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

Pharmacological background

Primidone (C₁₂H₁₄N₂O₂, molecular weight=218.25 Da, conversion factor from µg/mL to molar unti [µmol/L]=4.6) is a synthetic analogue of the antiepileptic drug phenobarbital, first synthesized by Bogue and Carrington in 1949 and associated with fewer sedative and cognitive side effects than phenobarbital [1, 2]. It is metabolized to phenobarbital and phenylethylmalonamide, however the anticonvulsant activity and mechanism are not yet fully understood [3]. Unlike phenobarbital, primidone does not interact directly with GABA-A receptors or ion channels, but it reduces high-frequency repetitive firing of neurons and alternates the movement of transmembrane sodium and calcium ions [4]. Moreover, the minor metabolite phenylethylmalonamide has been shown to potentiate the anticonvulsant activity of the major metabolite phenobarbital against tonic-clonic seizures and to decrease the myoclonic jerk threshold and may play an important role in the anticonvulsant activity of primidone [5]. Nevertheless, primidone has been shown to be effective in the treatment of essential tremor and is therefore used primarily for this neurological condition (level A recommendation) rather than for epilepsy [3, 6, 7]. However, it is still used in certain circumstances as it is the only anticonvulsant drug known to shorten QT-interval and correct cardiac arrhythmias [8, 9].

Common adverse effects are confusion, ataxia, and nausea and have been reported in 22–77 % of the patients. These side effects occur even at very low doses [10]. The established therapeutic range of primidone is 5.00–10.0 μg/mL [11]. Primidone is primarily metabolized in the liver mainly to phenobarbital and has a half-life of 7–22 h in adults [12]. The major metabolite, phenobarbital, is metabolized rather slowly and therefore accumulates to high concentrations that may be in the toxic range (>50 μg/mL) [5, 11]. Hence, monitoring of both primidone and phenobarbital is essential.

Analytical background

There are several in vitro diagnostic Conformitè Europeenne (CE) certified analytical solutions for therapeutic drug monitoring (TDM) of primidone and phenobarbital by LC-MS/MS. In addition, there are various LC-MS/MS methods published in literature that are used in research and development [13], [14], [15]. However, there is currently no reference measurement procedure (RMP) listed in the JCTLM database [16].

An RMP is understood to be free of systematic error, must be traceable to a higher order reference material and in accordance with ISO guidelines (ISO 17511) [17]. The Bureau International des Poids et Mesures (BIPM) and the National Metrology Institutes (NMI) provide metrological control to ensure the accuracy and reliability of RMPs and to provide traceable calibrator and control materials. If materials and/or methods are not provided by BIPM and NMIs, individual traceability approaches must be followed [18]. These systems are essential to improve inter-laboratory measurement accuracy and to reduce inter-laboratory and inter-test variability within routine laboratories.

To establish traceability to SI-units, quantitative nuclear magnetic resonance (qNMR) spectroscopy is used since it is a well-established primary ratio method for determining the mass fraction (absolute content; g/g) of an analyte in a single, non-destructive experiment. This technique offers the potential for complete molecular structure elucidation owing to a plethora of robust 1D and 2D NMR pulse sequences along with a linear response to the amount of analyte and direct traceability to the kilogram via qNMR internal standards [19, 20]. These features provide an unparalleled ability to determine the amount or “counts” of a specific analyte. The highest order qNMR internal standards are directly traceable to NIST benzoic acid 350b (coulometric) and/or NIST PS1 (benzoic acid; first primary qNMR standard) [21]. Furthermore, according to the latest IUPAC Technical Report, qNMR has been identified as a potential primary reference measurement technique that is well suited for the characterization of primary reference materials (PRMs) [22].

The measurement uncertainty of these systems should be no more than one third of the measurement uncertainty of routine methods and/or the biological variability [23]. As described in previous published RMPs [24], [25], [26], [27], the target measurement uncertainty (k=1) and the expanded measurement uncertainty (k=2) of the candidate RMP of primidone was determined to be less than 1.5 and 3.0 %, respectively [28].

Herein, we present a novel candidate RMP for primidone that meets the requirements of the ISO 15193 guideline [29]. This manuscript does only cover primidone, an RMP for phenobarbital will be covered in a separate report. The technical implementation of the test procedure, the qNMR-based reference material characterization, and the calculation of the measurement uncertainty are described in three supplemental materials. These supplementary materials provide sufficient detail to enable the RMP to be reproduced in other laboratories.

Chemicals and reagents

LC-MS grade methanol (CAS 67-56-1) and formic acid (CAS 64-18-6) were purchased from Biosolve (Valkenswaard, The Netherlands). Isopropanol was purchased from Riedel-de Haën (Seelze, Germany). Dimethyl sulfoxide (DMSO) (CAS 67-68-5, ACS reagent, ≥99.99 %), ammonium acetate (CAS 631-61-8, LC/MS grade), DMSO-d6 (CAS 2206-27-1), the analyte primidone (CAS 125-3-7, Art. No. P7295, Lot MKCM2420), and the qNMR internal standard 1,3,5-trimethoxybenzene (TraceCert^®, Lot Nr. BCBW3670, CAS 621-23-8) were bought from Sigma Aldrich (Taufkirchen, Germany). The deuterated internal standard (ISTD), [²H₅]-primidone (CAS 73738-06-4, Art. No. C629, Lot JF-ALS-19-080-P3) was purchased from Alsachim (Illkirch Graffenstaden, France). Water was produced in-house using a Millipore Milli-Q 3 UV system from Merck (Darmstadt, Germany). Native human serum (Art. No. S1-LITER) was obtained from Merck (Darmstadt, Germany), TDM-free human serum (multi-individual pooled; defined as surrogate matrix, ID No. 12095432001) was obtained from Roche Diagnostics GmbH (Mannheim, Germany). Native plasma matrix with different additives (lithium-heparin (Li-Heparin), dipotassium ethylene diamine tetraacetic acid (K₂-EDTA), and tripotassium dipotassium ethylene diamine tetraacetic acid (K₃-EDTA)) was obtained from leftover anonymized patient samples. Patient pools were prepared in accordance with the Declaration of Helsinki.

General requirements for laboratory equipment

A list of equipment used, and their requirements is given in Supplemental Material 1.

qNMR for determination of the purity of the standard materials

The single-pulse ¹HNMR measurements were performed on a JEOL 600 MHz NMR spectrometer equipped with an ultra-cool probe head, which provides 4–5 times sensitivity enhancement compared to normal room-temperature probe heads. For Primidone, single-pulse-1HNMR was utilized for the quantitation (Supplemental Material 2 Figure 1; methylene protons; CH₃CH ₂; δ=1.96 ppm; 2H; 1,3,5-trimethoxybenzene as qNMR internal standard; DMSO-d₆ as solvent) with an inter-scan delay of 70 s (Supplemental Material 2, Figure 1). The identity of these methylene protons was established based on the delta value, coupling constant and 2D-TOCSY spectrum. The differentiation of these protons from similar protons which could originate from the starting materials for the synthesis of Primidone viz., 5-Phenyl-5-ethyl-2-thiobarbituric Acid, 2-Ethyl-2-phenylmalonamide, 5-ethyl-2-methoxy-5-phenyl-dihydro-pyrimidine-4,6-dione and Diethyl 2-ethyl-2-phenylmalonate, was established by an extensive analysis of ¹³C{¹H} NMR, 2D-TOCSY, g-HSQCAD and CIGAR-HMBCAD spectra (Supplemental Material 2, Figures 3–5). Additional details about FID processing can be found in the supplementary information.

Preparation of calibrators and quality control samples

Two independent calibrator stock solutions were prepared by weighing 30.0 mg of primidone in tin boats on a microbalance. These tin boats were then transferred to 5 mL volumetric flasks and dissolved in DMSO to reach concentrations of 6.00 mg/mL. The concentrations of the stock solutions were calculated using the reference material’s purity of 99.8 ± 0.1 % (as determined by qNMR) and the amount weighed. Two working solutions were then prepared by diluting the stock solutions with DMSO in 5 mL volumetric flasks. These working and stock solutions were used to prepare spike solutions in DMSO. Finally, eight matrix-based calibrators were prepared by a volumetrically 1+99 dilution (v/v) into human serum matrix, uniformly distributed from 0.150 to 30.0 μg/mL (0.687–137 μmol/L) (see Figure 1).

Figure 1:

Schematic overview of set calibrator and control levels, which were chosen to allow an optimal coverage of measurement and therapeutic reference ranges. Black circles indicate calibrator spike solution, black triangles are QC spike solutions 1–4, the solid black line is the measurement range, the dashed line is the therapeutic reference range, and the black diamond is the alert level. Cal, calibrator; QC, quality control. Conversion factor µg/mL to µmol/L: 4.6.

A third independent stock solution was used to produce four levels of matrix-based quality control (QC) levels in the same way as described for the calibrators. The final concentrations were selected at four critical points, including levels above the limit of quantitation, below and within the therapeutic reference range, and at the laboratory alert level. Final concentration levels were 0.200, 2.00, 7.50 and 25.0 μg/mL (see Figure 1).

ISTD solution

To prepare the ISTD stock solution, approximately 1 mg of [²H₅]-primidone was dissolved in the appropriate volume of DMSO in the manufacturer’s container to obtain a 1000 μg/mL stock solution. The ISTD stock solution was stored at −20 °C for a maximum of 8 months. On each day of sample preparation, a fresh ISTD working solution was prepared by diluting the ISTD stock solution twofold. Specifically, 5 µL of ISTD stock solution was mixed with 95 µL of DMSO, followed by the addition of 3900 µL of Milli-Q water. The final concentration of the ISTD working solution was 1.25 μg/mL.

Sample preparation

Native human serum, TDM free serum (surrogate matrix) and plasma (Li-heparin, K₂-EDTA and K₃-EDTA) were used as sample matrix. 100 µL of the ISTD working solution was transferred into a 2 mL tube (Eppendorf Safe-Lock Tubes; Eppendorf, Hamburg, Germany) and 50 µL of the sample specimen (native sample/calibrator/QC) was added. 1000 µL of precipitation solution (75 % methanol [v/v]) was then added for protein precipitation. Finally, 20 µL of the supernatant was diluted 1+49 (v/v) with mobile phase A.

Liquid chromatography mass spectrometry

An Agilent 1290 Infinity II LC system (Santa Clara, California, USA) with a binary pump, a vacuum degasser, an autosampler at 7 °C, and a temperature-controlled column compartment at 40 °C was used for chromatographic separation. Analyte detection was performed using an AB Sciex Triple Quad 6500+ and Q-Trap 6500+ mass spectrometer (Framingham, Massachusetts, USA) equipped with a Turbo V ion source. Primidone was separated on an Agilent Zorbax Eclipse XDB-C18 column (100 × 3 mm, 3.5 µm, Santa Clara, California, USA). 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). Measurements were performed at a flowrate of 0.6 mL/min with a gradient over 8.0 min. The injection volume was set at 12 μL and a diverter valve was used to divert the eluent flow to the waste until 0.8 min and from 5.0 min to reduce contamination of the MS.

Primidone was detected in multiple reaction monitoring (MRM) mode and positive electrospray ionization mode (ESI+mode). The mass spectrometer parameters were optimized as follows: ion spray voltage of 5000 V; temperature of 400 °C; declustering potential of 50 V; dwell time of 50 ms. Nitrogen gas was used as curtain gas, collision gas, ion gas source 1, and ion gas source 2 and was set to 45, 11, 50, and 60 psi, respectively.

The quantifier mass transition is used as the basis of the quantitative method. Additionally, a qualifier transition was measured to check for interferences by monitoring the quantifier/qualifier ratios of clean system suitability test (SST) samples and clinical samples. The clinical sample and neat sample should not differ from each other by more than 20 %. Table 1 summarizes the quantifier and qualifier transitions and the remaining compound-dependent MS settings.

System suitability test

System sensitivity, chromatographic performance, and potential carry-over effect are tested prior to each analysis in an SST. Therefore, two neat samples, SST1 and SST2 (primidone concentration of the processed calibrator levels 1 and 8, respectively), were freshly prepared for each analysis. The signal-to-noise ratio of the quantifier for SST1 had to be ≥10 and the retention times for both SST samples had to be within 3.8 ± 0.5 min to pass the SST, to verify the sensitivity and chromatographic performance, respectively. Potential carry-over effects were examined by injecting the high-concentration sample SST2, followed by two solvent blanks. The analyte peak area observed in the first blank after the high-concentration sample had to be ≤20 % of the analyte peak area of calibrator 1 to pass the SST. This purity criterion applied also to all further blanks in the measurement campaigns.

Calibration and structure of analytical series

The system was calibrated using eight calibrators which were measured in increasing concentrations at the beginning and at the end of the analytical series. Both calibrations were used to generate the final calibration function by linear regression of the area ratios of the analyte and ISTD (y) against the analyte concentration (x) resulting in the function, y=x × A+b. Data evaluation was performed using the Analyst software (version 1.6.3 or higher) using the Intelli Quant algorithm. Retention times for primidone and its ISTD were 3.8 min and were integrated within a 30 s window. Peaks were processed with a smoothing factor of 3 and then integrated with a peak splitting factor of 2 and a noise percent of 80 % with a base sub window of 0.5 min.

Method validation

Assay validation and determination of measurement uncertainty were carried out based on the existing validation guidelines such as the Clinical & Laboratory Standard Guidelines C62A Liquid Chromatography-Mass Spectrometry Methods [30], the International Conference on Harmonisation guidance document Harmonised Tripartite Guideline Validation of Analytical Procedures: Text and Methodology Q2 (R1) [31] and the Guide to the expression of uncertainty in measurement (GUM) [32].

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-trimethoxybezene as a qNMR internal standard which is directly traceable to the primary qNMR standard NIST PS1. Furthermore, traceability to the SI unit of amount of substance (mole) is also included owing to referencing to NIST PS1. Since Planck’s constant is the main parameter for defining kilogram and 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 reference measurement procedure ideally suited for the characterization of primary reference materials [22]. Six individual experiments (Table 1 and Figure 2 of Supplemental Material 2), involving six individual weightings of the analyte and 1,3,5-trimethoxybenzene (qNMR ISTD) yield a final content value of 99.8 ± 0.1 % (k=1). The expanded uncertainty (k=2) can be obtained by multiplying the above-mentioned uncertainty by a factor of 2.

Selectivity

Baseline separation of primidone and its metabolite phenobarbital was achieved using reversed phase chromatography with an Agilent Zorbax Eclipse XDB-C18 column and a gradient elution scheme. In this setting a retention time of 3.8 min was realized for primidone. To evaluate the chromatographic selectivity towards phenobarbital, a polarity switch was used since primidone is detected in positive ESI mode and its metabolite phenobarbital in negative ESI mode. Primidone and phenobarbital were spiked at a concentration of 1.50 μg/mL into native human serum, surrogate serum, and Li-heparin plasma. As shown in Figure 2, the phenobarbital signal is well separated resulting in a chromatographic resolution (R) greater than 5.3 for all matrices tested. Evaluation of the primidone quantifier and qualifier transitions showed no interfering signals in the expected retention time window of primidone (3.8 min), neither in analysis-free matrices nor in native patient samples measured in the method comparison study (see Figure 3). In addition, no residual unlabeled analyte was observed in the ISTD [²H₅]-primidone.

Figure 2:

Primidone LC-MS/MS derived analytical readouts. Analyte on the left-hand side, primidone ISTD on the right-hand side. Chromatogram of primidone (black) and its metabolite phenobarbital (grey) both with concentrations of 1.50 μg/mL in native serum. ISTD, internal standard; LC-MS/MS, liquid chromatography-tandem mass spectrometry.

Figure 3:

Primidone LC-MS/MS derived analytical readouts. Analyte on the left-hand side, primidone ISTD on the right-hand side. (A) The lowest calibrator level peak (0.150 μg/mL) in native serum; (B) patient pool (n>5, 5.66 μg/mL). ISTD, internal standard; LC-MS/MS, liquid chromatography-tandem mass spectrometry.

Matrix effect and specificity

The possible occurrence of matrix-dependent effects by salts, proteins and phospholipids was minimized by diluting the sample after protein precipitation. The post-column infusion experiment showed that there was no visible matrix-related effect on the ionization of the primidone quantifier and qualifier transitions when processed matrix samples were injected.

To further demonstrate the absence of MEs, mean slopes and coefficients of determination of calibration series in different matrices (native serum, surrogate serum, and Li-heparin plasma) were compared with calibrations in neat solution. The slopes were found to be 0.538 (95 % CI 0.531 to 0.545) for neat solution, 0.543 (95 % CI 0.539 to 0.547) for native serum, 0.550 (95 % CI 0.545 to 0.554) for surrogate serum and 0.538 (95 % CI 0.534 to 0.542) for native Li-heparin plasma. The coefficient of determination was 1.00 independent of the matrix used for calibration. Thus, based on the similar slopes, the overlapping 95 % CI of the slopes, and the coefficients of determination it can be confirmed that no ME is present. Additionally, calibrator levels 1–8 in native human serum and Li-heparin plasma (n=6, sample preparations) were evaluated by applying the neat calibration as a standard. The mean relative bias across all concentrations ranged from −3.4 to 3.7 %, with a CV of ≤5.3 %.

Additionally, the absolute peak areas of the analyte and ISTD and the area ratio of matrix samples were compared to neat samples. There was no significant ME detected and the mean recoveries ranged from 99 to 102 % for the analyte and from 100 to 103 % for the ISTD independent of the matrix. The mean area ratios were between 97 and 102 % (Table 2).

Linearity

Linearity was determined for an extended range of 20 % at the lower and upper end of the range (0.120–36.0 μg/mL). The residuals were randomly distributed in a linear regression model which was therefore selected for assay calibration. The coefficients of determination of all individual calibrators were equal or better than 0.999. The serially diluted samples showed a linear dependence with a coefficient of determination of 0.999 and relative deviations of the samples ranging from −0.3 to 3.6 % with CVs ≤2.8 %.

Lower limit of the measuring interval and limit of detection

The LLMI of primidone is consistent with the lowest calibrator (0.150 μg/mL) and was evaluated by measuring spiked serum matrix samples. The CV (n=4, sample preparations) was found to be 1.5 % and relative bias (n=4, sample preparations) was 0.8 %. One sample was highlighted as an outlier according to Grubbs F [38] and therefore not considered. In addition, the LOD was estimated to be 0.0620 μg/mL.

Accuracy and precision

For the validation of accuracy and precision, spiked human serum and plasma samples at four concentration levels (0.200, 2.00, 7.50 and 25.0 μg/mL) were prepared.

Precision was evaluated with a multi-day validation experiment. Each concentration level was prepared sixfold by two different operators on five different days (n=12 measurements per day and n=60 measurements per 5 days). To estimate the overall variability of the method, various sources including between injections, preparations, calibrations, and days were determined using an ANOVA-based variance component analysis. Intermediate precision was found to be less than 2.5 % for all levels and matrices except for the lowest level in serum which was found to be ≤4.0 %. The repeatability CV was found to be ≤3.3 % for the lowest concentration level in serum and ranged from 1.0 to 1.8 % across all other concentration levels regardless of the matrix type (Table 3).

To evaluate the trueness of the method, spiked native serum and plasma samples were evaluated. The mean bias (n=6, sample preparations) within the measurement range (0.200, 2.00, 7.50 and 25.0 μg/mL) was between 0.1 and 3.9 % for serum levels and between −2.6 and 2.8 % for plasma levels. The bias of two high-concentration levels (35.0 and 50.0 μg/mL), which were diluted with native human serum before sample preparation, ranged from −1.8 to 1.9 % (Table 4).

Stability

The stability of processed samples was assessed by re-analyzing stored samples with freshly prepared serum samples. Autosampler stability at 7 °C was demonstrated for 9 days, with calculated recoveries ranging from 98 to 104 %. Moreover, the stability of spiked frozen serum calibrators and QC levels stored at −20 °C was examined after 33 days by comparing them to freshly prepared serum calibrators. The samples showed a recovery between 97 and 103 %.

Equivalence of results between independent laboratories

To demonstrate the equivalence between two independent laboratories, 82 samples (native serum patient samples, patient pools and samples) were analyzed at site 1 and 2. Two samples were excluded from the analysis as they were below the calibration range. The Passing-Bablok showed a regression equation with a slope of 1.01 (95 % CI 1.00 to 1.01), an intercept of −0.02 (95 % CI −0.07 to 0.01) and a correlation coefficient of 1.00, indicating an excellent agreement between the two laboratories (Figure 4A). The Bland-Altman analysis showed a mean deviation of 0.6 % (95 % CI interval from 0.2 to 1.0 %) and a data scatter which was independent from the primidone concentration. The 2S agreement was 3.6 % (lower limit CI interval from −3.7 to −2.3 %, upper limit CI interval from 3.5 to 4.9 %) (Figure 4B). The precision experiment at site 2 yielded good CVs for repeatability and intermediate precision (n=36) of ≤1.5 and ≤3.0 %, respectively, regardless of concentration level and matrix. These results show that the method is transferable and meets the requirements independent of the laboratory equipment or personnel.

Figure 4:

Results from the patient sample-based primidone 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=110 patients) 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 1.00–1.01) and an intercept of −0.02 (95 % CI −0.07 to 0.01). The Pearson correlation value was ≥999. (B) Bland-Altman plot for the method comparison study of the RMP (n=80, patient samples and patient pool samples) between two independent laboratories (Laboratory 1: Risch site, and Laboratory 2: Roche site). The inter-laboratory measurement bias was 0.6 % (95 % CI interval from 0.2 to 1.0 %) and the 2S interval of the relative difference was 3.6 % (lower limit CI interval from −3.7 to −2.3 %, upper limit CI interval from 3.5 to 4.9 %). CI, confidence interval; RMP, reference measurement procedure; SD, standard deviation.

Uncertainty of results

The total measurement uncertainty for single measurements of serum samples was found to be less than 2.1 %, except for the lowest level where it was 4.0 %. It was further reduced for the target value assignment by repeated measurements (n=6, three measurements on two days) where it ranged from 0.9 to 1.0 % regardless of the concentration level and sample type (Tables 5 and 6). The derived total uncertainty is multiplied by a coverage factor of k=2 to obtain an expanded uncertainty.