03774

EXPERIMENTAL STUDIES

Cellular and Ionic Mechanisms Underlying Erythromycin-InducedLong QT Intervals and Torsade de Pointes

CHARLES ANTZELEVITCH, PHD, ZHUO-QIAN SUN, MD, PHD, ZI-QING ZHANG, MD,GAN-XIN YAN, MD, PHD

Utica, New York

Objectives. This study sought to elucidate the cellular and ionicbasis for erythromycin-induced long QT syndrome.

Background. Erythromycin is known to produce long QTUintervals on the electrocardiogram (ECG) and to be associatedwith the development of torsade de pointes (TdP). The mecha-nisms responsible for the adverse effects of this widely usedantibiotic are not well defined.

Methods. The present study used microelectrode and whole-cellpatch-clamp techniques to assess the effects of erythromycin onepicardial, endocardial and M cells in transmural strips, arteri-ally perfused wedges and single myocytes isolated from the canineleft ventricle.

Results. In isolated strips, erythromycin (10 to 100 �g/ml)produced a much more pronounced prolongation of the actionpotential duration (APD) in M cells than in endocardial andepicardial cells, resulting in the development of a large dispersionof repolarization across the ventricular wall at slow stimulationrates. Erythromycin (50 to 100 �g/ml) induced early afterdepo-larizations (EADs) in cells in the M (20%) but not epicardial orendocardial regions in transmural strips of ventricular free wall.Erythromycin (100 �g/ml) also caused APD prolongation and a

transmural dispersion of repolarization, but not EADs, in intactarterially perfused wedges of canine left ventricle. These changeswere attended by the development of a long QT interval on thetransmural ECG. A polymorphic ventricular tachycardia closelyresembling TdP was readily and reproducibly induced aftererythromycin but not before. Whole-cell patch-clamp techniques,used to examine the effects of erythromycin on myocytes isolatedfrom the M region, showed a potent effect of the drug to inhibit therapidly activating component (IKr) but not the slowly activatingcomponent (IKs) of the delayed rectifier potassium current (IK).The inward rectifier current (IK1) was unaffected.

Conclusions. Our data demonstrate a preferential response ofM cells to the class III actions of erythromycin, due principally tothe effect of the drug to inhibit IKr in a population of cells largelydevoid of IKs. Our findings indicate that erythromycin thusproduces long QT intervals as well as a prominent dispersion ofrepolarization across the ventricular wall, setting the stage forinduction of TdP-like tachyarrhythmias displaying characteristicstypical of reentry.

(J Am Coll Cardiol 1996;28:1836–48)�1996 by the American College of Cardiology

The clinical syndrome of acquired long QT occurs in associa-tion with various pharmacologic agents, electrolyte abnormal-ities and bradycardic states (1–5). Most pharmacologic agentscapable of prolonging the QT interval appear to be capable ofcausing ventricular tachyarrhythmias, most notably a polymor-phic ventricular tachycardia known as torsade de pointes(TdP). A recent addition to the group is the macrolideantibiotic erythromycin (6–11). A recent study by Daleau et al.(12) demonstrates the effect of erythromycin to inhibit therapid component of the delayed rectifier current (IKr) inguinea pig myocytes of unknown transmural origin. Monopha-

sic action potentials recorded from the endocardial surface ofa Langendorff perfused guinea pig heart showed modestprolongation. A recent study by Rubart et al. (10) presentedindirect evidence suggesting that erythromycin-induced pro-longation of the action potential in canine Purkinje fibers isdue to inhibition of the delayed rectifier potassium current(IK). Their study also showed that ventricular endocardium(papillary muscle) was little affected by the drug. Thus, themechanism by which erythromycin prolongs ventricular repo-larization and causes long QT intervals and TdP is not welldefined.

Recent studies (5,13–15) have shown that M cells in thedeep structures of ventricular myocardium are primary targetsfor most agents that prolong repolarization in the canine heart.Epicardial and endocardial cells were found to be little affectedby concentrations of drug that cause marked prolongation,early afterdepolarization (EAD) and triggered activity in Mcells. The present study tests the hypothesis that the ability oferythromycin to induce long QT intervals, EADs, triggeredactivity and TdP is due principally to its ability to delayrepolarization of M cells in the deep structures of the canine

From the Masonic Medical Research Laboratory, Utica, New York. Thisstudy was supported by Grants HL37396 and HL47678 from the National Heart,Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland; bya fellowship grant from the New York Affiliate of the American HeartAssociation; and by grants from Dolgeville Lodge 796, Dolgeville, New York,and the Sixth and Seventh Manhattan Masonic Districts, New York, New York.E-4031 was kindly donated by EISAI Co., Ltd., Tokyo, Japan.

Manuscript received July 8, 1996, accepted August 13, 1996.Address for correspondence: Dr. Charles Antzelevitch, Masonic Medical

Research Laboratory, 2150 Bleecker Street, Utica, New York 13501. E-mail:[email protected].

JACC Vol. 28, No. 7December 1996:1836–48

1836

�1996 by the American College of Cardiology 0735-1097/96/$15.00Published by Elsevier Science Inc. PII S0735-1097(96)00377-4

Downloaded From: http://content.onlinejacc.org/ on 01/25/2015

heart through inhibition of one or both components of IK. Theexperimental protocols assess the differential responsiveness ofthe three principal cell types spanning the left ventricular wallto erythromycin using isolated tissue preparations, the devel-opment of polymorphic arrhythmias resembling TdP using anarterially perfused wedge of canine left ventricle and the ionicmechanism underlying the response to erythromycin using thewhole-cell patch-clamp technique applied to myocytes isolatedfrom the M region of the canine left ventricle. Our preliminaryresults have been reported in abstract form (16).

MethodsAction potential studies. Transmural strips (�2.0 � 1.0 �

0.1 to 0.2 cm) were isolated from the left ventricular free wallof canine hearts removed from anesthetized (30 mg/kg bodyweight sodium pentobarbital) mongrel male dogs. The tissuepreparations were obtained by razor blade shavings (Der-matome Power Handle no. 3293 with cutting head no. 3295,Davol Simon). The preparations were obtained near the baseof the left ventricle, and shavings were made either parallel orperpendicular to the surface of the free wall.

The preparations were placed in a tissue bath and allowedto equilibrate for at least 2 h while superfused with anoxygenated (95% oxygen/5% carbon dioxide) Tyrode’s solu-tion (37 � 0.5�C, pH � 7.35; composition: [mmol/liter] NaCl129, KCl 4, NaH2PO4 0.9, NaHCO3 20, CaCl2 1.8, MgSO4 0.5,D-glucose 5.5).

The tissues were stimulated at a basic cycle length (BCL)ranging between 500 and 8,000 ms using field or point stimu-lation delivered to the endocardial side through silver bipolarelectrodes insulated except at the tips. Transmembrane poten-tials were recorded from endocardial, epicardial and M regionsites using glass microelectrodes filled with 2.7 mol/liter KCl(10 to 20-M� direct current resistance) connected to a highinput impedance amplification system (World Precision Instru-ments). The signal was displayed on Tektronix oscilloscopes,amplified (model 1903A programmable amplifiers, CambridgeElectronic Design), digitally sampled at a rate of �7 kHz(Spike 2 acquisition module, Cambridge Electronic Design),stored on magnetic or other media (personal computer hard

drives, magnetic tapes and CD-ROM) and analyzed (Spike 2analysis module, Cambridge Electronic Design).

Arterially perfused wedge of canine left ventricle. Dogsweighing 20 to 25 kg received anticoagulation with heparin andwere anesthetized with pentobarbital (30 mg/kg intravenously).The chest was opened by means of a left thorocatomy, and theheart was excised, placed in a cardioplegic solution consistingof cold (4�C) Tyrode’s solution containing 8.5 mmol/literextracellular K� concentration ([K�]o) and transported to adissection tray. Transmural wedges with dimensions of 0.8 �0.9 � 0.8 cm to 1.2 � 2 � 1.1 cm (height � length � wallthickness) were dissected from the left ventricle. The tissue wascannulated through a small (�100 to 150-�m diameter) branchof the left anterior descending or other coronary artery andperfused with cardioplegic solution. The total period of timefrom excision of the heart to cannulation and perfusion of theartery was usually 4 min. The preparation was then placed ina small tissue bath and arterially perfused with Tyrode’ssolution of the following composition (mmol/liter): NaCl 129,KCl 4, NaH2PO4 0.9, NaHCO3 20, CaCl2 1.8, MgSO4 0.5,glucose 5.5, and 1 U/liter of insulin buffered with 95% oxygenand 5% carbon dioxide (temperature 36 � 1�C). The perfusatewas delivered to the artery by a roller pump (Cole ParmerInstrument Co.). Unperfused tissue, identified by its dark redappearance (erythrocytes not washed out) soon after thewedge was cannulated and perfused, was excised using a razorblade. In initial experiments with this preparation, we injectedtriphenyl tetrazolium chloride to identify ischemic regions (17)in wedge preparations that had been beating in the tissuechamber for a period of �5 h. No ischemic regions weredetected.

Perfusion pressure was monitored with a pressure trans-ducer (World Precision Instruments, Inc.) and maintainedbetween 40 and 50 mm Hg by adjustment of the perfusion flowrate. This pressure is considered normal for small arteries (100to 150-�m diameter) (18). Perfusion pressures 50 mm Hgwere difficult to maintain constant and were found to causeedema, whereas pressures 30 mm Hg sometimes caused mildischemia in endocardium accompanied by perceptible actionpotential and ST segment changes. The preparations remainedimmersed in the arterial perfusate, which was allowed to rise toa level 2 to 3 mm above the tissue surface (temperature 36 �1�C).

The left ventricular wedges were allowed to equilibrate inthe tissue bath until electrically stable, usually 1 h. Thepreparations were stimulated at BCLs ranging from 500 to5,000 ms using bipolar silver electrodes insulated except at thetips and applied to the endocardial surface.

A transmural electrocardiographic (ECG) signal was re-corded using extracellular silver chloride electrodes placednear the epicardial and endocardial surfaces of the preparationplugged into a differential direct current amplifier. Transmem-brane action potentials were simultaneously recorded from theepicardial, endocardial and M regions using three separateintracellular floating microelectrodes (direct current resistance10 to 20 M�) filled with 2.7 mol/liter KCl and connected to a

Abbreviations and Acronyms

APD � action potential durationAPD90 � action potential duration at 90% repolarizationBCL � basic cycle lengthEADs � early afterdepolarizationsECG � electrocardiogram, electrocardiographicIK � delayed rectifier K� currentIKr � rapidly activating component of delayed rectifier K� currentIKs � slowly activating component of delayed rectifier K� currentIK1 � inward rectifier K� current[K�]o � extracellular K� concentrationTdP � torsade de pointes

1837JACC Vol. 28, No. 7 ANTZELEVITCH ET AL.December 1996:1836–48 MECHANISMS OF ERYTHROMYCIN-INDUCED LQTS


high input impedance amplifier. Impalements were obtainedfrom the cut surface of the preparation at positions approxi-mating the transmural axis of the ECG recording. Amplifiedsignals were digitized, stored on magnetic media and CD-ROM and analyzed using Spike 2 (Cambridge ElectronicDesigns, Cambridge, England, U.K.). The electrical stability ofthe preparation was assessed in a series of seven experimentsin which action potential duration (APD) and QT intervalwere measured over a period of 4 h (Table 1).

Whole-cell patch-clamp studies. Myocytes were isolated byenzymatic dissociation as previously described (19). Briefly,adult male and female mongrel dogs were anesthetized withsodium pentobarbital (30 mg/kg intravenously); the heartswere quickly removed and placed in normal Tyrode’s solution.A wedge consisting of that part of the left ventricular free wallsupplied by the left circumflex coronary artery was excised. Theartery was cannulated and flushed with Ca2�-free Krebs buffersupplemented with 0.1% bovine serum albumin (fraction V,Sigma) and gassed with 95% oxygen/5% carbon dioxide for5 min at a rate of 12 ml/min. Perfusion was then switched to 75ml of calcium-free Krebs buffer containing 75 mg of bovineserum albumin and 37.5 of mg collagenase (CLS 2, 171 U/mg,Worthington) for 15 to 20 min at 37�C (95% oxygen/5% carbondioxide, with recirculation). After perfusion, thin slices oftissues were dissected from epicardium (1.5 mm from theepicardial surface), M region (2 to 7 mm from the epicardialsurface) and endocardium (2 mm from the endocardialsurface) using a dermatome. Shavings were made parallel tothe surface of the left ventricular free wall midway along theapicobasal axis. Tissues from each region were placed inseparate beakers; minced and incubated in fresh Krebs buffercontaining 0.5 mg/ml of collagenase, 3% bovine serum albuminand 0.3 mmol/liter CaCl2; and agitated with 95% oxygen/5%carbon dioxide. Incubation was repeated three to five times at15-min intervals with fresh enzyme solution. The supernatantfrom each digestion was filtered (220-�m mesh) and centri-fuged (200 to 300 rpm for 2 min). Cells were then stored in aHEPES-buffered Tyrode’s solution (see below) supplementedwith 0.5 mmol/liter Ca2� at room temperature for later use.

The Krebs buffer used in the cell dissociation procedurecontained (mmol/liter): NaCl 118.5, KCl 2.8, NaHCO3 14.5,KH2PO4 1.2, MgSO4 1.2, glucose 11.1. The composition of the

HEPES-buffered Tyrode’s solution was (mmol/liter): NaCl132, KCl 4, CaCl2 2, MgSO4 1.2, HEPES 20, glucose 11.1; pHwas adjusted with NaOH to 7.35. The Na�-, K�- and Ca2�-free external solution contained (mmol/liter): choline chloride140, MgCl2 2.0, HEPES (free acid) 20, glucose 11.1; pH wasadjusted to 7.35 with LiOH. Ethyleneglycol-bis-beta-aminoethylether-N,N�-tetraacetic acid (EGTA [0.5 mmol/liter]) was added in some experiments. The pipette solutioncontained (mmol/liter): potassium aspartate 125, KCl 20,MgCl2 1, adenosine triphosphate (Mg salt) 5, HEPES 5,EGTA 10; pH was adjusted to 7.1 to 7.2 with KOH.

Myocytes were superfused with a HEPES-buffered Tyrode’ssolution (aerated with 100% oxygen) at a flow rate of 2 to3 ml/min. Only relaxed, quiescent cells displaying clear crossstriations were used. All experiments were performed at 35 to37�C (�0.5�C).

The delayed rectifier current was measured using standardwhole-cell patch-clamp techniques. For this study, anAxopatch-1D amplifier with a CV-4 1/100 headstage (AxonInstruments) was used. Suction pipettes made of borosilicateglass (1.5 mm outer diameter and 1.1 mm inner diameter,Becton, Dickinson) were pulled on a Flaming-Brown typepipette puller (Sutter Instrument Co.) and heat polishedbefore use. Pipette tip resistances measured in Tyrode’s solu-tion (passed through a 0.22-�m sterile filter, Millipore Corp.)were 2 to 4 M� when filled with pipette solution. The junctionpotential between the pipette solution and Tyrode’s solutionwas zeroed before the formation of the membrane–pipette sealin normal Tyrode’s (�10 mV, pipette negative). This zeroingcreated an offset equal to the junction potential, but ofopposite sign, that remained after the establishment of whole-cell recording. All voltages in the patch-clamp experimentswere corrected for this offset. Once the suction pipette made agigaohm seal with the cell, the pipette capacitance was partiallyneutralized. The membrane was ruptured by applying addi-tional negative pressure.

The delayed rectifier current was measured by holding thecells at �40 mV, a potential at which Na� current is inacti-vated, and stepping to more positive voltages (�20 to �60mV). The Ca2� current was inhibited by the addition of 4.0�mol/liter nifedipine to the superfusate. Because 0 mmol/liter[K�]o has been shown to increase IKs and greatly diminish IKr

Table 1. Time Controls: Action Potential Duration and QT Intervals Measured in Arterially PerfusedLeft Ventricular Wedge Preparations at End of Equilibration Period (1 h) and at End of a 4-h Periodof Perfusion*

1 h 4 h

No.APD50 APD90 QT APD50 APD90 QT

Epi 186 � 23 230 � 20 188 � 22 231 � 23 7M 236 � 13 286 � 19 309 � 19 238 � 16 288 � 19 310 � 21 7Endo 215 � 7 259 � 9 214 � 11 260 � 10 4PF 267 � 38 356 � 42 273 � 38 361 � 44 5

*Basic cycle length 2,000 ms. Data presented are mean value � SD (ms). APD50 and APD90 � action potentialduration measured at 50% and 90% repolarization using floating microelectrodes; Endo � endocardium; Epi �epicardium; PF � Purkinje fiber.

1838 ANTZELEVITCH ET AL. JACC Vol. 28, No. 7MECHANISMS OF ERYTHROMYCIN-INDUCED LQTS December 1996:1836–48


(20–22), these conditions were used to assess the possibleeffects of erythromycin on IKs. In another series of experimentsdesigned to assess the relative effect of erythromycin on IKrversus IKs, we exposed the cells to erythromycin in the pres-ence and absence of E-4031, a specific IKr blocker (4 mmol/liter [K�]o and 4 �mol/liter nifedipine) (21,23). The transientoutward current was not blocked, but it had little influence onour measurement of IK because of its fast inactivation kinetics(24).

In another set of experiments, steady state current–voltage(I–V) relations were determined using 2-s hyperpolarizing anddepolarizing voltage steps applied from a holding potential of�40 mV. Nifedipine (4.0 �mol/liter) and ouabain (3.0 �mol/liter) were used throughout the course of these experiments toblock the Ca2� and Na�/K� pump currents, respectively; theNa� current was inactivated by the holding potential at�40 mV. Cell capacitance was calculated by integrating thearea under the uncompensated capacitance transient producedby a 10-mV hyperpolarizing step from 0 mV and dividing thisarea by the voltage step. The average access resistance (thesum of the pipette resistance and the residual resistance of theruptured patch) was 5.1 � 0.91 M� (mean � SD, n � 21),estimated by dividing the time constant tau of the decay of thecapacitance transient by the calculated cell membrane capac-itance (25). The membrane currents recorded in the presentstudy were 500 to 600 pA in most cases. Thus, the maximalvoltage error caused by the series resistance would be expectedto be on the order of 2 to 3 mV.

All measurements of IK reported were obtained between 5and 12 min after rupture of the plasma membrane. In previous

studies (19), we demonstrated that rundown of IK is notobserved for at least 15 min after membrane rupture.

A personal computer equiped with 12-bit analog-digital/digital-analog converters (DIGIDATA 1200, Axon Instru-ments, Inc.) was used for data acquisition and generation ofpulse template and command potentials for both current- andvoltage-clamp modes (pCLAMP software, Axon Instruments,Inc.). Currents were filtered with a four-pole Bessel filter at 0.5to 1 kHz and digitized at 1 kHz.

Statistics. Statistical analysis of the data was performedusing one-way analysis of variance coupled with the Scheffe orTukey procedure or with a Student t test (SigmaStat softwarepackage, Jandel Scientific).

Drugs. Erythromycin HCl (Sigma) was prepared freshdaily as a stock solution of 10 or 50 mg/ml. It was added to theTyrode’s perfusate to obtain final concentrations of 10 to 100�g/ml. Nifedipine (Calbiochem-Novabiochem Corp.) andE-4031 (Eisai Co., Ltd., Tokyo, Japan) were prepared freshbefore each use.

ResultsAction potential experiments. Figures 1 to 5 illustrate the

characteristics of the three principal cell types encountered inthe free wall of the canine left ventricle and their responsive-ness to erythromycin. Figure 1 shows the time course ofdevelopment of the actions of erythromycin on transmembraneactivity recorded in a transmural strip isolated from the canine

Figure 1. Time course of effects of erythromycin (Ery) ontransmembrane activity recorded from epicardial (Epi), endo-cardial (Endo) and deep subepicardial (M region) sites in atransmural strip of canine left ventricle. A to C, Each panelshows superimposed action potentials recorded before andafter 30 and 120 min of exposure to erythromycin (50 �g/ml).D, Graphic representation of the time course of erythromycin-induced changes in APD90 in the same experiment. BCL �8,000 ms.



left ventricle. Shown are recordings obtained from epicardial,endocardial and deep subepicardial (M region) sites beforeand during a 2-h period of exposure to 50 �g/ml of erythro-mycin. At slow stimulation rates, the effects of erythromycin toprolong the APD are greater in cells from the M region thanin those from the epicardial or endocardial regions of the leftventricular free wall. The APD prolongation reaches a quasi-steady state after 30 min of exposure to the drug, although avery slow progressive increase is observed in the succeeding90-min period. Qualitatively similar results were obtained intwo other experiments. On the basis of these results, wedesigned the protocols for assessing dose–response relation(Fig. 2).

Figure 2 illustrates the dose dependence of the actions oferythromycin to prolong the APD in the three cell types

recorded along the length of a transmural left ventricularpreparation. Transmembrane recordings were obtained beforeand 30 min after each increase in the concentration of the drug(10 to 100 �g/ml). The data demonstrate a clear dose-dependent effect of erythromycin to prolong the APD in the Mcell but little effect of the drug on the APD in epicardial andendocardial tissues.

The effect of erythromycin to delay repolarization of theaction potential is a sensitive function of rate. As Figure 3illustrates, the dependence of the APD on rate is greatlyexaggerated in the presence of erythromycin. This effect of thedrug is barely noticeable in epicardial cells, modest in endo-cardial cells and profoundly exaggerated in M cells. Erythro-

Figure 3. Rate dependence of effects of erythromycin (Ery) ontransmembrane activity recorded from epicardial (Epi), endocardial(Endo) and deep subepicardial (M cell) sites in a transmural strip ofcanine left ventricle. Each panel shows superimposed action potentialsrecorded at BCLs of 1,000, 2,000, 3,000, 4,000, 5,000 and 8,000 ms. A,Control. B, Recorded after 30 min of exposure to 10 �g/ml oferythromycin. C, Recorded 30 min after increasing concentration oferythromycin to 100 �g/ml.

Figure 2. Dose dependence of effects of erythromycin to prolong theaction potential at epicardial (Epi), endocardial (Endo) and deepsubepicardial (M region) sites in a transmural strip of canine leftventricle. Average value of APD90 is plotted as a function of erythro-mycin concentration. Data were collected after 30 min of exposure toeach concentration of drug. BCL � 8,000 ms. Data shown are meanvalue � SD.

Figure 4. Rate dependence of erythromycin (Ery)-induced prolonga-tion of APD at epicardial (Epi), endocardial (Endo) and deepsubepicardial (M cell) sites in transmural preparations of canine leftventricle. Each panel plots APD90 as a function of BCL, recordedunder control conditions, after 30-min exposure to 10 �g/ml oferythromycin and 30 min after increasing the concentration of eryth-romycin to 100 �g/ml. Data shown are mean value � SD.



mycin prolonged the APD of the M cell to 370 ms at a BCL of1,000 ms and to 1,800 ms at the slowest rate tested (BCL of8,000 ms).

Summary data collected from preparations in which APD-rate relations were examined in the absence and presence oferythromycin (10 and 100 �g/ml) are presented in Figure 4.The effect of erythromycin to markedly prolong the APD in theM cell is clearly reverse rate (or use) dependent, whereas themuch more modest effect of the drug to delay repolarization inepicardial and endocardial cells appears to be rate indepen-dent. These distinctions are maintained even when a singlepremature beat is used to examine the effects of erythromycinon the restitution of the APD at epicardial and M region sites(data not shown).

High concentrations of erythromycin (50 to 100 �g/ml)

induced EADs in cells recorded from the M region (2 of 10preparations) but not in those from the epicardial or endocar-dial regions of the canine left ventricular free wall. Figure 5Aillustrates an example of the differential effects of erythromycin(100 �g/ml) on the three cell types in a transmural preparation.At slow stimulation rates, the erythromycin-induced EADs inthe M cells gave rise to triggered responses that often propa-gated to neighboring epicardial and endocardial regions, thusgenerating extrasystolic activity. The rate dependence oferythromycin-induced triggered activity and the ability of thetriggered response originating in the M cell to reexcite ventric-ular epicardium are illustrated in Figure 5B.

Long QT intervals and TdP. The arrhythmogenic potentialof these disparate cellular actions of erythromycin was furtherevaluated using an arterially perfused wedge of canine leftventricle in which we are able to simultaneously record trans-membrane responses from several intramural sites along witha transmural ECG. Floating microelectrodes were used torecord action potentials from subendocardial Purkinje, Mregion and epicardial cells. Figure 6 graphically illustrates theeffect of erythromycin (100 �g/ml) to prolong the APD mea-sured at 90% repolarization (APD90) in the three cell types. Aswith the results obtained using tissue slices, the response of theM cell action potential is much greater than that of epicar-dium. The drug-induced prolongation of the Purkinje actionpotential is greater than that of the M cell, consistent with theresults previously reported by Rubart et al. (10). Not unexpect-edly, the QT interval changes parallel those of the M cellaction potential. Much of the difference is attributable toimpulse conduction time to the M region and the difference

Figure 5. Erythromycin induces EADs and triggered activity in cellslocated in the deep subepicardial (M region) but not in epicardial(Epi) or endocardial (Endo) cells. A, Transmembrane responsesrecorded from endocardial, deep subepicardial and epicardial sites ina transmural strip of canine left ventricle. Traces were recorded afterexposure of preparation to 100 �g/ml of erythromycin for 30 min;BCL � 2,000 ms. B, Rate dependence of erythromycin-inducedtriggered activity. Shown are transmembrane responses from deepsubepicardial and epicardial sites in a transmural canine left ventric-ular preparation recorded after 30 min of exposure to 100 �g/ml oferythromycin. Left, At a BCL of 2,000 ms, an EAD is apparent in theM region but not in epicardium. Middle, At a BCL of 3,000 ms, theEAD gives rise to a triggered response that occurs too early topropagate to the epicardial site. Right, At a BCL of 4,000 ms, a secondtriggered response appears that propagates successfully, causing reex-citation of the epicardial region of the preparation.

Figure 6. Summary data of effects of erythromycin (100 �g/ml) on theECG QT interval and APD90 simultaneously recorded from epicardial(Epi), M region (M cell) and subendocardial Purkinje cells in isolatedarterially perfused canine left ventricular wedge preparations. Thepreparations (n � 4) were paced from an endocardial site at a BCL of2,000 ms. Note that the change in QT interval parallels that of APD90of the M cell: The difference is largely accounted for by the timenecessary for the impulse initiated at the endocardial surface to reachthe M region (usually 10 to 20 ms). The APD90 of subendocardialPurkinje fibers was longer than the QT interval, suggesting little if anycontribution of the Purkinje system to the ECG.



between APD90 and the APD at 100% repolarization. Theseresults indicate that the erythromycin-induced long QT inter-val is attended by the development of a marked dispersion ofrepolarization across the ventricular wall, even in intact wedgesof left ventricle where the myocardial cells are electrically wellcoupled. As expected, electronic forces in the wedge serve todamp the dispersion of repolarization across the wall. Com-parison of APD90 values presented in Figures 4 and 6 (BCL2,000 ms) indicate that under control conditions, the averageAPD90 of the M cell is 37 ms shorter in the wedge than in thetissue slices (279 � 9 vs. 316 � 11 ms), whereas the APD90 ofthe epicardial cell is nearly 19 ms longer in the wedge than inthe tissue slices (231 � 10 vs. 212 � 9 ms) and that aftererythromycin (100 �g/ml), the average APD90 of the M cell is131 ms shorter in the wedge than in the tissue slices (357 � 11vs. 488 � 107 ms), whereas the APD90 of the epicardial cell isnearly 52 ms longer in the wedge than in the tissue slices(280 � 18 vs. 228 � 24 ms).

In a series of four experiments we used programmedstimulation to assess the arrhythmogenic potential of theseactions of erythromycin. Figure 7 presents an example ofinduction of an episode of polymorphic ventricular tachycardiain which the QRS complex is seen to twist about the isoelectricline, typical of TdP. A premature beat elicited at an S1–S2interval of 220 ms is observed to initiate what can best bedescribed as an intramural reentry showing a 4:3 Wenckebachconduction of the impulse from epicardium to the M region.

Figure 8 illustrates an example of a longer run of TdPrecorded in the same preparation. Erythromycin (100 �g/ml)produced a marked dispersion of repolarization across thewall, with the M cell manifesting much longer APDs thanepicardium. A premature stimulus applied to epicardium at anS1–S2 of 220 ms initiates a run of TdP that persists for �7 sbefore self-terminating. Similar results were obtained in threeof four preparations exposed to erythromycin (100 �g/ml). Inall cases, basic stimuli were applied to the endocardial surfaceat a BCL of 2,000 ms, and the premature beats were applied tothe epicardial surface at S1–S2 intervals of 200 to 270 ms.Although TdP was inducible at shorter BCLs (e.g., 1,500 ms),we did not systematically evaluate the stimulation criteria inthis study. In the three experiments in which TdP was readilyinduced with an S2 delivered to the epicardium, TdP could notbe induced when the S2 was delivered to the endocardial or Mregion (twice diastolic threshold intensity). When the stimulusintensity was increased to 5 to 10 times diastolic threshold, TdPcould be observed after endocardial stimulation (one of three).The only preparation that failed to manifest TdP was thesmallest of the four. Prominent EADs or EAD-induced trig-gered activity were never observed. Programmed stimulationapplied during the predrug control period or in our timecontrols (4 h of perfusion) failed to induce TdP (0 of 11preparations).

Ionic current experiments. The mechanism underlying theeffect of erythromycin to delay repolarization and to induceEAD and triggered activity in canine ventricular M cells wasprobed in a series of experiments using whole-cell patch-clamptechniques applied to myocytes isolated from the M region of

Figure 7. Erythromycin-induced TdP-like polymorphic ventriculartachycardia in an isolated arterially perfused canine left ventricularwedge. The preparation was paced from the endocardial surface at aBCL of 2,000 ms (S1). Shown are transmembrane responses obtainedfrom Purkinje, M region and epicardial (Epi) sites together with atransmural ECG, all simultaneously recorded after exposure of thepreparation to erythromycin (100 �g/ml) for 30 min. An extrastimulus(S2) applied to the epicardial surface at an S1–S2 interval of 220 mspropagates to the M region with a large delay because of the longerrefractory period of the midmyocardial region and initiates a polymor-phic tachyarrhythmia characterized by a 4:3 Wenckebach pattern ofconduction between epicardial and M cell recording sites. The se-quence of events is suggestive of an intramural reentry with beat tobeat variability in either conduction velocity or the path of thereentrant circuit, or both. Termination of the arrhythmia is due toblock of the impulse between epicardial and M regions.

Figure 8. A long episode of TdP induced by erythromycin. Samepreparation as in Figure 7. The recordings and stimulation protocolare also the same. Preferential prolongation of M cells in response toerythromycin (100 �g/ml for 45 min) resulted in prolongation of theQT interval (394 ms) and the development of a marked dispersion ofrepolarization across the ventricular wall. An extrastimulus (S2)applied to the epicardial (Epi) surface at an S1–S2 interval of 220 msinitiates a polymorphic tachyarrhythmia with characteristics typical ofTdP. The sequence of events is suggestive of a migrating reentry withvariable penetration into the M region and Purkinje network.



the canine left ventricular free wall. The primary focus of thesestudies was on the outward potassium currents IKr and IKs andthe inward rectifier current (IK1).

Effect of erythromycin on IK. IK was initially recorded incells bathed in Tyrode’s solution containing 4 mmol/liter [K�]oand 4 �mol/liter nifedipine using either 5,000- or 250-msdepolarizing pulses applied from a holding potential of �40mV to progressively more positive potentials (�20 to �60mV). Figure 9 shows the effect of erythromycin on IK, recordedusing 5,000-ms pulses. Figure 9A shows typical currents re-corded before and after the addition of erythromycin (100�g/ml). The drug significantly inhibits both the developing(Fig. 9B) and tail currents (Fig. 9C). Similar results wereobtained in another group of cells in which IK was measuredusing 250-ms pulses (Fig. 10). With both protocols, erythromycin-induced inhibition of the tail current is greater than that of thedeveloping current. These results are consistent with a predomi-nant effect of erythromycin to inhibit IKr.

Effect of erythromycin on IKr versus IKs. As a test of thishypothesis, we performed an envelope of tails test in thepresence and absence of erythromycin (100 �g/ml). If IK is dueto conductance of a single type of channel, the envelope of tailstest (26) predicts that the magnitude of tail currents recordedafter depolarizing pulses of progressively longer durationshould increase parallel to the time course of activation of thedeveloping outward current recorded during the pulse. Inother words, the ratio of tail current to time-dependentdeveloping current should be constant, regardless of the pulseduration. The data presented in Figure 11 are from midmyo-cardial cells depolarized from a holding potential of �40 to

�40 mV for pulse durations ranging between 200 and5,000 ms. The interpulse interval was 20 s, sufficiently long topermit complete deactivation of the current. In the absence oferythromycin, the ratio of the tail and developing currents isnearly constant (0.46) for pulse durations 1,000 ms butincreases progressively with shorter pulses. In the presence oferythromycin (100 �g/ml), the ratio of the tail and developingcurrents was constant (0.38) at all pulse durations. This valueis close to the predicted ratio (0.4) calculated from the ratio ofthe driving force at �40 and �40 mV for a nonrectifying,K�-selective outward current. Thus, the envelope of tails testwas satisfied only in the presence of 100 �g/ml of erythromycin.Once again the data point to a predominant effect of erythro-mycin to reduce IKr.

To assess the effect of erythromycin on IKs, we measuredthe effect of erythromycin on IK in the presence of themethanesulfonamide E-4031 (5 �mol/liter) and in Tyrode’ssolution containing 0 mmol/liter [K�]o. Both interventions areknown to largely eliminate IKr (19,21,27). Figure 12 (A and D)illustrates representative current tracings recorded in cellsbathed in Tyrode’s solution containing 5.0 �mol/liter E-4031(Fig. 12A) or 0 mmol/liter [K�]o (Fig. 12D) in the absence andpresence of erythromycin. From a holding potential of �40mV, the cells were depolarized to progressively more positivepotentials for 5,000 ms and then returned to �40 mV. Thedeveloping current as a function of test potential before andafter erythromycin (100 �g/ml) in the presence of 5.0 �mol/liter E-4031 (n � 7) or 0 mmol/liter [K�]o (n � 6) is plotted inFigure 12, B and E, respectively; the cumulative data for the

Figure 9. Effect of erythromycin on the IK in myocytes isolated fromthe M region of the canine left ventricle. A, Currents were elicited bythe voltage pulse protocol shown in top inset; from a holding potential�40 mV, the cells were depolarized to progressively more positivepotentials for 5,000 ms before (left) and after exposure to erythromy-cin (100 �g/ml) (right). Cells were bathed in Tyrode’s solutioncontaining 4 mmol/liter [K�]o and 4 �mol/liter nifedipine. B, Devel-oping currents measured at end of 5,000-ms pulse are plotted as afunction of test pulse voltage. C, Tail currents measured on return ofmembrane potential to �40 mV from the indicated test potentials.Data shown are mean value � SE. *p 0.05.

Figure 10. Effect of erythromycin on IKr in myocytes isolated from theM region of the canine left ventricle. A, Currents recorded during250-ms pulses from a holding potential of �40 to �20, 0, �20, �40and �60 mV before (left) and after exposure to erythromycin (100�g/ml) (right). Cells were bathed in Tyrode’s solution containing 4mmol/liter [K�]o and 4 �mol/liter nifedipine. B, Developing currentmeasured at end of 250-ms pulse as a function of test pulse voltage. C,Tail current measured on return of membrane potential to �40 mVfrom the indicated test potentials. Open circles � control; solidcircles � erythromycin (100 �g/ml). Data shown are mean value � SE.*p 0.05.



tail currents are plotted in Figure 12, C and F. These resultsindicate that 100 �g/ml of erythromycin exerts no effect on IKs.The absence of IKr was substantiated by the fact that theenvelope of tails test is satisfied under these conditions (notshown).

Effect of erythromycin on steady state I–V relation. Theeffect of erythromycin on IK1 was assessed in another series ofsix experiments. Steady state I–V relations were constructed byplotting the current levels recorded at the end of 2-s pulsesfrom a holding potential of �40 mV to test potentials rangingbetween �100 and 0 mV. Measurements were made in thepresence of nifedipine (4.0 �mol/liter) to inhibit inward Ca2�

current and ouabain (3.0 �mol/liter) to inhibit Na�/K� pumpcurrent before and after erythromycin (100 �g/ml). TheNa� current was inactivated by a holding potential of �40 mV.Figure 13A shows the representative current tracings recordedbefore and after exposure of the cell to erythromycin. Figure13B shows cumulative data of current density as a function of

test voltages before and after erythromycin. Clearly, the drugexerts no effect on steady state I–V relations, suggesting thaterythromycin has no effect on IK1.

DiscussionOur findings demonstrate for the first time an effect of

erythromycin, a widely prescribed macrolide antibiotic, toproduce prominent action potential prolongation and EAD-induced triggered activity in M cells but not epicardial orendocardial cells in the free wall of the canine left ventricle.The marked dispersion of repolarization created across theventricular wall is shown to set the stage for the developmentof polymorphic ventricular arrhythmias displaying the ECGmanifestation of TdP. Our voltage-clamp results point toerythromycin-induced block of IKr as a prominent mechanismcontributing to the prolongation of repolarization in the Mcells.

In the canine heart, endocardial, epicardial and M regionaction potentials differ principally with respect to repolariza-tion characteristics. Epicardial and M cell action potentialscommonly display a spike and dome configuration due to thepresence of a distinct early repolarization phase (phase 1). Inaddition, M cell action potentials differ from those of epicar-dial and endocardial cells with respect to phase 3, showing adelayed repolarization phase, especially at slow stimulationrates. A smaller contribution of IKs has been shown (19) toparticipate in determining this unique repolarization proper-ties of the M cell. This ionic distinction is also thought tocontribute to the exceptional sensitivity of the M cell to a widevariety of agents with class III actions (5,13,14). Our findingsindicate that erythromycin, like other APD-prolonging agents,targets the M cells in the deep structures of the ventricularmyocardium. The effects of erythromycin on the action poten-tial of the M cell are similar to those previously described (10)in canine Purkinje fibers. In both isolated Purkinje and M celltissues, erythromycin produces a major dose-dependent pro-longation of the APD, as well as EADs and EAD-inducedtriggered activity. The EADs were observed in only 20% of Mcell tissue slices. In contrast, erythromycin in concentrations aslarge as 100 �g/ml never produced EAD or triggered activityand only a modest prolongation of APD in canine ventricularepicardial and endocardial tissue slices. Because EAD activitywas observed in only 20% of M cell tissues and in 0% ofendocardial or epicardial preparations, it is not surprising thatthe drug produced no EADs in the arterially perfused leftventricular wedge, where the electrotonic influences of epicar-dium and endocardium would be expected to diminish theextent of APD prolongation and prevent the appearance ofEADs in the M region. The higher levels of interstitial [K�]0expected in the intact wall preparation may have also contrib-uted to the absence of erythromycin-induced EAD activity inthe arterially perfused left ventricular wedge preparation (28).

Action potential prolongation and EADs are known toresult from 1) a reduction in the availability of K� currents thatcontribute to repolarization (IK and IK1); 2) an increase in the

Figure 11. Effect of erythromycin on the envelope of tails test for IK.A, Superimposed current traces recorded before (Control) and afteradministration of 100 �g/ml of erythromycin during the voltage pulseprotocol shown in top inset. Developing current amplitude (Ik) wasmeasured at �40 mV at the end of test pulses of variable duration (200to 5,000 ms). Tail current amplitude (Ik tail) was measured at �40 mVafter each test pulse. B, Ratio of delayed rectifier tail current todeveloping current recorded before (Control) and after erythromycinis plotted as a function of pulse duration. Trace shown in inset wasrecorded in the presence of erythromycin. Data shown are meanvalue � SE.



availability of the inward Ca2� current; 3) a delay in Na�

current inactivation, giving rise to an increase in late Na�

current (5); or 4) an increase in the contribution of electro-genic current generated by the Na�–Ca2� exchanger (29).Rubart et al. (10) reported that in Purkinje fibers, erythromy-cin (100 �g/ml) dramatically prolongs the APD but does notaffect maximal rate of rise of the action potential upstroke(Vmax), action potential amplitude or developed tension. Theyconcluded that the effects of erythromycin on APD are un-likely to be mediated through actions of the drug on fast Na�

and Ca2� inward currents. Augmentation in the level oftetrodotoxin-sensitive slowly inactivating Na� current (lateNa� current or window current) (30–33) was also discountedby Rubart et al. (10) on the basis of the effects of erythromycinafter tetrodotoxin. Pretreatment of the Purkinje fibers witherythromycin was shown to antagonize the effect of dofetilide,an IKr blocker. On the basis of this observation, Rubart et al.suggested that the effect of erythromycin to prolong the actionpotential is through block of IK. In the present study, wedemonstrated a similar effect of erythromycin on the actionpotential of M cells and provide a direct test of the hypothesisthat IK block contributes to these changes.

In some species, the delayed rectifier potassium current iscomposed of two components or currents (IKr and IKs)(19,21,23). The two components are distinguished by theiractivation kinetics, rectification and sensitivity to methanesul-fonamide agents displaying class III actions, such as E-4031.Daleau et al. (12) recently showed that erythromycin blocks IKrbut not the inward Ca2� current or IK1 in guinea pig ventric-ular myocytes (unknown ventricular origin). The two compo-nents of IK are larger in guinea pig than in dog ventricularmyocytes under control conditions (19). Our data show thaterythromycin (100 �g/ml) inhibits IK elicited by stepping froma holding potential of �40 mV to progressively more positivepotentials in M cells (Figures 9 and 10). The envelope of tailstest is satisfied only in the presence of erythromycin (Figure11), suggesting that erythromycin is a potent IKr blocker in thecanine ventricle. Under conditions designed to eliminate IKr(5.0 �mol/liter E-4031 or 0 mmol/liter [K�]o), erythromycin(100 �g/ml) exerts no effect on IK or the envelope of tails test(Figure 12), indicating that IKs is unaffected by the drug. It iswell known that IK1 also plays an important role in theregulation of the APD. A lack of an effect of erythromycin onthe steady state I–V relations (Figure 13), indicates that

Figure 12. Effect of erythromycin on IKs. Myocytesisolated from the M region of the canine leftventricle were bathed in Tyrode’s solution contain-ing either 5 �mol/liter E-4031 (left) or 0 mmol/liter[K�]o (right) to eliminate IKr. A and D, Currenttraces recorded during voltage protocols illustratedin top insert before (top) and after addition of 100�g/ml of erythromycin (bottom). B and E, Devel-oping current recorded at the end of 5,000-ms testpulse as a function of test pulse voltage. C and F,Plot of tail currents measured on return of mem-brane potential to �40 mV as a function of testpulse voltage. Data shown are mean value � SE.



inhibition of IK1 does not contribute to the effects of the drugto prolong the APD in the canine ventricle. In summary, ourwhole-cell patch-clamp data point to IKr block as a mechanismresponsible for the effect of erythromycin to delay repolariza-tion and induce EAD activity, although they do not exclude thepossible participation of other mechanisms.

Physiologic and clinical implications. Our findings showthat the actions of erythromycin to markedly prolong APDsand induce EAD and triggered activity in a select population ofcells (M cells) in ventricular myocardium can lead to thedevelopment of a prominent dispersion of repolarization andrefractoriness within the ventricle, setting the stage for theinduction of tachyarrhythmias such as TdP. Our results ob-tained using the arterially perfused left ventricular wedgesuggest circus movement reentry as the basis for the mainte-nance of TdP. There are several lines of evidence that point toreentry as a likely mechanism: 1) the presence of a markeddispersion of repolarization and refractoriness after erythro-mycin; 2) the ability to most easily induce the arrhythmia using

a single premature beat introduced at the site of earliestrepolarization (epicardium); and 3) the occurrence of main-tained arrhythmic activity only in larger preparation. These areall well established hallmarks of circus movement reentry. Theabsence of any other obvious arrhythmogenic sources such asEAD-induced triggered activity, delayed afterdepolarization(DAD)-induced triggered activity or abnormal automaticity(based on direct recording or expected behavior) also leavesreentry as the most likely mechanism.

The ability to induce TdP in the wedge using programmedelectrical stimulation (PES) applied to epicardium is consistentwith similar observations made in in vivo models of TdP.Programmed electrical stimulation-induced TdP is the rulerather than the exception in recently developed in vivo modelsof acquired long QT syndrome and TdP (34,35). In the clinic,the onset of TdP has long been known to follow a short–long–short cycle length sequence (36,37). Recent clinical reportsindicate that a sudden moderate acceleration from an initiallyslow heart rate when followed by an intrinsic or extrinsicextrasystole holds the highest risk for induction of TdP inpatients with long QT syndrome (38) as well as in animalmodels with acquired long QT syndrome (34,39–41). Thebehavior of our preparation is consistent with both clinical andexperimental observations.

The absence of prominent EADs or EADs-induced trig-gered activity in the left ventricular wedge preparations inwhich TdP was readily and reproducibly inducible arguesagainst the hypothesis that EAD-mediated triggered activity isresponsible for the maintenance of TdP (5). Our resultsprovide support for the hypothesis that TdP is maintainedthrough a reentrant mechanism but leaves open the possibilitythat the arrhythmia may be induced (precipitated) by anEAD-triggered response, as suggested by Rubart et al. (10) orby an extrasystole from some other source (intrinsic or extrin-sic).

Elucidation of the specific mechanism or mechanisms un-derlying the induction and maintenance of TdP requiresfurther studies in which a greater number of intracellular,extracellular or monophasic action potentials are simulta-neously recorded from the arterially perfused wedge prepara-tion. Alternatively, voltage-sensitive dye techniques can beused to probe the complexities of TdP in this preparation. Bothapproaches are currently underway. Preliminary results usingvoltage-sensitive dye recording techniques to generate highresolution maps indicate that intramural reentry is responsiblefor the maintenance of TdP in wedge preparations pretreatedwith IKr blockers (Yan G-X, Rosenbaum D, Akar F, Antzel-evitch C. Unpublished observations).

The disproportionate prolongation of the M cell actionpotential in response to erythromycin may underlie the pro-longation of the QTU interval on the ECG and the develop-ment of TdP in some patients given the drug (6–9,11,42–44),especially those with the congenital long QT syndrome (45).

Summary. We examined the effects of erythromycin inisolated myocytes, tissues and intact left ventricular wall prep-arations so as to integrate information at these various levels.

Figure 13. Effect of erythromycin on the steady state I–V relationsobtained using the voltage step protocol shown at top. A, Represen-tative current traces recorded before (left) and after addition oferythromycin (right) during voltage steps from a holding potential of�40 mV to test potentials ranging between �70 and �100 mV for 2 s.B, Steady state I–V relations established by plotting the current levelmeasured at the end of a 2-s pulse as a function of the test voltage(�100 to 0 mV, 10-mV increments) before (Control) and aftererythromycin (100 �g/ml). Myocytes are from the M region of thecanine left ventricle. Nifedipine (4 �mol/liter) and ouabain (3 �mol/liter) were present throughout the experiment. Data shown are meanvalue � SE.



The results demonstrate how inhibition of an outward currentcan lead to important heterogeneities of repolarization intissues spanning the left ventricular wall, thus setting the stagefor arrhythmias with all the characteristics of TdP. The studyprovides a direct link between the degree of IKr inhibition(similar to that found in patients with LQTS with the chromo-some 7 or HERG [human ether-a-go-go-related gene] defect)and arrhythmogenesis.

Limitations of the study. The concentrations of erythromy-cin used in our study (10 to 100 �g/ml) are higher than thosenormally measured in human plasma (46). This fact notwith-standing, use of these levels in our experimental protocols isnot without clinical relevance for the following reasons: 1)Plasma concentrations of erythromycin have not been mea-sured in the vast majority of cases of erythromycin-inducedlong QT intervals and TdP in the acquired long QT syndrome;2) serum concentrations in healthy volunteers with normalliver function have been shown (46) to reach levels as high as30 �g/ml; 3) in the dozen or more cases described recently(47–49), long QT syndrome usually develops in patients alsotaking disopyramide, terfenadine and amiodarone and immu-nosuppressants, such as pentamidine, each individually capa-ble of prolonging the QT interval; 4) the effect of IKr blockers,of which erythromycin is one, to produce long QT intervals andto precipitate TdP is generally greatly enhanced by hypokale-mia and hypomagnesemia as well as by various forms of heartdisease, including congenital long QT syndrome.

The effect of these agents and disease states to potentiatethe QT-prolonging effects of erythromycin is largely anecdotalbecause of the small number of reported clinical cases. How-ever, numerous clinical and animal studies have demonstratedthe ability of similar drugs, electrolyte imbalances and diseasestates to predispose patients or animals receiving IKr blockersto the development of long QT intervals and TdP.

Although we have interest in assessing these interactions,we considered it more important to establish a baseline for thedose–response relation and to gain an understanding of themechanisms responsible. The experimental protocols used inthis study are in keeping with these objectives.

Our study demonstrates significant prolongation of thecanine M cell action potential with an erythromycin concen-tration as low as 10 �g/ml (the lowest concentration studied[Fig. 2]), with dramatic prolongation of the action potential atconcentrations of 50 and 100 �g/ml. Rubart et al. (10) ob-served a similar response of isolated canine Purkinje fibersover a range of 20 to 200 �g/ml; whereas Daleau et al. (12)observed a much more modest increase in endocardialmonophasic APD in the guinea pig heart using an erythromy-cin concentration of 75 �g/ml (100 �mol/liter).

Erythromycin inhibition of the hepatic P450 metabolic path-way may further predispose some patients to long QTUintervals and TdP because many agents that cause long QTUintervals are metabolized by this route (e.g., terfenadine,astemizole and disopyramide) (50–53).

We are grateful to Judy Hefferon, Tengxian Liu and Robert Goodrow fortechnical assistance.

References1. Jackman WM, Friday KJ, Anderson JL, Aliot EM, Clark MA, Lazzara R.

The long QT syndromes: a critical review, new clinical observations and aunifying hypothesis. Prog Cardiovasc Dis 1988;2:115–72.

2. Moss AJ, Robinson JL. Long QT syndrome. Heart Dis Stroke 1992;1:309–14.

3. Schwartz PJ, Moss AJ, Vincent GM, Crampton RS. Diagnostic criteria forthe long QT syndrome: an update. Circulation 1993;88:782–4.

4. El-Sherif N. Early afterdepolarizations and arrhythmogenesis: experimentaland clinical aspects. Arch Mal Coeur 1991;84:227–34.

5. Antzelevitch C, Sicouri S. Clinical relevance of cardiac arrhythmias gener-ated by afterdepolarizations: the role of M cells in the generation of U waves,triggered activity and torsade de pointes. J Am Coll Cardiol 1994;23:259–77.

6. McComb JM, Campbell NPS, Cleland JGF. Recurrent ventricular tachycar-dia associated with QT prolongation after mitral valve replacement and itsassociation with intravenous administration of erythromycin. Am J Cardiol1984;54:922–3.

7. Guelon D, Bedock B, Chartier C, Haberer JP. QT prolongation andrecurrent “torsades de pointes” during erythromycin lactobionate infusion[abstract]. Am J Cardiol 1986;58:666.

8. Freedman RA, Anderson KP, Green LS, Mason JW. Effect of erythromycinon ventricular arrhythmias and ventricular repolarization in idiopathic longQT syndrome. Am J Cardiol 1987;59:168–9.

9. Nattel S, Talajic M, Lemery R, Roy D. Erythromycin induced long QTsyndrome: concordance with quinidine and underlying cellular electrophysi-ologic mechanism. Am J Med 1990;89:235–8.

10. Rubart M, Pressler ML, Pride HP, Zipes DP. Electrophysiological mecha-nisms in a canine model of erythromycin-associated long QT syndrome.Circulation 1993;88:1832–44.

11. Farrar HC, Walsh-Sukys MC, Kyllonen K, Blumer JL. Cardiac toxicityassociated with intravenous erythromycin lactobionate: two case reports anda review of the literature. Pediatr Infect Dis J 1993;12:688–91.

12. Daleau P, Lessard E, Groleau M, Turgeon J. Erythromycin blocks the rapidcomponent of the delayed rectifier potassium current and lengthens repo-larization of guinea pig ventricular myocytes. Circulation 1995;91:3010–6.

13. Sicouri S, Antzelevitch C. Afterdepolarizations and triggered activity de-velop in a select population of cells (M cells) in canine ventricular myocar-dium: the effects of acetylstrophantidin and Bay K 8644. PACE 1991;14:1714–20.

14. Sicouri S, Antzelevitch C. Drug-induced afterdepolarizations and triggeredactivity occur in a discrete subpopulation of ventricular muscle cell (M cells)in the canine heart: quinidine and digitalis. J Cardiovasc Electrophysiol1993;4:48–58.

15. Antzelevitch C, Sicouri S, Lukas A, Nesterenko VV, Liu DW, Di Diego JM.Regional differences in the electrophysiology of ventricular cells: physiolog-ical and clinical implications. In: Zipes DP, Jalife J, editors. CardiacElectrophysiology: From Cell to Bedside. Philadelphia: WB Saunders,1994:228–45.

16. Zhang ZQ, Antzelevitch C. Erythromycin produces prominent action poten-tial prolongation and early afterdepolarization (EAD)-induced triggeredactivity in M but not epicardial or endocardial regions of the canine ventricle[abstract]. Circulation 1993;88 Suppl I:I-327.

17. Fishbein MC, et al. Early phase acute myocardial infarct size quantification:validation of the triphenyl tetrazolium chloride tissue enzyme stainingtechnique. Am Heart J 1995;101:593–600.

18. Chilian WM, Eastham CL, Marcus ML. Microvascular distribution ofcoronary vascular resistance in beating left ventricle. Am J Physiol 1986;251:H779–88.

19. Liu DW, Antzelevitch C. Characteristics of the delayed rectifier current (IKrand IKs) in canine ventricular epicardial, midmyocardial and endocardialmyocytes: a weaker IKs contributes to the longer action potential of the Mcell. Circ Res 1995;76:351–65.

20. Sanguinetti MC, Jurkiewicz NK: Role of external Ca2� and K� in gating ofcardiac delayed rectifier K� currents. Pflugers Arch 1992;420:180–6.

21. Sanguinetti MC, Jurkiewicz NK. Two components of cardiac delayed



rectifier K� current: differential sensitivity to block by class III antiarrhyth-mic agents. J Gen Physiol 1990;96:195–215.

22. Scamps F, Carmeliet E: Delayed K� current and external K� in singlecardiac Purkinje cells. Am J Physiol 1989;257:C1086–92.

23. Sanguinetti MC, Jurkiewicz NK. Delayed rectifier outward K� current iscomposed of two currents in guinea pig atrial cells. Am J Physiol 1991;260:H393–9.

24. Liu DW, Gintant GA, Antzelevitch C. Ionic bases for electrophysiologicaldistinctions among epicardial, midmyocardial, and endocardial myocytesfrom the free wall of the canine left ventricle. Circ Res 1993;72:671–87.

25. Sigworth FJ. Electronic design of the patch clamp. In: Sakmann B, Neher E,editors. Single-Channel Recording. New York: Plenum, 1983:3–35.

26. Noble D, Tsien RW. Outward membrane currents activated in the plateaurange of potentials in cardiac Purkinje fibers. J Physiol (Lond) 1969;200:205–31.

27. Sanguinetti MC, Jurkiewicz NK, Scott AL, Siegl PKS. Isoproterenol antag-onizes prolongation of refractory period by the class III antiarrhythmic agentE-4031 in guinea pig myocytes. Circ Res 1991;68:77–84.

28. Yang T, Roden DM. Extracellular potassium modulation of drug block ofIKr: implications for torsade de pointes and reverse use-dependence. Circu-lation 1996;93:407–11.

29. Szabo B, Sweidan R, Rajagopalan C, Lazzara R. Role of Na�: Ca2�

exchange current in Cs�-induced early afterdepolarizations in Purkinjefibers. J Cardiovasc Electrophysiol 1994;5:933–44.

30. Attwell D, Cohen IS, Eisner DA, Ohba M, Ojeda C. The steady-statetetrodotoxin-sensitive (“window”) sodium current in cardiac Purkinje fibers.Pflugers Arch 1979;379:137–42.

31. Gintant GA, Daytner NB, Cohen IS. Slow inactivation of a tetrodotoxin-sensitive current in canine cardiac Purkinje fibers. Biophys J 1984;45:509–12.

32. Carmeliet E. Slow inactivation of the sodium current in rabbit cardiacPurkinje fibers. Pflugers Arch 1987;408:18–26.

33. Kiyosue T, Arita M. Late sodium current and its contribution to actionpotential configuration in guinea pig ventricular myocytes. Circ Res 1989;64:389–97.

34. Vos MA, Verduyn SC, Gorgels APM, Lipcsei GC, Wellens HJ. Reproduc-ible induction of early afterdepolarizations and torsade de pointes arrhyth-mias by d-sotalol and pacing in dogs with chronic atrioventricular block.Circulation 1995;91:864–72.

35. Verduyn SC, Vos MA, Gorgels APM, van der Zande J, Leunissen JDM,Wellens HJ. The effect of flunarizine and ryanodine on acquired torsades depointes arrhythmias in the intact canine heart. J Cardiovasc Electrophysiol1995;6:189–200.

36. Coumel P. Early afterdepolarizations and triggered activity in clinicalarrhythmias. In: Rosen MR, Janse MJ, Wit AL, editors. Cardiac Electro-physiology: A Text Book. Mount Kisco (NY): Futura, 1990:387–411.

37. Moss AJ. Long QT syndrome. In: Podrid PJ, Kowey PR, editors. CardiacArrhythmia: Mechanisms, Diagnosis and Management. Baltimore: Williams& Wilkins, 1995:1110–20.

38. Locati EH, Maison-Blanche P, Dejode P, Cauchemez B, Coumel P. Spon-taneous sequences of onset of torsade de pointes in patients with acquiredprolonged repolarization: quantitative analysis of Holter recordings. J AmColl Cardiol 1995;25:1564–75.

39. Cobbe SM, Hoffman E, Ritzenhoff A, Brachmann J, Kubler W, Senges J.Action of sotalol on potential reentrant pathways and ventricular tachyar-rhythmias in conscious dogs in the late postmyocardial infarction phase.Circulation 1983;68:865–71.

40. Habbab MA, El-Sherif N. TU alternans, long QTU, and torsade de pointes:clinical and experimental observations. PACE 1992;15:916–31.

41. Sasyniuk BI, Brunet S. Torsade de pointes induced by quinidine, d-sotalol,and E-4031 in the isolated rabbit heart: importance of interval dependentdispersion of repolarization [abstract]. PACE 1995;18 Suppl:II-904.

42. Keren A, Tzivoni D, Gavish D, et al. Etiology, warning signs and therapy oftorsades de pointes. Circulation 1981;64:1167–74.

43. Guelon D, Bedock B, Chartier C, Haberer JP. QT prolongation andrecurrent “torsade de pointes” during lactobionate erythromycin infusion.Am J Cardiol 1986;58:666.

44. Schoenenberger RA, Haefeli WE, Weiss P, Ritz RF. Association of intra-venous erythromycin and potentially fatal ventricular tachycardia with QTprolongation (torsades de pointes). BMJ 1990;300:1375–6.

45. Freedman RA, Anderson KP, Green LS, Mason JW. Effect of erythromycinon ventricular arrhythmias and ventricular repolarization in idiopathic longQT syndrome. Am J Cardiol 1987;59:168–9.

46. Austin KL, Mather LE, Philpot CR, McDonald PJ. Intersubject anddose-related variability after intravenous administration of erythromycin.Br J Clin Pharmacol 1980;10:273–9.

47. Nightingale SL. New antiarrhythmic approved. JAMA 1992;268:3418.48. Brandriss MW, Richardson WS, Barold SS. Erythromycin-induced QT

prolongation and polymorphic ventricular tachycardia (torsade de pointes):case report and review. Clin Infect Dis 1994;18:995–8.

49. Barold SS. Torsade de pointes induced by erythromycin. Intelligence Re-ports in Cardiac Pacing and Electrophysiology 1994;13:5.

50. Woosley RL, Chen YW, Freiman JP, Gillis RA. Mechanism of the cardio-toxic actions of terfenadine. JAMA 1993;269:1532–6.

51. Honig PK, Woosley RL, Zamani K, Conner DP, Cantilena LR. Changes inthe pharmacokinetics and electrocardiographic pharmacodynamics of terfe-nadine with concomitant administration of erythromycin. Clin PharmacolTher 1992;52:231–8.

52. Echizen H, Kawasaki H, Chiba K, Tani M, Ishizaki T. A potent inhibitoryeffect of erythromycin and other macrolide antibiotics on the mono-N-dealkylation metabolism of disopyramide with human liver microsomes.J Pharmacol Exp Ther 1993;264:1425–31.

53. Roden DM. The study of pharmacokinetics and pharmacodynamics as a toolfor understanding mechanisms of antiarrhythmic drug action. In: BreithardtG, Borggrefe M, Camm J, Shenasa M, editors. Antiarrhythmic Drugs:Mechanisms of Antiarrhythmic and Proarrhythmic Actions. Berlin:Springer-Verlag, 1995:290–9.



Documents

03774