Violent behavior is a common medical indication for emergency sedation of patients who are intoxicated with ethanol or other drugs. There is continuing controversy concerning the best sedative to use for this population, but most physicians choose either a butyrophenone neuroleptic agent or a benzodiazepine sedative. Haloperidol and lorazepam are frequently selected drugs from each class, respectively. Both agents, however, have significant side-effect profiles. Lorazepam, like most benzodiazepines, produces dose-dependent respiratory depression. This effect is increased by coincident ethanol intoxication (Ativan Product Labeling; Wyeth-Ayerst, Philadelphia, PA, U.S.A.). Haloperidol, on the other hand, is commonly believed to have few cardiovascular or respiratory effects, even during cardiogenic shock (1-4). Current clinical practice has seen increasing use of butyrophenones for chemical restraint of agitated patients in both emergency departments and intensive care units.
Concomitant with the increased clinical use of high-dose haloperidol for sedation, several case reports have documented cardiac conduction abnormalities, ventricular arrhythmias, and sudden death after administration of haloperidol (5-8). Although product labeling warns that haloperidol should be avoided in patients with diabetes and underlying cardiovascular disease, there is little basic research concerning this claim, and practitioners frequently ignore these warnings (9). Clinical studies conducted in the emergency department found that haloperidol consistently produces mild decreases in heart rate and blood pressure; these same studies also report a ≤9% incidence of clinically significant hypotension (arterial blood pressure <90/60 mm Hg), even at low doses (10-12). These studies were unable to identify subpopulations at risk for hypotension.
Because ethanol intoxication is so common in violent patients in the emergency department, it seemed reasonable to suspect it as a cofactor in both haloperidol-associated arrhythmias and in haloperidol-associated hypotension. Ethanol alone is widely recognized to cause cardiodepression and arrhythmias in both humans and experimental animals (13-16), and previous case reports of haloperidol-induced arrhythmias have speculated that chronic ethanol abuse is a risk factor for the development of torsade de pointes(17,18). Clinical investigation of the interaction between ethanol and haloperidol is difficult, however, because cardiovascular monitoring is often not feasible in violently agitated, alcohol-intoxicated patients until after they have been sedated. Given the difficulty of monitoring agitated patients, it is possible that the cardiodepressant and proarrhythmic effects of haloperidol are unrecognized, and later decompensation is clinically ascribed to other causes (e.g., coincident cocaine intoxication, hemorrhagic shock). Accordingly, this study was designed to investigate the role of haloperidol in altering cardiac conduction and contraction during acute ethanol intoxication. The interaction of lorazepam and ethanol also was investigated because lorazepam is another commonly used sedative.
Rat heart isolation and perfusion
Male Sprague-Dawley (Charles River Labs, MA) rats weighing 400-500 g were injected with 2,500 units of sodium heparin intraperitoneally and then anesthetized with sodium pentobarbital. The hearts were rapidly excised and immediately placed in iced normal saline. The hearts were perfused by using the Langendorff technique with modified Krebs-Henseleit-Bicarbonate buffer (KHB). KHB solution was made fresh daily with distilled, deionized water and twice filtered. Its composition (in millimoles): NaCl, 118; KCl, 4.7; NaHCO3, 21; CaCl2, 1.25; MgSO4, 1.2; KH2PO4, 1.2; and d-glucose, 11. Total [Na+] = 140; total [K+] = 5.6. It was gassed with 95% O2 and 5% CO2, which produced a pO2 = 600-650 mm Hg and pCO2 = 35-40 mm Hg. Immediately after initiating perfusion, the pulmonary artery was cut to allow free ejection from the right ventricle, and a stab incision was made in the cardiac apex to allow left ventricular drainage. The hearts then underwent either mechanical or electrophysiologic testing. Simultaneous measurement of these parameters was not possible because of small heart size.
Isolated hearts were studied by using the Langendorff technique. Hearts were stabilized on KHB buffer at 60 mm Hg perfusion pressure, and the resultant coronary flow was held constant. Left ventricular-generated pressure [LVGP = left ventricular systolic pressure (LVSP) - left ventricular end-diastolic pressure (LVEDP)] was determined by a saline-filled cellophane balloon connected to a Gould P23 pressure transducer. The balloon was inserted via the mitral valve and filled with saline until the end-diastolic pressure (EDP) was zero during stabilization, and balloon volume was held constant thereafter. If the intrinsic heart rate decreased to <330, the hearts were paced from the left ventricle by using constant-voltage square-wave pulses 0.6 ms wide. Concentration-response effect of ethanol was investigated for a range of 20-200 mM (100-1,000 mg/dl), and for haloperidol and lorazepam, in concentrations ranging from 100 to 750 ng/ml (≈300-2,000 nM for each drug). Concentration responses were again examined for haloperidol and lorazepam with the addition of ethanol at 20 and 65 mM. Four to six hearts were studied for each concentration in each group. Mechanical measurements were made after 20 min perfusion with each drug or drug + ethanol.
Coronary perfusion pressure was maintained at 60 mm Hg with a Masterflex pump and the inflow reservoir connected to a Gould P23 transducer. Buffer temperature was adjusted to 35.5°C. Chloride silver recording electrodes were placed on the heart to record three orthogonal surface electrograms (19). An small incision was made in the right atrium, well away from the sinus node (SN), and a bipolar chloride silver electrode was placed on the septum in the area of the tricuspid valve to record the His bundle electrogram (HBE).
Paired platinum needles were used as pacing electrodes. These were plunged into the right ventricular free wall and into the right atrium in the area of the SN. The heart was stimulated with constant-current square-wave pulses, 0.7 ms duration, with the stimulus current adjusted to twice threshold (typically 1-2 mA). Electrograms were band-pass filtered between 0.3 Hz and 1 kHz. His bundle signals were low-pass filtered at 1 kHz, and the high-pass filter was adjusted between 0.3 and 100 Hz to provide the best signal-to-noise ratio. Pacing protocols were generated by a micro-computer by using custom software. Hearts that displayed inconsistent P wave structure or abnormal PR conduction after placement of electrodes were discarded.
Electrophysiologic studies (EPS)
Data were collected on intrinsic rhythm and intervals. The heart was then paced from the atria with trains of 15 pulses. The cycle length was decreased from 180-40 ms by 20-ms decrements. This was followed by ventricular pacing by using the same cycle lengths. The AV effective refractory period (AVERP) was determined to the nearest millisecond by pacing from the atria at 180-ms cycle length for 15 beats (S1) and introducing successively premature stimuli (S2) at the end of the train. S2 was decremented by 20 ms from 160 to 40 ms. The longest S2 interval that did not conduct through the AV node was noted. This range was investigated in 5-ms increments and then in 1-ms increments until the AVERP was determined to the nearest millisecond. A minimum of 20 s was allowed between trains (60-100 spontaneous beats) for heart recovery. In addition to the AVERP, the atrioventricular-His (AH) interval, the His-ventricular (HV) interval, and the sinus node recovery time (SNRT) were measured at different atrial drive rates (180, 140, 100 ms). The AH conduction time (AH interval) was defined as the interval between the first evidence of atrial activity from the surface electrograms and the initial deflection of the HBE. The HV conduction time (HV interval) was defined as the interval between the initial deflection of the HBE and the first evidence of ventricular depolarization from the surface electrograms. The SNRT was defined as the interval between the first spontaneous P wave and the last stimulus in atrial train. The corrected SNRT (CSNRT) is the SNRT less the spontaneous sinus cycle length. No meaningful data could be collected concerning QT interval or QRS width because the QRS complex blends into the T wave in rat hearts (20).
EPS perfusion protocol
For electrophysiological studies, hearts were instrumented and stabilized (≈30 min total time) with KHB perfusion, at which time two basal control (BC) measurements were obtained. Hearts were then switched to a second perfusion of KHB ± ethanol (65 mM; AAPER, Shelbyville, KY, U.S.A.) for 40 min, at which time measurements were repeated. Perfusate was then switched a third time to KHB ± ethanol with either lorazepam (900 nM; Wyeth-Ayerst, Philadelphia, PA, U.S.A.), or haloperidol (800 nM, Solo-Pak Laboratories, Elk Grove Village, IL, U.S.A.). This third perfusion was continued for 20 min, at which time the third set of measurements was obtained (i.e., EPS indices representing drug effect). Thus hearts received one of six perfusion protocols: time control (TC, KHB throughout), ethanol + haloperidol (E during the second perfusion, EH during the third perfusion), ethanol + lorazepam (E during second perfusion, EL during the third perfusion), ethanol control (EC, ethanol during the second and third perfusion), haloperidol control (HC, haloperidol during the third perfusion only), and lorazepam control (LC, lorazepam during the third perfusion only). Each group contained six to eight hearts.
All data were tested for normality before application of statistical tests. Basal control data were analyzed by using one-way analysis of variance (ANOVA) among groups to exclude starting point differences for all measures. A paired t test was used to examine the change in each variable compared with BC. Only one index per heart is reported for mechanical studies (LVGP); therefore the null hypothesis was rejected with p < 0.05 for this measure. Because five measurements were examined on each heart in EPS studies, a Bonferroni correction was used; therefore p < 0.01 had to be reached to reject the null hypothesis at the 0.05 level. Data are presented as means ± SEM. All statistical analysis was performed by using SAS for Windows (Cary, NC, U.S.A.).
The range in baseline LVGP values was 69-81 mm Hg for all hearts. There was no significant difference in BC LVGP among groups. Perfusion of ethanol alone produced no significant depression in LVGP until perfused in a concentration >200 mM, at which point LVGP decreased -15 ± 5 mm Hg below BC. Ethanol alone tended to decrease end-diastolic pressures (-2.5 ± 2.0 mm Hg at 65 mM; p = 0.14). Lorazepam and haloperidol individually demonstrated significant decreases in LVGP at the same concentration for both drugs (750 ng/ml; Fig. 1A and B). Lorazepam at the highest concentration studied (2,000 nM; 750 ng/ml) increased the EDP significantly (13 ± 4 mm Hg); haloperidol did not affect EDP significantly at any concentration studied. When ethanol (20 and 65 mM) was added to lorazepam, no change in the concentration-response for LVGP was observed (Fig. 1A). However, when ethanol (20 and 65 mM) was added to haloperidol perfusion, a dose-dependent potentiation in the negative inotropy of haloperidol was evident (Fig. 1B). Ethanol at 20 and 65 mM lowered the threshold concentration at which haloperidol produced significant reduction in LVGP to 1,500 and 900 nM, respectively (500 and 300 ng/ml; Fig. 1B). The magnitude of both the positive and negative first derivatives of LVP (±dP/dt) decreased proportionately and in parallel with changes in LVGP with haloperidol-ethanol exposure. Neither the LV EDP nor the coronary flow changed significantly with haloperidol-ethanol exposure. At the end of the experiment, coronary artery perfusion pressures and resistances were not significantly changed versus BC in any group.
There were no significant differences among groups for any BC measurement. Baseline data are shown in Table 1. There were also no significant differences between basal measurements and the data collected after the second perfusion (40 min ethanol perfusion, except for group TC). Data are presented with emphasis on the change between BC and after the third perfusion period.
Sinus node. The combination of haloperidol and ethanol (EH) significantly decreased the intrinsic heart rate by 74 ± 12 beats/min (-25% vs. BC). Haloperidol alone decreased the heart rate by 55 ± 10 beats/min (-20% vs. BC). There was no statistical difference between groups EH and HC. Neither ethanol control (EC), lorazepam control (LC), nor the combination (EL) significantly decreased spontaneous heart rate (Fig. 2a). There was a trend toward decreased heart rate for all hearts with time, which may reflect catecholamine depletion in the excised heart after repeated pacing, but this trend did not reach significance.
There was a trend toward increased CSNRT (Fig. 2b) in the EH group, but this did not reach significance among groups, probably because our methods were not adequately sensitive to discriminate between atrial ectopic pacemaker sites and signals originating from the SN. The raw data showed both very long and very short CSNRT for the EH group, which is typical of SN entry or exit block or both, in which the tissue surrounding the SN is refractory to stimulation originating either from within the SN (exit block) or from outside the SN (entry block) (21).
AV node. Haloperidol without ethanol increased the AVERP by 67 ± 12 ms, an 80% increase over baseline. The combination of haloperidol and ethanol increased AVERP by 55 ± 8 ms (+60% vs. BC). There were no significant differences between the other groups (Fig. 3a). Increased AV nodal refractoriness is supported by the measurement of AH interval at different atrial rates (Fig. 3b). Shorter cycle lengths produced increasing AH intervals in the HC and EH groups. Most hearts in all groups demonstrated the Wenckebach phenomenon at cycle lengths of 100 ms, so these data were excluded from the analysis of AH intervals.
Second- and third-degree AV block (AVB) developed in some of the hearts in the HC group (two of six), as well as in the EH group (two of six) when paced at 140 ms basic cycle length. No other groups developed AVB at either 180 or 140 ms. Data from hearts that developed AVB were excluded from analysis, although their inclusion would have increased the difference between the two haloperidol groups and the four other groups. If hearts developed AVB during the basic drive cycle of the AVERP determination, they were scored as 180-ms refractory times. This also tended to decrease the difference between these hearts and hearts not in AVB. Thus the differences in AV nodal function presented in Fig. 3a and b reflect a conservative estimate of the change observed in the AVERP of the HC and EH groups.
His-Purkinje system. The His-Purkinje system was evaluated by measuring the HV intervals at different atrial drive cycle lengths. The combination of ethanol and haloperidol produced a 30% increase in HV interval at the spontaneous heart rate and a 62% increase when the drive cycle was shortened to 140 ms. Increasing HV intervals were observed at decreasing cycle lengths, although they plateaued at 140 ms, probably because of increasing refractoriness in the AV node, which delayed stimuli delivered to the His bundle (i.e., increasing functional refractory period of the AV node). None of the other drug treatments significantly altered HV conduction (Fig. 4, HV intervals).
Our study demonstrates that ethanol potentiates haloperidol-mediated depression of cardiac conduction and contraction in the isolated rat heart. In the clinical setting, hypotension that follows haloperidol administration is usually attributed to blockade of α1-adrenergic receptors (10,11,22). In an ethanol-intoxicated patient, however, haloperidol also could produce clinically significant depression in LV function. Figure 1B demonstrates that ethanol, present in a concentration commonly measured in the blood of intoxicated persons, significantly altered the concentration-response curve for haloperidol in the isovolumetrically contracting heart. This shift suggests that ethanol can augment the biologic effect of haloperidol on a specific receptor. Haloperidol is recognized as a weak L-type calcium channel antagonist (5,23,24), and ethanol is known to blunt systolic calcium currents in cardiomyocytes (25,26), so it appears logical that this drug combination would produce an additive negative inotropic effect. However, our findings do not fully support a classic paradigm in which ethanol increases the haloperidol antagonism of L-channels. Specifically, ethanol did not exaggerate the effect of haloperidol on AV nodal refractoriness or on intrinsic heart rate (Figs. 2a, 3a, and b), and coronary artery resistance did not decrease with the drug combination compared with BC (data not shown). On the other hand, because the LVEDP did not increase, and peak negative dP/dt was not altered out of proportion to changes in LVGP, myocardial calcium overload seems unlikely to have occurred. Taken together, it appears plausible that ethanol and haloperidol produced additive and relatively selective depression of systolic calcium transients in ventricular myocytes without typical L-channel antagonistic features or gross mechanical evidence of impaired reuptake of calcium during diastole.
In addition to a producing a negative inotropic effect that could cause hypotension, the combination of ethanol and haloperidol could significantly alter the ability of the heart to increase heart rate in response to physiologic reflexes. First, ethanol and haloperidol reduced spontaneous heart rate with a trend toward increasing CSNRT. The all-or-none characteristic of the CSNRT data suggests that ethanol + haloperidol decreased heart rate by producing either SN entry or exit block (Fig. 2a and b). Second, haloperidol, with or without ethanol, increased AVERP and AV conduction interval, potentially limiting maximal attainable ventricular heart rate (Fig. 3a and b). Third, ethanol + haloperidol significantly delayed impulse conduction through the His-Purkinje system (Fig. 4). This global depression in cardiac impulse formation and conduction may explain why recent clinical studies have reported a normal heart rate in patients with haloperidol-related hypotension (3,7,11).
The depression in His-Purkinje conduction caused by ethanol + haloperidol intoxication is of particular clinical importance. Because no significant change in HV interval occurred with either ethanol or haloperidol alone, it suggests that there is a synergistic action of these two agents on the Na+ channel. Haloperidol is known to block Na+ channels in excitable tissues (27), whereas ethanol does not (28,29). Therefore it appears that ethanol augments haloperidol-induced Na+ channel antagonism. Electrophysiologic data from the intact, isolated heart provide limited insight into the mechanism of the effect of ethanol on haloperidol-mediated Na+ channel antagonism. Nonetheless, it is clear that HV prolongation increased during pacing to greater than the native heart rate, indicating use-dependent Na+ channel blockade. Haloperidol blocks inactivated Na+ channels with a 2:1 stoichiometry (27), and ethanol may alter this stoichiometry or the kinetics of binding. Alternatively, the prolongation by ethanol of the action potential (28) may increase duration of Na+ channel inactivation, favoring greater haloperidol binding. Regardless of the precise mechanism of action, clearly ethanol and haloperidol could produce inhomogeneous conduction through the His-Purkinje system, potentially contributing to arrhythmogenesis. This concern would be particularly important in the emergency department setting where agitated patients are frequently cointoxicated with drugs known to induce triggered activity, such as cocaine or tricyclic antidepressants. In contrast, lorazepam demonstrated no depression of the His-Purkinje system or the sinus or AV nodes, showed no potentiation with ethanol, and as such, may provide a safe alternative to haloperidol for chemical sedation.
Recent clinical reports have focused attention on haloperidol as an agent that prolongs the QT interval and has the potential to cause torsade de pointes (5,8,30,31). It is generally held that torsade de pointes results from abnormalities in repolarization resulting from potassium channel dysfunction (32). The QT interval was not reported in this study because, in rat hearts, the T wave blends into the QRS complex, which renders measurements of the QT interval unreliable (20). Clinical case reports, however, suggest that ethanol use may be a risk factor for the development of haloperidol-associated torsades de pointes (17,18). This hypothesis warrants further investigation.
This study was designed to determine whether ethanol changes the relative cardiovascular potency of two drugs commonly used for chemical restraint. This study does not allow interpretation regarding the in vivo relevance of the concentrations used for haloperidol and lorazepam. Both drugs are lipophilic, and it is probable that their cardiac effects were amplified because they were dissolved in a protein-free buffer. By using the most relevant pharmacokinetic data available, it appears that a concentration of 300 ng/ml corresponds to a haloperidol dose of 1-2 mg/kg, i.v., and a lorazepam dose of 0.1-0.2 mg/kg, i.v. (33,34). In fact, in anesthetized dogs, these doses produce an ≈20% decrease in contractile force-a decrement similar to that observed at a concentration of 500 ng/ml for both drugs in the present study (Fig. 1)(22). Thus the concentrations used here are probably functionally equivalent to concentrations achieved after high-dose chemical sedation for both drugs (9). It is unlikely that the osmotic effects of ethanol contributed to the observed depression in LVGP for two reasons. First, ethanol is freely permeable across cell membranes and is unlikely to produce a transmembrane osmotic gradient. Second, ethanol alone produced no change in mechanical function until perfused at a concentration of 200 mM.
1. Goldney RD, Spence ND, Bowes JA. The safe use of high dose neuroleptics in a psychiatric intensive care unit. Aust N Z J Psychiatry
2. Tesar ET, Murray GB, Cassem NH. Use of high-dose intravenous haloperidol in the treatment of agitated cardiac patients. J Clin Psychopharmacol
3. Kaneko S, Edwards JG, Goldie A, Fukushima Y, Sato T. Effect of haloperidol decanoate on the cardiovascular system. Jpn J Psychiatry Neurol
4. Sanders KM, Stern TA. Management of delirium associated with use of the intraaortic balloon pump. Am J Crit Care
5. Aunsholt NA. Prolonged Q-T interval and hypokalemia caused by haloperidol. Acta Psychiatr Scand
6. Mehta D, Mehta S, Petit J, Shriner MD. Cardiac arrhythmia and haloperidol. Am J Psychiatry
7. Cruz FG, Thiagarajan D, Harney JH. Neuroleptic malignant syndrome after haloperidol therapy. Southern Med J
8. Kriwisky M, Perry GY, Tarchitsky D, Gutman Y, Kishon Y. Haloperidol-induced torsade de pointes. Chest
9. Sanders KM, Murray GB, Cassem NH. High-dose intravenous haloperidol for agitated delirium in a cardiac patient on intra-aortic balloon pump. J Clin Psychopharmacol
10. Thomas HJ, Schwartz E, Petrilli R. Droperidol versus haloperidol for chemical restraint of agitated and combative patients. Ann Emerg Med
11. Clinton JE, Sterner S, Steinmachers Z, Ruiz E. Haloperidol for sedation of disruptive emergency department patients. Ann Emerg Med
12. Lenehan GP, Gastfriend DR, Stetler C. The use of haloperidol in the management of agitated or violent, alcohol intoxicated patients in the emergency department. J Emerg Nurs
13. Ettinger PO, Wu CF, De La Cruz C, Weisse AB, Ahmed SS, Regan TJ. Arrhythmias and the “holiday heart”: alcohol associated cardiac rhythm disorders. Am Heart J
14. Dyer AR, Stamler J, Paul O, et al. Alcohol consumption, cardiovascular risk factors, and mortality in two Chicago epidemiological studies. Circulation
15. Brackett DJ, Gauvin DV, Lerner MR, Holloway FA, Wilson MF. Dose- and time-dependent cardiovascular responses induced by ethanol. J Pharmacol Exp Ther
16. Ganz V. The acute effect of alcohol on the circulation and on the oxygen metabolism of the heart. Am Heart J
17. Metzger E, Freidman R. Prolongation of the corrected QT and torsades de pointes cardiac arrhythmia associated with intravenous haloperidol in the medically iII. J Clin Psychopharmacol
18. Faigel DO, Metz DC, Kochman ML. Torsade de pointes complicating the treatment of bleeding esophageal varices: association with neuroleptics, vasopressin, and electrolyte imbalance. Am J Gastroenterol
19. Stark G, Stark U, Tritthart HA. Assessment of the conduction of the cardiac impulse by a new epicardiac surface and stimulation technique (SST-ECG) in Langendorff perfused mammalian hearts. J Pharmacol Methods
20. Detweiler DK. Electrocardiographic monitoring in toxicological studies: principles and interpretations. Adv Exp Med Biol
21. Josephson ME. Clinical cardiac electrophysiology. Williams and Wilkins, 1993;2:74-95.
22. Brannan MD, Riggs JJ, Hageman WE, Pruss TP. A comparison of the cardiovascular effects of haloperidol, thioridazine and chloropromazine HCl. Arch Int Pharmacodyn Ther
23. Galizzi J, Fosset M, Romey G, Laduron P, Lazdunski M. Neuroleptics of the diphenylbutylpiperidine series are potent calcium channel inhibitors. Proc Natl Acad Sci U S A
24. Carpentier RG, Gallardo-Carpentier A. Acute and chronic effects of ethanol on sinoatrial electrophysiology in the rat heart. J Cardiovasc Pharmacol
25. Danzinger RS, Sakai M, Capogrossi MC, Spurgeon HA, Hansford RG, Lakatta EG. Ethanol acutely and reversibly supresses excitation-contraction
coupling in cardiac myocytes. Circ Res
26. Kojima S, Wu ST, Wikman-Coffelt J, Parmley WW. Acute effects of ethanol on cardiac function and intracellular calcium in perfused rat heart. Cardiovasc Res
27. Ogata N, Narahashi T. Block of sodium channels by psychotropic drugs in single guinea-pig cardiac myocytes. Br J Pharmacol
28. Williams ES, Mirro MJ, Bailey JC. Electrophysiological effects of ethanol, acetaldehyde, and acetate on cardiac tissues from dog and guinea pig. Circ Res
29. Carpentier RG, Gallardo-Carpentier A. Effect of ethanol on guinea pig ventricular action potentials. J Electrocardiol
30. Wilt JL, Minnema AM, Johnson RF, Rosenblum AM. Torsade de pointes associated with the use of intravenous haloperidol. Ann Intern Med
31. Zee-Cheng CS, Mueller CE, Seifert CF, Gibbs HR. Haloperidol and torsade de pointes. Ann Intern Med
32. Tan HL, Hou CJY, Lauer MR, Sung RJ. Electrophysiological mechanisms of the long QT interval syndromes and torsade de pointes. Ann Intern Med
33. Holley FO, Magliozzi JR, Stanski DR, Lombrozo L, Hollister LE. Haloperidol kinetics after oral and intravenous doses. Clin Pharmacol Ther
34. Greenblatt DJ, Shader RI, Franke K, et al. Pharmacokinetics and bioavailability of intravenous, intramuscular and oral lorazepam in humans. J Pharm Sci