Studies in animal models, as well as clinical experience with amiodarone and sotalol, suggest that action potential prolongation may be a useful antiarrhythmic mode of action. A number of agents that produce this class III effect are currently under development. The single greatest liability for further development of this group of drugs is the occasional, and apparently unpredictable, development of exaggerated QT prolongation and polymorphic ventricular tachycardia (torsades de pointes). Available data suggest that QT interval prolongation is not a good indicator of whether or not a class III antarrhythmic will suppress a target arrhythmia; however, exaggerated QT prolongation is a predictor of torsades de pointes. Further studies to delineate the mechanism underlying the development of torsades de pointes might lead to safer and more effective antiarrhythmic drugs. (Am J Cardiol 1993;72:44B-449B) The concept of action potential prolongation as a useful antiarrhythmic mode of action was first advanced over 2 decades ago with the initial characterization of the electrophysiologic actions of amiodarone and sotalol.[1-3] This class III mode of antiarrhythmic drug action has since been supported by the generally favorable clinical experience with these 2 compounds and by the Cardiac Arrhythmia Suppression Trial (CAST),[4] which indicated that potent sodium channel blockers could increase arrhythmic mortality. Preclinical studies conducted in the past decade have provided a further strong rationale for selective action potential prolongation as a desirable mode of antiarrhythmic drug action. For example, in the dog model developed by Lucchesi, flecainide is, at best, no more effective than placebo in preventing ischemia-related ventricular fibrillation.[5] In contrast, several of the agents to be discussed, including sematilide, racemic d,l and d-sotalol, and amiodarone, are all > 50-60% effective in preventing ventricular fibrillation (VF), compared with approximately 10% efficacy for placebo.[6-8] Data from a number of laboratories have demonstrated that sodium channel block by lidocaine or encainide is associated with a marked increase in the energy required to defibrillate the dog heart.[9-11] In contrast, such compounds as N-acetyl-procainamide (NAPA), clofilium, and cesium, whose predominant action is to block potassium currents and thereby prolong cardiac action potentials, all appear to decrease the energy required to defibrillate the heart. Moreover, potassium channel block appears to make ventricular fibrillation more difficult to establish in the dog heart.[12] Finally, action potential prolongation usually has no effect on or increases cardiac contractility.[13] This is in contrast to sodium channel block, which usually depresses, or at best does not alter, cardiac contractility. The clinical use of action potential prolongation as a viable mode of antiarrhythmic drug action is thus well supported by preclinical and clinical data. A major concern regarding the development of selective action potential prolongation, however, is the occasional, and generally unpredictable, occurrence of marked QT prolongation and polymorphic ventricular tachycardia (torsades de pointes), which could offset any advantage conferred by class III drugs. For this report, clinical managers of each of the compounds listed in Table I were contacted. They have provided the data presented here, much of which has not yet appeared in the peer-reviewed literature. [TABULAR DATA I OMITTED] STRUCTURAL CONSIDERATIONS Although amiodarone and sotalol prolong cardiac action potentials, they also exert other, clinically important pharmacologic effects. Amiodarone blocks calcium channels and inactivated sodium channels in the heart and exerts a noncompetitive antiadrenergic effect[14-16]; sotalol is also a [Beta] blocker.[2,3,17] The d-isomer of sotalol also prolongs cardiac action potentials and is only a weak [Beta] blocker.[17,18] One of the first compounds to be recognized as a relatively pure action potential-prolonging agent was NAPA, the major metabolite of procainamide. Jaillon and Winkle[19] showed in the late 1970s that whereas infusion of procainamide prolonged both the HV interval (an index of sodium channel block) and the QT interval (an index of cardiac repolarization), NAPA prolonged only the QT interval. Subsequent in vitro studies showed that only high concentrations of NAPA exerted any effect on [V.sub.max] (an index of sodium channel activity) in canine cardiac Purkinje fibers or ventricular muscle.[20] In contrast, relatively low concentrations markedly prolonged action potential duration. Clinical trials with NAPA revealed a relatively high incidence of gastrointestinal side effects and low efficacy against premature ventricular contractions and nonsustained ventricular tachycardia (common endpoints for assessment of antiarrhythmic drug action in the early 1980s).[21-23] (NAPA remains under investigation as an antiarrhythmic agent, but will not be further discussed here.) The structure of NAPA and that of sotalol provided a template for synthesis of sematilide (Figure 1).[24] The aromatic methanesulfonanilide motif of sotalol and sematilide is repeated in dofetilide, E4031, ibutilide, L706,000, and WAY123,398. As will be discussed, these compounds, except ibutilide, appear to act through a similar mechanism at the level of the ion channel. An important determinant of potency may be the length of aliphatic substitutions at a tertiary nitrogen position, in analogy to structurally simpler potassium channel blocking agents such as tetraethylammonium, its analogues, and clofilium. Not all drugs discussed here have this methanesulfonanilide aromatic motif. Some, such as almokalant and MS551, have substituted for the methanesulfonanilide another electron-withdrawing group such as CN or [NO.sub.2]. In other cases, no structural similarity to sotalol or NAPA is apparent. As will be discussed, multiple ion channel mechanisms contribute to cardiac repolarization; until the molecular mechanisms are known whereby compounds such as those shown in Figure 1 block individual ion channels, the structural determinants of ion channel block remain conjectural. It is increasingly recognized that the enantiomers of drugs with chiral centers may have dissimilar pharmacologic properties. Thus, characterization of individual enantiomers of drugs being developed as isomers assumes increasing importance in drug development. Among compounds shown in Figure 1, sotalol, ibutilide, and almokalant each have chiral centers; however, preliminary data suggest no major differences in action potential prolongation for the enantiomers of these compounds. L706,000 is actually being developed as the individual enantiomer shown in Figure 1. Of the compounds presented in Table I, development of WAY 123,398 and terikalant (another compound that, like L706,000, was investigated as the individual isomer) is not being pursued. The compounds whose development is furthest advanced (as of October 1992) are sematilide, d-sotalol, almokalant, and dofetilide, all of which are in late Phase 2 or early Phase 3 trials (i.e., efficacy studies in patients with arrhythmias). PHARMACOKINETICS Available data indicate that most of the compounds currently under development have relatively simple pharmacokinetics (Table I). In general, they are not subject to extensive hepatic metabolism and do not have active metabolites. Ibutilide is an exception. It undergoes extensive first-pass hepatic metabolism, with formation of 2 inactive metabolites. Because small changes in hepatic metabolism could result in marked variability in plasma concentrations of the parent drug, it is not likely that ibutilide will be developed for long-term oral therapy. Nevertheless, development of an intravenous compound for the acute management or arrhythmias seems possible. In vitro data suggested that terikalant is a substrate for 1 particular hepatic enzyme, P450IID6. Because P450IID6 is functionally absent in approximately 7% of whites and blacks,[25] it is conceivable that terikalant therapy would have caused very high drug concentrations and a risk of attendant toxicity in poor metabolizers. MECHANISMS OF ACTION Cardiac repolarization is a balance between inward and outward ion currents. Inward currents, generally carried through sodium or calcium channels, gradually decrease during an action potential, whereas outward currents, generally carried through potassium channels, gradually increase. Action potentials can be prolonged by augmenting inward current or by decreasing outward current. With the exception of ibutilide, the compounds discussed here block potassium currents. Ibutilide appears to act by augmenting inward current through sodium channels.[26] The structural similarities between ibutilide and other methanesulfonanilide compounds suggest that the molecular structure of sodium channels is similar to that of the potassium channels the other compounds block. Recent advances in molecular biology and in application of whole cell voltage clamp to a variety of cardiac preparations have demonstrated that heart cells contain multiple subtypes of potassium channels, each of which may be a target for drug action. Multiple potassium channel phenotypes have been recognized. Delayed rectifier currents develop and are sustained with time, whereas inactivating channels (in the heart also referred to as transient outward current) develop with time and, with sustained depolarization, turn off ('inactivate'). Other potassium channel phenotypes (such as the inward rectifier, which sets resting membrane potential) are largely time independent within the time frame of a cardiac action potential. Our laboratory and others have shown that delayed rectifier current in guinea pig myocytes is actually composed of multiple channel subtypes, which have now been identified as a lanthanum-sensitive, E4031-sensitive, rapidly activating component ([I.sub.Kr]) and a more slowly activating component ([I.sub.Ks]).[27,28] cDNA encoding a protein that, when expressed in Xenopus oocytes, gives rise to an ([I.sub.Ks]) -like current has been detected in guinea pig and human heart.[29] In addition, at least 2 cDNAs encoding inactivating channels have been isolated from the human ventricle.[30] Thus, multiple potassium channel subtypes may exist in the human heart, each of which might be a target for drug action. Some drugs, such as E4031, dofetilide, sematilide, WAY123,398, L706,000, NAPA, and almokalant appear to target [I.sub.Kr] only.[28, 31] Others, such as quinidine or amiodarone, block [I.sub.Kr], [I.sub.Ks], and multiple transient outward phenotypes.[32, 33] Sotalol and its d-isomer block both [I.sub.Kr] and transient outward currents in Purkinje tissue.[17] Terikalant appears to target the inward rectifier current. The exact ionic target for other compounds listed in Table I is not yet known. As these details of molecular mechanisms of action become better understood, it is conceivable that a channel or combination of channels whose targeting is desirable, or undesirable, in certain disease states may be identified. MECHANISMS OF TORSADES DE POINTES: IMPLICATIONS FOR DRUG DEVELOPMENT In conducting system (Purkinje) tissue, action potential prolongation frequently results in an abnormality of terminal repolarization, referred to as an early afterdepolarization (EAD) (Figure 2). The ionic mechanism underlying the genesis of EADs in Purkinje tissue is not established. Cells that develop this secondary plateau also frequently develop oscillatory electrical activity, as shown in Figure 2.[34] These triggered upstrokes are thought to underlie the development of torsades de pointes. This understanding of the likely mechanism of torsades de pointes can provide a basis for rational therapy for this clinical problem and, conceivably, help identify screening maneuvers for newer compounds that might lack this action. For example, the triggered upstrokes are, in many circumstances, calcium dependent.[35] Drugs that block calcium channels in this voltage range might thus be less likely to produce torsades de pointes. Bepridil or lidoflazine could serve as a template for such a drug, although both have been associated with torsades de pointes.[36-38] Because catecholamines can enhance calcium current, drugs that interfere with catecholamine-augmented increases in calcium current (e.g., [Beta] blockers) might also inhibit the development of torsades de pointes. It should therefore be interesting to compare the incidence of torsades de pointes during therapy with d-sotalol or its analogues with that observed during treatment with racemic sotalol. Of the compounds described here, torsades de pointes has occurred to date during therapy with sotalol, d-sotalol, ibutilide, sematilide, dofetilide, and almokalant. Another way to reduce the probability of torsades de pointes might be to develop drugs that block potassium channels in a more positive voltage range than that associated with EADs. Drugs that markedly prolong action potentials in ventricular tissue but do not prolong, or even shorten, action potentials in Purkinje tissue, might also be unlikely to produce torsades de pointes. It is conceivable that the low incidence of torsades de pointes during amiodarone therapy is determined by this mechanism.[39] In fact, drugs such as flecainide prolong action potentials in ventricular muscle (presumably by potassium channel block) and shorten them in Purkinje tissue by virtue of their potent sodium channel-blocking actions.[40, 41] Flecainide has other liabilities, but torsades de pointes is not a feature of its toxicity. Embryotoxicity in the rat may represent another variation on the theme of toxicity related to triggered activity.[42] Early studies with almokalant, dofetilide, and d-sotalol demonstrated that if pregnant rats were fed drug at midpregnancy, the fetuses died. Electrophysiologic studies have demonstrated that these compounds have no effect on cardiac action potentials of adult rats; however, exposure of fetal rat tissue to class III drugs results in marked action potential prolongation and triggered activity. Thus, fetal loss due to class III drugs may be attributable to cardiotoxicity, conceivably even torsades de pointes, in utero. There is no systematic experience with class III drugs in human pregnancy, although compounds such as quinidine, which also are associated with torsades de pointes, are thought to be generally safe. PREDICTION OF EFFICACY AND TOXICITY Information from the sotalol data base, as well as emerging information gathered during workup of other compounds, has raised a number of interesting issues with regard to appropriate clinical evaluation. Many patients were treated on a 'compassionate use' basis with sotalol, and a variety of methods were used to evaluate treatment efficacy. In the subset with sustained ventricular tachycardia, about 70% of individuals judged to be complete respondents to sotalol were still receiving the drug after 1 year, regardless of the means used to evaluate drug efficacy (generally, Holter monitoring or electrophysiologic testing). Interestingly, approximately the same percentage of patients judged to be partial responders or even nonresponders still remained on sotalol after 1 year. Whether this apparent failure of acute response to predict long-term outcome accurately reflects the [Beta] -blocking actions of sotalol or its action potential-prolonging effects may well be determined by experience with other, newer compounds. Data from a smaller sematilide trial raise a similar issue. The recurrence rate of arrhythmias among patients discharged on sematilide because electrophysiologic testing had predicted that the drug would be effective was relatively high (approximately 20%). It is not known whether this reflects a changing substrate in a population with advanced disease, late proarrhythmia, a failure of electrophysiology testing to respond accurately to this compound, or a combination of mechanisms. Examining QT interval prolongation may be an obvious alternative method of evaluating the efficacy of class III drugs. However, as information from both the sotalol and the sematilide data bases and more limited experience with intravenous dofetilide[43] makes clear, this is not an appropriate strategy. Although each of these compounds prolongs the QT interval to some extent, the degree to which the QT interval is prolonged in an individual patient does not predict whether or not arrhythmias are suppressed in that patient. Given current practice standards, the major concern with any new antiarrhythmic drug is not so much efficacy as safety. The potential exists for most class III agents currently in development to share the liability of exaggerated action potential prolongation and resultant triggered activity, manifest in patients as torsades de pointes. It is not currently possible to predict which patients are likely to develop torsades de pointes, although we can identify individuals who appear to be at particularly high risk. These include patients with well-recognized factors such as bradycardia, hypokalemia, hypomagnesemia, and concomitant use of other QT-prolonging agents, exposure to relatively high dosages (especially of intravenously administered drug), marked QT prolongation at baseline (although this concept has never been formally tested), and, for reasons that are not completely apparent, recent conversion from atrial fibrillation to sinus rhythm.[44-46] CONCLUSION The pharmaceutical industry has embarked on a large-scale experiment to determine if action potential prolongation is a desirable mode of antiarrhythmic drug action. The clinical features and mechanisms underlying a major potential class liability, torsades de pointes, are now being elucidated. Further developments in understanding the molecular mechanisms of action of these compounds and the individual ion channels they target should lead to identification of currently available drugs, or newly developed compounds, that are safer and more effective than the antiarrhythmic agents currently available. REFERENCES [1] Singh BN, Williams EMV. The effect of amiodarone, a new antianginal drug, on cardiac muscle. Br J Pharmacol 1970;39:657-667. [2] Strauss HC, Bigger JT, Hoffman BF. Electrophysiological and beta-receptor blocking effects of MJ 1999 on dog and rabbit cardiac tissue. Circ Res 1970;26:661-678. [3] Singh BN, Vaughan Williams EM. A third class of anti-arrhythmic action: effects on atrial and ventricular intracellular potentials, and other pharmacological actions on cardiac muscle, of MJ 1999 and AH 3747. Br J Pharmacol 1970;39:675-689. [4] Echt DS, Liebson PR, Mitchell LB, Peters RW, Obias-Manno D, Barker AH, Arensberg D, Baker A, Friedman L, Greene HL, Huther ML, Richardson DW, CAST Investigators. Mortality and morbidity in patients receiving encainide, flecainide, or placebo, N Engl J Med 1991;324:781-788. [5] Kou WH, Nelson SD, Lynch JJ, Montgomery DG, DiCarlo L, Lucchesi BR. Effect of flecainide acetate on prevention of electrical induction of ventricular tachycardia and occurrence of ischemic ventricular fibrillation during the early postmyocardial infarction period: evaluation in a conscious canine model of sudden death. J Am Coll Cardiol 1987;9:359-365. [6] Lynch JJ, Wilber DJ, Montgomery DG, Hsieh TM, Patterson E. Lucchesi BR. Antiarrhythmic and antifibrillatory actions of the levo- and dextrorotatory isomers of sotalol. J Cardiovasc Pharmacol 1984;6:1132-1141. [7] Lynch JJ Jr, Heaney LA, Wallace AA, Gehret JR, Selnick HG, Stein RB. Suppression of lethal ischemic ventricular arrhythmias by the class III agent E4031 in a canine model of previous myocardial infarction. J Cardiovasc Pharmacol 1990;15:764-775. [8] Chi L, Mu D, Driscoll EM, Lucchesi BR. Antiarrhythmic and electrophysiologic actions of CK-3579 and sematilide in a conscious canine model of sudden coronary death. J Cardiovasc Pharmacol 1990;16:312-324. [9] Echt DS, Black JN, Barbey JT, Coxe DR, Cato EL. Evaluation of antiarrhythmic drugs on defibrillation energy requirements in dogs: sodium channel block and action potential prolongation. Circulation 1989;79:1106-1117. [10] Dorian P, Fain ES, Davy JM, Winkle RA. Lidocaine causes a reversible, concentration-dependent increase in defibrillation energy requirements. J Am Coll Cardiol 1986;8:327-332. [11] Fain ES, Dorian P, Davy JM, Kates RE, Winkle RA. Effects of encainide and its metabolites on energy requirements for defibrillation. Circulation 1986;73:1334-1341. [12] Bacaner MB, Clay JR, Shrier A, Brochu RM. Potassium channel blockade: a mechanism for suppressing ventricular fibrillation. Proc Natl Acad Sci USA 1985;82:2223-2227. [13] Tande PM, Refsum H. Class III antiarrhythmic action linked with positive inotropy: effects of the d- and l-isomer of sotalol on isolated rat atria at threshold and suprathreshold stimulation. Pharmacol Toxicol 1988;62:272-277. [14] Nishimura M, Follmer CH, Singer DH. Amiodarone blocks calcium current in single guinea pig ventricular myocytes. J Pharmacol Exp Ther 1989;251:650-659. [15] Mason JW, Hondeghem LM, Katzung BG. Block of inactivated sodium channels and of depolarization-induced automaticity in guinea pig papillary muscle by amiodarone. Circ Res 1984;55:277-285. [16] Polster P, Broekhuysen J. The adrenergic antagonism of amiodarone. Biochem Pharmacol 1975;25:131-134. [17] Carmeliet E. Electrophysiologic and voltage clam analysis of the effects of sotalol on isolated cardiac muscle and Purkinje fibers. J Pharmacol Exp Ther 1985;232:817-825. [18] Funck-Brentano C, Silberstein DJ, Wood AJJ, Roden DM, Woosley RL. d-Sotalol-induced changes in heart rate are not due to [Beta] -adrenergic blockade. Circulation 1988;78:SII499. [19] Jaillon P. Winkle RA. Electrophysiologic comparative study of procainamide and N-acetylprocainamide in anesthetized dogs: concentration-response relationships. Circulation 1979;60:1385-1394. [20] Dangman KH, Hoffman BF. In vivo and in vitro antiarrhythmic and arrhythmogenic effects of N-acetyl procainamide. J Pharmacol Exp Ther 1981;217:851-862. [21] Roden DM, Reele SB, Higgins SB, Smith R, Oates JA, Woosley RL. Antiarrhythmic efficacy, pharmacokinetics and safety of N-acetylprocainamide in man: comparison to procainamide. Am J Cardiol 1980;46:463-468. [22] Winkle RA, Jaillon P, Kates RE, Peters F. Clinical pharmacology and antiarrhythmic efficacy of N-acetylprocainamide. Am J Cardiol 1981;47:123-130. [23] Kluger J, Leech S, Reidenberg MM, Lloyd V, Drayer DE. Long term antiarrhythmic therapy with acetylprocainamide. Am J Cardiol 1981;48:1124-1130. [24] Lumma WC Jr, Wohl RA, Davey DD, Argentieri Tm, DeVita RJ, Gomez RP, Jain VK, Marisca AJ, Morgan TK Jr, Reiser HJ, Sullivan ME, Wiggins J, Wong SS. Rational design of 4-[(methylsulfonyl)amino] benzamides as class III antiarrhythmic agents. J Med Chem 1987;30:755-758. [25] Eichelbaum M, Gross AS. The genetic polymorphism of debrisoquine/sparteine metabolism--clinical aspects. Pharmacol Ther 1990;46:377-394. [26] Lee KS. Ibutilide, a new compound with potent class III antiarrhythmic activity, activates a slow inward [Na.sup.+] current in guinea pig ventricular cells. J Pharmacol Exp Ther 1992;2,262:99-108. [27] Balser JR, Bennett PB, Roden DM. Time-dependent outward current in guinea pig myocytes: gating kinetics of the delayed rectifier. J Gen Physiol 1990;96:835-863. [28] Sanguinetti MC, Jurkiewicz NK. Two components of cardiac delayed rectifier [K.sup.+] current. Differential sensitivity to block by class III antiarrhythmic agents. J Gen Physiol 1990;96:195-215. [29] Folander K, Smith JS, Antanavage J, Bennett C, Stein RB, Swanson R. Cloning and expression of the delayed rectifier IsK channel from neonatal rat heart and diethylstilbestrol-primed rat uterus. Proc Natl Acad Sci USA 1990;87:2975-2979. [30] Tamkun MM, Knoth KM, Walbridge JA, Kroemer H, Roden DM, Glover DM. Molecular cloning and characterization of two voltage-gated [K.sup.+] channel cDNAs from human ventricle. FASEB J 1991;5:331-337. [31] Turgeon J, Bennett PB, Roden D. N-acetylprocainamide (NAPA) blocks an early component of outward potassium current. Circulation 1990;82(suppl III):528. [32] Balser JR, Bennett PB, Hondeghem LM, Roden DM. Suppression of time-dependent outward current in guinea pig ventricular myocytes: actions of quinidine and amiodarone. Circ Res 1991;69:519-529. [33] Imaizumi Y, Giles WR. Quinidine-induced inhibition of transient outward current in cardiac muscle. Am J Physiol 1987;253:H704-H708. [34] Roden DM, Hoffman BF. Action potential prolongation and induction of abnormal automaticity by low quinidine concentrations in canine Purkinje fibers: relationship to potassium and cycle length. Circ Res 1985;56:857-867. [35] January CT, Riddle JM. Early afterdepolarizations: mechanism of induction and block: a role for L-type [Ca.sup.2+] current. Circ Res 1989;64:977-990. [36] Leclercq JF, Kural S, Valere PE. Bepridil et torsades de pointes. Arch Mal Coeur 1983;76:341-348. [37] Kennelly M. Comparison of lidoflazine and quinidine in prophylactic treatment of arrhythmias. Br Heart J 1977;39:540-546. [38] Campbell RM, Iansmith DHS, Woosley RL, Roden DM. Lack of triggered automaticity despite repolarization abnormalities due to bepridil or lidoflazine. PACE Pacing Clin Electrophysiol 1990;13:30-36. [39] Mattioni TA, Zheutlin TA, Sarmiento JJ, Parker M, Lesch M, Kehoe RF. Amiodarone in patients with previous drug-mediated torsades de pointes. Ann Intern Med 1989;111-574-580. [40] Ikeda N, Singh BN, Davis LD, Hauswirth O. Effects of flecainide on the electrophysiologic properties of isolated canine and rabbit myocardial fibers. J Am Coll Cardiol 1985;5:303-310. [41] Follmer CH, Colatsky TJ. Block of delayed rectifier potassium current, [I.sub.K], by flecainide and E-4031 in cat ventricular myocytes. Circulation 1990;82:289-293. [42] Abrahamsson C, Carlsson L, Duker G. Action potential prolongation and early afterdepolarizations induced by class III antiarrhythmic agents in the fetal but not in the adult rat heart. J Mol Cell Cardiol 1992;24(suppl V):S33. [43] Echt DS, Lee JT, Murray KT, Vorperian V, Borganelli SM, Crawford DM, Roden DM. A randomized, double-blind, placebo-controlled dose-ranging study of intravenous UK-68,798 (dofetilide) in patients with inducible sustained ventricular tachyarrhythmias. Circulation 1991;84(suppl II):714. [44] Kay GN, Plumb VJ, Arciniegas JG, Henthorn RW, Waldo AL. Torsades de pointes: the long-short initiating sequence and other clinical features: observations in 32 patients. J Am Coll Cardiol 1983;2:806-817. [45] Roden DM, Woosley RL, Primm RK. Incidence and clinical features of the quinidine-associated long QT syndrome: implications for patient care. Am Heart J 1986;111:1088-1093. [46] Jackman WM, Friday KJ, Anderson JL, Aliot EM, Clark M, Lazzara R. The long QT syndromes: a critical review, new clinical observations and a unifying hypothesis. Prog Cardiovasc Dis 1988;31:115-172. Dan M. Roden, MD From the Division of Clinical Pharmacology, Vanderbilt University, Nashville, Tennessee. Supported in part by grants from the United States Public Health Service (HL46681, GM31304) and an endowment from Daiichi Pharmaceutical.