Stereoselective halofantrine disposition and effect: concentration-related QTc prolongation

Abstract

Aims

1) To characterize the variability of multiple-dose halofantrine pharmacokinetics over time in healthy adults, 2) to correlate the pharmacodynamic measure electrocardiographic (ECG) QT interval with (+)- and (−)-halofantrine plasma concentration and 3) to evaluate the safety and tolerance of halofantrine hydrochloride given over time to healthy adults.

Methods

Twenty-one healthy subjects were enrolled and 13 completed the study (180 days). Subjects received either 500 mg of racemic halofantrine once daily in the fasted state for 42 days, or placebo, and then halofantrine washout was documented for the following 138 days. Pharmacokinetic and pharmacodynamic (ECG QTc) measurements were obtained.

Results

Mean accumulation half-times (days) for halofantrine were: 7.0 ± 4.8 [(+)-halofantrine] and 7.3 ± 4.8 [(−)-halofantrine]. Mean steady-state concentrations were: 97.6 ± 52.0 ng ml⁻¹ [(+)-halofantrine] and 48.5 ± 20.8 [(−)-halofantrine]. Steady-state oral clearance was: 139 ± 73 l h⁻¹ [(+)-halofantrine] and 265 ± 135 l h⁻¹ [(−)-halofantrine]. Peak plasma concentrations of both (+)- and (−)-halofantrine were attained at 6 h and maximal ECG QTc prolongation was at 4–8 h following drug administration. Fourteen of 16 subjects who received active drug had ECG QTc prolongation that was positively correlated with both (+)- and (−)-halofantrine concentration. The five subjects who received placebo had no demonstrable change in ECG QTc throughout the study.

Conclusions

Halofantrine accumulates extensively and shows high intersubject pharmacokinetic variability, is associated with concentration-related ECG QTc prolongation in healthy subjects, and is clinically well tolerated in this subject group.

Introduction

Halofantrine hydrochloride is useful for the treatment of uncomplicated malaria and is effective when used against both chloroquine-sensitive and chloroquine-resistant Plasmodium falciparum infections [1–3]. Halofantrine pharmacokinetics have been described in healthy volunteers [4–6] and patients with malaria [6–8]. Following oral administration, the extent of halofantrine absorption is thought to be quite limited and highly variable. Absolute bioavailability has been determined in beagle dogs, and in the fasted state oral bioavailability is about 10% [9]. Co-administration with food markedly increases exposure after oral administration of halofantrine [9, 10]. The mechanism for this may be not only enhanced drug absorption but also redistribution due to halofantrine lipoprotein binding secondary to lipoprotein concentration changes with food ingestion [10]. The reported rate of elimination has been quite variable among subjects or patients in any given study and between studies, with reported elimination half-lives ranging from 1.8 to 15.8 days. Halofantrine has a chiral centre and is administered as a racemic mixture. A single report in which the isomers were separated indicated plasma concentrations of (+)-halofantrine may be higher than those of (−)-halofantrine in patients who received a three dose treatment regimen [11]. In vitro study of halofantrine has suggested biotransformation to desbutylhalofantrine, thought to be the major metabolite, is via CYP3A4 [12].

Although halofantrine has been used with good effect to treat multidrug-resistant P. falciparum [13], its use has been associated with electrocardiographic QT prolongation [14–17]. The relationship of this finding to torsade de pointes cardiac dysrhythmia in patients with congenital long QT syndrome was suggested [18]. In the initial reports syncope, sudden death, and torsade de pointes had been noted [14–16]. However, no evaluation for the possibility of congenital long QT syndrome in these patients was reported. These data have been replicated and confirmed by prospective study [19, 20] with QTc prolongation positively correlated to racemic halofantrine plasma concentration [19]. A more recent case report of ventricular fibrillation associated with QTc prolongation and halofantrine exposure in which the patient was successfully resuscitated and noted to later have a normal QTc interval supports the likelihood that halofantrine exposure can be associated with ventricular cardiac dysrhythmia in the absence of genetic QT abnormality [20].

A preliminary report suggested halofantrine cardiotoxicity is associated with (+)-halofantrine, with (−)-halofantrine lacking this effect in a patch clamp electrophysiological model [22]. This finding has been indirectly supported by X-ray crystallography [23].

The objective of the present study was to describe the stereoselective pharmacokinetics of halofantrine during extended once-daily dosing of racemic drug and examine the relationship of this exposure to electrocardiographic QTc prolongation in healthy adult volunteers. The dosage regimen was designed to approximate a schedule that would be used for malaria prophylaxis.

Methods

Subjects and drug administration

The study was performed at the Georgetown University Medical Center Clinical Research Center. Each subject signed a written informed consent to participate in the protocol, which was approved by the Georgetown University Institutional Review Board and the U.S. Army Human Subjects Research Review Board. Twenty-one civilian male and female volunteers between the ages of 18–45 years who met inclusion criteria for the study participated. Subject age was 34 ± 8 years (mean±s.d.) (range=21–44 years), the mean weight was 74 ± 8 kg (range=63–96 kg); and the mean height was 68 ± 5 in (range=52–73 in). Eight subjects were Caucasian, 13 were Black, 18 were male and three were female (Table 1).

Table 1.

Halofantrine subject demographics.

Subject number	Race	Gender	Age (years)	Weight (kg)	Height (in)	Duration of participation
1	W	M	23	64	67	Completion
2	B	M	21	96	68	34 days
3⋆	W	F	23	68	70	Completion
4	B	M	27	68	71	37 days
5	B	M	35	68	70	28 days
6⋆	W	M	26	77	70	Completion
7	W	M	33	73	68	Completion
8	B	M	28	64	67	10 days
9	W	M	38	75	71	Completion
10	B	F	39	75	65	54 days
11	B (Hispanic)	F	43	66	52	45 days
12⋆	B	M	37	82	71	Completion
13⋆	W	M	35	63	67	Completion
14	B	M	43	75	68	40 days
15	B	M	28	71	63	Completion
16	B	M	44	77	73	44 days
17	B	M	21	68	70	1 day
18	B	M	36	68	69	62 days
19	W	M	43	82	71	Completion
20	B	M	22	75	70	Completion
21⋆	W	M	38	92	69	Completion

^⋆denotes placebo.

Exclusion criteria consisted of any history of serious medical problems, including heart disease of any type. Females were required to use effective contraceptive measures.

Medical history was obtained and physical examination was performed on all subjects, and they were documented to be in good health. Lab screening included CBC, SMA20, urinalysis, HIV and hepatitis screens. In addition a 12-lead electrocardiogram was obtained to document QTc < 450 ms. Subjects were within 20% of ideal body weight.

The study was conducted under placebo-controlled double-blind conditions with active drug : placebo allocated in a 4 : 1 ratio. Drug and placebo tablets were obtained from Smith, Kline, Beecham (Welwyn Garden City, Herts, UK). Daily oral doses of 500 mg of racemic halofantrine hydrochloride were administered to subjects for 42 consecutive days: 21 days as an inpatient at the Georgetown University Medical Center Clinical Research Center (CRC), and 21 days during which the subject reported for daily evaluation to the Clinical Research Center and was administered the halofantrine dose. Each day prior to drug administration, subjects were asked about their general state of well-being and asked to report any change in how they felt. Subjects were fasted for at least two hours prior to and following each oral dose, which was given in the morning when they were inpatients and during their daily morning visits when they were outpatients. Follow-up visits after the last dose of halofantrine (from days 44–180) were conducted with periodic pharmacokinetic sampling and electrocardiographic QTc determination at the CRC on study days 44, 45, 48, 51, 54, 57, 72 and 180.

Pharmacokinetic sampling and analysis

Blood samples were taken daily from subjects for determination of (+)- and (−)-halofantrine and desbutylhalofantrine plasma concentrations during the 21 days of in-patient status, and then when follow-up visits were conducted. On study days 4, 7, 14 and 21, blood samples were taken every 2 h for 12 h after the drug was administered, and on study days 1 and 42, at ½, 1, 2, 3, 4, 6, 8, 10 and 12 h after drug dosing. On subject day 1 and all other sampling days a sample was taken 0.5 h before the drug was given (predose). Blood samples were centrifuged promptly in a refrigerated centrifuge, and plasma was stored at −70 °C until analysis for drug and metabolite concentrations. Plasma drug concentrations were determined by high-performance liquid chromatography with fluorescence detection at the Drug Studies Unit, the University of California at San Francisco, under the direction of Dr Emil Lin, using a modification of the method of Gimenez et al. [24]. The separation column was a 5 µm Chiral AD (25 cm × 4.6 mm) (Daicel Chemical Industries, Inc.); the mobile phase was hexane/ethanol/2-butanol/diethylamine (100/1.5/1/0.1); and the flow rate was 0.3 ml min⁻¹. Detector excitation wavelength was 300 nm and emission wavelength 375 nm. The internal standard was desmethylimipramine and retention times were (+)-halofantrine: 15.0 min (−)-halofantrine: 16.5 min, internal standard: 28.5 min, (+)-N-desbutylhalofantrine: 32.5 min, and (−)-N-desbutylhalofantrine: 38.0 min. Lower limit of quantification was 1.0 ng ml⁻¹ for each of the analytes and both within-and between-day reproducibility were CV ≤ 5%.

Electrocardiograms (ECG)

ECG QTc analysis was obtained from 12-lead ECG at 50 mm sec⁻¹ paper speed from leads I, aVF, and V2 as previously reported by Woosley and Sale et al. [25, 26]. ECGs were taken immediately prior to each blood sample used for halofantrine determination. Three consecutive cardiac cycles were measured and averaged to establish QT interval and RR interval measurements. Based on the RR interval, QTc was calculated either according to the Fridericia [27] correction (RR < 500 ms), or the Bazett formula [28] (RR > 500 ms). The QT duration was determined by measuring on the lead with the earliest QRS complex, from Q to the end of the longest T-wave of the three leads. The end of the T-wave was found by drawing a tangent from the steepest part of the down slope of the T-wave to the isoelectric line. The QT measurements were based on a modification of the method of Lepeschkin et al. [29]. The mean QT and RR measurements during the screening and predose phase served as the baseline from which prolongation of QTc was determined.

Pharmacokinetic data analysis

(+)- and (−)-halofantrine and N-desbutylhalofantrine accumulation were determined by using iterative nonlinear least squares regression analysis to fit the following function to the data:

where C is the plasma drug concentration at time t (in days) after the start of the multidose treatment, C_ss is the steady-state concentration at time t→infinity, and k is the first-order rate constant for (+)- or (−)-halofantrine and/or N-desbutylhalofantrine accumulation. Accumulation half-lives were calculated from the slope of the nonlinear regression analysis determined by the above equation (accumulation half-time=ln(2)/slope).

Steady-state oral clearance of halofantrine, utilizing the mean concentration values obtained from predose blood samples for study days 25–42, was determined by dividing the halofantrine dose (corrected for the hydrochloride salt) rate per unit time by the steady-state plasma concentration of halofantrine. Concentrations of the isomers of N-desbutylhalofantrine are reported [30, 31], but clearance to this metabolite was not calculated as the extent of halofantrine conversion down this pathway is unknown.

Pharmacodynamic data analysis

Pharmacodynamic analysis was done utilizing a linear model of the form:

in which E is the measured effect (ECG QTc), E_o is the baseline QTc (y-intercept), S is the slope, and C is the halofantrine isomer concentration [32]. Data for ECG QTc were fitted to the linear model by use of least-squares regression analysis. Statistical analysis was by paired Student's t-test when comparing within-subject baseline QTc to the longest QTc measured after the dose on the first and 22nd days of treatment, and by correlational analysis when relating measured QTc to halofantrine isomer concentration.

Results

Twenty-one subjects participated in the trial, and 11 completed the entire study. The demographics of the subjects are detailed in Table 1. Subject 17 received only two doses of drug, and although safety information was collected, this subject was not included as one of the 20 with sufficient pharmacokinetic/pharmacodynamic data for evaluation. The other nine subjects who failed to complete the study were included in the data analysis. Drop-out from the study was for a variety of personal reasons with none due to adverse events. Three adverse events, namely gastroenteritis, skin rash, and severe headaches, did occur. The gastroenteritis was associated with previous food ingestion, although drug exposure could not be excluded. The skin rash disappeared with the subject who completed the study still taking drug. Therefore, it seemed unlikely to be related to drug exposure. The headaches lasted for several days during drug administration and may have been associated with halofantrine exposure.

Peak plasma concentrations for (+)-halofantrine and (–)-halofantrine were reached at 6 h after dosage on all study days for which samples were taken at hourly intervals. Mean steady-state concentrations of (+)-halofantrine exceeded those of (−)-halofantrine, and therefore the mean steady-state oral clearance of (+)-halofantrine was much lower than that of (−)-halofantrine (mean±s.d.; 139 ± 73 vs 265 ± 135 l h⁻¹). The difference between means for these clearance values is 126 l h⁻¹ (95% C.I. 59, 193). Accumulation half-life (days) for (+)-halofantrine was 7.0 ± 4.8; (range=0.8–18.3) and for (−)-halofantrine 7.3 ± 4.8; (range=0.95–14.3) (Figure 1). Plasma concentration of (−) N-desbutylhalofantrine (M−) accumulated throughout the 42 days of multidose treatment and did not reach steady-state. As the concentrations of both halofantrine isomers increased, the QTc interval was prolonged in 14 of the 15 subjects receiving halofantrine who had sufficient data for analysis and not prolonged in any of the five subjects who received placebo. One subject (number 17, Table 1) received only two doses of halofantrine and had insufficient data for analysis.

Figure 1.

Accumulation of (+)-halofantrine and (−)-halofantrine (—) over a once-daily 42 day dosing period in a representative subject. Lines are curve-fit pharmacokinetic functions. All points represent the predose ‘trough’ concentration.

At a once daily oral dose of racemic 500 mg halofantrine hydrochloride, drug exposure was associated with significant lengthening of QTc intervals. The mean QTc interval for the first 22 days of treatment was 423 ms (range=392–433 ms). Before the first treatment, the mean QTc interval was 400 ± 14 ms; 0.5 h after the first treatment, the QTc interval was 392 ± 16 ms; and after 4 h it reached 407 ± 25 ms (P < 0.19). However, after 22 days of treatment, the mean QTc interval reached 423 ± 20 ms (P = 0.001). The difference between means for pretreatment and 22 day QTc is 23 ms (95% C.I. 17.5,28.5). Peak QTc intervals were reached at 4–8 h after the drug was administered on most study days. This 22 day interval was chosen as the most reliable for evaluation because all data were obtained during the period of hospital confinement. In addition this represented maximal prolongation for all subjects receiving halofantrine, with no further prolongation noted during the out of hospital dosing period.

Fitting the linear pharmacodynamic model to QTc interval and halofantrine isomer concentration for each individual who received active drug resulted in a range of correlations with a positive correlation coefficient, most of which were statistically significant for all subjects but one (number 05) (see Table 2; Figures 3–5).

Table 2.

Linear regression analysis of halofantrine isomer concentration vs electrocardiographic QTc interval.

Subject	(+)-Halofantrine vs QTc interval			(−)-Halofantrine vs QTc interval
	r	95% CI	P value	r	95% CI	P value
01	0.361	0.132,0.553	0.0025	0.308	0.061,0.520	0.0153
02	0.511	0.265,0.694	0.0002	0.447	0.183,0.651	0.0014
04	0.334	0.070,0.554	0.0141	0.272	−0.004,0.509	0.0532
05	0.021	−0.262,0.300	0.8867	−0.067	−0.354,0.231	0.6637
07	0.255	0.018,0.465	0.0355	0.120	−0.124,0.350	0.3347
08	0.569	0.241,0.780	0.0016	0.429	0.059,0.696	0.0246
09	0.238	0.003,0.448	0.0470	0.189	−0.048,0.406	0.1174
10	0.525	0.328,0.678	0.0000	0.487	0.281,0.650	0.0000
11	0.339	0.106,0.537	0.0051	0.268	0.026,0.480	0.0305
14	0.522	0.293,0.694	0.0000	0.352	0.079,0.576	0.0126
15	0.379	0.160,0.563	0.0010	0.472	0.260,0.641	0.0000
16	0.555	0.362,0.703	0.0000	0.415	0.182,0.604	0.0008
18	0.455	0.241,0.627	0.0001	0.349	0.119,0.544	0.0036
19	0.168	−0.081,0.397	0.1853	0.131	−0.119,0.365	0.3035
20	0.567	0.364,0.719	0.0000	0.487	0.257,0.665	0.0001

Figure 3.

a) Linear regression of (+)-halofantrine concentration and QTc interval for subject 2 with a typical concentration-QTc relationship. b) Linear regression of (−)-halofantrine concentration and QTc for subject 2.

Figure 5.

a) Linear regression of (+)-halofantrine concentration and QTc interval for subject 5, the individual with no concentration-QTc relationship. b) Linear regression of (−)-halofantrine concentration and QTc for subject 5.

Figure 4.

(A) Linear regression of (+)-halofantrine concentration and QTc interval for subject 20, the individual with the strongest positive concentration-QTc relationship. (B) Linear regression of (−)-halofantrine concentration and QTc for subject 20.

Discussion

In this study, the concentration of (+)-halofantrine exceeded the concentration of (−)-halofantrine by about 2-fold during once daily treatment with oral halofantrine hydrochloride. N-desbutylhalofantrine concentrations were much lower for (+)-N-desbutylhalofantrine (M +) than for (−)-N-desbutylhalofantrine (M−) (Figure 2). The half-life of accumulation was highly variable but not different between (+)- and (−)-halofantrine. Their values are consistent with previously published data for elimination half-life [4–8], which could not be usefully evaluated in this study due to variability in plasma concentration data during the washout phase. The extremely high oral clearance of halofantrine probably reflects its low bioavailability following oral administration in the fasted state [5, 9, 10].

Figure 2.

a) Mean concentrations of (+)-halofantrine (▴) and (−)-halofantrine (▵) for all subjects. Points show all halofantrine concentrations, including the concentration peaks on days 1, 4, 7, 14, 21, and 42. b) Mean concentrations of (+)-N-desbutylhalofantrine (•) and (−)-N-desbutylhalofantrine (○) for all subjects. Data points are the same shown in panel a.

A highly variable but positive correlation between (+)-and (−)-halofantrine concentrations and QTc prolongation was noted for 14 of the 15 subjects in whom sufficient data were available for analysis (Table 2). The correlation appeared to be stronger for (+)-halofantrine than (−)-halofantrine, which may be consistent with indirect evidence suggesting increased cardiotoxic potential of the (+)-isomer [22, 23], or simply because variations in (+)-halofantrine concentrations were greater.

These findings indicate that in young healthy subjects once daily racemic halofantrine dosing leads to extensive accumulation of (+)-halofantrine and steady state concentration of (−)-halofantrine is about one-half that of (+)-halofantrine. In 14 out of the 15 subjects drug concentration was positively correlated with electrocardiographic QTc prolongation but with marked intersubject variability. Whether one halofantrine isomer is more associated with QTc prolongation than the other cannot be determined from these data, as the concentrations of both isomers covary following administration of racemic halofantrine. No arrhythmic events were observed during this study. However, based on previous case reports and the consistent QTc prolongation noted in the present study, the risk of halofantrine-induced torsade de pointes should be evaluated in a larger population being treated for the prophylaxis for malaria.