Differentiating Drug-Induced Multichannel Block on the Electrocardiogram: Randomized Study of Dofetilide, Quinidine, Ranolazine, and Verapamil

Transcription

1 nature publishing group Clinial Trial Differentiating Drug-Indued Multihannel Blok on the Eletroardiogram: Randomized Study of Dofetilide, Quinidine, Ranolazine, and Verapamil L Johannesen 1,2, J Viente 1,3, JW Mason 4, C Sanabria 4, K Waite-Labott 4, M Hong 5, P Guo 5, J Lin 5, JS Sørensen 6, L Galeotti 1, J Florian 6, M Ugander 1,2, N Stokbridge 7 and DG Strauss 1,2 Blok of the herg potassium hannel and prolongation of the QT interval are preditors of drug-indued torsade de pointes. However, drugs that blok the herg potassium hannel may also blok other hannels that mitigate torsade risk. We hypothesized that the eletroardiogram an differentiate the effets of multihannel drug blok by separate analysis of early repolarization (global J ) and late repolarization (global ). In this prospetive randomized ontrolled linial trial, 22 subjets reeived a pure herg potassium hannel bloker (dofetilide) and three drugs that blok herg and either alium or late sodium urrents (quinidine, ranolazine, and verapamil). The results show that herg potassium hannel blok equally prolongs early and late repolarization, whereas additional inward urrent blok (alium or late sodium) preferentially shortens early repolarization. Charaterization of multihannel drug effets on human ardia repolarization is possible and may improve the utility of the eletroardiogram in the assessment of drug-related ardia eletrophysiology. Fourteen drugs have been removed from the market worldwide beause they inrease the risk for torsade de pointes, 1 a ventriular arrhythmia that an ause sudden ardia death. Drugs that inrease the risk for torsade an be identified by assessing whether they blok the human ether-à-go-go-related gene (herg) potassium hannel (an outward urrent) and prolong the QT interval on the eletroardiogram (ECG). 1,2 In response, the US Food and Drug Administration (FDA) requires almost all new drugs to undergo thorough QT studies. 3 A positive thorough QT study does not prelude a drug from regulatory approval but almost always inreases the required ardia safety evaluations during later stages of drug development. 3,4 Beause of the inreased risk and ost of developing drugs that blok the herg potassium hannel and/or prolong QT, some of these ompounds are dropped from development, sometimes inappropriately. 1 Some drugs blok the herg potassium hannel and prolong QT with minimal torsade risk beause they also blok alium and/or sodium hannels (inward urrents). The most notable example is amiodarone, whih auses substantial QT prolongation but has a low risk of torsade. 5 This is likely beause bloking inward urrents an prevent early afterdepolarizations, whih trigger torsade. 6 8 Previous studies have foused on apturing hanges in T-wave morphology, 9 11 but so far these efforts have been foused primarily on deteting the presene of herg potassium hannel blok. 12 Using data from 34 thorough QT studies, we demonstrated previously that multihannel blok an be deteted on the ECG and that not all QT prolongation is equal. 13 That analysis suggested that herg potassium hannel blok prolongs both early repolarization (J : end of QRS to global peak of T-wave) and late repolarization ( : global peak to end of T-wave), whereas alium and late sodium urrent blok preferentially shorten early repolarization (Figure 1). The preferential effet of alium and late sodium urrents on early repolarization is 1 Offie of Siene and Engineering Laboratories, Center for Devies and Radiologial Health, US Food and Drug Administration, Silver Spring, Maryland, USA; 2 Department of Clinial Physiology, Karolinska Institutet and Karolinska University Hospital, Stokholm, Sweden; 3 BSICoS Group, Aragón Institute of Engineering Researh (I3A), IIS Aragón, University of Zaragoza, Zaragoza, Aragon, Spain; 4 Spaulding Clinial Researh, West Bend, Wisonsin, USA; 5 Frontage Laboratories, Exton, Pennsylvania, USA; 6 Division of Pharmaometris, Offie of Clinial Pharmaology, Offie of Translational Sienes, Center for Drug Evaluation and Researh, US Food and Drug Administration, Silver Spring, Maryland, USA; 7 Division of Cardiovasular and Renal Produts, Offie of New Drugs, Center for Drug Evaluation and Researh, US Food and Drug Administration, Silver Spring, Maryland, USA. Correspondene: DG Strauss (david.strauss@fda.hhs.gov) Reeived 5 May 214; aepted 9 July 214; advane online publiation 17 September 214. doi:1.138/lpt Clinial pharmaology & Therapeutis 1

2 Cellular AP Surfae ECG R Calium, late sodium J First ell s AP Last ell s AP Q S QRS Depolarization J- Early repolarization -T end Late repolarization Figure 1 An illustration of a ventriular ation potential (AP) and the orresponding surfae eletroardiogram (ECG). Arrows pointing into the ation potential are inward urrents (alium and late sodium) and arrows pointing out denote outward urrents (human ether-à-go-go-related gene (herg) potassium). Bloking the alium or late sodium urrent primarily shortens the early parts of repolarization (J ), whereas herg potassium hannel blok prolongs both early (J ) and late repolarization ( ). onsistent with these inward urrents being ative during the early repolarization phase of the ation potential. 13 However, the prior study 13 was limited by the fat that in vitro ion hannel data were not available for all drugs and the risk of torsade de pointes was not known for the drugs studied. Therefore, we designed a prospetive randomized ontrolled linial trial, funded by the FDA s Critial Path Initiative, to assess the ECG effets of multiple marketed drugs that either blok the herg potassium hannel alone or do so while also bloking alium and sodium hannels. The seleted drugs inluded four strong herg potassium hannel blokers with varying degrees of sodium and alium hannel blok: dofetilide, quinidine, ranolazine, and verapamil. Dofetilide is a strong pure herg potassium hannel bloker 14 with a high torsade risk. 15 The seond drug is quinidine, whih is also a strong herg potassium hannel bloker, but in addition to bloking the herg potassium hannel it also bloks alium and sodium hannels at high onentrations, 14 and torsade has been observed to our more frequently at lower plasma quinidine onentrations The last two drugs, ranolazine and verapamil, both blok not only the herg potassium hannel but also the late sodium urrent (ranolazine) or L-type alium hannel (verapamil), likely explaining why they are both assoiated with a low risk of torsade. 8,19 Administration of these four drugs to the same subjets enables the haraterization of ECG signatures of pure herg potassium hannel blok as ompared with multihannel blok. We T herg potassium T end Study Highlights WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC? 33 The QT interval is a sensitive biomarker of drug-indued herg potassium hannel blok and torsade de pointes risk; however, it is not speifi. Charaterization of the effets of additional inward urrent blok (alium or late sodium) on ardia repolarization may improve risk assessment beause inward urrent blok an offset the proarrhythmi effets of herg potassium hannel blok. WHAT QUESTION DID THIS STUDY ADDRESS? 33This study tested the hypothesis that herg potassium hannel blok prolongs both J (early repolarization) and (late repolarization) intervals on the ECG, whereas the addition of alium or late sodium urrent blok preferentially shortens J. WHAT THIS STUDY ADDS TO OUR KNOWLEDGE 33 This prospetive linial study demonstrated that pure herg potassium hannel blok equally prolongs both J and, whereas additional inward urrent blok (alium or late sodium) preferentially shortens J. HOW THIS MIGHT CHANGE CLINICAL PHARMACOLOGY AND THERAPEUTICS 33 Charaterization of multihannel drug effets on human ECGs is possible and may hange ardia safety assessment by onfirming omprehensive prelinial ion hannel assessments and influene dosing strategies for drugs. Table 1 Baseline harateristis All subjets (N = 22) Demographi Age (years) 26.9 ± 5.5 Female 11 (5%) Body mass index (kg/m 2 ) 23.1 ± 2.6 Vital signs Systoli blood pressure (mm Hg) 17.1 ± 8.5 Diastoli blood pressure (mm Hg) 59.7 ± 7.2 Heart rate (bpm) 56.8 ± 6.4 ECG PR interval (ms) ± 21.6 QRS duration (ms) 97.4 ± 6.7 J (ms) ± 19.8 (ms) 73.1 ± 6.4 (ms) ± 17.1 Continuous variables are represented as mean ± SD. ECG, eletroardiogram. hypothesized that herg potassium hannel blok prolongs both the J and intervals, whereas the addition of alium or late sodium urrent blok preferentially shortens the J interval. 2

3 a Dofetilide b Quinidine Dofetilide ± 95% Cl (ng/ml) 2 1 Quinidine ± 95% Cl (µg/ml) Ranolazine d Verapamil Ranolazine ± 95% Cl (µg/ml) Verapamil ± 95% Cl (ng/ml) Figure 2 Measured plasma onentrations (mean ± 95% onfidene interval) for (a) dofetilide, (b) quinidine, () ranolazine, and (d) verapamil. RESULTS Twenty-two healthy subjets (11 females) partiipated in this randomized ontrolled linial trial with a mean age of 26.9 ± 5.5 years and a mean body mass index of 23.1 ± 2.6 kg/ m 2 ; see Table 1 for baseline harateristis. All ompleted the study, exept one subjet who withdrew prior to the last treatment period. There were no unexpeted treatment-related adverse events. Pharmaokineti analysis The results of the pharmaokineti analysis are shown in Figure 2 for eah drug: dofetilide (a), quinidine (b), ranolazine (), and verapamil (d). Dofetilide and quinidine exhibited similar pharmaokineti profiles, with maximum onentration ourring at 2.5 h (range: 1 4. h) for dofetilide and at 2. h (.5 4. h) for quinidine, and with similar half-lives: (mean ± SD) 7.2 ± 1.1 h (dofetilide) and 7.8 ± 1.5 h (quinidine). The maximum onentrations for dofetilide and quinidine were 2.7 ±.3 ng/ml and 1.8 ±.4 µg/ml, respetively. Ranolazine peaked later at 4. h (1. 14 h), with a onentration of 2.3 ± 1.4 µg/ml, and had a half-life of 7.5 ± 4. h. Finally, verapamil peaked at 1. h (.5 2. h), with a plasma onentration of 13.3 ± 75.8 ng/ml, and had a half-life of 1.4 ± 3.2 h. Dofetilide: Pure herg potassium hannel blok prolongs early and late repolarization equally Dofetilide prolonged the heart rate orreted global QT () interval by 79.3 ms (95% onfidene interval (CI): ms, P <.1; Figure 3a), with equal prolongation of the heart rate orreted J (J : 39.5 ( ) ms, P <.1) and (4. ( ) ms, P <.1). Similarly, the onentration-dependent analysis showed that the prolongation by dofetilide equally affeted J and (J : 14.1 ( ) ms per ng/ml, P <.1; : 14.5 ( ) ms per ng/ml, P <.1; P =.89 for J vs. ; Figure 4a). These findings support the notion that herg potassium hannel blok prolongs both J and. Clinial pharmaology & Therapeutis 3

4 a Dofetilide (herg blok) b Quinidine (herg>alium>sodium blok) ± 95% CI (ms) 5 25 J ± 95% CI (ms) 5 25 J Ranolazine (late sodium>herg blok) d Verapamil (alium>herg blok) 2 2 ± 95% CI (ms) 1 ± 95% CI (ms) 1 J 1 J J Figure 3 Drug-indued hanges (mean ± 95% onfidene interval) for the plaebo-orreted hange from baseline (ΔΔ) of (gray), J (orange), and (blue) for (a) dofetilide, (b) quinidine, () ranolazine, and (d) verapamil. The y-axis in eah row of panels has been saled to enhane interpretation. Quinidine: strong herg potassium hannel blok with additional alium and sodium hannel blok prolongs late more than early repolarization Similar to dofetilide, quinidine prolonged the global by 78.1 ms ( ms, P <.1; Figure 3b); however, quinidine was more strongly assoiated with prolongation of than of J (J : 29.1 ( ) ms, P <.1; T end : 49.8 ( ) ms, P <.1; Figure 3b). Similarly, the onentration-dependent analysis showed a stronger relationship for as ompared with J (J : 11.6 ( ) ms per µg/ml, P =.8; : 29.9 ( ) ms per µg/ml; P <.1; P =.25 for J vs. ; Figure 4b). The preferential prolongation of assoiated with quinidine is likely due to alium hannel blok reduing the J prolongation assoiated with herg potassium hannel blok. Ranolazine: late sodium urrent blok opposes herg effets on early repolarization The global prolongation assoiated with ranolazine was 12.6 ms ( ms, P <.1; Figure 3), primarily via prolonged (J : 3.3 ( 3.4 to 9.9) ms, P =.34; : 8.8 ( ) ms, P =.13; Figure 3). A similar observation was seen in the onentration-dependent analysis, in whih ranolazine onentrations were assoiated with an inrease in (4.4 ( ) ms per µg/ml, P <.1; Figure 4); however, there was no assoiation with J (P =.42) (P <.1 for J vs. ; Figure 4). 4

5 a Dofetilide (herg blok) b Quinidine (herg>alium>sodium blok) T end J 5 25 T end J Conentration (ng/ml) Conentration (µg/ml) Ranolazine (late sodium>herg blok) d Verapamil (alium>herg blok) 2 T end 2 T end 1 1 J 1 J Conentration (µg/ml) Conentration (ng/ml) J T end Figure 4 Drug-indued hanges (mean ± 95% onfidene interval) for the plaebo-orreted hange from baseline (ΔΔ) for (gray), J (orange), and (blue) from model preditions vs. plasma onentrations for (a) dofetilide, (b) quinidine, () ranolazine, and (d) verapamil. For larity, the observed data are not shown in this figure but are inluded in Supplementary Figure S1 online. The range of the x-axis is 1.5 times the observed maximum onentration. The y-axis in eah row of panels has been saled to enhane interpretation. In eah plot, the line represents the predited mean effet of the linear model and the shaded area represents 95% onfidene intervals. These findings suggest that late sodium urrent blok primarily redues the J prolongation assoiated with herg potassium hannel blok. Verapamil: strong alium blok opposes effets of herg blok on early and late repolarization Verapamil did not ause a signifiant hange in global (5.2 ( ) ms, P =.15; Figure 2d), J ( 2.4 ( ) ms, P =.49; Figure 2d), or (4.8 ( ) ms, P =.18; Figure 3d). Similar observations were made in the onentration-dependent analysis (Figure 3d). The lak of prolongation of J and with verapamil suggests that strong alium blok may attenuate the effets of herg potassium hannel blok on both J and. Effet of drugs on PR, QRS, and heart rate The PR interval is most ommonly prolonged by slowing ondution through the atrioventriular node, where primarily alium urrent regulates depolarization. 2 Verapamil, a strong alium hannel bloker, was related to PR prolongation (32.1 ( ) ms, P <.1), whih was onentration dependent (.2 (.2.3) ms per ng/ml, P <.1). Quinidine, whih bloks the alium hannel, 14 trended toward PR prolongation (5.1 (.3 1.5) ms, P =.65); however, there was no onentration dependene (P =.91). The lak of PR prolongation observed in this study with quinidine despite its alium hannel blok is likely beause of ompeting autonomi effets on atrioventriular ondution. 21 Consistent with minimal alium hannel blok, only a small amount of PR prolongation was observed Clinial pharmaology & Therapeutis 5

6 a Dofetilide (herg blok) b Ranolazine (late sodium>herg blok) 4 T end 4 2 J 2 T end J Conentration (ng/ml) Conentration (µg/ml) J T end Figure 5 Drug-indued hanges in J and for (a) a pure herg potassium hannel bloker (dofetilide) and (b) a herg + late sodium urrent bloker (ranolazine). This zoomed plot of the onentrations of dofetilide that produe an amount of prolongation omparable to that of ranolazine shows the ability of J and to detet multihannel blok. herg, human ether-à-go-go-related gene. for ranolazine (6.5 ( ) ms, P =.18 and 1.1 (.1 2.) ms per µg/ml, P =.33). 19 No PR prolongation was observed for dofetilide, onsistent with a lak of alium hannel blok. 14 Bloking the sodium urrent inreases QRS duration by slowing ventriular ondution. 22 Quinidine is a sodium hannel bloker, but at the onentrations observed in this study, quinidine is expeted to blok the sodium urrent only minimally. 14 Quinidine only trended toward QRS prolongation (2.1 (.2 to 4.3) ms, P =.71), and there was no orrelation between plasma levels and QRS inrease (P =.95). Ranolazine is also a sodium hannel bloker, but it has weak peak sodium urrent blok, 19 and was assoiated with only a slight QRS prolongation (2.7 (.5 4.9) ms, P =.18); however, there was no orrelation between plasma levels and QRS (P =.6). Neither dofetilide nor verapamil is assoiated with sodium hannel blok. 14 Dofetilide produed no QRS prolongation, and verapamil produed a slight QRS prolongation (2.6 (.4 4.8) ms, P =.2); however, there was no positive relationship between onentration and QRS (P =.39). Finally, verapamil and quinidine aused an inrease in heart rate of 9.4 bpm (( ), P <.1) and 9.8 bpm (( ), P <.1), respetively. In both ases, there was a relationship between plasma onentration and heart rate (verapamil:.7 (.5.9) bpm per ng/ml, P <.1; quinidine: 4.7 ( ) bpm per µg/ml, P <.1). Dofetilide did not ause any hange in heart rate (P =.12), and ranolazine inreased heart rate slightly (4.2 ( ) bpm, P =.7); however, there was no relationship between onentration and heart rate (P =.22). DISCUSSION This prospetive randomized ontrolled linial trial demonstrated that by separating the QT interval into early repolarization (global J ) and late repolarization (global ), a pure herg potassium hannel bloking drug with high torsade risk an be differentiated from multihannelbloking drugs that also blok inward urrents (alium or late sodium) during repolarization. Deteting additional alium or late sodium urrent blok is of importane beause both the alium and sodium urrent support early afterdepolarizations, whih an trigger torsade de pointes. 6 8 Thus, being able to detet multihannel effets using the ECG may have impliations for the future of ardia safety evaluation of drugs and linial dosing strategies. The results of this study demonstrate that pure herg potassium hannel blok (dofetilide) equally prolongs J and, whereas additional alium and late sodium urrent blok (quinidine, ranolazine, and verapamil) preferentially shorten J. Of note, the pure herg potassium hannel bloker dofetilide and the multihannel bloker quinidine aused equal prolongation. Despite the omparable prolongation, dofetilide and quinidine were assoiated with different effets on J and, suggesting that the interval annot differentiate pure herg potassium hannel blok from multihannel blok. As an example of this, Figure 5 shows a omparison between onentrations of dofetilide and ranolazine that produe equal amounts of prolongation. From this figure, it is lear that the interval annot differentiate multihannel blok, whereas evaluation of J and intervals provides insight into multihannel blok that is of relevane for ardia safety evaluation. These findings suggest that future studies should onsider reporting the ratio of the hanges in J and. It is notable that dofetilide and quinidine aused substantial prolongation (~78 ms) in the present study. These results are onsistent with those from prior single-dose studies of dofetilide 6

7 and quinidine. Coz et al. 23 reported a slope of dofetilide onentration vs. of 31 ms per ng/ml, ompared with 29 ms per ng/ml in our study. Benton et al. 24 reported a slope of quinidine onentration vs. of 29 ms per µg/ml in men and 42 ms per µg/ml in women, ompared with 43 ms per µg/ml in our study. However, a separate study of dofetilide 75 µg twie daily for 4 days only observed a prolongation of ~6 ms. 25 This ould be due to differenes in measurement of the end of the T-wave, differenes in study design, or dereased sensitivity to dofetilide over time as reported in the dofetilide label. 26 The lak of prolongation with verapamil suggests that when alium hannel blok is stronger than herg potassium hannel blok, it may also attenuate the effet of herg potassium hannel blok on. These findings with verapamil are onsistent with those in the rabbit heart, in whih hanges in QT and were not observed at onentrations similar to those observed in this study. 27 Previously, alium hannel blok has been proposed to be deteted by an inrease in the PR interval, 13 whih was the ase for verapamil in this study but not for quinidine. The reason for the lak of PR prolongation with quinidine is likely due to quinidine s ompeting autonomi effets on atrioventriular node ondution, as quinidine has been shown to prolong PR in heart transplant patients. 21 Therefore, PR prolongation is not always present with alium hannel blok, and other markers suh as J and might prove more universally useful to detet the presene of alium hannel blok. Interestingly, these drug-indued ECG signature patterns have been observed previously with eletrolyte abnormalities and with geneti abnormalities of ion hannels. As early as the 195s, 28 it was observed that hypokalemia prolongs the QT interval and flattens and widens the T-wave. This is in ontrast to QT prolongation from hypoalemia, whih auses ST segment lengthening without T-wave hanges. 28 Hypokalemi QT prolongation has been attributed to inreasing the rate at whih herg potassium hannels inativate. 29,3 Similarly, the rate of inativation for alium hannels is dependent on alium onentrations and has been shown to be the reason for hyperalemia-indued ation potential shortening. 31 Thus, hypokalemia reates the equivalent of herg potassium hannel blok, whereas hyperalemia reates the equivalent of alium hannel blok. Similarly, geneti abnormalities in the herg potassium hannel (ongenital long QT type 2) ause flattening and widening of the T-wave, as with drug-indued herg potassium hannel blok. Geneti abnormalities in the sodium hannel (ongenital long QT type 3) ause prolongation of the ST segment with a normal T-wave. 32 Therapeuti impliations of multiple ion hannel effets With pure herg potassium hannel blok, there is likely a diret relationship between inreasing plasma drug onentration and torsade risk. For example, with dofetilide, there is a relationship among dose,, and torsade risk. 33 However, with quinidine, torsade has been observed to our more frequently at lower plasma quinidine onentrations. 16 This has been onfirmed in prelinial models in whih the number of quinidine-indued arrhythmias was greater at lower rather than higher quinidine onentrations, 17,18 whih is assoiated with more alium and late sodium urrent blok. The development of torsade requires an eletrial trigger and then a substrate to support re-entry. 34 Beause of the differential expression of ion hannels throughout the heart, herg potassium hannel blok aentuates dispersion of repolarization, thus reating the substrate for re-entry. 35 The trigger for torsade is believed to be early afterdepolarizations, whih an our as a result of inreased alium or sodium urrent during repolarization. 6,7 Thus, alium and late sodium urrent blok an prevent early afterdepolarizations (the trigger for torsade). The interval, as measured in a preordial lead, has been proposed as a measure of transmural dispersion of repolarization. 36 The relationship between transmural dispersion and is subjet to ontroversy as it has been proposed that instead reflets total dispersion and not transmural dispersion. 37,38 has, however, been shown to be prolonged by herg potassium hannel blok and to be longer in long QT2 patients (abnormalities in the herg potassium hannel) as ompared with long QT1 (abnormalities in the slow potassium hannel) and long QT3 (abnormalities in the sodium hannel) patients. 13,39 In this study, we quantified globally on the vetor magnitude derived lead, whih is likely more onsistent than measuring in a single lead in the presene of omplex T-wave patterns. Many drugs remain on the market with a known torsade risk, inluding numerous antibiotis and antimalarial, antiviral, psyhiatri, onology, and ardia drugs. 4 This raises the question of whether adding a late sodium- or alium urrent bloking drug to a herg potassium hannel bloking drug ould derease torsade risk. This onept has been evaluated in anine studies, in whih mexiletine (late sodium urrent bloker) dereased the torsade risk assoiated with sotalol (strong herg potassium hannel bloker). 41 Mexiletine has also been evaluated as a potential gene-speifi therapy for patients with ongenital long QT syndrome. In patients with ongenital long QT type 3 (abnormalities in the sodium hannel), it has been shown that administering mexiletine shortened their QT intervals. 42 Similar observations have been made with ranolazine. 43 In the mexiletine study, the investigators also evaluated the effets of mexiletine treatment in patients with long QT type 2 (abnormalities in the herg potassium hannel). They observed a numerial derease in that was not statistially signifiant; however, only six patients were studied. In addition, mexiletine has been shown to derease the interval in a subjet with Timothy syndrome (abnormalities in the alium hannel). 44 Furthermore, the potential antiarrhythmi effet of mexiletine was evaluated in linial studies in whih mexiletine was administered in ombination with quinidine. When mexiletine was administered on its own, there were no signifiant hanges in, but when mexiletine was administered to patients already reeiving quinidine, there was a signifiant shortening of the quinidine-indued prolongation. 45,46 These studies suggest that oadministering an inward urrent bloker with a herg potassium hannel bloker ould offset Clinial pharmaology & Therapeutis 7

8 I avr V1 V4 II avl V2 V5 III avf V3 V6 Transform to vetorardiogram QRS loop Vetor magnitude lead T end T-wave loop Figure 6 Method for assessment of the global and T end. The 12-lead eletroardiogram is transformed to the vetorardiogram via a published transformation matrix. 55 The peak and end of the T-wave are loated in the vetor magnitude lead. The end of the T-wave is determined using the tangent method, 56 whih involves loating the intersetion between the line through the terminal desending part of the T-wave and the isoeletri line. the herg potassium hannel bloking effets, whih should be evaluated in further studies. Limitations Two of the drugs (dofetilide and quinidine) are assoiated with massive inreases and T-wave hanges, inluding nothing, whih an make and T end determination diffiult. However, there was strong agreement between two independent assessments of all ECGs and less than 5 ms differene in 98.6% of the ECGs. It should also be reognized that beause the QT interval is subjet to rate dependeny and delayed adaptation to heart rate, it is likely that the J and intervals are as well. Rate dependeny for both the J and T end intervals has been established previously, but the effet on at near-resting heart rates is minimal, 13,47 and thus the interval was not orreted for heart rate. It is possible that using individualized heart rate orretion would lead to results different from those using populationbased heart rate orretion (i.e., the same orretion fator for all subjets) as was done in this study for QT and J. Different heart rate orretion methods are likely to influene results only for drugs that signifiantly affet heart rate. 48 In our study, quinidine and verapamil inreased heart rate by 1 and 9 bpm, respetively. For quinidine, it is unlikely that a different heart rate orretion method would hange the results beause J,, and prolongation were strongly assoiated with quinidine onentration, and the effet of quinidine on was almost twie that of its effet on J. For verapamil, it is possible that an individualized orretion for heart rate ould ause different results. Finally, the potential effets of delayed adaptation of repolarization to heart rate (hysteresis) were minimized by measuring the intervals in ECGs at stable heart rates. 49 Conlusion This study supports the hypothesis that ECG measures of early repolarization (global J ) and late repolarization (global ) an differentiate pure herg potassium hannel blok assoiated with a high torsade risk from ombined herg potassium hannel and inward urrent blok (alium or late sodium), whih may lower torsade risk. By ontrast, evaluating only the interval does not detet multihannel blok. This linial study investigating the ECG signatures of multihannel blok is one of three efforts studying potential approahes to improve the urrent regulatory paradigm of fousing almost exlusively on herg potassium hannel blok and. This inludes a Comprehensive in vitro Proarrhythmia Assay (CiPA), 5 in whih the effets of drugs on multiple ion hannels would be assessed, and the use of detailed ECG olletion in early linial studies with exposure response analysis. 51 Ultimately, future approahes will likely use a more mehanisti approah to evaluate the risk for drug-indued torsade de pointes. METHODS Clinial study design. We onduted a randomized ontrolled five-way single-dose rossover linial trial in 22 healthy volunteers (11 females) at a phase I linial researh unit (Spaulding Clinial, West Bend, WI). The study was approved by the US Food and Drug Administration 8

9 Researh Involving Human Subjets Committee and the loal institutional review board. All subjets gave written informed onsent. The drugs evaluated in this study were dofetilide (5 µg, Tikosyn, Pfizer, New York, NY), quinidine sulfate (4 mg, Watson Pharma, Corona, CA), ranolazine (1,5 mg, Ranexa, Gilead, Foster City, CA), and verapamil hydrohloride (12 mg, Heritage Pharmaeutials, Edison, NJ). The inlusion riteria required subjets to be of general good health as determined by a physiian, without a history of heart disease or unexplained synope or a family history of long QT syndrome; to be years of age, weigh at least 5 kg, and have a body mass index of kg/ m 2 ; and to be able to read and understand the informed onsent. In addition, subjets were exluded if they had more than 1 etopi beats during a 3-h ontinuous ECG reording at sreening. In the morning of eah treatment period, the subjets reeived one of the four drugs or plaebo under fasting onditions. There was a 7-day washout period between eah 24-h treatment period, so subjets reeived treatment on days 1, 9, 17, 25, and 33. Prior to dosing, a ontinuous 12-lead ECG reorder (Surveyor, Mortara Instrument, Milwaukee, WI) using the Mason Likar 52 eletrode onfiguration was onneted to eah subjet. The ontinuous ECG reordings were made at 5 Hz and with an amplitude resolution of 2.5 µv. From the ontinuous reording, three repliate 1-s ECGs (pre- and postdose) were extrated at 16 predefined time-points (predose and.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, 6, 7, 8, 12, 14, 24 h postdose), during whih the subjets were resting in a supine position for 1 min. After eah ECG extration time-point period, a blood sample was drawn for pharmaokineti analysis. Plasma drug onentration was determined using a validated liquid hromatography with tandem mass spetrosopy method by Frontage Laboratories (Exton, Philadelphia, PA). ECG assessment. At eah of the 16 time-points, three optimal ECGs were extrated with stable heart rates and maximum signal quality using Antares software (AMPS, New York, NY). 49 This resulted in a total of 48 ECGs per subjet per treatment period and 5,28 planned ECGs in total, whih were all evaluated with omputerized interval annotations on high-resolution images by the ECG reader, who was blinded to treatment and time. The same ECG reader evaluated all ECGs from the same subjet and determined P onset, QRS onset, and QRS offset using lead II. To quantify the global repolarization intervals, two ECG readers blinded to treatment and time determined the global peak and end of the T-wave independently using previously developed software 53,54 on the vetor magnitude of the vetorardiogram (obtained from the Mason Likar 12-lead ECG by applying the Guldenring transformation matrix). 55 The end of the T-wave was determined using the tangent method, 56 whih involves loating the intersetion between the line through the terminal desending part of the T-wave and the isoeletri line (Figure 6). This approah of using the tangent method in the vetor magnitude lead produes more onsistent measurements of the QT interval. 57 Of note, the U-wave was not inluded in the measurement of the end of the T-wave. Notably, globally measured is different as ompared with measured in a preordial lead, whih has been proposed as a measure of transmural dispersion. 36 The relationship between T end and transmural dispersion, is, however, subjet to disussion. 37,38 was measured globally, whih is likely more onsistent in the presene of omplex T-wave patterns. was defined as the first disernable peak in the T-wave. Disagreements on a T-wave being measureable, presene of a noth, or a differene of more than 5 ms in either or T-wave offset were re-reviewed and adjudiated by an expert ECG reader. This was the ase for 73 ECGs (or ~1.4%). From the measured fiduial points, the PR, QRS, J (QRS offset to global T-wave peak), (global to T end ), and QT intervals were obtained (based on the global T end ). J was orreted for heart rate using a oeffiient obtained from a previous analysis of pooled subjets from 34 thorough QT studies 13 (J = J /RR.58 with RR in seonds), and QT was orreted with Frideriia s orretion 58 ( = QT/RR 1/3 with RR in seonds). Although heart rate dependeny for has been reported, 47 rate orretion was not performed for. Rate orretion was not done beause previous studies, inluding a pooled analysis of subjets from 34 thorough QT studies, have shown that at resting heart rates exhibits minimal heart rate dependeny. 13,47 Statistial analysis. The plaebo-orreted hange from baseline was omputed using PROC MIXED in SAS 9.3 (SAS Institute, Cary, NC). The hange from baseline for eah ECG biomarker (e.g., the average, ) by time-point was the dependent variable, for whih baseline was defined as the average predose value. Sequene, period, time, treatment, and an interation between treatment and time were inluded as fixed effets, and subjet was inluded as a random effet. In addition, exposure response analysis was performed with a linear mixed-effets model to evaluate the relationship between plasma drug onentrations and ECG measurements. 59 Differenes in J and for eah drug were ompared using a paired t-test in R (R Foundation for Statistial Computing, Vienna, Austria). P values <.5 were onsidered statistially signifiant. SUPPLEMENTARY MATERIAL is linked to the online version of the paper at ACKNOWLEDGMENTS We thank Jill Coker, FDA extramural linial studies oordinator, and staff at Spaulding Clinial Researh for exeuting the study. This projet was supported in part by appointments to the Researh Partiipation Programs at the Oak Ridge Institute for Siene and Eduation through an interageny agreement between the Department of Energy and the FDA. The opinions presented here are those of the authors. No offiial support or endorsement by the FDA is intended nor should be inferred. The mention of ommerial produts, their soures, or their use in onnetion with material reported herein is not to be onstrued as either an atual or implied endorsement of suh produts by the FDA. This projet was supported by the FDA s Critial Path Initiative and Offie of Women s Health. AUTHOR CONTRIBUTIONS D.G.S., L.J., J.V., J.W.M., C.S., K.W.-L., M.H., P.G., J.S.S., L.G., J.L., J.F., M.U., and N.S. wrote the manusript. D.G.S., L.J., J.V., J.W.M., L.G., J.F., and N.S. designed the study. D.G.S., L.J., J.V., J.W.M., C.S., K.W.-L., M.H., P.G., and J.L. performed the study. D.G.S., L.J., J.V., J.W.M., C.S., K.W.-L., M.H., P.G., J.S.S., L.G., J.L., J.F., M.U., and N.S. analyzed the data. CONFLICT OF INTEREST J.W.M., K.W.-L., and C.S. are employees of Spaulding Clinial Researh, and M.H., P.G., and J.L. are employees of Frontage Laboratories, whih are ontrat researh organizations. The other authors delared no onflit of interest. 214 Amerian Soiety for Clinial Pharmaology and Therapeutis 1. Stokbridge, N., Morganroth, J., Shah, R.R. & Garnett, C. Dealing with global safety issues: was the response to QT-liability of non-ardia drugs well oordinated? Drug Saf. 36, (213). 2. Trudeau, M.C., Warmke, J.W., Ganetzky, B. & Robertson, G.A. HERG, a human inward retifier in the voltage-gated potassium hannel family. Siene 269, (1995). 3. The International Conferene on Harmonisation of Tehnial Requirements for Registration of Pharmaeutials for Human Use. Guidane for industry E14 Clinial Evaluation of QT/ interval prolongation and proarrhythmi potential for non-antiarrhythmi drugs. < guidaneomplianeregulatoryinformation/guidanes/um73153.pdf >(25). Aessed 14 Marh Malik, M., Garnett, C.E. & Zhang, J. Thorough QT studies: questions and quandaries. Drug Saf. 33, 1 14 (21). 5. Hohnloser, S.H., Klingenheben, T. & Singh, B.N. Amiodarone-assoiated proarrhythmi effets. A review with speial referene to torsade de pointes tahyardia. Ann. Intern. Med. 121, (1994). 6. January, C.T. & Riddle, J.M. Early afterdepolarizations: mehanism of indution and blok. A role for L-type Ca2+ urrent. Cir. Res. 64, (1989). 7. Wu, L. et al. Augmentation of late sodium urrent unmasks the proarrhythmi effets of amiodarone. Cardiovas. Res. 77, (28). 8. Aiba, T. et al. Cellular and ioni mehanism for drug-indued long QT syndrome and effetiveness of verapamil. J. Am. Coll. Cardiol. 45, 3 37 (25). Clinial pharmaology & Therapeutis 9

10 9. Andersen, M.P., Xue, J.Q., Graff, C., Kanters, J.K., Toft, E. & Struijk, J.J. New desriptors of T-wave morphology are independent of heart rate. J. Eletroardiol. 41, (28). 1. Aar, B., Yi, G., Hnatkova, K. & Malik, M. Spatial, temporal and wavefront diretion harateristis of 12-lead T-wave morphology. Med. Biol. Eng. Comput. 37, (1999). 11. Couder, J.P. Measurement and regulation of ardia ventriular repolarization: from the QT interval to repolarization morphology. Philos. Trans. A. Math. Phys. Eng. Si. 367, (29). 12. Malik, M. Drug-indued hanges in the T-wave morphology. Drug Saf. 32, (29). 13. Johannesen, L. et al. Improving the assessment of heart toxiity for all new drugs through translational regulatory siene. Clin. Pharmaol. Ther. 95, (214). 14. Kramer, J. et al. MICE models: superior to the HERG model in prediting Torsade de Pointes. Si. Rep. 3, 21 (213). 15. Redfern, W.S. et al. Relationships between prelinial ardia eletrophysiology, linial QT interval prolongation and torsade de pointes for a broad range of drugs: evidene for a provisional safety margin in drug development. Cardiovas. Res. 58, (23). 16. Roden, D.M., Woosley, R.L. & Primm, R.K. Inidene and linial features of the quinidine-assoiated long QT syndrome: impliations for patient are. Am. Heart J. 111, (1986). 17. Antzelevith, C. et al. The M ell: its ontribution to the ECG and to normal and abnormal eletrial funtion of the heart. J. Cardiovas. Eletrophysiol. 1, (1999). 18. Wu, L. et al. Role of late sodium urrent in modulating the proarrhythmi and antiarrhythmi effets of quinidine. Heart Rhythm 5, (28). 19. Antzelevith, C. et al. Eletrophysiologial effets of ranolazine, a novel antianginal agent with antiarrhythmi properties. Cirulation 11, (24). 2. Narimatsu, A. & Taira, N. Effets of atrio-ventriular ondution of aliumantagonisti oronary vasodilators, loal anaesthetis and quinidine injeted into the posterior and the anterior septal artery of the atrio-ventriular node preparation of the dog. Naunyn. Shmiedebergs. Arh. Pharmaol. 294, (1976). 21. Mason, J.W., Winkle, R.A., Rider, A.K., Stinson, E.B. & Harrison, D.C. The eletrophysiologi effets of quinidine in the transplanted human heart. J. Clin. Invest. 59, (1977). 22. Harmer, A.R., Valentin, J.P. & Pollard, C.E. On the relationship between blok of the ardia Na+ hannel and drug-indued prolongation of the QRS omplex. Br. J. Pharmaol. 164, (211). 23. Le Coz, F., Funk-Brentano, C., Morell, T., Ghadanfar, M.M. & Jaillon, P. Pharmaokineti and pharmaodynami modeling of the effets of oral and intravenous administrations of dofetilide on ventriular repolarization. Clin. Pharmaol. Ther. 57, (1995). 24. Benton, R.E., Sale, M., Flokhart, D.A. & Woosley, R.L. Greater quinidineindued interval prolongation in women. Clin. Pharmaol. Ther. 67, (2). 25. Démolis, J.L., Funk-Brentano, C., Ropers, J., Ghadanfar, M., Nihols, D.J. & Jaillon, P. Influene of dofetilide on QT-interval duration and dispersion at various heart rates during exerise in humans. Cirulation 94, (1996). 26. Pfizer Laboratories Division of Pfizer. Tikosyn (Dofetilide) Capsule [pakage insert]. < label/213/2931s7lbl.pdf> (211). Aessed 8 August Liu, T. et al. Differentiating eletrophysiologial effets and ardia safety of drugs based on the eletroardiogram: a blinded validation. Heart Rhythm 9, (212). 28. Marriott, H.J.L. Pratial Eletroardiography (Williams & Wilkins, Baltimore, MD 1954). 29. Yang, T., Snyders, D.J. & Roden, D.M. Rapid inativation determines the retifiation and [K+]o dependene of the rapid omponent of the delayed retifier K+ urrent in ardia ells. Cir. Res. 8, (1997). 3. Samps, F. & Carmeliet, E. Delayed K+ urrent and external K+ in single ardia Purkinje ells. Am. J. Physiol. 257, C186 C192 (1989). 31. Grandi, E., Pasqualini, F.S., Pes, C., Corsi, C., Zaza, A. & Severi, S. Theoretial investigation of ation potential duration dependene on extraellular Ca2+ in human ardiomyoytes. J. Mol. Cell. Cardiol. 46, (29). 32. Moss, A.J. et al. ECG T-wave patterns in genetially distint forms of the hereditary long QT syndrome. Cirulation 92, (1995). 33. Pfizer. Risk Evaluation and Mitigation Strategy (REMS) Doument. < PostmarketDrugSafetyInformationforPatientsandProviders/UCM pdf (213). Aessed 1 May Yap, Y.G. & Camm, A.J. Drug indued QT prolongation and torsades de pointes. Heart 89, (23). 35. Antzelevith, C. & Fish, J. Eletrial heterogeneity within the ventriular wall. Basi Res. Cardiol. 96, (21). 36. Antzelevith, C. et al. Does provide an index of transmural dispersion of repolarization? Heart Rhythm 4, (27). 37. Opthof, T. et al. Dispersion of repolarization in anine ventrile and the eletroardiographi T wave: T p e interval does not reflet transmural dispersion. Heart Rhythm 4, (27). 38. Xia, Y., Liang, Y., Kongstad, O., Holm, M., Olsson, B. & Yuan, S. T end interval as an index of global dispersion of ventriular repolarization: evaluations using monophasi ation potential mapping of the epi- and endoardium in swine. J. Interv. Card. Eletrophysiol. 14, (25). 39. Kanters, J.K. et al. T(peak)T(end) interval in long QT syndrome. J. Eletroardiol. 41, (28). 4. AZCERT. CredibleMeds. < (214). Aessed 14 Marh Chézalviel-Guilbert, F., Davy, J.M., Poirier, J.M. & Weissenburger, J. Mexiletine antagonizes effets of sotalol on QT interval duration and its proarrhythmi effets in a anine model of torsade de pointes. J. Am. Coll. Cardiol. 26, (1995). 42. Shwartz, P.K. et al. Long QT syndrome patients with mutations of the SCN5A and HERG genes have differential responses to Na+ hannel blok and to inreases in heart rate. Impliations for gene-speifi therapy. Cirulation 92, (1995). 43. Moss, A.J., Zareba, W., Shwarz, K.Q., Rosero, S., MNitt, S. & Robinson, J.L. Ranolazine shortens repolarization in patients with sustained inward sodium urrent due to type-3 long-qt syndrome. J. Cardiovas. Eletrophysiol. 19, (28). 44. Gao, Y. et al. Inhibition of late sodium urrent by mexiletine: a novel pharmotherapeutial approah in timothy syndrome. Cir. Arrhythm. Eletrophysiol. 6, (213). 45. Giardina, E.G. & Wehsler, M.E. Low dose quinidine mexiletine ombination therapy versus quinidine monotherapy for treatment of ventriular arrhythmias. J. Am. Coll. Cardiol. 15, (199). 46. Duff, H.J., Mithell, L.B., Manyari, D. & Wyse, D.G. Mexiletine quinidine ombination: eletrophysiologi orrelates of a favorable antiarrhythmi interation in humans. J. Am. Coll. Cardiol. 1, (1987). 47. Smetana, P., Bathvarov, V., Hnatkova, K., John Camm, A. & Malik, M. Sex differenes in the rate dependene of the T wave desending limb. Cardiovas. Res. 58, (23). 48. Garnett, C.E. et al. Methodologies to haraterize the QT/orreted QT interval in the presene of drug-indued heart rate hanges or other autonomi effets. Am. Heart J. 163, (212). 49. Badilini, F., Vaglio, M. & Sarapa, N. Automati extration of ECG strips from ontinuous 12-lead holter reordings for QT analysis at presheduled versus optimized time points. Ann. Noninvasive Eletroardiol. 14 (suppl. 1), S22 S29 (29). 5. Sager, P.T., Gintant, G., Turner, J.R., Pettit, S. & Stokbridge, N. Rehanneling the ardia proarrhythmia safety paradigm: a meeting report from the Cardia Safety Researh Consortium. Am. Heart J. 167, (214). 51. Darpo, B. et al. The IQ-CSRC prospetive linial Phase 1 study: Can early QT assessment using exposure response analysis replae the thorough QT study?. Ann. Noninvasive Eletroardiol. 19, 7 81 (214). 52. Mason, R.E. & Likar, I. A new system of multiple-lead exerise eletroardiography. Am. Heart J. 71, (1966). 53. Johannesen, L., Viente, J., Galeotti, L. & Strauss, D.G. ECGlib: Library for proessing eletroardiograms. Comput Cardiol 4, (213). 54. Viente, J., Johannesen, L., Galeotti, L. & Strauss, D.G. ECGlab: User friendly ECG/VCG analysis tool for researh environments Comput Cardiol 4, (213). 55. Guldenring, D. et al. Transformation of the Mason Likar 12-lead eletroardiogram to the Frank vetorardiogram. Conf. Pro. IEEE Eng. Med. Biol. So. 212, (212). 56. Lepeshkin, E. & Surawiz, B. The measurement of the Q T interval of the eletroardiogram. Cirulation 6, (1952). 57. Diamant, U.B., Winbo, A., Stattin, E.L., Rydberg, A., Kesek, M. & Jensen, S.M. Two automati QT algorithms ompared with manual measurement in identifiation of long QT syndrome. J. Eletroardiol. 43, 25 3 (21). 58. Frideriia, L.S. Die Systolendauer im Elektrokardiogramm bei normalen Menshen und bei Herzkranken. Ata Media Sandinavia 53, (192). 59. Garnett, C.E. et al. Conentration-QT relationships play a key role in the evaluation of proarrhythmi risk during regulatory review. J. Clin. Pharmaol. 48, (28). 1