1. Introduction
Plasmodium falciparum causes almost all of the >1.5 million world-wide deaths from malaria each year [
1,
2]. Parenteral artesunate is the treatment of choice for severe malaria in many countries as recommended by the WHO [
2,
3] and is effective against chloroquine resistant strains of
P. falciparum malaria. Artesunate {butanedioic acid, [3
R-(3α,5αβ,6β,8aβ, 9α,10α,12β,12a
R*)] -mono(decahydro-3,6,9-trimethyl-3,12-epoxy-12H-pyrano[4,3-j]-1,2-benzodioxepin-10-yl) ester} is a semi-synthetic derivative of artemisinin {(3
R,5a
S,6
R,8a
S,9
R,12
S,12a
R)-octahydro-3,6,9-trimethyl-3,12-epoxy-12
H pyrano[4,3-
j]-1,2-benzodioxepin-10(3
H)-one} which was isolated in 1972 from the plant qinghao (
Artemisia annua). Artemisinin derivatives are nitrogen-free sesquiterpenes with endo-peroxide linkages that are responsible for the antimalarial activity. Artemisinins are thought to cause free radical damage to parasite membranes through binding with malarial proteins by scavenging iron and heme, both essential for parasite survival [
4]. Recently, artemisinin (ARN) was reported to inhibit
P. falciparum PfATP6, a calcium dependent ATPase [
5]. Artesunate (AS) is a safe drug with few side effects and has a global marketplace [
6]. Due to high relapse rates of short-course artesunate monotherapy, patients with uncomplicated malaria are generally treated with AS partnered with a long-acting antimalarial such as mefloquine or piperaquine [
7].
As a hemisuccinate ester of dihydroartemisinin (DHA), AS is rapidly biotransformed to DHA {[3R-(3α,5aβ,6β,8aβ,9α,10α,12β,12a
R*)]-decahydro-10-hydroxy-3,6,9-trimethyl-3,12-epoxy-12H-pyrano[4.3-j]-1,2-benzodioxepin} which is an active metabolite (
Figure 1). The rate of transformation and the bioavailability depend on the route of administration, and multiple effective routes including intravenous, intramuscular, oral, and rectal give the drug broad clinical applications. Previously, the standard bioanalytical method to measure AS and DHA from clinical samples relied on HPLC with reductive electrochemical detection systems; however, these were complicated and required large sample volumes [
8,
9]. Currently, most research groups use liquid chromatography with mass spectrometry detection (LC-MS or LC-MS/MS), a highly sensitive method with much smaller sample requirements. However, the extraction methods for AS and DHA were either liquid-liquid or solid phase extraction (SPE) which are complex and time-consuming [
8,
9,
10,
11,
12] or otherwise expensive, as in the case of 96-well plate-based SPE [
13]. Although rapid, this method requires a 96-well autosampler, not commonly available in most pharmacology laboratories.
Here we report the development of a simple and rapid method for the quantification of both AS and DHA simultaneously from human plasma with ARN as internal standard (IS), based on one-step protein precipitation in acetonitrile (ACN) followed by LC-MS analysis. This method proved to be sensitive and selective with a wide detection range and a low detection limit. We present data on long-term stability of AS and DHA stored in plasma samples, stability of DHA α/β epimers as well as potential interactions with other antimalarials commonly used in the treatment of malaria such as chloroquine and quinine and AS partner drugs with long elimination half-life such as mefloquine.
Figure 1.
Full scan mass spectrum of (a) AS, (b) DHA and (c) ARN internal standard extracted from human plasma. Artesunate and DHA were detected by LC-MS using electrospray ionization in positive ion mode and single ion recording. Both analytes and IS formed ammonium adducts [M+NH4]+ having unique m/z ratios of 402, 302 and 300, respectively; m/z ratios at 267 and 284 are common fragments for both AS and DHA.
Figure 1.
Full scan mass spectrum of (a) AS, (b) DHA and (c) ARN internal standard extracted from human plasma. Artesunate and DHA were detected by LC-MS using electrospray ionization in positive ion mode and single ion recording. Both analytes and IS formed ammonium adducts [M+NH4]+ having unique m/z ratios of 402, 302 and 300, respectively; m/z ratios at 267 and 284 are common fragments for both AS and DHA.
3. Conclusions
We report a novel assay for the accurate and selective quantification of AS and its metabolite DHA in plasma samples based on one-step protein precipitation followed by LC-MS assay. Our method is simple, reproducible, sensitive, and satisfies the practical requirements of conducting antimalarial clinical trials in austere settings. The method has been validated to quantify both compounds over a concentration range (LLOQ–ULOQ) of 3.20–3,000 nM (1.23–1,153 ng/mL) for AS, with a LOD of 1.02 nM (0.39 ng/mL) and 5.33–5,000 nM (1.52–1,422 ng/mL) for DHA, with a LOD of 0.44 nM (0.13 ng/mL). The method has been validated to measure both compounds at the concentrations outside the calibration range up to 200 µM. The approach is suitable for clinical samples from patients treated with IV AS formulation. We have extended the stability data by showing that both AS and DHA in plasma remained stable for up to one year at -80 °C within ±15% of the initial concentration measured at time zero. In addition, the method described here detected β-DHA as a major epimer and showed that it was stable up to 24 h in an autosampler (10 °C). There was no requirement for extra time of more than half a day to allow equilibration of the α/β DHA epimers, as reported by others. It was observed that plasma haemoglobin levels greater than 0.09 g/dL compromised the quantification of AS and DHA. More investigation is needed to address this issue and the use of stable isotopic labeled internal standard when available may be warranted.
4. Experimental
4.1. Reagents
Artesunate or artesunic acid (AS, WR256283, Bottle# BQ38641, MW 384.425), dihydroartemisinin (DHA, WR253997AN, Bottle# ZW60909, MW 284.355) and artemisinin (ARN, WR49309, Bottle# BN97453, MW 282.332) as well as mefloquine (MQ), quinine (QN) and chloroquine (CQ) were obtained from the Experimental Therapeutics Chemical Inventory System of the Walter Reed Army Institute of Research (Silver Spring, MD, USA). Artesunate and DHA had purities of 99.5% and 98%, respectively. Acetonitrile (Optima grade) was purchased from Fisher (Waltham, MA, USA), glacial acetic acid (HPLC grade) from J.T. Baker (Phillipsburg, NJ, USA) and ammonium acetate (99.999%) was obtained from Aldrich (St. Louis, MO, USA). Ultra-pure analytical grade Type 1 water was obtained from Simplicity 185 Water Purification System (Millipore Corporation, Billerica, MA, USA) freshly before use for the preparation of standards and aqueous solutions. Outdated human plasma (Lot# BB-100319) was obtained from a local blood bank using whole blood with citric phosphate dextrose, anticoagulant (Red-cross, Bangkok, Thailand). De-identified patient samples for method validation were kindly obtained from Colonel Peter J. Weina from the Walter Reed Army Institute of Research (Silver Spring, MD, USA).
4.2. Instrumentation/Equipments
The chromatographic system consisted of a reversed phase column (XTerra MS C18, 3.5 µm, 2.1 × 50 mm) and a pre-column of the same material (2.1 × 10 mm) mounted on an Alliance 2695 Liquid Chromatography System (autosampler set at 10 °C and column compartment set at 25 °C) equipped with a single quadrupole Micromass ZQ (MM1) mass spectrometry detector and MassLynx/QuanLynx® v.4.0 software was used for data acquisition and quantification (Waters Corporation, Milford, MA, USA). A nitrogen generator (NM30LA, Peak Scientific, Billerica, MA, USA) was used with the MS system for cone gas flow. A micro-analytical balance (AX-26 Comparator, MettlerToledo, Columbus, OH, USA), and a high-speed centrifuge (Eppendorf 5415D or 5417R, Hamburg, Germany) were used for the preparation of standards and samples. Haemoglobin was measured in whole blood by Sysmex XT2000i (Chuo-ku, Kobe, Japan). A −80 °C freezer (Revco ULT 2586-9-V38, Waltham, MA, USA) with circular temperature recording pen was used for long-term storage of analytical samples. The freezer was physically checked once daily (except weekends) and the recording pad changed weekly.
4.3. Standards
4.3.1. Standard curves
Stock solution of AS, DHA and ARN (used as internal standard) of 5 mM were prepared in 100% ACN, serially diluted in 50% ACN and finally in human plasma or 50% ACN (for neat drug analysis). All the standards were accurately weighed on a micro-analytical balance. Combined AS and DHA working solutions were prepared in 50% ACN so that when spiked in human plasma (1/25 dilution) the resulting final AS/DHA concentrations were 3.20/5.33, 8.00/13.3, 40.0/66.7, 100/167, 250/417, 750/1,250, 1,500/2,500, 3,000/5,000 nM. Stock solutions of other antimalarials were prepared at 10 µg/mL and diluted to 1 µg/mL in 50% ACN solution. All solutions were prepared in polypropylene tubes and micro-centrifuge tubes. The stock solutions were stored in -80 °C freezer.
4.3.2. Quality controls
The working solutions for quality control were prepared from the same stock solution as the standard curve. The working solutions were freshly diluted 25-fold each, using the same sample matrix as that of the standard curve. The final concentrations of AS/DHA in the solutions for analysis were 10.0/16.0, 500/800, and 2,500/4,000 nM. Each quality control solution was prepared in triplicate and analyzed.
4.3.3. Internal standard
The working solution of ARN at 200 µM in 50% ACN was diluted to 15 µM using human plasma. The resulting solution (4 µL) was added to all the samples and vortex before analysis. The total %ACN in plasma sample was less than 2%.
4.4. Sample preparation
All standards, QC and analytical samples in either plasma or 50% ACN solution were handled in the same manner, using Eppendorf micro-centrifuge tubes. Artemisinin (15 µM, 4 µL) was added as IS to all samples (100 µL) after which the drugs were extracted from plasma by protein precipitation using 2 volumes of ice-cold ACN (e.g. 100 µL of sample and 200 µL of ACN), vortex for 1 min, and centrifugation at 10,000 RPM for 10 min. A 100 µL aliquot of the clear supernatants were transferred to HPLC glass vials from which the autosampler made 5 µL injections into the LC-MS system (The remaining solution was stored in the refrigerator for later repeat analysis within 24 h when necessary). For the 50% ACN matrix, the sample solution was diluted 3-fold with two volumes of ACN before injecting 5 µL into the LC-MS system.
4.5. LC-MS conditions
The mobile phase contained 6.25 mM ammonium acetate, pH 4.5 and the buffer-acetonitrile-water was delivered via the LC’s quaternary gradient system. The buffer contribution was fixed at 8% with initial mobile phase composition of 20% ACN and the remaining volume made up with water. The initial composition was held for 1 min to remove polar solvent peaks, changed to 40% ACN in 0.5 min, held for 4.5 min to elute both analytes, changed to 60% ACN in 1 min, held for another 1 min to elute the ARN, changed back to 20% ACN in 1 min and held for the last 2 min to equilibrate the column for following run. The flow rate was at 0.4 mL/min throughout the analysis time of 12 min (including injection delay).
The mass spectrometer was set in the positive electrospray ionization mode and single ion recording. The parameters were optimized for each analyte and IS by infusion of 500 ng/mL in initial mobile phase composition at 10 µL/min. The voltage for capillary, cone, extractor and RF lens were set at 0.60 (kV), 12, 1.00, and 0.4 V respectively. The source and cone temperature were set at 120 and 20 °C. The nitrogen generator was set at 100 psi to generate a cone and desolvation gas flow set at 30 and 300 L/h respectively. For the analyzer, both LM1 and HM1 resolution were set at 13 with the ion energy set at 0.6 and the multiplier set at 650 V.
4.6. Quantification method
For quantification, the peak areas of AS, DHA and ARN were integrated using the MassLynx/QuanLynx chromatography software. The response was calculated from the ratio of the peak areas of the analyte:IS multiplied by the concentration of IS and plotted against concentration in plasma. The responses-concentrations of the standard curve were fitted to the polynomial equation including origin using 2nd order curve type and weighting (Equation 1). Least square regression analysis was performed and the resulting parameters were used to back calculate the concentration of the standard and calculate the concentration of the samples.
4.7. Validation
Full validation [
10,
11] was performed without IS. After modifying the method by using ARN as IS, a partial validation was performed for system suitability, recovery, matrix effect, linearity, LLOQ, ULOQ, intra-assay and inter-assay variability for accuracy and precision, post-preparative stability of AS and DHA, effect of haemoglobin (Hb) on the measurement of AS and DHA, and cross-validation using clinical plasma samples following IV AS (2.4 mg/kg).
4.7.1. System suitability
The system suitability test was done using solutions of 10 and 100 nM to each analyte and 200 nM IS prepared in 50% ACN solution. The solutions were injected in consecutive sequences of ten replicates. The peak area responses were measured and the injection precision (%CV) determined.
4.7.2. Selectivity
Human plasma from six donors were extracted and analyzed for endogenous peak that might interfere with the analytes peak. The detection limit (LOD) was determined at the same time by determining the noise near the region where the analytes eluted using the average noise multiplied by 3. The resulting response was then extrapolated from the standard curve to calculate the concentration at LOD. Interference from other xenobiotics (such as antimalarials: mefloquine, quinine and chloroquine) often co-administered or used in combination with AS were evaluated at 1 µg/mL in 50% ACN. When there was no interference in the neat drug solution, the drugs were spiked in plasma with the analytes at 10 and 100 nM and analyzed in six replicates.
4.7.3. Linearity test
Standard curves were prepared using triplicates of each AS/DHA concentration in plasma covering the range of at 3.20/5.33 to 3,000/5,000 nM. The analysis was done using ARN as IS at 600 nM and performed one curve at a time. The response of each analyte was obtained from the peak area ratio of the analyte:IS and multiplied by the concentration of IS. The standard curve was constructed by plotting the response against its concentration in plasma, linearly fitted, and weighed by 1/concentration. The linear equation parameters of the standard curve were reported as an average of 3 individual curves.
4.7.4. Accuracy and precision
The intra-assay and inter-assay accuracy and precision of the method was evaluated at 5 different AS/DHA concentrations in plasma: LLOQ, 3xLLOQ, medium, 80% of ULOQ and ULOQ. Six replicates for each concentration were measured for the intra-assay test. Triplicate analyses per concentration per day over different days were analyzed for the inter-assay test. The concentrations of AS/DHA were 3.20/5.33, 8.00/13.3, 10.0/16.0, 500/800, 2,500/4,000, 3,000/5,000 nM. The accuracy was determined by the deviation of the mean value of the found concentration from the theoretical value. The coefficients of variation for each tested concentration was calculated to determine the precision. Following the analysis of the highest concentration (ULOQ–3,000/5,000 nM), an injection of 50% ACN was run to evaluate carry-over effect. For concentration over the ULOQ, the sample dilution with the same matrix as the sample was evaluated at 20 and 200 nM in plasma by performing a series of 10-fold dilution until the concentration was in the working range – between LLOQ and ULOQ. The analytes were extracted from plasma samples and analyzed in six replicates.
4.7.5. Recovery and matrix effect
The recovery was determined at three concentrations: 50, 500, and 2,500 nM of AS and DHA spiked in 50% ACN, and compared to plasma samples. The sample process and analyses of analytes were performed in the same manner for both matrixes. The response of the analytes in 50% ACN was considered as 100%. The matrix effect from plasma of eight donors was evaluated by injecting the extracted plasma samples during post-column infusion of the analytes (100 nM) and IS (200 nM) in 50% ACN at 10 µL/min.
4.7.6. Stability
All stability determinations were performed by using a set of samples prepared from a freshly made stock solution of both analytes in analytes-free, interference-free plasma. The stock solution of both analytes was diluted in 50%ACN and spiked in the blank plasma, in triplicates, at 50 and 2,500 nM. The analytes in spiked plasma samples were then extracted and analyzed in triplicates per concentration except where specified. The test and freshly prepared solution were then compared.
Stability of stock solution in 50% ACN was determined in three sets of aliquots stored - at room temperature (25 °C) for 6 h, refrigerated temperature (8 °C) for 1 and 2 weeks, and frozen temperature (−80 °C) for 1 and 2 months. Post-preparative stability was determined by extracting five aliquots of the spiked plasma samples at each concentration, the supernatant was then stored in the autosampler (10 °C) for 0, 6, 12, 16, 20, and 24 h. The post-preparative stability tests were performed for each analyte alone to evaluate hydrolysis of AS to DHA and its epimerization to α-DHA and β-DHA. The test was also evaluated in the combination of AS/DHA.
In plasma, freeze/thaw stability of AS/DHA was evaluated by using three aliquots of the spiked plasma samples at each concentration, stored at −80 °C for 24 h and completely thawed unassisted at room temperature. The analytes in spiked plasma samples were extracted and analyzed after the samples underwent up to three cycles. The short-term room temperature (bench top) stability was tested at 0, 2, 4, and 6 h. Spiked plasma samples were extracted, analyzed and quantified against freshly prepared standard curves in plasma. For long-term storage stability test, the spiked plasma samples in six replicates were stored at -80 °C for 0, 1, 2, 6, and 12 months and followed the same procedure as the short-term stability test.
4.8. Application to clinical plasma samples
Blood samples from severe malaria patients are sometimes haemolyzed. The haemolytic products related to sample collection and/or malaria infection may degrade artemisinin derivatives [
20]. Since the major haemolytic product is Hb, its effect on the measurement of AS and DHA was evaluated.
The Hb level in whole blood was measured before separation of plasma and after replacing the plasma with the exact same volume of water and vortexing for 5 min to lyse the red blood cells (RBC). The Hb level in the lysed RBC (100% RBC) was the same as that of whole blood. The 100% RBC was serially diluted with water to 40, 30, 10, 5, and 1% (v/v) and 14 µL of each %RBC suspension was spiked into 0.7 mL plasma to generate five different levels of Hb in plasma samples as to cover the range of colors seen in clinical samples (+1 to +5). The Hb levels in plasma were calculated from the original level in 100%RBC. At each Hb level in plasma (including control without Hb) AS and DHA were spiked in the plasma at 50 and 2500 nM in triplicates per concentration. The analytes were then extracted, and analyzed by LC-MS.