Simultaneous determination of topiramate, carbamazepine, oxcarbazepine and its major metabolite in human plasma by SFC-ESI-MS/MS with polarity switching: Application to therapeutic drug monitoring

2.1 Chemicals and reagents

Standard samples of TPM, CBZ, OXC, MHD and diazepam (IS) were obtained from the National Institutes for Food and Drug Control (Beijing, China). HPLC-grade methanol and ethyl acetate were purchased from Fisher Scientific (Pittsburgh, USA) and Yuwang Chemical Co., Ltd. (Jinan, China), respectively. Both CO₂ (⩾99.99%) and N₂ (99.999% purity) were supplied by Shenyang Qianzhen Chemical Gas Co., Ltd. (Shenyang, Liaoning, China). Other reagents were of analytical grade and obtained commercially. Freshly blank human plasmas were purchased from the Blood Transfusion Center (Shenyang, China) and stored at −20 °C until required.

2.2 Sample collection

Twenty blood samples were taken from March to April in 2015 from the General Hospital of Nanjing Command PLA. The blood specimens were taken for routine diagnostic purposes. The collection time was half an hour before the next drug administration, which was the minimum concentration time. Blood samples, about 2 mL were collected and then transferred into heparinized vacuum tubes and centrifuged to obtain plasma at 13,000 rpm for 10 min at 37 °C. A batch of samples was measured daily at Fuzhou General Hospital of Nanjing Command PLA with the EMIT method (Luo et al., 2011), while the others were immediately stored at −20 °C until analyzed by the SFC-ESI-MS/MS method within one month. This study was approved by the Ethics Committee of Nanjing Command PLA. All patients gave their informed consents.

2.3 Instrument, chromatographic and MS conditions

The ACQUITY UPC² system with binary solvent manager, sample manager, column manager and convergence chromatography manager was supplied by Waters Corp., Milford, MA, USA. The separation was performed by injecting a 5 μL sample onto a UPC^2TM BEH, 2-EP column (100 mm × 3 mm, 1.7 μm) in less than 2.5 min. The mobile phase consisted of carbon dioxide and methanol at a flow rate of 1.0 mL/min in a gradient elution mode, while pure methanol was used as the compensation solvent at a flow rate of 0.2 mL/min. The gradient program was as follows: 0–1.6 min, 12% MeOH in CO₂; 1.6–1.75 min, 12–7%; 1.75–2.3 min, 7%; 2.3–2.5 min, 7–12%. The auto sampler was maintained at 4 °C while the column temperature was 50 °C. The backpressure of the system was maintained at 13.79 MPa. TPM, CBZ, OXC, MHD and IS were successfully separated under these conditions with retention times of 1.25, 1.54, 1.50, 1.97 and 1.04 min, respectively. The UPC² system connected with the mass detector by a shunt of fixed proportion which was supplied by Waters Corporation. The monitoring was carried out using a triple-quadrupole tandem mass spectrometric detector (Waters Corp., Milford, MA, USA). Detection was achieved by an electrospray ionization (ESI) interface in multiple reaction monitoring (MRM) mode with constant polarity switching. The full scan product ion spectra of analytes and IS are shown in Fig. 1. The optimal MS conditions were as follows: the source temperature and desolvation temperature were maintained at 150 °C, 400 °C; the cone and nitrogen flow rates were set at 10 L/h and 800 L/h, respectively; and the capillary voltage was 2.5 kV. The optimized MRM fragmentation transitions and MS parameters for each analyte are listed in Table 1. All the collected data were processed using MassLynx™ NT 4.1 software with a QuanLynx™ program (Waters Corp., Milford, MA, USA).

Close

Figure 1

The fullscan product ion spectra of (A) CBZ, (B) OXC, (C) MHD, (D) TPM and (E) IS.

Export to PPT

2.4 Preparation of standards, internal standard and quality control samples

Independently, an appropriate amount of analyte in methanol was used to obtain stock standard solutions, with a concentration of 4.0 mg/mL for TPM, CBZ, OXC and MHD, which were serially diluted using methanol to obtain the working standard solutions at several concentrations. In addition, stock standard solutions of IS (1.0 mg/mL) were dissolved in methanol and further diluted to yield 1.0 μg/mL solutions. Furthermore, seven combined spiked solutions were prepared daily by mixing the working standard solutions of the analytes with final concentrations of 0.8, 1.6, 4.0, 20, 40, 200, 400 μg/mL for TPM, 0.1, 0.2, 1.0, 5.0, 15, 75, 150 μg/mL for CBZ, 0.1, 0.2, 0.8, 4.0, 8.0, 40, 80 μg/mL for OXC and 5.0, 10, 25, 50, 100, 250, 500 μg/mL for MHD. Calibration standards were prepared by the addition of 10 μL of appropriate combined spiked solutions to 100 μL blank plasma, giving final designated plasma concentration ranges of 0.08–40 μg/mL for TPM, 0.01–15 μg/mL for CBZ, 0.01–8.0 μg/mL for OXC and 0.5–50 μg/mL for MHD. The QC samples were also independently prepared by spiking 100 μL blank plasma with 10 μL of corresponding combined solutions, resulting in final plasma concentrations of 0.16, 2.0 and 32.0 μg/mL for TPM, 0.02, 0.5 and 12 μg/mL for CBZ 0.02, 0.4 and 6.4 μg/mL for OXC, 1.0, 5.0 and 40 μg/mL for MHD corresponding to the three different concentration levels. All the stock solutions were stored at −20 °C until required for further dilution.

2.5 Sample preparation

Plasma sample preparation was performed by simple LLE without nitrogen evaporation to dryness and reconstitution. Aliquots of 100 μL blank plasma sample were transferred to 5 mL Eppendorf tubes, then 10 μL methanol, 10 μL IS (1 μg/mL) and 150 μL water were added and vortex-mixed for 1 min prior to the addition of the extraction solvent. This was followed by the addition of 1000 μL ethyl acetate and vortexed for another 3 min. After centrifugation at 13,000 rpm for 5 min, the upper organic layer was directly transferred to a vial and 5 μL was injected into the SFC-MS/MS system.

2.6 Method validation

Validation was performed with respect to the selectivity, linearity and sensitivity, precision and accuracy, matrix effect, extraction recovery and stability.

The selectivity was evaluated by comparing the chromatograms of blank plasma with simulated plasma samples at LLOQ level. The lower limit of quantification (LLOQ) samples were prepared and validated with a relative error (RE) within ±20% and RSD lower than 20%. Calibration curves were prepared as described above and processed by the internal calibration curve method and weighted by 1/x. The accuracy and precision were determined based on the analysis of six replicates of QC samples within the same day (n = 6; intra-day assay) and over eighteen replicates of QC samples on different days (n = 18; inter-day assay). The precision was presented as the relative standard deviation (RSD) while the accuracy was validated with the relative error (RE). Recovery was assessed by comparing the mean peak areas of AEDs in plasma samples spiked before LLE to the peak areas in samples spiked after LLE. The matrix effect was determined by comparing the mean peak areas of AEDs in plasma samples spiked after LLE to the peak areas of analytes added directly to methanol. In order to evaluate the stability of the analytes in plasma samples, stability tests were conducted using QC samples of three different concentration levels under different conditions: short-term (kept at room temperature for 8 h), long-term (stored at −20 °C for 30 days), freeze–thaw (three cycles), and post-preparative (kept in an auto-sampler maintained at 4 °C for about 12 h), respectively.

2.7 Practical application and method comparison

In order to demonstrate that the developed SFC-ESI-MS/MS was applicable for TDM studies, both the SFC-ESI-MS/MS and EMIT methods were compared by evaluating clinical samples (Luo et al., 2011). In addition, Bland–Altman plots were used to assess the bias between these two methods. Comparison studies of clinical specimens were not carried out for OXC and TPM because they are difficult to obtain and are not routinely determined at this hospital.

3.1 Optimization of chromatographic and mass spectrometry conditions

Several factors affecting both the chromatographic performance and ionization efficiency were explored, such as the compositions of the mobile phase, kinds of modifier, flow-rate, column temperature, and backpressure.

In the current study, several proportions (95:5, 90:10, 85:15, v/v) of CO₂/methanol under isocratic conditions and several gradient elution modes were investigated. With the gradient elution mode described above, sharper peaks, better chromatographic resolutions and suitable retention times for all the analytes were obtained compared with the other elution modes. For the composition of mobile phase, both methanol and acetonitrile were investigated, and the results indicated that the addition of methanol could obtain shaper peaks, better elution and larger responses of targets than applying acetonitrile. Due to the fact that supercritical carbon dioxide possessed better diffusivity compared with organic solvents, the flow rate of the mobile phase with a good chance of either increasing or decreasing did not change the column efficiency (Li et al., 2015). Two kinds of columns, ACQUITY UPC^2TM HSS C₁₈ SB column (100 × 3 mm, 1.8 μm) and UPC^2TM BEH, 2-EP column (100 × 3 mm, 1.7 μm), were investigated to optimize the chromatographic peak shape and separation of analytes. The UPC^2TM BEH, 2-EP column was shown to be more suitable offering better chromatographic performances. Thus, due to the narrower peak, shorter analysis time and higher response achieved for all the analytes, 50 °C was proved to be the optimal column temperature out of 40, 45, 50 and 55 °C. The compensation solvent could improve the mass response of the analytes which was in accordance with a previous study (Liu et al., 2014). Methanol and the different concentrations of formic acid (0.1%, 0.2%) in methanol were investigated, the results indicated that methanol was the best choice due to the negative ionization switching mode of TPM, and the addition of formic acid suppressed the ionization of TPM leading to a lower response.

The effect of back-pressure (12.41, 13.10, 13.79, 15.17 MPa) was also studied. However, the back-pressure had little significant influence on the chromatographic performance, so a back-pressure of 13.79 MPa was set for the best instrumental performance, which was adjusted by the Waters engineers.

In order to validate a sensitive SFC-MS/MS technology, an electrospray ionization (ESI) interface that operated in polarity switching mode under MRM conditions was examined. Parameters, such as the capillary voltage, collision energy and cone voltage, were also adjusted, and a stronger and more stable response of ion transitions was obtained. TPM was determined by using the negative ion mode, while the other AEDs were quantified in positive ion mode.

3.2 Sample preparation

LLE is a very efficient method because of the lower polarity of the AEDs. To optimize the LLE procedure so as to obtain higher recoveries, cleaner extracts and negligible matrix effects, some factors affecting the extraction recovery, such as the type and volume of the extracting solvent, the centrifugation speed and time, and vortexed time were taken into account. The immiscible extraction solvents, ethyl acetate, dichloromethane and methyl tert-butyl ether, were investigated. The recoveries for all analytes were within the acceptable range for both ethyl acetate and methyl tert-butyl ether. Nevertheless, methyl tert-butyl ether is much more toxic and pungent than ethyl acetate, so ethyl acetate was selected as the optimum solvent. In addition, the volume of extraction solvent (in the range of 800–1200 μL) was studied and 1000 μL was determined to be the optimum volume in terms of recovery. Because the upper organic layer was directly injected into the SFC-MS/MS system, the drug concentration in the upper organic layer is relatively low. So, we selected 100 μL as the volume of human plasma so as to maintain a higher sensitivity. Each tube, after addition of the extraction solvent, was thoroughly vortexed and centrifuged. The effect of the centrifugation speed and time, as well as the vortexed time, was investigated in terms of the recovery. According to the obtained results, 3 min was selected as the vortexed time with the best efficiency. The obtained results indicated that these parameters were distinctly ineffective and so, 13,000 rpm and 5 min were selected as the optimized centrifuge rate and time, respectively. In addition, the effect of adding water to plasma was studied, and a high dilution of plasma with 150 μL water was found to eliminate any matrix effect as far as possible.

3.3 Method validation

Regarding selectivity, the representative chromatograms obtained from the analysis of blank plasma and the corresponding spiked plasma at the LLOQ level are shown in Fig. 2. As can be seen from the chromatograms, none of the matrix components interfered with the analytes or the IS at their corresponding retention times, which indicates that the assay has acceptable selectivity.

Close

Figure 2

Representative MRM chromatograms of CBZ, OXC, MHD, TPM and IS in human plasma: (A) a blank plasma; (B) a blank sample spiked with the analytes and IS (at LLOQ level).

Export to PPT

Calibration curves were linear over the plasma concentration ranges of 0.08–40 μg/mL for TPM, 0.01–15 μg/mL for CBZ, 0.01–8 μg/mL for OXC and 0.5–50 μg/mL for MHD. The respective linear regression equations of TPM, CBZ, OXC and MHD were as follows: y = 0.069x + 0.008 (r² = 0.9967), y = 32.43x + 0.2580 (r² = 0.999), y = 5.214x + 0.041 (r² = 0.9912) and y = 0.049x + −0.021 (r² = 0.9947), where y represents the peak area ratio of the analyte to IS and x is the concentration of the analyte. The regression coefficients (r²) of all calibration curves were greater than 0.99. The lowest limit of quantification was found to be 0.08, 0.01, 0.01 and 0.5 μg/mL for TPM, CBZ, OXC and MHD, respectively, which was markedly lower than the target plasma concentrations reported, indicating that this method was sensitive enough for all analytes.

The values for the intra- and inter-day precision and accuracy are presented in Table 2. All the results were within the acceptable ranges. Further investigations demonstrated that the novel method was accurate, reliable and reproducible.

The obtained results for the matrix effect and extraction recovery of the three levels of QC samples for all the analytes are listed in Table 3. The extraction recovery and matrix effect of the IS were 81.0 ± 4.9%, and 90.9 ± 1.2%, respectively. All the obtained values of the matrix effect were between 85% and 115%, which indicated this method effectively removed interfering substances from biological matrices.

The stability of all analytes under different conditions is shown in Table 4. Under all the tested conditions, the results were less than ±15% deviation from the nominal concentration, which allowed us to conclude that all analytes were proved to be stable and the method was reliable.

3.4 Practical application

Data distribution obtained by both SFC-ESI-MS/MS and EMIT methods is shown in the Bland-Altman graph, where the average difference was between 0.97 and −3.58, and the standard deviation was −1.3 (Fig. 3). The CBZ blood concentrations determined by EMIT were higher than those obtained by SFC-ESI-MS/MS. Representative chromatograms of two different authentic samples are shown in Fig. 4. We have demonstrated that the SFC-ESI-MS/MS approach was applicable for the determination of authentic samples with good specificity, precision and accuracy.

Close

Figure 3

Bland-Altman’s graph of CBZ blood concentration determined by EMIT and SFC-ESI-MS/MS method.

Export to PPT

Close

Figure 4

Representative chromatograms of real samples from two different patients with carbamazepine: (A) 8.49 μg/mL; (B) 4.25 μg/mL.

Export to PPT

EMIT and SFC-ESI-MS/MS methods were used to determine CBZ blood concentrations. The Bland-Altman graph of the CBZ blood concentrations exhibited a negative bias, which is consistent with the literature references (Liu et al., 1993; Shibat et al., 2012). The greater degree of EMIT was due to cross-reaction of the EMIT reagent with other interfering substances such as structural analogs, its major metabolite and the matrix itself (Burianova and Borecka, 2015).

3.5 Limitations

The developed SFC-ESI-MS/MS approach with many advantages will be significant for monitoring the above AEDs. However, this method has the following shortcomings: Firstly, the accuracy of quantitation may be compromised by the lack of dedicated isotopically labeled internal standards. Therefore, the test is not appropriate for patients who have taken or been administered diazepam, within 7 days prior to the sample collection or during the study. In addition, this study has not been evaluated with authentic samples for the drugs included in the panel, other than CBZ.