Standards and reagents
LTG (98% purity) was purchased from HePeng (Shanghai, China) Biotech, Ltd. OXC (Lot, 100657-201102, 99.8% purity) and fluconazole (FCZ, internal standard, IS) (Lot, 100314-201204, 99.8% purity) were acquired from the National Institutes for Food and Drug Control (Beijing, China). MHD (Lot, 20140827, 97% purity) was obtained from Nanjing Chemlin Chemical Industry Company, Ltd. Methanol and acetonitrile were purchased from Fisher Scientific (Waltham, USA). Acetic acid and potassium dihydrogen phosphate were obtained from Sinopharm Chemical Reagent Company, Ltd. Human blank plasma was provided by healthy volunteers and checked to ensure they did not contain LTG, MHD, OXC, or FCZ. Ultrapure water was produced by Millipore ultra-pure water system (Bedford, USA).
Equipment and chromatographic conditions
Experiment was carried out on Dionex UltiMate 3000 (Thermo Fisher, USA) with a PDA3000 UV Detector (Thermo Fisher, USA) equipped with an Acclaim™ C18 column (Thermo, 4.6 × 150 mm, 5 μm particles). Instrument control and data acquisition were performed by the Chromeleon™ software (Thermo Fisher, version 7).
A mixture of potassium dihydrogen phosphate buffer (50 mM) and methanol (61:39) was used for separation at 1.0 mL/min. The column oven was maintained at 37 °C. LTG, OXC, and MHD were monitored at 210 nm.
Preparation of stock and working solutions
Two sets of separately weighted stock solutions (2400 mg/L for LTG, OXC, and MHD) were used to prepare calibration working solutions and quality control (QC) working solutions by dissolving appropriate amount of drugs in methanol-acetonitrile (50:50, v/v). Working solutions for calibration samples were prepared at 12, 24, 60, 150, 300, and 600 mg/L (for LTG, OXC, and MHD). Working solutions for QC samples were prepared at 12, 30, 120, 240, and 480 mg/L (for LTG, OXC, and MHD). IS stock solution was prepared by dissolving FCZ in methanol at 4400 mg/L, and IS working solution was 440 mg/L. All stock and working solutions were stored at − 80 °C and thawed to room temperature before use.
Preparation of calibration and QC samples
Ten microliters of analytes working solution and 10 μL of IS working solution were added to 50 μL of analytes- and IS-free blood plasma, then 130 μL of methanol (1% acetic acid) was added following a 5-min vortex-mixing and a 2-min centrifugation at 12000×g for protein precipitation. Sixty microliters of the supernatant was transferred into a sample vial, and 20 μL was injected for analysis. Thus, calibration samples for curve fitting were prepared at 2.4, 4.8, 12, 30, 60, and 120 mg/L for LTG, MHD, and OXC, and QC samples were prepared at 2.4, 6, 24, 48, and 96 mg/L for LTG, MHD, and OXC. The final concentration was 22 mg/L for IS.
Sample extraction and patient enrollment
This study was approved by the Ethics Committee of Beijing Tiantan Hospital, Capital Medical University, Beijing, Peoples’ Republic of China. Informed consent was obtained from all individual participants included in the study. About 1 to 2 mL of venous blood was obtained from patients with epilepsy on OXC and/or LTG therapy. After 5 min centrifugation at 5000×g, 50 μL of the blood plasma was transferred and mixed with 10 μL of IS, then 140 μL of methanol (1% acetic acid) was added, and then a 5-min vortex-mixing and a 2-min centrifugation at 12,000×g were performed for protein precipitation. Subsequently, 60 μL of the supernatant was transferred into the sample vial, and 20 μL was injected for analysis.
Method validation
Method validation was performed in accordance with the EMEA and FDA guidelines including selectivity, lower limit of quantitation (LLOQ), carry-over, linearity, accuracy, precision, recovery, and stability (FDA, 2013; EMEA, 2011).
Selectivity and LLOQ
The selectivity was assessed by comparing the peak area of analytes- and IS-free plasma from six individuals with those of LLOQ samples. The peak area of interfering peaks in blank plasma should be less than 20% of that of the LLOQ sample for analytes (FDA, 2013; EMEA, 2011). The lowest concentration of the calibration curve (2.4 mg/L for LTG, MHD, and OXC) was defined as the LLOQ. LLOQ is reliable when its bias and precision are within ± 20% and < 20%, respectively.
Linearity and carry-over
Least square method was used for linear regression for all analytes. A weighting factor of 1/x2 was chosen for its lower relative error and higher r2 for linear regression (Almeida et al., 2002). The linear regression equations and correlation coefficients were calculated. Carry-over was evaluated by injecting a blank sample immediately following the highest concentration of the calibration sample. Acceptable carry-over is achieved when the peak area of interfering peaks in blank plasma was less than 20% of those in the LLOQ sample for analytes, and 5% for the IS (FDA, 2013; EMEA, 2011).
Accuracy and precision
The inter-day and intra-day accuracy and precision were estimated by analyzing five replicates of QC samples at 2.4, 6, 24, 48, and 96 mg/L for all analytes on 12 days, and the bias and coefficient of variation (CV%) were calculated. The accuracy was acceptable when bias was within ± 15% (± 20% for LLOQ), and the acceptable precision was obtained when CV was less than 15% (20% for LLOQ).
Recovery
Two batches of QC samples at 2.4, 6, 24, 48, and 96 mg/L for LTG, OXC, and MHD were prepared to evaluate the recovery as follows (FDA, 2013; EMEA, 2011): (A) LTG, OXC, MHD, and IS spiked in blank plasma from six individuals with extraction; (B) LTG, OXC, MHD, and IS in post-extracted blank plasma from the same six individuals. The ratios of (Aanalyte/Banalyte) × 100% were defined as recovery while the ratios of (Aanalyte/Banalyte)/(AIS/BIS) × 100% were defined as IS normalized recovery. The recovery was reliable when IS normalized recovery was consistent at all QC levels.
Stability
Stability was assessed by using four levels of QC samples. The bias between post-stored samples and the QC samples prepared at the same day was used to evaluate the stability of analytes during sample preparation, storage, and analysis: in plasma: 25 °C for 6 h and 9 h, three freeze-thaw cycles from − 80 to 25 °C, and − 80 °C for 15 days; post-extraction: 25 °C for 10 h and 24 h, 4 °C for 24 h, and − 80 °C for 17 days. Analytes were stable when the bias of QC samples was within ± 15%.
The influence of other co-medicated AEDs on the measurement of LTG, OXC, and MHD
The antiepileptic drugs commonly used in China are LTG, OXC, carbamazepine (CBZ), phenobarbital (PB), phenytoin (PHT), levetiracetam, topiramate, ethosuximide, and valproate. The influence of valproic acid, levetiracetam, topiramate, and ethosuximide on the measurement of these analytes was not evaluated because all of them were lack of ultra-violet absorption. To make sure the commonly used AEDs did not influence the determination of LTG, OXC, and MHD, samples were prepared by adding working solutions (concentration 24 mg/L for LTG, OXC, and MHD) into the plasma obtained from patients taking PB, PHT, or CBZ (the plasma concentrations of these co-medicated AEDs were all within the therapeutic window) and analyzed within a runtime of 30 min and 15 min, respectively. Then, two blank samples (methanol only) were injected immediately.
The UHPLC-MS/MS method
The UHPLC-MS/MS method was performed on an Acquity UHPLC H-Class (Waters, MA, USA) tandem 5500 QTRAP mass system (AB SCIEX, CA, USA). The calibration range was 0.06 to 12 mg/L for LTG and 0.03 to 6 mg/L for MHD using LTG-13C3 and MHD-D4 as internal standards. The quantitative ion pair was m/z 256.0 > 144.9 for LTG, m/z 255.1 > 194.1 for MHD, m/z 259.0 > 144.9 for LTG-13C3, and m/z 259.1 > 198.1 for MHD-D4. The separation was achieved by using methanol (A, 0.1% formic acid) and water (B, 0.1% formic acid) at 0.4 mL/min under gradient elution with a run time of 3 min. Calibrators at 0.03, 0.06, 0.3, 0.6, 3, and 6 mg/L for MHD and 0.06, 0.12, 0.6, 1.2, 6, and 12 mg/L for LTG and QC samples at 0.09, 0.225, 2.25, and 4.5 mg/L for MHD and 0.18, 0.45, 4.5, and 9 mg/L for LTG were prepared by spiking 10 μL working solution and 10 μL IS into 100 μL drug- and IS-free blood plasma and vortex-mixed and centrifuged with 300 μL methanol for precipitation. The supernatant was diluted 10 times, and then, 1 μL of the supernatant was injected for analysis. Ten-fold dilution by plasma or water did not affect the measurement of analytes in LC-MS/MS method, by doing that, the method could cover the clinical plasma concentration range for all analytes.
Statistical analysis
Statistical analysis was performed by using SPSS (SPSS Inc., Chicago, IL, USA, version 17) and MedCale (MedCale Software bvba, Ostend, Belgium, version 15.8) software. Statistical significance was defined as a P value less than 0.05.
Application and comparison
A total of 186 samples for LTG (25 samples for MHD) were analyzed by the HPLC-UV and the UHPLC-MS/MS methods. The equation and the correlation coefficient were evaluated by Passing-Bablok regression (a linear regression procedure with no special assumptions regarding the distribution of the samples and the measurement errors) and Pearson test for both two methods (Passing & Bablok, 1983). The mean value of the two measurements was evaluated by paired samples t test or Wilcoxon test. Bland-Altman plot was used to supplement the results of the Passing-Bablok regression and evaluate the agreement of the two methods (Bland & Altman, 1986; Bland & Altman, 1999).