Pharmacokinetics of Hydroxychloroquine and Its Clinical Implications in Chemoprophylaxis against Malaria Caused by Plasmodium vivax

Subject and study design.The current study consisted of two parts. The first was a PK study which aimed to obtain an adequate model of the PKs of HCQ after the oral administration of HCQ sulfate. The second was a clinical comparison study that was conducted in order to compare the real plasma concentrations from individuals who were orally medicated with HCQ sulfate to the simulated values from the PK model in these studies (Fig. 2).

• Open in new tab
• Download powerpoint

FIG. 2.

Overall study flow.

In our studies, a total of 431 plasma samples were prepared from 22 healthy subjects and 69 patients with vivax malaria for three different studies (studies Ia, Ib, and II). In study Ia, a single dose of HCQ sulfate of 400 mg (310 mg as HCQ) was administered to six healthy South Korean male volunteers and serial, frequent blood samples (8 ml each) were collected at time zero (prior to drug administration) and at 0.5, 1, 2, 3, 4, 6, 8, 12, 24, 48, 72, 144, and 288 h after the administration of HCQ sulfate. This provided the data for selection of the PK structural model. In study Ib, another 16 healthy adult subjects were administered an oral dose of HCQ sulfate of 400 mg; and blood samples were collected at 0, 3, 72, and 168 h after drug administration. In study II, 69 civilian patients with vivax malaria were administered HCQ sulfate at 800 mg and were then administered HCQ sulfate at 400 mg at 6, 24, and 48 h afterward. These patients had febrile illness, and all of them were diagnosed with vivax malaria from May to October 2003 by microscopic examination of peripheral blood smears stained with Giemsa. After the diagnosis was made, the patients were administered HCQ sulfate under the supervision of a physician. Blood samples were drawn from each patient at time zero (the baseline); 24, 48, and 72 h after administration of the first dose; and 6 to 9 days after administration of the last dose.

To evaluate the reasons for prophylactic failures, a clinical comparison study was conducted in which blood samples were collected from 61 soldier patients who had developed vivax malaria from 2000 to 2003, despite the administration of prophylactic doses of HCQ sulfate. The blood samples were used to measure the plasma concentrations of HCQ. The plasma concentrations of HCQ were compared to the simulated time-concentration profiles for the prophylactic medication of HCQ sulfate based on the PK model developed from the current PK studies.

In these studies, one brand of HCQ sulfate was used to treat all subjects. Blood samples were centrifuged at 250 × g for 10 min at 4°C, and the plasma obtained was immediately stored in polypropylene tubes at −70°C until further analysis.

Human use protocol.All protocols for these studies were reviewed and approved by the Institutional Review Board of the Gil Medical Center (Incheon, Republic of Korea), and all the procedures were conducted in accordance with the recommendations of the Declaration of Helsinki on biomedical research involving human subjects. The subjects in studies Ia and Ib were proved to be healthy after comprehensive medical examinations, including a review of their medical histories, physical examination, determination of vital signs, 12-lead electrocardiography, and routine clinical laboratory tests within the 3 weeks before the administration of HCQ sulfate. All the subjects were within 15% of their ideal body weight and had no history of smoking or heavy drinking within 3 months of the study. None of these subjects had taken any medicine within 7 days prior to the commencement of the study. The subjects were not allowed to smoke, consume alcoholic beverages, or ingest caffeine-containing beverages and/or food during the study. They were also instructed to refrain from vigorous activities. The subjects fasted from 10 h prior to the dosing of HCQ sulfate through 4 h after the dosing. All subjects gave written informed consent before any procedures related to this study were performed. The informed consent included information on the regimen, the blood collection schedule, medical examination, the efficacy and possible side effects of the drug, retraction from participation, management of the private database of the volunteers, etc.

Measurement of plasma drug concentrations.HCQ, DCQ, and BDCQ were provided by the U.S. Centers for Disease Control and Prevention (Atlanta, GA). The internal standard, 2,3-diaminoaphthalene, was obtained from Sigma (St. Louis, MO). Plasma concentrations of HCQ and its metabolites, DCQ and BDCQ, were measured by a validated reversed-phase high-performance liquid chromatography method, as described by Easterbrook (13), with slight modifications. In brief, 0.4 ml of each plasma sample was mixed with 50 μl of a distilled water solution of the internal standard (0.1 μg/ml) and 250 μl of 1 M ammonium solution. Extraction was performed with 6 ml of diethyl ether. The supernatant obtained after centrifugation was desiccated with a Speed-Vac apparatus at −80°C. One hundred twenty microliters of the mobile phase was injected into the sample, and the mixture was vortex mixed. Then, 90 μl of the sample was injected into the high-performance liquid chromatograph. Chromatography was performed on a Capcell Pak C₁₈ column (particle size, 5 μm; 4.6 by 150 mm) at room temperature at a flow rate of 1.0 ml/min. The compounds were quantified with a fluorescence detector set at an excitation wavelength of 320 nm and an emission wavelength of 370 nm. The mobile phase consisted of acetonitrile and 0.02 M phosphate buffer (389:1,000, vol/vol; pH 4.9). The method was validated in the range of 10 to 2,000 ng/ml (10, 20, 50, 100, 200, 500, 1,000, 2,000 ng/ml) for HCQ and 5 to 200 ng/ml (5, 10, 20, 50, 100, 200 ng/ml) for DCQ and BDCQ. Intra- and interassay coefficients of variation varied from 3.1% to 5.2% and from 3.5 to 7.3%, respectively, for HCQ at 10, 50, 1,000, and 2,000 ng/ml; from 7.8% to 12.2% and from 8.1 to 11.8%, respectively, for DCQ at 5, 20, 100, and 200 ng/ml; and from 7.2% to 12.5% and from 7.8 to 12.7%, respectively, for BDCQ at 5, 20, 100, and 200 ng/ml. The intra- and interassay accuracies were less than 15% for all the compounds. The lower limits of quantification were 10 ng/ml and 5 ng/ml for DCQ and BDCQ, respectively, and the intra- and interday coefficients of variation were less than 20% for all compounds. Concentrations below the lower limit of quantification prior to HCQ administration were considered to be 0 ng/ml.

PK analysis. (i) Noncompartmental analysis.Serial plasma concentration data for HCQ, DCQ, and BDCQ from six healthy subjects were analyzed by noncompartmental methods with the WinNonlin (version 5.2) program (Pharsight Corporation, Mountain View, CA). The numbers of data used in the analysis, excluding the concentrations below the limit of quantification, were 77, 59, and 66 for HCQ, DCQ, and BDCQ, respectively. The maximum drug concentrations in plasma (C_max) and the time to C_max were determined directly from the observed values. The terminal elimination rate constant (λ_z) was estimated by linear regression of the log-linear decline of at least three individual plasma time-concentration data. The terminal half-life (t_1/2) was calculated for each individual as follows: t_1/2 = ln(2)/λ_z. The area under the concentration-time curve (AUC) from time zero to the last measurable time (AUC_last) was calculated by the linear-log linear trapezoidal method. The AUC from time zero extrapolated to infinity (AUC_inf) was also calculated by using a combination of the linear-log linear trapezoidal method and extrapolation to infinity by using λ_z and the last observed concentration.

(ii) Analysis by mixed-effect modeling.Plasma concentration data for HCQ from all 91 subjects were analyzed by mixed-effect modeling by using the NONMEM (version VI) program (GloboMax Limited Liability Company, Hanover, MD). The PK parameters were estimated with NONMEM subroutines ADVAN4 and TRANS4 by use of the FOCE (first-order conditional estimation) with INTERACTION method. The parameters for a specific subject are described by equation 1: $$mathtex$$\[P_{i}{=}P_{TV}{\times}\mathrm{exp}({\eta}_{i})\]$$mathtex$$(1) where P_TV is the typical value of a parameter, and η_i is a normally distributed variable with zero mean.

The residual error model was characterized by use of the combined-error mode, as described by equation 2: $$mathtex$$\[C_{\mathrm{obs}}{=}C_{\mathrm{pred}}{+}(C_{\mathrm{pred}}{\varepsilon}_{1}){+}{\varepsilon}_{2}\]$$mathtex$$(2) where C_obs is the observed concentration, C_pred is the predicted concentration, and ε₁ and ε₂ are zero mean normally distributed variables.

Various compartmental models and error models were assessed, guided by a graphical assessment of the optimum fit properties and statistical significance criteria. The covariates tested for HCQ PKs were age, sex, body weight, height, and disease status (healthy or malarial). To identify a potentially significant covariate, random permutation tests were conducted over 1,000 times for each variable or combination of variables. A likelihood ratio test was used to discriminate between the hierarchical models at a P value of ≤0.05, based on the fact that the distribution of the −2 log likelihood of the models approximately follows a chi-square distribution. Standard diagnostic plots, including the observed values of the dependent variable versus the individual predicted values and the individual predicted values versus the individual weighted residuals, were used for the detection of optimum fit capabilities. Other diagnostics were the objective function value and the standard error of the parameter estimates. To evaluate the stability of the model and to confirm the result, bootstrapping with wings was conducted with the NONMEM program (27). A total of 2,000 bootstrap runs were performed, and from the resultant parameter distributions, the 95% confidence intervals of the parameter estimates were obtained as 2.5th and 97.5th percentiles. The modeling process was facilitated by use of the Asan software tool for NONMEM, which is an interface for NONMEM based on text editor and the R program (19).

HCQ time-concentration profiles at the dosage used for the prophylaxis of malaria (HCQ sulfate at 400 mg a week) were simulated by using the NONMEM (version VI) program and the fixed- and random-effect parameter estimates. The 95% prediction intervals from the simulation were compared to the actual plasma HCQ concentration data that were obtained from vivax malaria patients who developed the disease, despite the previous prophylactic administration of HCQ sulfate.

Subjects.PK studies were conducted with 91 subjects in three different substudies (studies Ia, Ib, and II). Study Ia consisted of 6 male healthy individuals, and study Ib consisted of 15 healthy male and 1 healthy female individuals. Study II consisted of 49 male and 20 female patients with vivax malaria. The demographic characteristics of the subjects from each substudy are shown in Table 1.

PK analysis.In the noncompartmental analysis with six healthy subjects (study Ia), AUC_inf and C_max were 102.3 ± 60.8 nmol · h/ml (mean ± standard deviation) and 1.22 ± 0.40 nmol/ml, respectively, for HCQ; 37.7 ± 16.9 nmol · h/ml and 0.06 ± 0.03 nmol/ml, respectively, for DCQ; and 53.6 ± 44.5 nmol · h/ml and 0.36 ± 0.64 nmol/ml, respectively, for BDCQ (Fig. 3; Table 2). The measure of clearance divided by the measure of bioavailability (CL/F) and the volume of distribution based on the terminal elimination phase divided by the bioavailability of HCQ were calculated to be 12.0 ± 6.8 liters/h and 2,851 ± 2,147 liters, respectively.

• Open in new tab
• Download powerpoint

FIG. 3.

Plasma concentrations (mean and standard deviation) of HCQ and its metabolites in the six healthy subjects in study Ia before and after the administration of a single oral dose of HCQ sulfate at 400 mg.

The population PK analysis of HCQ was conducted with 431 plasma concentration data from all the 91 subjects in the PK studies by using the NONMEM (version VI) program. The model that best described the typical time course of the plasma HCQ concentrations was a two-compartment linear model with first-order absorption. The model was improved significantly by adding an absorption lag from the depot compartment to the central compartment (change in the objective function value, 32.5 [from 3,334.8 to 3,302.3]). No covariate was included in the final model, since a significant tendency on graphics between the interindividual variabilities of each fixed-effect parameter estimate and the various demographic variables was not observed, and none of the variables significantly reduced the objective function value when they were incorporated into the model. CL/F was estimated to be 10.9 liters/h, and from this value, an average steady-state concentration of 0.51 M (170 ng/ml) was predicted. The population PK parameter estimates of HCQ and basic diagnostic plots are presented in Table 3 and Fig. 4, respectively.

• Open in new tab
• Download powerpoint

FIG. 4.

Diagnostic plots for the final population PK model of HCQ in the pharmacokinetic studies (studies Ia, Ib, and II). Solid line, fit by local regression (loess); dashed line, line of identity.

Simulation of a time-concentration profile of HCQ.By using the final PK model developed in this study, 1,000 simulations were performed for the plasma concentrations of prophylactic doses of HCQ (HCQ sulfate at 400 mg a week) with the NONMEM (version VI) program. These results were compared to the actual plasma concentration data for HCQ for 61 soldier patients with vivax malaria (clinical comparison study) who became infected, despite the previous weekly chemoprophylactic administration of 400 mg HCQ sulfate for more than 4 weeks as a preventive measure. The plasma concentrations of HCQ in 43 patients (71%) were found to be below the lower bounds of the 95% prediction interval, and those in 5 patients (8%) were found to be higher than the upper bounds of the 95% prediction interval. The plasma concentrations were within the range of the 95% prediction interval in only 13 patients (21%) (Fig. 5). On the other hand, the plasma concentrations of HCQ were below the average predicted concentrations in 48 patients (79%), whereas they were above the average predicted concentrations in 13 patients (21%).

• Open in new tab
• Download powerpoint

FIG. 5.

Comparison of the actual plasma concentrations of HCQ in 61 soldier patients in the clinical comparison study who were infected with malaria parasites despite chemoprophylaxis for longer than 4 weeks to the simulated plasma time-concentration profiles of HCQ after oral administration of HCQ sulfate with a prophylactic dose of 400 mg/week.