations from ECGs inent QT interval, n+1. The left panel shows the effects in the rabbit receiving infusion ofthe compound inducing TdP, respectively.

Page 5: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

Fig. 4. Maximal change in QT interval plotted as a function of maximal change in beat-by-beat QT interval variability (STV) in methoxamine-sensitized anaesthetised rabbitsadministered compounds inducing torsades de pointes (TdP, filled squares) or notinducing TdP (No TdP, open circles).

Table 2hERG-blocking potency and calcium and sodium channel binding affinity.

logIC50/pKi (M)

IKr (hERG) ICaL INa

TdPlow (n=5) 5.8±0.37 6.0±0.47 5.8±0.47TdPintermed (n=7) 5.5±0.40 5.3±0.64 5.6±0.51TdPhigh (n=10) 5.5±0.44 5.4±0.40 5.1±0.40⁎

⁎ pb0.05 vs TdPlow compounds.

44 I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

3.3. Effects of mixed ion channel blockers with intermediate proarrhythmicpotential on QT interval prolongation and beat-by-beat QT intervalvariability

In rabbits (n=14) administered 7 different compounds giving riseto intermediate TdP incidence (i.e., 50%), the baseline QT interval andthe STV did not significantly differ between rabbits experiencing TdPand rabbits free from TdP during the infusion (172±13ms and 163±9 ms and 0.9±0.5 and 0.8±0.3 ms, respectively, Fig. 5). Although notstatistically significant, the drug infusion was associated with asomewhat more pronounced QT interval prolongation in rabbitsdeveloping TdP than in the group of rabbits completing the studyprotocol (i.e., 30 min infusion) without experiencing TdP. Hence, in theTdP susceptible rabbits the QT interval was prolonged to 262±48msand in the non-susceptible rabbits to 234±27ms, respectively (Fig. 5).In contrast, the STV discriminated rabbits experiencing TdP from rabbitsfree of TdP. Thus, in the 7 TdP rabbits the STV increased to 8.7±7.3 ms(pb0.001 vs baseline) as compared to 2.5±1.1 ms (pN0.05) in the 7rabbits without TdP induction (Fig. 5).

3.4. hERG-blocking potency and calcium and sodium channel bindingaffinity

Although the TdPlow compounds generally were somewhat lesspotent as hERG blockers and showed somewhat higher affinity forICaL and INa channels, the only statistically significant differenceobtainedwas for INa affinity.MeanpKi for Na affinitywas 5.8±0.47 µM

Fig. 5. QT interval (left panel) and beat-by-beat QT interval variability (STV, right panel)intermediate proarrhythmic potential in methoxamine-sensitized rabbits. The figure illustraexperiencing TdP (TdP) during drug infusion. Shown are mean±SD, *, **; pb0.05 and 0.01

and 5.1±0.41 µM for TdPlow and TdPhigh compounds, respectively(pb0.05, Table 2).

4. Discussion

Drug-induced proarrhythmia is of major concern in the develop-ment of new therapeutic agents. A substantial number of drugcandidates under development or drugs already on the market havebeen associated with TdP, a potentially life-threatening polymorphicventricular tachycardia occasionally occurring secondary to a delay ofventricular repolarisation (Redfern et al., 2003). Such a drug-inducedrepolarisation delay is almost always a consequence of blockade of thehuman ether-a-go-go-related gene (hERG) channel, the molecularcorrelate of IKr (Redfern et al., 2003). The block results in a delay ofthe final phase of the repolarisation of the ventricular action potential,revealed as a prolongation of the QT interval in the surface ECG. Thus,drug-induced prolongation of the QT interval has been considered apredictor for risk of TdP, an assumption that has been challenged(Shah & Hondeghem, 2005; Thomsen et al., 2006).

The present study using a sensitive animal model of TdP showsthat drug-induced prolongation of the QT interval does not inevitablyresult in TdP. In contrast, an increase in the beat-by-beat variabilityof the QT interval was more closely associated with the induction ofTdP than the magnitude of the QT prolongation. In contrast to otheranimal studies in which snapshots of ECG or monophasic action po-tential recordings were analyzed for drug-induced repolarisationvariability, we measured the beat-by-beat variability of the QT intervalcontinuously throughout the experiment (Lengyel, Varró, Tábori, Papp,&Baczkó, 2007;Michael, Dempster, Kane, &Coker, 2007; Thomsenet al.,2004; Vincze et al., 2008). This discrepancymay be a reasonwhy othersusing theα-adrenoceptor stimulated rabbit model of TdP have failed topredict drug-induced TdP through beat-by-beat QT interval assess-ments (Michael et al., 2007; Vincze et al., 2008). Furthermore, ouranalysis of beat-by-beat QT interval variability did not include non sinusbeats. However, the present observations are in line with the originalfindings by Hondeghem and colleagues in the AV-blocked Langendorff-perfused rabbit heart showing that instability of the action potential

before infusion (Pre-drug) and during infusion with investigational compounds withtes the difference in QT intervals and STV between rabbits not experiencing (No TdP) orvs Pre-drug. For further details, see Material and methods.

Page 6: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

45I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

duration strongly correlates to drug-induced proarrhythmia liability(Hondeghemet al., 2001). Subsequently, several other studies assessingdrug-induced proarrhythmia in Langendorff-perfused rabbit hearts, inanaesthetised rabbits, in anaesthetised dogs with chronic AV-block andin conscious dogs have come to similar conclusions (Carlsson et al.,2009; Hanton, Yvon, & Racaud, 2008; Hondeghem & Hoffmann, 2003;Thomsen et al., 2004; Thomsen et al., 2006). Furthermore, availableclinical data suggest a relationship between repolarisation liability andTdP. Thus, Svernhage et al. showed that in 4 patients experiencing TdPfollowing administration of the IKr-blocking antiarrhythmic agentalmokalant, the proarrhythmia was preceded by a pronounced increaseof the beat-by-beat QT interval variability not seen in matched-controlcases without TdP (Svernhage et al., 1998). In a recent study, beat-by-beat QT interval variability was demonstrated to discriminate patientswith a history of drug-induced TdP from matched-control patients(Hinterseer et al., 2008). The variability was suggested to have asuperior proarrhythmic predictive power to that of the QTc interval andwas put forward as a useful non-invasive marker for patients at risk fordrug-related TdP.

A factor thatmay reduce the proarrhythmic potential of IKr-blockingdrugs is mixed ion channel blocking activity. Thus, even if a drugpotently blocks the IKr channel, ancillary features such as blockade ofinwardly directed depolarising sodium and/or calcium currents maycounteract excessive repolarisationdelay and triggeringof TdP (Amos etal., 2001; Andersson, Abi-Gerges, & Carlsson, 2010; Bril et al., 1996; Orthet al., 2006; Wu et al., 2005). Amiodarone, an effective antiarrhythmicagent and a prototype mixed ion channel blocker with high potency toblock IKr profoundly prolongs the QT interval but yet rarely causes TdP.Likewise, verapamil potently blocks IKr without inducing TdP, mostlikely a result of its additional L-type calcium channel blockingcharacteristics (Redfern et al., 2003). Yet another example is ranolazine,a novel antianginal agent blocking IKr and prolongs the QT interval butthat has not been associated with TdP (Scirica et al., 2007). The presentfinding that compounds which can be characterised as mixed ionchannel blockers (Geller et al., 2009; Persson et al., 2005; Wu et al.,2005) differ in proarrhythmic potential is likely due to varying degreesof block of IKr in relation to block of the sodium current and the L-typecalcium current. Ion channel blocking potency and affinity data fromthe present study identified a higher sodiumchannel affinity for TdPlowas compared to TdPhigh compounds. Assuming that this affinity corre-sponds also to blocking potency, the present data support earlier find-ings that sodium channel block is attenuating IKr related proarrhythmia(Carlsson, Drews, Duker, & Schiller-Linhardt, 1993; Carlsson et al., 2009;Orth et al., 2006), possibly bymitigating excessive QT prolongation. It isconcluded from the present series of experiments in a sensitive rabbitmodel of TdP that increased beat-by-beat QT interval variabilityprecedes drug-induced TdP, as opposed to what was seen at similarincreases of the QT interval duration induced by TdPlow agents. Inaddition, assessment of this potential proarrhythmia marker may beuseful in discriminating highly proarrhythmic compounds fromcompounds with a low proarrhythmic potential.

Acknowledgements

The authors would like to thank Lena Löfberg, Gunilla Linhardt andBirgit Andersson for invaluable technical assistance.

References

Amos, G. J., Abrahamsson, C., Duker, G., Hondeghem, L., Palmer, M., & Carlsson, L.(2001). Potassium and calcium current blocking properties of the novelantiarrhythmic agent H 345/52 — Implications for proarrhythmic potential.Cardiovascular Research, 49, 351−360.

Andersson, B., Abi-Gerges, N., & Carlsson, L. (2010). The combined ion channel blockerAZD1305 attenuates late Na current and IKr-induced action potential prolongationand repolarization instability. Europace, 12, 1003−1010.

Bril, A., Gout, B., Bonhomme, M., Landais, L., Fiavre, J. F., Linee, P., et al. (1996). Combinedpotassium and calcium channel blocking activities as a basis for antiarrhythmicefficacy with low proarrhythmic risk: Experimental profile of BRL-32872. Journal ofPharmacology and Experimental Therapeutics, 276, 637−646.

Brown, G. B. (1986). 3H-batrachotoxin-A-benzoate binding to voltage-sensitive sodiumchannels: Inhibition by the channel blockers tetrodotoxin and saxitoxin. Journal ofNeuroscience, 6, 2064−2070.

Carlsson, L. (2008). The anaesthetised methoxamine-sensitised rabbit model oftorsades de pointes. Pharmacology & Therapeutics, 119, 160−167.

Carlsson, L., Almgren, O., & Duker, G. (1990). QTU-prolongation and torsades de pointesinduced by putative class III antiarrhythmic agents in the rabbit — Etiology andinterventions. Journal of Cardiovascular Pharmacology, 16, 276−285.

Carlsson, L., Amos, G., Andersson, B., Drews, L., Duker, G., & Wadstedt, G. (1997).Electrophysiological characterization of the prokinetic agents cisapride andmosapride — Implications for proarrhythmic potential? Journal of Pharmacologyand Experimental Therapeutics, 282, 220−227.

Carlsson, L., Andersson, B., Linhardt, G. S., & Löfberg, L. (2009). Assessment of the ionchannel-blocking profile of the novel combined ion channel blocker AZD1305 andits proarrhythmic potential vs dofetilide in the methoxamine-sensitised rabbit invivo. Journal of Cardiovascular Pharmacology, 54, 82−89.

Carlsson, L., Drews, L., Duker, G., & Schiller-Linhardt, G. (1993). Attenuation ofproarrhythmias related to delayed repolarization by low-dose lidocaine in theanaesthetised rabbit. Journal of Pharmacology and Experimental Therapeutics, 267,1076−1080.

Cheng, C. S., Alderman, D., Kwash, J., Dessaint, J., Patel, R., Lescoe, M. K., et al. (2002). Ahigh throughput hERG potassium channel function assay: An old assay with a newlook. Drug Development and Industrial Pharmacology, 28, 177−191.

Geller, J. C., Egstrup, K., Kulakowski, P., Rosenqvist, M., Aunes Jansson, M., Berggren, A.,et al. (2009). Rapid conversion of persistent atrial fibrillation to sinus rhythm byintravenous AZD7009. Journal of Clinical Pharmacology, 49, 312−322.

Hanton, G., Yvon, A., & Racaud, A. (2008). Temporal variability of QT interval andchanges in T wave morphology in dogs as markers of the clinical risk of drug-induced proarrhythmia. Journal of Pharmacological and Toxicological Methods, 57,194−201.

Hinterseer, M., Thomsen, M. B., Beckmann, B., Pfeufer, A., Schimpf, R., Wichmann, H. E.,et al. (2008). Beat-to-beat variability of QT intervals is increased in patients withdrug-induced long-QT syndrome: A case control pilot study. European HeartJournal, 29, 185−190.

Hondeghem, L., Carlsson, L., & Duker, G. (2001). Instability and triangulation of theaction potential predict serious proarrhythmia, but APD prolongation is antiar-rhythmic. Circulation, 103, 2004−2013.

Hondeghem, L. M., & Hoffmann, P. (2003). Blinded test in isolated female rabbit heartreliably identifies action potential duration prolongation and proarrhythmic drugs:Importance of triangulation, reverse use dependence, and instability. Journal ofCardiovascular Pharmacology, 41, 14−24.

Lengyel, C., Varró, A., Tábori, K., Papp, J. G., & Baczkó, I. (2007). Combinedpharmacological block of IKr and IKs increases short-term QT interval variabilityand provokes torsades de pointes. British Journal of Pharmacology, 151, 941−951.

Michael, G., Dempster, J., Kane, K. A., & Coker, S. J. (2007). Potentiation of E-4031-induced torsade de pointes by HMR1556 or ATX-II is not predicted by actionpotential short-term variability or triangulation. British Journal of Pharmacology,152, 1215−1227.

Orth, P. M. R., Hesketh, J. C., Mak, C. K. H., Yang, Y., Lin, S., Beatch, G. N., et al.(2006). RSD1235 blocks late INa and suppresses early afterdepolarizationsand torsades de pointes induced by class III agents. Cardiovascular Research,70, 486−496.

Persson, F., Carlsson, L., & Duker, G. (2005). Blocking characteristics of hERG, hNav1.5and hKvLQT1/hminK after administration of the novel antiarrhythmic compoundAZD7009. Journal of Cardiovascular Electrophysiology, 16, 329−341.

Redfern, W. S., Carlsson, L., Davis, A. S., Lynch, W. G., MacKenzie, I., Palethorpe, S., et al.(2003). Relationships between preclinical cardiac electrophysiology, clinical QTinterval prolongation and torsade de pointes for a broad range of drugs: Evidencefor a provisional safety margin in drug development. Cardiovascular Research, 58,32−45.

Reynolds, I. J., Snowman, A. M., & Snyder, S. H. (1986). (−)[3 H)desmethoxy-verapamillabels multiple calcium channel modulators in brain and skeletal musclemembranes: Differentiation by temperature and dihydropyridine. Journal ofPharmacology and Experimental Therapeutics, 237, 731−738.

Scirica, B. M., Morrow, D. A., Hod, H., Murphy, S. A., Belardinelli, L., Hedgepeth, C. M.,et al. (2007). Effect of ranolazine, an antianginal agent with novel electrophysi-ological properties, on the incidence of arrhythmias in patients with non-ST-segment-elevation acute coronary syndrome: Results from theMetabolic Efficiencywith Ranolazine for Less Ischemia in Non-ST-Elevation Acute Coronary Syndrome— Thrombolysis in Myocardial Infarction 36 (MERLIN-TIMI 36) RandomizedControlled Trial. Circulation, 116, 1647−1652.

Shah, R. R., & Hondeghem, L. M. (2005). Refining detection of drug-inducedproarrhythmia: QT interval and TRIaD. Heart Rhythm, 2, 758−772.

Svernhage, E., Houltz, B., Blomström, P., Brachmann, J., Crijns, H., Jensen, S., et al. (1998).Early electrocardiographic signs of drug-induced torsades de pointes. Annals ofNoninvasive Electrophysiology, 3, 252−260.

Tang, W., Kang, J., Wu, X., Rampe, D., Wang, L., Shen, H., et al. (2001). Development andevaluation of high throughput functional assay methods for hERG potassiumchannel. Journal of Biomolecular Screening, 6, 325−331.

Thomsen,M. B., Matz, J., Volders, P. G. A., & Vos,M. A. (2006). Assessing the proarrhythmicpotential of drugs: Current status of models and surrogate parameters of torsades depointes arrhythmias. Pharmacology & Therapeutics, 112, 150−170.

Page 7: Beat-by-beat QT interval variability, but not QT prolongation per se, predicts drug-induced torsades de pointes in the anaesthetised methoxamine-sensitized rabbit

46 I. Jacobson et al. / Journal of Pharmacological and Toxicological Methods 63 (2011) 40–46

Thomsen, M. B., Verduyn, S. C., Stengl, M., Beekman, J. D. M., de Pater, G., van Opstal, J.,et al. (2004). Increased short-term variability of repolarization predicts d-sotalol-induced torsades de pointes in dogs. Circulation, 10, 2453−2459.

Vincze, D., Farkas, A. S., Rudas, L., Makra, P., Csík, N., Leprán, I., et al. (2008). Relevance ofanaesthesia for dofetilide-induced torsades de pointes in α-adrenoceptor-stimulated rabbits. British Journal of Pharmacology, 153, 75−89.

Wang, W. Q., Robertson, C., Dhalla, A. K., & Belardinelli, L. (2008). Antitorsadogeniceffects of (±)-N-(2, 6-dimethyl-phenyl)-(4[2-hydroxy-3-(2-methoxyphenoxy)

propyl]-1-piperazine (ranolazine) in anesthetized rabbits. Journal of Pharmacologyand Experimental Therapeutics, 325, 875−881.

Wu, Y., Carlsson, L., Liu, T., Yan, G. X., & Kowey, P. R. (2005). Assessment of theproarrhythmic potential of the novel antiarrhythmic agent AZD7009 and dofetilide inexperimental models of torsades de pointes. Journal of Cardiovascular Electrophysiol-ogy, 16, 898−904.