An isotope dilution-liquid chromatography-tandem mass spectrometry (ID-LC-MS/MS)-based candidate reference measurement procedure for the quantification of carbamazepine-10,11-epoxide in human serum and plasma

Chemicals and reagents

LC-MS grade methanol (CAS 67-56-1) and formic acid were (CAS 64-18-6) purchased from Biosolve (Valkenswaard, The Netherlands). Dimethyl sulfoxide (DMSO) (CAS 67-68-5, ACS reagent, ≥99.99 %), ammonium acetate (CAS 631-61-8, LC/MS grade), carbamazepine (CAS 298-46-4, Art. No. PHR1067, Lot No. LRAC1961) and 1,3,5-trimethoxybenzene (CAS 621-23-8, Art. No. 74599, Lot No. BCBW3670) were purchased from Sigma Aldrich (Taufkirchen, Germany). Isopropanol (HPLC grade) was bought from Riedel-de Haën (Seelze, Germany). The standard carbamazepine-10,11-epoxide (CAS 36507-30-9, Art. No. MM0076.01, Lot Nr. W988078) was bought from LGC Mikromol (Teddington, UK) and its ISTD [²H₈]-carbamazepine-10,11-epoxide (CAS 36507-30-9 unlabeled, Art. No. C3529, Lot No. TM-ALS-13-234-P1) was purchased from Alsachim (Illkirch-Graffenstaden, France). The secondary metabolite 10,11-dihydro-10-hydroxycarbamazepine (CAS 29331-92-8, Art. No. 18467, Lot No. 0489614) was purchased from Cayman (Ann Arbor, Michigan US) and 10,11-dihydro-10,11-dihydroxycarbamazepine (CAS 58955-93-4, Art. No. D449040, Lot No. 1-NWW-177-2) was bought from TRC Canada (North York, Canada). 5 mm NMR tubes were obtained from Eurisotop GmbH (Hadfield, United Kingdom). Native human serum (Art. No. S1-LITER) was obtained from Merck (Darmstadt, Germany), TDM-free human serum (multi-individual pooled; surrogate matrix, ID No. 12095432001) from Roche Diagnostics GmbH (Mannheim, Germany). Native plasma matrix (Li-heparin, K2-EDTA and K3 EDTA) was obtained from anonymized, leftover patient samples. Water was purified using a Millipore Milli Q 3 UV system from Merck (Darmstadt, Germany).

General requirements for laboratory equipment

Certified and calibrated equipment was used. The minimum sample weight for the microbalance used (XPR2, Mettler Toledo, Columbus, OH, USA) was determined according to the United States Pharmacopeial Convention (USP) guidelines (USP Chapters 41 and 1251). Positive displacement pipettes were used to measure organic solvents and serum. Volumetric glassware (class A volumetric flasks) that fulfilled the criteria of ISO 1042 and USP were used to prepare stock and spike solutions.

qNMR for determination of the purity of the standard materials

qNMR experiments were performed on a Jeol 600 MHz NMR (Jeol Ltd, Tokyo, Japan) equipped with a helium-cryoprobe head. A single-pulse ¹HNMR was utilized for the quantitation (epoxide ring protons; δ=4.36 ppm; 2H; 1,3,5-trimethoxybenzene as qNMR ISTD; DMSO-d ₆ as solvent) with an inter-scan delay of 70 s. The epoxide resonance was utilized as the quantitative signal because it is the only signal well-separated from the 1H-signals of reactants from which is synthesized, viz., carbamazepine, 10,11-dihydrocarbamazepine, 10,11-dihydro-5H-dibenzo[b,f]azepine-5-carboxamide, iminostilbene, 5-cyano-5H-dibenzo[b,f]azepine and 1-phenyl-1H-indole. We also measured 2D-TOCSY in order to eliminate the presence of unknown impurities. Even the aforementioned hypothetical presence of these molecules would not interfere with our qNMR methodology. Further details about NMR acquisition and processing parameters are available in the Supplementary Material 2 (Table 1, Figures 1 and 2).

Figure 1:

Schematic overview of carbamazepine-10,11-epoxide calibrator and control levels chosen to allow optimal coverage of measurement and therapeutic ranges. Black circles, calibration samples 1–8; black triangles, QC levels 1–4; black diamond, alert level; black line, measurement range; dashed line, therapeutic reference range. Conversion factor µg/mL to µmol/L: 4.0.

Figure 2:

LC-MS/MS readouts of a native human serum sample containing 0.800 µg/mL of each carbamazepine-10,11-epoxide (peak 1), carbamazepine (peak 2), 10,11-dihydro-10,11-dihydroxycarbamazepine (peak 3), 10,11-dihydro-10-hydroxycarbamazepine (peak 4). SRM ion traces of (A) carbamazepine-10,11-epoxide (peak 1); peak 5 shows the in-source dehydration of 10,11-dihydro-10,11-dihydroxycarbamazepine to oxcarbazepine. (B) Carbamazepine (peak 2); peak 6 showing in-source fragmentation of 10,11-dihydro-10-hydroxycarbamazepine to carbamazepine. (C) 10,11-Dihydro-10,11-dihydroxycarbamazepine. (D) 10,11-Dihydro-10-hydroxycarbamazepine.

Figure 3:

Carbamazepine-10,11-epoxide LC-MS/MS derived analytical readouts. (A) Chromatogram of a matrix blank showing the analyte SRM ion trace (left) and the ISTD SRM ion trace (right). (B) Chromatogram of the lowest calibrator level with a concentration of 0.0400 μg/mL spiked in native serum; analyte (left) and ISTD (right). (C) Patient sample with a concentration of 0.678 μg/mL, showing peaks of (1) dehydration of 10,11-dihydro-10,11-dihydroxycarbamazepine to oxcarbazepine (2) carbamazepine-10,11-epoxide and (3) unknown interference; analyte (left) and ISTD (right). ISTD, internal standard.

Preparation of calibrators and quality control samples

Two independent calibrator stock solutions were prepared by weighing 16 mg of carbamazepine-10,11-epoxide in tin boats on a microbalance (XPR2, Mettler Toledo, Columbus, Ohio, USA) and dissolving it in 5 mL DMSO using volumetric flasks. The concentrations of the stock solutions were calculated considering the purity of the reference material determined by qNMR (99.7±0.2 %) and the amount weighed, resulting in stock solutions with a concentration of 3.20 μg/mL. These stock solutions were further diluted with DMSO to prepare working solutions, which were used together with the stock solutions to prepare eight calibrator spike solutions. Final matrix-based calibrators, uniformly distributed from 0.0400 to 12.0 μg/mL (0.159–47.6 μmol/L, see Figure 1), were then prepared by a volumetric 1+99 dilution (v/v) of the calibrator spike solutions in native human serum matrix.

Using a third independent stock solution, four levels of matrix-based quality control (QC) levels were prepared following the same procedure as described for the calibrator levels. The QC concentrations were set at four crucial control points: above the lower limit of quantification, below and within the therapeutic reference range, and at the laboratory alert level (refer to Figure 1). Final concentrations of the QC levels were 0.120, 0.450, 1.50 and 6.00 μg/mL.

Internal standard stock solution

For the preparation of [²H₈]-carbamazepine-10,11-epoxide ISTD stock solution, 1 mg was dissolved by a volumetric addition of DMSO to obtain a concentration of 1,000 μg/mL. The stock solution was stored at −20 °C. [²H₈]-carbamazepine-10,11-epoxide ISTD working solution was freshly prepared each day by mixing 95 µL of DMSO with 5 µL of the ISTD stock solution, followed by the addition of 3,900 µL Milli-Q water to reach a final concentration of 1.25 μg/mL.

Sample preparation

Native human serum, surrogate serum, and plasma (Li-heparin, K2-EDTA, and K3-EDTA) can be used as sample matrix. In a 2 mL tube (Eppendorf Safe-Lock Tubes), 100 µL of ISTD working solution was mixed with 50 µL of sample specimen (native sample/calibrator/QC). 1,000 µL 75 % methanol (v/v) was then added for protein precipitation. After centrifugation, 20 µL of the supernatant was mixed with 980 µL Milli-Q water (1+19 [v/v]).

Liquid chromatography mass spectrometry

The chromatographic separation was performed on an Agilent 1290 Infinity II LC system (Santa Clara, California, USA) consisting of a binary pump, a vacuum degasser, an autosampler set at 7 °C, and a column compartment maintained at 40 °C. A chromatographic gradient with a total run time of 9 min at a flow rate of 0.6 mL/min was developed to separate the target analyte carbamazepine-10,11-epoxide from the parent drug carbamazepine and the secondary metabolites (10,11-dihydro-10-hydroxycarbamazepine and 10,11-dihydro-10,11-dihydroxycarbamazepine) using a Luna Omega Polar C18 column (100x3 mm, 3 μm, Lane Cove, Australia). The mobile phases consisted of 2 mM ammonium acetate in Milli-Q water with 0.1 % formic acid (A) and methanol+2 mM ammonium acetate in Milli-Q-water 95+5 (v/v) with 0.1 % formic acid (B). Contamination of the MS was reduced by using a diverter valve to switch the eluent flow until 0.5 min and from 7.0 min to the waste.

Carbamazepine-10,11-epoxide was detected in multiple reaction monitoring (MRM) mode using an AB Sciex Triple Quad 6500+ or a Q-Trap 6500+ MS (Framingham, Massachusetts, USA) with a Turbo V ion source operating in positive electrospray ionization mode (ESI+ mode). An ion spray voltage of 5,000 V and a temperature of 400 °C were applied. As curtain gas, collision gas, ion gas source 1, and ion gas source 2 nitrogen was used and was set at 45 psi, 11 psi, 50 psi, and 60 psi, respectively. For all mass transitions the collision cell entrance potential was set at 10 V and the collision cell exit potential at 15 V. A second specific mass transition (qualifier) was monitored to eliminate the possibility of interference from other substances present in native matrix samples. This was achieved by comparing the quantifier/qualifier ratios of neat System Suitability Test (SST) samples and native matrix samples. The difference between the ratios should not exceed 20 %. The SRM transitions and the remaining MS settings are summarized in Table 1.

System suitability test

System sensitivity, chromatographic performance, and potential carry-over were evaluated by performing an SST prior to each analysis. The concentration levels of SST sample 1 and 2 were equivalent to the corresponding analyte concentration of processed calibrator level 1 and 8, respectively. To pass the SST, the signal-to-noise ratio of the quantifier transition for SST sample 1 had to be ≥10. The retention time of both SST samples had to be within 3.6±0.5 min. Moreover, the carry-over effect was evaluated by injecting SST sample 2 followed by injecting two solvent blanks. The analyte peak area observed in the blank after the injection of SST sample 2 had to be ≤20 % of the analyte peak area of the calibrator 1 level.

Calibration, structure of analytical series and data processing

The final calibration function was generated by measuring the calibration levels in increasing concentration at the beginning and the end of the analytical series. A quadratic regression of the area ratios of the analyte and ISTD (y) against the analyte concentration (x) was used to obtain the function y=a₀+a₁x+a₂x² using both calibrations. Data processing was performed using Analyst^® software (version 1.6.3 or higher) with the IntelliQuant algorithm. Carbamazepine-10,11-epoxide and [²H₈]-carbamazepine-10,11-epoxide eluted at a retention time of 3.6 min and were integrated within a 30 s window. Peak integration included a smoothing factor of 3 and a peak splitting factor of 2. The noise percent was set to 90 % with a base sub window of 0.5 min.

Method validation

Assay validation and determination of measurement uncertainty were performed based on existing validation guidelines such as Clinical & Laboratory Standards Institute’s C62A Liquid Chromatography-Mass Spectrometry Methods [31], the International Council of Harmonization’s (ICH) guidance document Harmonised Tripartite Guideline Validation of Analytical Procedures: Text and Methodology Q2 (R1) [32], the Guide to the expression of uncertainty in measurement (GUM) [33]. A more detailed description on the estimation of measurement uncertainty was published previously by Taibon et al. [34].

Traceability to SI units

Traceability to the SI unit of mass (kilogram), the most important parameter for a reference measurement method, has been established by the utilization of 1,3,5-trimethoxybenzene as a qNMR ISTD which is directly traceable to NIST PS1 (primary qNMR standard). Furthermore, traceability to the SI unit of amount of substance (mole) is also included owing to NIST PS1. Since Planck’s constant is now the main parameter for defining kilogram and Avogadro’s constant for mole, qNMR methodology is best suited to fulfill both the traceability chains. Additionally, as per the latest IUPAC Technical Report, qNMR has been stated as a potential primary RMP ideally suited for the characterization of primary reference materials [17]. Six individual experiments (Supplementary Figure 2; Table 1) involving six individual weightings, yield a final absolute content value of 99.7±0.2 % (k=1). The extended uncertainty (k=2) can be obtained by multiplying the above-mentioned uncertainty by a factor of 2.

Selectivity

Baseline separation of carbamazepine-10,11-epoxide, carbamazepine, 10,11-dihydro-10-hydroxycarbamazepine, and 10,11-dihydro-10,11-dihydroxycarbamazepine was achieved on a reversed phase column (Phenomenex Luna Omega Polar C18) showing resolutions greater than 2.0 in all tested matrices (Figure 2). The chromatogram of different matrices (pools of analyte-free native human serum, surrogate serum, and human Li-heparin plasma) showed no signals at the expected retention time.

In some patient samples, a signal was observed during the method comparison study at a retention time of −0.4 min relative to carbamazepine-10,11-epoxide as well as at a retention time of +0.3 min relative to carbamazepine-10,11-epoxide (Figure 4). The signal at +0.3 min occurred only in two of 110 serum samples, whereas the other signal was observed in all native serum and plasma samples. The signal eluting prior to the target analyte may be explained by the in-source dehydration of the metabolite 10,11-dihydro-10,11-dihydroxycarbamazepine to oxcarbazepine, a congener isobaric to carbamazepine-10,11-epoxide (schematically shown in Figure 3). The chemical nature of the isobaric component eluting later than the target analyte was not identifiable. Since all signals are baseline separated the quantification of carbamazepine-10,11-epoxide is not affected (Figure 4). The labelled ISTD is considered suitable for the intended use as it did not contain any unlabeled analyte.

Figure 4:

Schematic overview of the in-source dehydration of 10,11-dihydro-10,11-dihydroxycarbamazepine to oxcarbazepine.

Matrix effect and specificity

The sample preparation protocol was based on protein precipitation followed by a high dilution step to minimize possible matrix effects. Matrix independency was demonstrated by the post column infusion experiment, as no change in the ionization field was observed at the expected retention times in native human serum, surrogate serum, and native plasma matrix (Li-heparin, K2-EDTA and K3-EDTA plasma).

Moreover, matrix effects were investigated by comparing slopes and coefficients of determination of calibration curves in different matrices. The slopes of carbamazepine-10,11-epoxide (n=6, sample preparations) were 0.867 (95 % CI 0.854 to 0.879) for neat solution, 0.876 (95 % CI 0.862 to 0.890) for native serum matrix, 0.874 (95 % CI 0.858 to 0.890) for surrogate serum matrix, and 0.878 (95 % CI 0.863 to 0.892) for Li-heparin plasma matrix. Since the 95 % CIs overlap, meaning that they are not significantly different, a matrix effect can be excluded. The coefficients of determination were 1.00 regardless of the calibration matrix. The relative bias for all matrices tested ranged from −0.8 to 2.3 % with CVs of ≤3.2 % compared to neat solution, except for calibrator 1 (0.0400 μg/mL) in surrogate serum, which showed a CV of 5.7 %.

Comparing analyte peak areas, ISTD peak areas as well as area ratios of matrix samples with neat samples no matrix effects were observed. Carbamazepine-10,11-epoxide and [²H₈]-carbamazepine-10,11-epoxide peak areas showed mean relative recoveries to neat samples from 94 to 105 % and from 96 to 104 %, respectively. The corresponding comparison of mean area ratios ranged from 97 to 102 % (Table 2).

Linearity

The linearity was evaluated by analyzing six native serum calibration curves with an extended measuring range of 20 % at both lower and higher concentrations, resulting in a linearity over the range from 0.0320 to 14.4 μg/mL. The residuals were not randomly distributed in a linear regression model, therefore a quadratic model was applied. The coefficients of determination were 1.00 for all individual calibration curves.

Based on the selected regression model, the dilution series was evaluated and showed a linear dependence with a coefficient of determination of 1.00. The relative deviation (n=6, sample preparations) ranged from −0.2 to 2.4 % with CVs of ≤2.5 %.

Lower limit of the measuring interval and limit of detection

The LLMI was determined using samples spiked with the concentration of the lowest calibrator (0.0400 μg/mL). The relative bias showed a deviation of 2.1 % and CV of 4.0 %. The LOD was estimated to be 0.0111 μg/mL. The LOD’s intensity is approximately one-fourth of the average peak height of calibrator 1.

Accuracy and precision

To estimate precision, accuracy, and trueness, spiked native human serum and Li-heparin plasma samples were used at four concentration levels (0.120, 0.450, 1.50, 6.00 μg/mL). Moreover, two high-concentrated native human serum samples (14.0 and 20.0 μg/mL) were diluted with native human serum prior to the sample preparation to define the accuracy of high-concentration samples (dilution 1 and 2).

Intermediate precision and repeatability were determined using an ANOVA-based variance component analysis including variances as between-day, -calibration, -preparation, and -injection and resulted in CVs ≤2.1 and ≤1.8 %, respectively, regardless of the concentration levels and matrix. The measurement variability of patient samples was found to be indistinguishable from spiked materials (Table 3).

Trueness was evaluated by comparing the mean measured concentrations to the nominal sample concentrations. The mean bias (n=6, sample preparations, two operators) of native human serum samples ranged from 1.4 to 2.5 %, while Li-heparin plasma samples showed a mean bias between 1.4 and 3.5 %. Diluted samples (dilution 1 and 2) showed comparable results ranging from 3.2 to 3.4 % (Table 4).

Stability

The stability of processed samples at 7 °C was verified for 6 days. The mean recovery was 99.4 %. The stability of spiked calibrator and QC samples stored at −20 °C was evaluated for 35 days with a mean recovery of 103.2 %.

Equivalence of results between two independent laboratories

Equivalence between two independent laboratories was demonstrated performing a method comparison study with 140 samples (native serum and Li-heparin plasma patient samples, patient pools, and spiked samples). From these samples, five samples were below the calibration range and two samples could not be measured due to insufficient sample volume and were therefore excluded from the analysis (total n=133 samples). Passing–Bablok regression analysis showed a very good agreement between the two laboratories, resulting in a regression equation with a slope of 1.01 (95 % CI 1.00 to 1.02) and an intercept of −0.002 (95 % CI –0.003 to 0.009) (Figure 5A). The Pearson correlation coefficient was found to be ≥0.999. The plot of the relative Bland–Altman analysis confirmed the good agreement between the two laboratories and showed that the data scatter was independent of analyte concentration with a 2S agreement of 7.1 % (lower limit CI interval of −7.7 to −5.6 %, upper limit CI interval of 6.5–8.7 %). The resulting bias was found to be 0.5 % with a 95 % CI interval from ranging from −0.2 to 1.1 % (Figure 5B).

Figure 5:

Results from the carbamazepine-10,11-epoxide method comparison study performed between two independent laboratories. (A) Passing–Bablok regression plot including the Pearson regression analysis for the method comparison study of the RMP (n=133 samples) between the independent laboratories (Site 1: Risch; Site 2: Roche). Passing–Bablok regression analysis resulted in a regression equation with a slope of 1.01 (95 % CI 1.00 to 1.02) and an intercept of −0.002 (95 % CI –0.003 to 0.009). The Pearson correlation value was ≥0.999. (B) Bland–Altman plot for the method comparison study of the RMP (n=133 samples) between two independent laboratories (laboratory 1: Risch site, and laboratory 2: Roche site). The inter-laboratory measurement bias was 0.5 % (95 % CI interval from −0.2 to 1.1 %) and the 2S interval of the relative difference was 7.1 % (lower limit CI interval from −7.6 to −5.6 %, upper limit CI interval from 6.5 to 8.7 %).

Precision evaluation at Site 2 showed a mean intermediate precision CV ≤3.4 % and a repeatability CV ≤3.2 % comparable to Site 1. Due to a laboratory error within day three, level 4 was evaluated with n=30 measurements, while all other samples were evaluated with n=36 measurements. The data scatter was in good agreement with the error propagation between the two independent measurements and the method is transferable regardless of the laboratory equipment or personnel.

Uncertainty of results

Total measurement uncertainties and expanded measurement uncertainties for single measurement ranged from 1.6 to 2.2 % and from 3.2 and 4.5 %, respectively (Table 5). The derived total measurement uncertainty is multiplied by a coverage factor of k=2 to obtain an expanded uncertainty that represents a confidence level of 95 %, assuming normal distribution.