Quantification of Antimalarial Bisthiazolium Compounds and Their

Clinical Chemistry 51:3 593– 602 (2005)

Drug Monitoring and Toxicology

Quantification of Antimalarial Bisthiazolium Compounds and Their Neutral Bioprecursors in Plasma by Liquid Chromatography-Electrospray Mass Spectrometry Olivier Nicolas,1† Christine Farenc,2† Miche`le Calas,3 Henri J. Vial,2 and Françoise Bressolle1*

6.5–1309.8 ␮g/L for T4; 20 –2000 ␮g/L for TE4c) according to a quadratic equation. The accuracy ranged from 85% to 113.1%, and the imprecision from 2.2% to 15%. The mean extraction recoveries were 87%, 98%, and 80% for T3, T4, and TE4c, respectively. The lower limit of quantification was 6.4 ␮g/L for the two bisthiazolium compounds, whereas it was 20 ␮g/L for TE4c, the related lipophilic prodrug. Conclusion: This highly specific and sensitive method is suitable for analyzing samples collected during preclinical pharmacokinetic studies in rats and to determine the percentage binding of T3 and T4 to human plasma proteins.

Background: A new class of antimalarial drugs targeting membrane biogenesis during intraerythrocytic Plasmodium falciparum development has been identified. The bisthiazolium salts T3 and T4 have superior in vitro and in vivo parasite-killing properties and need to be monitored. Methods: We used a liquid chromatography– electrospray ionization mass spectrometry method (positive mode) to quantify two bisthiazolium compounds (T3 and T4) and a related prodrug (TE4c) in human and rat plasma. Verapamil was used as internal standard. Verapamil and the TE4c compound were characterized by protonated molecules at m/z 455.7 and m/z 725.7, respectively. T3 and T4 were detected through two ions [M2ⴙ/ 2] at m/z 227.7 and m/z 241.8 and by their adducts with trifluoroacetic acid [MⴙTFA]ⴙ at m/z 568 and m/z 596, respectively. The sample clean-up procedure involved solid-phase extraction. HPLC separation was performed on a reversed-phase column, using a water–acetonitrile gradient, with both solvents containing TFA. Stability under various conditions was also investigated. Results: The peak-area ratios (drugs/internal standard) were linked to concentrations (6.4 –1282 ␮g/L for T3;

© 2005 American Association for Clinical Chemistry

Malaria is one of the world’s major infectious diseases and is prevalent in both tropical and subtropical regions. Its burden is increasing because of drug and insecticide resistance as well as social and environmental changes (1 ). A limited number of antimalarial drugs are available, but their efficacy against Plasmodium falciparum, the most lethal malaria-causing parasite, is jeopardized by spreading drug resistance (2 ). Thus, the development of new chemotherapies is urgently needed, particularly compounds that work through new independent mechanisms of action and are structurally unrelated to existing antimalarial agents. During the last 10 years, a novel pharmacologic target, i.e., the phospholipid metabolism of Plasmodium, has been identified. Compounds mimicking the structure of choline, a precursor required for phosphatidylcholine synthesis by the parasite, inhibit de novo phosphatidylcholine biosynthesis and are highly active against multiresistant P. falciparum malaria (3–10 ). Three generations of compounds have been synthesized. The first two consist of mono- or bisquaternary ammonium salts (6, 7 ) and ami-

1 Clinical Pharmacokinetic Laboratory, Faculty of Pharmacy, University Montpellier I, Montpellier, France. 2 Unite´ Mixte de Recherche 5539, Centre National de la Recherche Scientifique, University Montpellier II, Montpellier, France. 3 Unite´ Mixte de Recherche 5810, Centre National de la Recherche Scientifique, Universities Montpellier I and II, Montpellier, France. †Olivier Nicolas and Christine Farenc contributed equally to this work. *Address correspondence to this author at: Laboratoire de Pharmacocine´tique Clinique, Faculte´ de Pharmacie Montpellier, 15 Avenue Charles Flahault, BP 14491, 34093 Montpellier Cedex 5, France. Fax 33-4-6754-8075; e-mail [email protected]. Received September 7, 2004; accepted November 30, 2004. Previously published online at DOI: 10.1373/clinchem.2004.042580

593

594

Nicolas et al.: LC/ESI-MS Quantification of Bisthiazolium Antimalarials

dine and guanidine compounds, respectively. These compounds have outstanding efficacy both in vitro and in vivo against P. falciparum and P. cynomolgi, a P. vivaxrelated parasite, with a high therapeutic index (3 ). The bisquaternary ammonium salts showed the highest activities. The third generation consists of neutral prodrugs that deliver bisthiazolium salts with antimalarial activity when present in nanomolar concentrations (11 ). In plasma, the prodrug is rapidly converted into the active drug, and the prodrugs have also been found to offer protection in a murine model of malaria as well as in P. cynomolgi-infected Rhesus monkeys (11 ). The rapid clearance of parasitemia, the absence of recrudescence, and the capacity of curing highly infected animals at a moderate dose (11 ) highlight that the bisthiazolium compounds T3 and T4 and their bioprecursors have very promising clinical potential for malaria treatment and need to be monitored. In this report, we describe an analytical method to simultaneously quantify the TE4c prodrug and T4, its enzyme-converted active metabolite, along with T3, a T4-like bisthiazolium compound in which methoxy groups are replaced by hydroxyl functions. T3 could be a possible metabolite of T4 in the body. The structures of these compounds are presented in Fig. 1 of the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol51/issue3/. The lack of ultraviolet absorbance and the presence of two quaternary ammonium moieties in both T3 and T4 prompted us to investigate the potential of electrospray ionization mass spectrometry (ESI-MS).4 Verapamil was selected as internal standard. The method was validated according to validation procedures, parameters, and acceptance criteria (12–15 ) based on US Pharmacopoeia XXIII guidelines (13 ) and the recommendations of Shah et al. (14 ). ESI-MS was used to assay samples from a preclinical study performed in rats and to determine the percentage binding of T3 and T4 to human plasma proteins.

Materials and Methods chemicals and reagents T3 {1,12-bis[4-methyl-5-(2-hydroxyethyl)-thiazol-3-ium-3yl]dodecane dibromide; Mr 614.7}, T4 {1,12-bis[4-methyl5-(2-methoxyethyl)-thiazol-3-ium-3-yl]dodecane dibromide; Mr 642.8}, and TE4c {N,N⬘-diformyl-N,N⬘-di[(1Z)-1-methyl2-S-thiobenzoyl-4-methoxybut-1-enyl]-1,12-diaminododecane; Mr 725} were obtained from the Laboratoire des Aminoacides, Peptides et Proteines (Universities Montpellier I and II) (16 ) and were stored at ambient temperature with protection from light.

4 Nonstandard abbreviations: ESI-MS, electrospray ionization mass spectrometry; TFA, trifluoroacetic acid; QC, quality control; LC, liquid chromatography; NMR, nuclear magnetic resonance; RSD, relative standard deviation; and LLOQ, lower limit of quantification.

During optimization of the analytical method, different internal standards were tested: T1 [3-dodecyl-5-(2-hydroxyethyl)-4-methyl-1,3-thiazol-3-ium bromide], T2 [3-dodecyl-5(2-methoxyethyl)-4-methyl-1,3-thiazol-3-ium bromide], G25 [1,16-hexadecamethylene bis-(N-methylpyrrolidinium) dibromide], and verapamil (see Fig. 2 of the online Data Supplement). The first three drugs are of the same chemical series as the test drugs. T1 and T2 are the monocationic analogs of T3 and T4, respectively. On the basis of the results for extraction recovery, accuracy, imprecision, and specificity (Table 1 of the online Data Supplement), verapamil was chosen as internal standard. The internal standard and the trifluoroacetic acid (TFA) were obtained from Sigma. Verapamil was stored protected against light at room temperature (20 °C). HPLC-grade acetonitrile, methanol, and dichloromethane were purchased from Merck. For the method validation, blood samples from healthy volunteers (Etablissement Français du sang) and Sprague–Dawley rats (Charles River Laboratories) were collected in sodium heparin, and plasma was obtained by centrifugation at 1000g for 10 min. For each species, pooled drug-free plasma samples were aliquoted, frozen at ⫺30 °C, and then used during the study in the preparation of standards and qualitycontrol (QC) samples. Stock solutions of T3 and T4 (64 and 65.5 mg/L, respectively, expressed in the form of bis-charged compounds) were prepared in distilled water. Stock solutions of TE4c (100 mg/L) and verapamil (250 mg/L) were prepared in acetonitrile– distilled water (50:50 by volume) and in methanol– distilled water (1:100 by volume), respectively. For each compound, two separate stock solutions were prepared: one was used for preparation of the calibration curve, and the second was used for the preparation of QC samples.

instrumentation ESI mass spectra were recorded on a Platform II quadrupole mass spectrometer (Micromass) fitted with an electrospray ion source. The mass spectrometer was calibrated in the positive ion mode by use of a mixture of NaI and CsI. Voltages were set at ⫹3.5 kV for the capillary and ⫹0.5 kV for the skimmer lens. The source was heated at 120 °C. Nitrogen was used as the nebulizing and drying gas at 15 and 250 L/h, respectively. The sampling cone voltage was set at 30 V. An Alliance 2690 LC system (Waters) equipped with an autosampler was used to deliver the mobile phase, which was continuously degassed. Liquid chromatography (LC)/ESI-MS experiments were carried out with a C18 end-capped Xterra column [30 ⫻ 2.1 mm (i.d.)], packed with 3.5-␮m particles (Waters), with a LC flow rate of 400 ␮L/min. The postcolumn flow was split to yield a flow rate in the source of 35 ␮L/min. A linear gradient from water containing 1 mL/L TFA to acetonitrile containing 1 mL/L TFA as organic

Clinical Chemistry 51, No. 3, 2005

modifier was applied over 8 min. The column was then washed for 1 min at the final gradient condition, brought back to the initial condition over 1 min, and reequilibrated for 4 min. The total run cycle was 14 min. Chromatography was performed at 20 °C. Mass spectrometric data were acquired in the singleion recording mode. Verapamil and the TE4c prodrug were characterized by the protonated molecules [M⫹H]⫹ at m/z 455.7 and m/z 725.7, respectively (Fig. 1, A and B). To increase the precision and accuracy of the method at low concentrations and the specificity, the T3 and T4 compounds were detected by use of two ions, the quaternary ammonium salts [M2⫹/2] at m/z 227.7 and m/z 241.8 and by the adducts with TFA [M⫹TFA]⫹ at m/z 568 and m/z 596, respectively (Fig. 1, C and D). To study the structure of the TE4c prodrug, we recorded 1H-nuclear magnetic resonance (NMR) spectra on a Bruker 400 MHz instrument using dimethyl sulfoxide-d6 solutions.

working solutions Stock solutions of T3 and T4 were diluted in distilled water to obtain 11 working calibrators ranging from 0.16 to 32 mg/L and 0.164 to 32.8 mg/L, respectively. The TE4c stock solution was diluted in acetonitrile– distilled

595

water (50:50 by volume) to obtain 11 working calibrators ranging from 0.25 to 50 mg/L. The stock solution of verapamil was diluted 20-fold (12.5 mg/L) in distilled water before use. An unextracted working solution containing 1.3 mg/L each of T3 and T4, 2 mg/L TE4c, and 2.5 mg/L verapamil in acetonitrile–1 mL/L TFA was injected before each run to check the performance of the LC/ESI-MS system.

preparation of calibration curves and qc samples Calibrators were prepared by adding 20 ␮L of the appropriate working solutions into 0.5 mL of drug-free plasma. The effective concentrations of T3 and T4 in plasma (expressed in the form of bis-charged compound) were 6.4, 12.8, 32.1, 64.2, 128.2, 320.6, 641.2, and 1282 ␮g/L and 6.5, 13.1, 32.7, 65.5, 131, 327.4, 654.9, and 1309.8 ␮g/L, respectively. For TE4c, the concentrations were 20, 50, 100, 200, 500, 1000, and 2000 ␮g/L. QC samples used in the validation were prepared in the same way as the calibrators, by mixing drug-free plasma samples with appropriate volumes of working solutions to achieve three different concentrations (T3: 16, 160.3, and 961.9 ␮g/L; T4: 16.4, 163.7, and 982.3 ␮g/L; TE4c: 25, 250, and 1500 ␮g/L).

Fig. 1. Mass spectra (scan mode) for verapamil (A), TE4c (B), T3 (C), and T4 (D).

596


sample preparation procedure Sample pretreatment involved a solid–liquid extraction. The 1-mL Waters Oasis HLB cartridges (30 mg of sorbent; mean particle diameter, 30 ␮m) were conditioned with 1 mL of methanol followed by 1 mL of distilled water before use. In a 5-mL glass tube, a 0.5-mL aliquot of plasma was mixed with 0.5 mL of water containing 10 mL/L TFA and 20 ␮L of internal standard solution (12.5 mg/L). The mixture was vortex-mixed for 15 s and centrifuged for 15 min at 1500g, and the supernatant was loaded on the cartridge. The column was then rinsed with 1 mL of distilled water. The elution was carried out with two 1-mL volumes of acetonitrile containing 1 mL/L TFA to eluate polar compounds (T3, T4, and verapamil), and then with two 1-mL volumes of dichloromethane to eluate the lipophilic prodrug TE4c. The eluate fractions were collected in a 10-mL glass tube and evaporated to dryness under a stream of nitrogen for 30 min at 40 °C. The dried residue was reconstituted in 100 ␮L of acetonitrile containing 1 mL/L TFA and placed in an ultrasonic bath for 1 min. A 3-␮L volume was injected into the HPLC system for analysis.

data analysis From recorded peak areas, the ratio of each analyte to internal standard was calculated. The peak-area ratios were linked to the concentrations of each analyte in plasma according to a quadratic process as: y ⫽ ax2 ⫹ bx ⫹ c. The regression curve was not forced through zero. Calibration curve equations were used to calculate “backcalculated” concentrations for the calibrators. The backcalculated values were statistically evaluated. The gaussian distribution of the residuals (difference between nominal and back-calculated concentrations) was verified. In addition, the mean residual values (mean predictor error) were computed and compared with zero (Student t-test); the 95% confidence intervals were also computed.

stituted extracts in the two studied matrices/reference solutions) were as follows: T3, 1.01 (CV ⫽ 5.0%); T4, 1.05 (CV ⫽ 3.2%); TE4c, 1.00 (CV ⫽ 4.1%), and verapamil, 1.06 (CV ⫽ 5.3%). Thus, no ion suppression was observed. We investigated interference from endogenous compounds by analyzing extracts from 10 different batches of blank human and rat plasma. We also verified the possible interference by other drugs that might be taken concomitantly with the test drugs. The following drugs were checked: acetylsalicylic acid, acetaminophen, ibuprofen, dextropropoxyphen, fluoxetine hydrochloride, bromazepam, oxazepam, zolpidem, pindolol, atenolol, digoxin, meprobamate, theophylline, valproic acid, chloroquine, quinine, amodiaquin, mefloquine, sulfadoxine, and pyrimethamine.

imprecision and accuracy We assessed both intra- and interassay imprecision and accuracy by analyzing QC samples at the above-mentioned three concentrations against calibration curves. For intraassay studies, we performed replicate analyses (n ⫽ 6) of each QC sample on the same day. For interassay studies, we analyzed each of the three QC samples once a day on different days (n ⫽ 6 –7). The accuracy was expressed as the mean measured concentration/theoretical concentration ⫻ 100. Imprecision is given as the relative standard deviation (RSD). To test whether it is possible to apply the method to samples whose concentrations are higher than the highest calibration point, we prepared QC samples containing the T3 and T4 compounds at 1300, 3250, and 6500 ␮g/L (expressed in the form of bis-charged compound). We then diluted these samples 2-, 5-, and 10-fold with drugfree human or rat plasma to bring the concentrations within the range of the calibration curves. Each analysis was performed six times at each concentration, using calibration curves and QC samples. The observed concentrations were reported and compared with the nominal concentrations.

ion suppression and specificity studies The absence of ion suppression was demonstrated by the method of Matuszewski et al. (17 ). To investigate potential ion suppression effects attributable to the matrix, we treated 10 different batches of each drug-free matrix (human and rat) as described above. The reconstituted extracts (100 ␮L of acetonitrile–1 mL/L TFA) were then enriched with the three drugs (at two different concentrations) and the internal standard to final nominal concentrations of 0.8 and 4.8 mg/L (T3 and T4), 1.25 and 7.5 mg/L (TE4c), and 2.5 mg/L (internal standard). A reference solution containing 100 ␮L of acetonitrile–1 mL/L TFA was also enriched with the four drugs to the same nominal concentrations. The reconstituted extracts and reference solutions were injected into the LC/MS system. Peak areas obtained from the reconstituted extracts were compared with the corresponding peak areas produced by the reference solutions. The mean area ratios (recon-

determination of the lower limit of quantification The lower limit of quantification (LLOQ) was defined as the lowest concentration that could be determined with an accuracy within 80 –120% and an imprecision ⱕ20% on a day-to-day basis (11–14 ). To determine the analytical error in the LLOQ, we used QC samples containing the compounds of interest.

extraction efficiency We determined the extraction recoveries of the three analytes by use of QC samples (T3: 16, 160.3, and 961.9 ␮g/L; T4: 16.4, 163.7, and 982.3 ␮g/L; TE4c: 25, 250, and 1500 ␮g/L) prepared as described above, except that the internal standard was added to the supernatant before evaporation. The area ratios were then compared with those obtained from reference samples. Reference samples


were prepared by adding the analytes, the internal standard, and an adequate volume of acetonitrile to drug-free plasma extracts to obtain the same final concentration of drugs as in the plasma samples. Three replicates of each concentration were extracted. The extraction efficiency was also determined for the internal standard.

stability study For stability studies in the two matrices, QC samples (T3: 16, 160.3, and 961.9 ␮g/L; T4: 16.4, 163.7, and 982.3 ␮g/L; TE4c: 25, 50, 250, and 1500 ␮g/L) were used. The stabilities of the three compounds in human and rat plasma samples were evaluated during storage and during all steps of the analytical method. The short-term stability in plasma was assessed after 0.5, 1, 2, 4, and 6 h of storage at both routine laboratory conditions (20 °C with exposure to daylight) and at 4 °C. The long-term stability in frozen plasma (⫺30 °C) was determined by periodic analysis over a span of 2 months. Enriched samples were analyzed immediately after preparation (reference values) and at selected time intervals after storage over the study period. Before their analyses, frozen samples were brought to room temperature and vortex-mixed well. Each determination was performed in triplicate. For the two bisthiazolium compounds (T3 and T4), we also investigated the freeze-thaw stability. Enriched plasma samples were analyzed immediately after preparation and on a daily basis after repeated freezingthawing cycles at ⫺30 °C on 3 consecutive days. We determined the run-time stability of processed samples at room temperature and at 4 °C for 24 h after extraction for each calibration point. Compounds were considered as stable when losses were ⬍15%.

percentage binding of t3 and t4 compounds to human plasma proteins In a first step, we determined the extent of plasma protein binding of T3 and T4 compounds by an ultrafiltration procedure using the Amicon Micropartition System (YMT-1 ultrafiltration membrane). We separated free drug from protein-bound drug by subjecting the system to low-speed centrifugation for 30 min (990g at 37 °C). Over the concentration range studied (60 –1300 ␮g/L), significant nonspecific adsorption (30 – 40%) of the two compounds to the membrane occurred. In view of this problem, we chose an ultracentrifugation method (18 ). The lack of proteins in the supernatant was assessed by a bicinchoninic acid protein assay (Sigma-Aldrich). We determined the time necessary for binding of T3 (64.2 and 1282 ␮g/L) and T4 (65.5 and 1309.8 ␮g/L) to plasma proteins to reach equilibrium in a preliminary study. For this, multiple enriched plasma samples (10 mL) were placed in a water bath at 37 °C for 0.5–3 h. At selected time intervals, one of the 10-mL samples was removed. From each 10-mL sample, 8 mL of plasma was re-

597

moved and immediately centrifuged at 100 000g for 24 h at 4 °C (Beckman Optima LE70 with a 50 Ti rotor); the free drug was then quantified in the supernatant. The remaining 2 mL were used to quantify the total drug concentration. For each concentration, three replicates were assayed according to the above procedure. We studied the percentage binding of the two drugs to plasma proteins, using plasma samples containing 64.2, 128.2, 320.6, 641.2, 1282, and 2564 ␮g/L T3 and 65.5, 131, 327.4, 654.9, 1309.8, and 2619.6 ␮g/L T4. Samples were placed at 37 °C and then centrifuged according the same procedure as described above. Measurements were performed in triplicate.

pharmacokinetic study The method was used to quantify T3 and T4 in Sprague– Dawley rat plasma after intraperitoneal administration of 6.4 and 23.7 mg/kg of the two drugs, expressed in the form of bis-charged compound, respectively. This research adhered to NIH guidelines (19 ). The animal study was approved by the local Animal Use Committee. Blood samples (two samples per rat) were drawn in heparincontaining polypropylene tubes at the following timepoints (six animals per timepoint), before administration and 10, 20, and 30 min, and 1, 2, 4, 8, 12, 24, and 36 h after drug administration. The first blood sample was drawn by venipuncture from the tongue vein; the second blood sample was collected after sacrifice of the animal by section of the carotid artery. Animals were anesthetized with diethyl ether 2 min before sampling. Blood samples were centrifuged at 4 °C (2000g for 20 min). The resulting plasma was transferred to polypropylene tubes and stored at ⫺30 °C until assay. Pharmacokinetic parameters were computed from the mean concentrations at each time points by use of Pk-fit software (20, 21 ).

Results retention times and specificity Representative chromatograms are shown in Fig. 2. Under the chromatographic conditions described above, peaks were adequately separated. During the 6 months of validation, observed retention times were 4.5 min (RSD ⫽ 1.6%) for T3, 5.0 min (RSD ⫽ 1.9%) for T4, 8.8 min (RSD ⫽ 2.1%) for TE4c, and 5.9 min (RSD ⫽ 1.7%) for the internal standard. For the TE4c compound, some chromatographically separable isomeric forms were observed, which corresponded to rotational isomers. Each end of the bioprecursor molecule can adopt various planar and stable conformations that result from a 180-degree rotation around some bonds. For example, rotation around the N–CHO bond (bond a in Fig. 1 of the online Data Supplement) and around the N–C bond of the ethylenic carbon (bond b in Fig. 1 of the online Data Supplement) leads to equilibria between several conformers. To corroborate the importance of these rotations, the effect of

598


Fig. 2. Typical chromatograms (single-ion recording mode) of blank human plasma enriched with verapamil (internal standard) at 2.5 mg/L (A), blank human plasma enriched with T3 at 0.32 mg/L (B), blank human plasma enriched with T4 at 0.32 mg/L (C), and blank human plasma enriched with TE4c at 0.5 mg/L (D). For LC/MS conditions, see the text.

temperature on the chromatogram and on the 1H-NMR spectrum was examined. As expected, when the temperature of the mobile phase was increased to 40 °C, peaks corresponding to the different conformers merged to give just one peak on the LC/MS chromatograms, which is consistent with a faster rotation about the above-mentioned bonds at higher temperatures than at room temperature. On the 1H-NMR spectrum at room temperature, the aldehydic proton (CHO) showed two singlets, 7.77 and 8.00 ppm (in a 5:1 ratio), each one corresponding to one rotamer. Similarly, the methyl group protons (CH3 on the ethylenic carbon) showed two singlets, one per rotamer, at 2.08 and 1.98 ppm (in a 5:1 ratio). At low temperature, the conversion between two rotamers is slow, allowing detection of the chemical shifts of each group of protons on each rotamer. Increasing the tem-

perature from 20 to 120 °C induced a coalescence of each of the two singlets (7.88 ppm for CHO and 2.00 ppm for CH3), displaying a faster rotation (Table 1). Thus, in quantifying TE4c, we used the sum of the isomer peaks. As shown in Fig. 3, no peaks attributable to the matrix interfered at the retention time of the analytes. No interference was found with all tested drugs.

Table 1. Chemical shifts of two groups of protons on the 1 H-NMR spectrum of the TE4c prodrug. Temperature Chemical shifts of H

In CHO In NOC(CH3)AC

20 °C

120 °C

7.77 and 8.00 ppm (in 5:1 ratio) 2.08 and 1.98 ppm (in 5:1 ratio)

7.88 ppm 2.00 ppm

599


Fig. 3. Chromatograms of blank human plasma at m/z 455.7 (A), m/z 725.7 (B), m/z 227.7 ⫹ 568 (C), and m/z 241.8 ⫹ 596 (D).

drug/detector response relationship No significant differences appeared between calibration curves constructed with rat plasma and human plasma. Thus, we report only results of calibration curves constructed with human plasma. The quadratic regression indicated mean (SD) coefficients of determination of 0.9992 (0.90 ⫻ 10⫺3) for T3 (n ⫽ 10), 0.9970 (0.57 ⫻ 10⫺2) for T4 (n ⫽ 14), and 0.9994 (0.57 ⫻ 10⫺3); (n ⫽ 10) for TE4c. Mean parameters of the quadratic equations were as follows: for T3, a ⫽ ⫺1.05 ⫻ 10⫺8; b ⫽ 0.895 ⫻ 10⫺3 (RSD ⫽ 14%); c ⫽ ⫺0.0092; for T4, a ⫽ ⫺9.42 ⫻ 10⫺7; b ⫽

4.55 ⫻ 10⫺3 (RSD ⫽ 19%); c ⫽ ⫺0.157; and for TE4c, a ⫽ ⫺2.2 ⫻ 10⫺8; b ⫽ 2.05 ⫻ 10⫺4 (RSD ⫽ 21%); c ⫽ ⫺0.0048. For each point on the calibration curves, the concentrations were back-calculated from the corresponding quadratic equation, and mean (SD) values were computed. The results are presented in Table 2. Linear regression of the back-calculated concentrations vs the nominal ones provided a unit slope and an intercept equal to zero. The distribution of the residuals showed random variations, the number of positive and negative values being approximately equal. Moreover, they were gaussian distributed

Table 2. Back-calculated concentrations from calibration curves constructed in human plasma.a T3 (n ⴝ 10) Theoretical concentration, ␮g/L

RSD, %

6.4 12.8 32.1 64.2 128.2 320.6 641.2 1282

20 17 6.9 12 6.0 7.7 5.8 0.9

a

T4 (n ⴝ 14)

Recovery, %

Theoretical concentration, ␮g/L

RSD, %

100.0 111.7 106.5 95.3 99.2 100.8 101.3 99.7

6.5 13.1 32.7 65.5 131 327.4 654.9 1309.8

18 13 8.9 13 12 5.1 4.8 5.6

Expressed as the bis-charged compounds.

TE4c (n ⴝ 10)

Recovery, %


RSD, %

Recovery, %

93.8 95.4 95.1 102.6 97.6 100.4 98.9 97.8

20 50 100 200 500 1000 2000

20 13 12 8.7 4.2 4.6 3.5

96.5 110.4 107.2 100.7 102.5 103.2 103.1

600


Table 3. Accuracy and imprecision of the method. T3a Theoretical concentration, ␮g/L

Human plasma Within-day (n ⫽ 6)b

Between-dayc

Rat plasma Between-day (n ⫽ 6)

Imprecision, %

T4a

Accuracy, %


Imprecision, %

TE4c

Accuracy, %


Imprecision, %

Accuracy, %

16.0 160.3 961.9

7.8 5.4 7.4

107.5 94.6 92.6

16.4 163.7 982.3

7.0 2.5 2.2

87.2 97.9 90.2

25 250 1500

12 8.0 7.0

103.0 92.0 95.5

16.0 160.3 961.9

15 12 9.2

113.1 104.2 104.4

16.4 163.7 982.3

11 12 9.0

92.1 100.3 94.2

25 250 1500

12 15 6.7

106.5 99.6 101.9

16.0 160.3 961.9

14 5.6 6.8

98.8 102.8 100.0

16.4 163.7 982.3

14 10 4.2

104.9 93.9 98.6

25 250 1500

3.5 12 1.8

88.4 98.4 85.0

a

Expressed as the bis-charged compounds. Number of replicates. c n ⫽ 7 for T3 and T4; n ⫽ 9 for TE4c. b

and centered around zero. The mean values of residuals were not statistically different from zero (⫺0.935, 2.91, and ⫺2.83 ␮g/L for T3, T4, and TE4c, respectively), and the 95% confidence intervals included zero (⫺5.81 to 3.94 for T3; ⫺2.98 to 8.80 for T4; ⫺5.94 to 0.28 for TE4c, respectively).

imprecision, accuracy, extraction efficiency, and lloq The accuracy and imprecision results are shown in Table 3. Dilution has no influence on the performance of the method. After 2-, 5-, and 10-fold dilutions, the imprecisions for the results were 8.8 –9.3%, 7.8 – 8.2%, and 9.1– 9.6%, respectively. The corresponding accuracy values were 94 –100%, 94 –96%, and 98 –100%, respectively.

For T3 and T4 (n ⫽ 9 per matrix), the mean (SD) extraction recoveries were 87 (5.3)% and 98 (1.7)%, respectively. The extraction recovery was 80 (6.3)% for TE4c (n ⫽ 9 per matrix). These recoveries were not statistically different over the range of concentrations studied. We also determined the extraction recovery of the internal standard, which was 83 (6)% (n ⫽ 3 per matrix). The extraction recoveries did not differ statistically according to the matrix studied. In the two matrices (0.5 mL of plasma), the LLOQ was established as 6.4 ␮g/L for the two bisthiazolium compounds and 20 ␮g/L for the lipophilic prodrug TE4c. At these concentrations, accuracy and imprecision were 94 – 100% and 18 –20%, respectively.

Fig. 4. Mean (SD; error bars) plasma concentration-vs-time curves after single intraperitoneal administrations of 6.4 mg/kg T3 (A) and 23.7 mg/kg T4 (B) in rats. Each concentration is the mean value computed from six rats.

601


stability Stock solutions of T3, T4, and TE4c did not reveal any appreciable degradation after 1 month of storage at 4 °C. After bench-top storage at room temperature and at 4 °C for 6 h, T3 (whatever the concentration studied) and T4 (163.7 and 982.3 ␮g/L) were stable in plasma, at both 20 and 4 °C, for 6 h. At each time studied, we observed no statistical difference compared with the reference values. At a concentration of 16.4 ␮g/L, the T4 compound was stable for 4 h. Regarding the TE4c prodrug, at both 20 and 4 °C, rapid conversion from prodrug to active drug occurred. After 6 h of storage, at low concentrations (25 and 50 ␮g/L), mean measured concentrations were 0% (20 °C) and 25% (4 °C) of the original concentrations. At the highest concentrations (250 and 1500 ␮g/L), measured concentrations were 56.5% (20 °C) and 70.1% (4 °C) of the original concentrations. When stored at ⫺30 °C for 2 months, T3, T4, and TE4c were stable, with all samples retaining ⬎92% of their original measured concentrations. In extracts originating from both rat and human plasma (i.e., in acetonitrile containing 1 mL/L TFA), after sample pretreatment, T3, T4, and TE4c were stable for at least 24 h at both room temperature and 4 °C, whatever the concentration studied. These extracts cannot be kept stored at ⫺20 °C because of precipitation of the T3 and T4 compounds. Finally, we determined that three freeze-thaw cycles were well tolerated, with no significant losses of the compounds (⬍10%).

percentage binding of t3 and t4 to human plasma proteins Binding studies of the compounds to plasma proteins, studied by ultracentrifugation, indicated that stable equilibrium was reached after 2 h at 37 °C for the two bisthiazolium compounds. The mean (SD) percentages of unbound fraction were 80.3 (7.6)% for T3 and 69.8 (7.3)% for T4.

pharmacokinetic study in sprague– dawley rat Semilogarithmic plots of mean (SD) plasma concentration–time curves after single intraperitoneal administrations of 6.4 mg/kg T3 and 23.7 mg/kg T4 are shown in panels A and B, respectively, of Fig. 4. Data were consistent with a two-compartment model. After intraperitoneal administration, the absorption was rapid; the maximum concentration was reached ⬃10 min after drug administration. The total clearance (5.0 vs 1 L 䡠 h⫺1 䡠 kg⫺1) and the volume of distribution (44.0 vs 16.7 L/kg) were higher for the T4 compound than for the T3 compound. These two parameters were computed taking into account bioavailabilities of 100% and 63% (data not shown), respectively. After T4 administration, the T3 compound was not detected in plasma samples, indicating that T3 is not a metabolite of T4 in rats.

Discussion In this study, we validated a LC/MS method to simultaneously quantify two bisthiazolium compounds, T3 and T4, and the TE4c prodrug in human and rat plasma. The high polarities and ionic natures of T3 and T4 complicated the selective sample clean-up, which is probably the main handicap for their determination in biological samples. The analytical procedure therefore includes solid-phase extraction. Regarding the TE4c prodrug, caution is required during sample handling to avoid rapid conversion to the T4 drug in the plasma. This transformation involves plasma enzyme activity, e.g., esterases or thioesterases (Vial L, Bressolle F, manuscript in preparation). Acidification of plasma and storage at low temperature prevented conversion of the TE4c prodrug into the active T4 drug. The method was used to evaluate the binding of T3 and T4 to human plasma proteins. The mean percentages of unbound fraction were high, ⬃80% for the T3 compound and ⬃70% for the T4 compound, and were independent of the concentration studied. The applicability of this assay was demonstrated in a pharmacokinetic study carried out in rats receiving intraperitoneal administrations of the T3 and T4 compounds. The total clearance and the volume of distribution were five- and threefold higher, respectively, for the T4 compound than for the T3 compound. These results could be explained by the higher lipophilicity of the T4 compound resulting from the presence of a methyl moiety on the hydroxyl ethyl group (Fig. 1 of the online Data Supplement).

We gratefully acknowledge Prof. J.L. Aubagnac, Dr. C. Enjalbal, P. Sanchez, and Prof. J. Martinez, Laboratoire des Aminoacides, Peptides et Proteines, Unite´ Mixte de Recherche 5810, Universities Montpellier I and II, for their analytical assistance. This study was supported by the European Community (QLK2-CT-2000-01166), Ministe`re de l’Education Nationale et Recherche Scientifique (PAL⫹), and the United Nations Development Programme/World Bank/WHO special program for Research and Training in Tropical Diseases.

References 1. Greenwood B, Mutabingwa T. Malaria in 2002. Nature 2002;415: 670 –2. 2. Wongsrichanalai C, Pickard AL, Wernsdorfer WH, Meshnick SR. Epidemiology of drug-resistant malaria. Lancet Infect Dis 2002;2: 209 –18. 3. Ancelin ML, Calas M, Bompart J, Cordina G, Martin D, Ben Bari M, et al. Antimalarial activity of 77 phospholipid polar head analogs: close correlation between inhibition of phospholipid metabolism and in vitro Plasmodium falciparum growth. Blood 1998;91: 1426 –37. 4. Ancelin ML, Calas M, Vidal-Sailhan V, Herbute S, Ringwald P, Vial HJ. Potent inhibitors of Plasmodium phospholipid metabolism with

602

5.

6.

7.

8.

9.

10.

11.

12.


a broad spectrum of in vitro antimalarial activities. Antimicrob Agents Chemother 2003;47:2590 –7. Ancelin ML, Calas M, Bonhoure A, Herbute S, Vial HJ. In vivo antimalarial activities of mono- and bis quaternary ammonium salts interfering with Plasmodium phospholipid metabolism. Antimicrob Agents Chemother 2003;47:2598 – 605. Calas M, Ancelin ML, Cordina G, Portefaix P, Piquet G, VidalSailhan V, et al. Antimalarial activity of compounds interfering with Plasmodium falciparum phospholipid metabolism: comparison between mono- and bisquaternary ammonium salts. J Med Chem 2000;43:505–16. Calas M, Cordina G, Bompart J, Ben Bari M, Jei T, Ancelin ML, et al. Antimalarial activity of molecules interfering with Plasmodium falciparum phospholipid metabolism. Structure-activity relationship analysis. J Med Chem 1997;40:3557– 66. Vial HJ, Ancelin ML, Philippot JR, Thuet MJ. Biosynthesis and dynamics of lipids in plasmodium-infected mature mammalian erythrocytes. Blood Cells 1990;16:531–55. Vial HJ, Calas M. Inhibitors of phospholipid metabolism. In: Rosenthal PJ, ed. Antimalarial chemotherapy: mechanism of action, resistance, and new directions in drug discovery. Totowa, NJ: Humana Press, 2000:347– 65. Wengelnik K, Vidal V, Ancelin ML, Cathiard AM, Morgat JL, Kocken CH, et al. A class of potent antimalarials and their specific accumulation in infected erythrocytes. Science 2002;295: 1311– 4. Vial H, Wein S, Farenc C, Kocken C, Nicolas O, Ancelin ML, et al. Precursors of bisthiazolium are potent antimalarials. Proc Natl Acad Sci U S A, 2004;101:15458 – 63. US Food and Drug Administration. Guidance for industry. Bioana-

13. 14.

15.

16.

17.

18.

19. 20.

21.

lytical method validation. May 2001. http://www.fda.gov/cder/ guidance/index.htm (accessed May 2004). United States Pharmacopoeia XXXIII. Rockville, MD: The United States Pharmacopoeia Convention, 2003:2439. Shah VP, Midha KK, Dighe S, McGilveray IJ, Skelly JP, Yacobi A, et al. Analytical methods validation: bioavailability, bioequivalence and pharmacokinetic studies. J Pharm Sci 1992;81:309 –12. Bressolle F, Bromet-Petit M, Audran M. Validation of liquid chromatographic and gas chromatographic methods. Application to pharmacokinetics. J Chromatogr B 1996;686:3–10. Vial H, Calas M, Bourguignon JJ, Ancelin ML, Vidal V, Rubi E, inventors. Prećurseurs de sels de bis-ammonium quaternaire et leurs applications comme prodrogues ayant une activite´ antiparasitaire. World Patent 99 09471 applied for by CNRS, 21 July 1999; WO 0105742 A1, published January 25, 2001. Matuszewski BK, Constanzer ML, Chavez-Eng CM. Matrix effect in quantitative LC/MS/MS analyses of biological fluids: a method for determination of finasteride in human plasma at picogram per milliliter concentrations. Anal Chem 1998;70:882–9. Barre J, Chamouard JM, Houin G, Tillement, JP. Equilibrium dialysis, ultrafiltration, and ultracentrifugation compared for determining the plasma-protein-binding characteristics of valproic acid. Clin Chem 1985;31:60 – 4. National Institutes of Health. Principles of laboratory animal care, revised. NIH publication 85-23. Bethesda, MD: NIH, 1985. Farenc C, Fabreguette JR, Bressolle F. Pk-fit: a pharmacokinetic/ pharmacodynamic and statistical data analysis software. Comput Biomed Res 2000;33:315–30. Research and Development for Population Pharmacokinetics (RDPP). Pk-fit computer program, Ver. 2.1. Montpellier, France: RDPP, 1999.