Journal of Chromatography B, 962 (2014) 102–108
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Vigabatrin in dried plasma spots: Validation of a novel LC–MS/MS method and application to clinical practice Nad¯a Kostic´ a , Yannis Dotsikas b , Nebojˇsa Jovic´ c , Galina Stevanovic´ c , And¯elija Malenovic´ a,∗ , Mirjana Medenica d a
University of Belgrade, Faculty of Pharmacy, Department of Drug Analysis, Vojvode Stepe 450, Belgrade, Serbia University of Athens, School of Pharmacy, Department of Pharmaceutical Chemistry, Panepistimioupoli Zografou, Athens, Greece c University of Belgrade, Faculty of Medicine, Clinic of Neurology and Psychiatry for Children and Youth, Dr Suboti´ca 6a, Belgrade, Serbia d University of Belgrade, Faculty of Pharmacy, Department of Physical Chemistry and Instrumental Methods, Vojvode Stepe 450, Belgrade, Serbia b
a r t i c l e
i n f o
Article history: Received 5 March 2014 Accepted 17 May 2014 Available online 27 May 2014 Keywords: Vigabatrin Chloroformates Dried plasma spots Mass spectrometry
a b s t r a c t This paper presents a LC–MS/MS method for the determination of antiepileptic drug vigabatrin in dried plasma spots (DPS). Due to its zwitterionic chemical structure, a pre-column derivatization procedure was performed, aiming to yield enhanced ionization efficiency and improved chromatographic behaviour. Propyl chloroformate, in the presence of propanol, was selected as the best derivatization reagent, providing a strong signal along with reasonable run time. A relatively novel sample collection technique, DPS, was utilized, offering easy sample handling and analysis, using a sample in micro amount (∼5 L). Derivatized vigabatrin and its internal standard, 4-aminocyclohexanecarboxylic acid, were extracted by liquid-liquid extraction (LLE) and determined in positive ion mode by applying two SRM transitions per analyte. A Zorbax Eclipse XDB-C8 column (150 × 4.6 mm, 5 m particle size) maintained at 30 ◦ C, was utilized with running mobile phase composed of acetonitrile: 0.15% formic acid (85:15, v/v). Flow rate was 550 L/min and total run time 4.5 min. The assay exhibited excellent linearity over the concentration range of 0.500–50.0 g/mL, which is suitable for the determination of vigabatrin level after per os administration in children and youths with epilepsy, who were on vigabatrin therapy, with or without co-medication. Specificity, accuracy, precision, recovery, matrix-effect and stability were also estimated and assessed within acceptance criteria. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Vigabatrin (4-amino-5-hexenoic acid) is an antiepileptic drug of newer generation, indicated predominantly as adjunctive therapy for patients with refractory complex partial seizures, as well as monotherapy for pediatric patients with infantile spasms [1]. It is also considered as a first-choice drug in the treatment of West syndrome, particularly in cases associated with tuberous sclerosis [2]. The suggested mechanism of vigabatrin action is irreversible inhibition of ␥-amino butyric acid (GABA) transaminase, the enzyme responsible for the metabolic degradation of the GABA. As a result, level of GABA in the central nervous system is increased [3,4]. Although vigabatrin is produced as a racemic mixture, it was shown that only S-(+) enantiomer is pharmacologically active [1]. Pharmacokinetic properties of the active enantiomer are not affected by the presence of R-(−) vigabatrin, since chiral inversion
∗ Corresponding author. Tel.: +381 11 3951 333; fax: +381 11 3972 840. ´ E-mail address: [email protected] (A. Malenovic). http://dx.doi.org/10.1016/j.jchromb.2014.05.037 1570-0232/© 2014 Elsevier B.V. All rights reserved.
in vivo is not observed [5]. Vigabatrin is eliminated in urine without any significant metabolic reaction, and therefore there is no significant metabolite which should be determined in biological fluids. A measurement of vigabatrin concentration is very important, especially because of visual impairment, associated with field defects, the most severe adverse affect observed during vigabatrin therapy. Therefore, a suitable analytical protocol for vigabatrin determination could be very useful in therapeutic drug monitoring, including evaluation of toxicity risks, as well. Literature survey revealed a certain number of articles dealing with different methodologies applied for the quantification of vigabatrin alone, or in combination with other antiepileptic drugs. However, regarding mass spectrometric (MS) methods, several articles applying gas chromatography (GC) and liquid chromatography (LC) were described. Pharmacokinetics of the individual enantiomers of vigabatrin was studied mainly by using stereospecific GC–MS protocols [6–8]. The simultaneous determination of vigabatrin and gabapentin in serum was obtained by a GC–MS method [9]. For the direct quantification of vigabatrin in plasma, one LC–MS/MS protocol was applied [10]. For the
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
analysis of vigabatrin with pregabalin and gabapentin [11], as well as in the combination with 21 other antiepileptic drugs [12], two UPLC–MS/MS protocols were described. Vigabatrin, pregabalin and gabapentin were determined in human serum using liquid chromatography of hydrophilic interactions with MS detection [13]. All previous LC–MS/MS protocols were characterized by sample preparation based on protein precipitation and lack of derivatization reaction. This combination led to a relatively short preparation time, however it is also responsible for not so clean extracts and unsatisfactory retention in RP chromatographic systems. DPS present a relatively novel alternative sample collection technique referring to the collection and preservation of plasma samples on filter paper. When combined with conventional plasma samples, DPS offer certain advantages regarding sample storage, transfer, stability and avoidance of contamination. In combination with LC–MS/MS, DPS provide an attractive approach for quantitative determinations in bioanalysis and thus the number of published methods is increasing [14–20]. Although dried blood spots (DBS) constitute the majority of applications of Dried Matrix Spots, especially when neonates are participants, DPS seem to have a promising future, due to two major advantages: (i) lack of hematocrit effect, the main concern in DBS analysis, and (ii) pharmacokinetic data mainly refer to plasma. The aim of this study was to develop the first LC–MS/MS method for the determination of vigabatrin in DPS. Due to its zwitterionic structure and coexistence of both positive and negative charge, which greatly affects ionization efficiency, a derivatization reaction based on alkyl chloroformate was performed [21]. It was recently shown that the length of alkyl moiety, originating from derivatization reagents, affects both MS signal and retention time in C8 column [22]. The selection of propyl chloroformate/n-propanol among all examined combinations (ranging from methyl to butyl chloroformate and from methanol to n-butanol) seems to provide the best compromise between signal enhancement and vigabatrin retention time/peak shape. Novel method was validated in accordance to guidelines for bioanalytical method validation [23,24] and its suitability was proven by analyzing samples obtained from epileptic children and youths. 2. Materials and methods 2.1. Chemicals and reagents Reference standards of vigabatrin and internal standard 4aminocyclohexanecarboxylic acid were obtained from British Pharmacopeia Commission Laboratory (Teddington, UK) and Alfa Aesar (Chembiotin, Athens, Greece), respectively. Acetonitrile (MS grade), n-propanol (HPLC grade), propyl chloroformate, pyridine, chloroform, ethyl acetate and formic acid were purchased from Sigma-Aldrich (St. Louis, USA). Hydrochloric acid was obtained from Lach-Ner (Neratovice, Czech Republic), while sodium hydroxide and n-hexane were acquired from J.T. Baker (Deventer, Holland) and LGC Promochem GmbH (Wesel, Germany), respectively. Whatman 903 filter papers were purchased from GE Healthcare (NJ, USA). Aqueous solutions were prepared with de-ionized and doubledistilled water (Resistivity >18 M) from Simplicity 185 (Millipore, Billerica, USA). Plasma samples were obtained from Clinic of Neurology and Psychiatry for Children and Youth in Belgrade. 2.2. Instrumentation Sample preparation was carried out by using Vortex—Genie 2T (Scientific Industries, Inc., New York, USA), Reacti—Vap III evaporation unit (Thermo Fisher Scientific Inc., San Jose, USA) and EBA 20 Hettich zentrifugen (DJB Labcare Ltd, England). Disks from DPS
103
were obtained by a 5 mm diameter non-commercial puncher. A TSQ Quantum Access MAX triple quadrupole mass spectrometer (Thermo Fisher Scientific Inc., San Jose, USA), equipped with electrospray ionization source (ESI), was utilized for tandem mass spectrometric detection. The chromatographic analysis was performed using Accela Thermo Scientific system, consisted of Accela pump and autosampler. Optimal conditions for the mass spectrometric and chromatographic parameters were defined by implementation of a step-by-step optimization strategy, as described in our previous paper [25]: position D for probe mount, 2.0 for micrometer settings, 0.05 m/z for scan width, 0.01 s for scan time, 4500 V for spray voltage, 50 and 10 arbitrary units for sheath and auxiliary gas pressure, respectively, 290 ◦ C for capillary temperature, 350 ◦ C for vaporizer temperature and 1.0 mTorr for collision pressure. The analytes were detected by monitoring the precursor → product ion transitions, using selected reaction monitoring (SRM) scan mode. The selected m/z transitions with corresponding collision energy values were 257.95 → 113.04 (15 eV) and 257.95 → 155.05 (7 eV) for derivatized vigabatrin and 272.03 → 127.01 (17 eV) and 272.03 → 212.00 (7 eV) for derivatized IS. All runs were performed using Zorbax Eclipse XDB-C8 column (150 × 4.6 mm, 5 m particle size) maintained at 30 ◦ C. The injection volume was 20 L with total run time of 4.5 min and running mobile phase was acetonitrile: 0.15% formic acid (85:15, v/v) with flow rate of 550 L/min. Data acquisition was performed using Xcalibur 1.2 software (Thermo Fisher Scientific Inc., San Jose, USA). 2.3. Sample preparation 2.3.1. Preparation of calibration standards and quality control/method validation samples Stock solutions of vigabatrin and internal standard (IS) were prepared by dissolving accurately weighted reference substances into 0.1 mol/L HCl to obtain a concentration of 500 g/mL and 100 g/mL, respectively. Working solutions were prepared by serial dilution of vigabatrin stock solution in 0.1 mol/L HCl with the following concentration levels: 500, 200, 100, 50.0, 20.0, 10.0 and 5.00 g/mL, while working solution for IS was prepared at concentration of 0.500 g/mL in 0.1 mol/L HCl. Another stock solution of vigabatrin (500 g/mL) was prepared after separate weighting for the preparation of quality control/method validation (QC/MV) working solutions in four levels: 350, 75.0, 15.0 and 5.00 g/mL in 0.1 mol/L HCl. Calibration curve consisted of seven non-zero vigabatrin standards prepared by a 10-fold dilution of respective working solutions in drug-free human plasma in 1.5 mL eppendorf tubes with final concentrations: 50.0, 20.0, 10.0, 5.00, 2.00, 1.00 and 0.500 g/mL. Final concentrations for QC/MV samples after 10fold dilution in drug-free plasma were 35.0 g/mL (QC3 /MV3 ), 7.50 g/mL (QC2 /MV2 ), 1.50 g/mL (QC1 /MV1 ) and 0.500 g/mL (QCL /MVL ). QC samples were used as the criterion of accepting or rejecting analytical runs, while MV samples were used for calculation of assay accuracy and precision. 2.3.2. Preparation of dried plasma spots DPS samples were prepared by spiking specific amount of standard/QC/unknown plasma sample onto filter paper. 40 L of each sample were carefully added to the center of the preprinted circle of labelled cards. In order to provide accurate and reliable results, spreading of plasma should be within the boundaries of the printed circle. Cards were allowed to dry at room temperature for 2–3 h, protected from light and humidity. Then they were packed into a plastic bag containing desiccant and stored in refrigerator till analysis.
104
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
2.4.2. Calibration curve A linear regression model (y = ax + b) was evaluated, where x is the concentration of vigabatrin and y corresponds to the areas ratios of derivatized vigabatrin to derivatized IS. Due to heteroscedasticity, a suitable weighting factor had to be estimated, based on the least sum of percentage relative error (%RE) per linear regression.
Fig. 1. Structures of underivatized (left) and derivatized (right) vigabatrin (a) and IS (b).
Plasma volume, extracted from each spot after punching, can be calculated using the following equation:
Vdisk
2
d 2 d/2 A = disk Vspot = Vspot = Vspot 2 Aspot D D/2
where V stands for volume, A for area, d represents the extracted plasma spot disc diameter and D the dried plasma spot diameter. Taking into account that d = 5 mm and D = 14 mm, the calculated plasma volume corresponding to each punched disc was ∼5 L. 2.3.3. Patients’ samples The current study was in accordance with Declaration of Helsinki and approved by Ethical Committee of the Clinic of Neurology and Psychiatry for Children and Youth in Belgrade. Signed informed consent was obtained by patients and/or parents or caregivers if appropriate. Patients having epilepsy or West syndrome received commercially available vigabatrin tablets/oral solution with or without co-medication. DPS from patients’ samples were prepared after collection of 500 L blood into tubes containing EDTA-K3 as anticoagulant. Blood was centrifuged for 5 min at 4000 rpm (1538 g) and then plasma supernatant was spread on card, creating four spots per filter paper. 2.3.4. DPS extraction and derivatization procedure A 5 mm diameter DPS disc from the center of the spot was punched out in 1.5 mL eppendorf tubes, followed by addition of 100 L of working IS solution. Extraction of analytes took place while tubes were vortex-mixed for 10 min. Derivatization reaction was performed by adding 150 L NaOH (0.33 mol/L), 80 L of n-propanol in pyridine solution (77:23, v/v) and 50 L propyl chloroformate reagent solution (mixture of propyl chloroformate, chloroform and n-hexane in ratio 17.4:71.6:11.0, v/v). LLE of newly formed derivatives was performed with 500 L of ethyl acetate. After vortex-mixing, all solutions were left to relax for 10 min and then 450 L of the upper layer were transferred to new vials and evaporated to dryness under nitrogen. Dry extracts were reconstituted with 500 L of acetonitrile and placed into autosampler at 10 ◦ C for LC–MS/MS analysis. Structures of underivatized and derivatized vigabatrin and IS are presented in Fig. 1. 2.4. Validation procedure 2.4.1. Selectivity Selectivity was tested with blank samples obtained from six different sources of plasma. Each blank sample was tested for interferences caused by potentially co-administered drugs, such as valproic acid, phenobarbital, lamotrigine, pregabalin, topiramate, levetiracetam, carbamazepine, as well as commonly prescribed drugs like paracetamol and amoxicillin. Selectivity was tested at 0.500 g/mL (LLOQ level).
2.4.3. Accuracy and precision Within-run and between-run accuracy and precision were estimated by analysis of five sets of MV samples at four levels during five different days. Accuracy was expressed as % recovery, while precision was expressed as % relative standard deviation (%RSD) at each MV level. 2.4.4. Carryover Carryover evaluation was estimated during analytical run by injecting blank solvent solution after calibrator of the highest concentration. 2.4.5. Extraction recovery Extraction recovery from DPS was calculated by comparing the peak area values of extracted and non-extracted MV samples, at three levels. Non-extracted samples were treated by applying derivatization reaction in 5 L vigabatrin solutions, prepared in 0.1 mol/L HCl and reconstituted with blank extract. Recovery test samples with concentrations corresponding to MV1 , MV2 and MV3 levels were analyzed in triplicate. 2.4.6. Matrix effect Matrix effect phenomenon was quantitatively determined at concentrations corresponding to MV1 , MV2 and MV3 levels. Two sets of vigabatrin samples in 0.1 mol/L HCl (5 L) underwent derivatization and extraction procedures in triplicate, as previously described. Samples of the first set (SOL) were reconstituted with pure acetonitrile, while the other samples were reconstituted with extracts originating from blank DPS (ME). %Matrix effect, per each investigated concentration level, is expressed as ME/SOL area ratio values multiplied by 100. 2.4.7. Stability studies Stability experiments were performed as part of the method validation protocol, in order to evaluate short-term and long-term stability, working solution and stock solution stability, as well as the stability of derivatized drug in autosampler. In each case, sample stability was validated by calculating the mean variation in percent. Autosampler, short-term and long-term stability were investigated by using two sample types, one with low to medium concentration of 2.00 g/mL (SL ) and one with medium to high of 40.0 g/mL (SH ). During stability testing all samples were analyzed in triplicate. Short-term stability was estimated by comparing fresh DPS with DPS standing at room temperature for 4 h. To evaluate long-term stability, 4 months old DPS, kept in refrigerator at 2–8 ◦ C, were analyzed versus freshly prepared vigabatrin DPS. As for autosampler stability, samples analyzed at the beginning of an analytical run were reanalyzed after 15 h, considering the case of analysis of 200 samples, which is the maximum number that the autosampler can accommodate during one sequence. Stock and working solutions stability (stored at 2–8 ◦ C) were estimated for the respective periods (3 months and 2 weeks). Old solutions diluted with 0.1 mol/L HCl in order to reach concentrations of 1 g/mL and 0.5 g/mL for stock and working stability testing, respectively were analyzed versus freshly prepared solutions.
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
105
Fig. 2. Representative SRM chromatograms of vigabatrin (top) and IS (bottom) obtained from a blank (a), LLOQ (b) and patient (c) sample.
2.4.8. Incurred sample reanalysis (ISR) The analytical method was also tested with ISR. All 12 samples were reanalyzed and the ISR results were compared with the original data, according to the following formula: (repeat sample − original sample) × 100/mean value. At least 67% of the samples should be within ±20%. 2.4.9. Dillution test QC sample in concentration of 90 g/mL in plasma was prepared and 40 L of this solution was placed on the filter paper. Sample preparation procedure was performed as already described and sample was reconstituted with 500 L of acetonitrile and additional 1000 L of acetonitrile used to reconstitute zero samples. Five replicates were prepared and tested as QC samples with three times dilution from nominal value, at concentration of 30 g/mL. Deviation from theoretical concentration, as well as %RSD should not exceed 15%.
resulted in increased signal and improved chromatographic behaviour in C8 columns [22]. Using just one disc from DPS, corresponding to ∼5 L of plasma, was proven to be adequate for achieving the desired lower limit of quantification (LLOQ) value of 0.500 g/mL. Derivatized vigabatrin and IS were detected in positive ESI mode, while quantitative analysis was performed via 2 SRM transitions per analyte. In the case of vigabatrin, parent molecular ion at m/z 257.95 underwent fragmentation to form product ions at m/z 113.04 and 155.05. Regarding IS, molecular ion [M + H]+ at m/z 272.03 was fragmented generating two strongest product ions at m/z 127.01 and 212.00. The optimized chromatographic conditions provided sharp, symmetric and Gaussian peaks for both analyte and IS. The retention times were 3.54 min and 3.76 min for derivatized vigabatrin and IS, respectively.
3. Results and discussion
3.1. Validation results
Mass spectrometry, in combination with dried matrix spots for sample collection, is obtaining increased popularity in modern bioanalysis. Inherent MS/MS sensitivity along with enhanced plasma concentrations in real samples, enable application of dried spots for the quantitation of vigabatrin. Implementation of derivatization reaction employing propyl chloroformate/n-propanol combination
3.1.1. Selectivity No interference was observed in the retention times of both analytes when analyzing blank DPS from various sources. Chromatogram of blank plasma sample is presented in Fig. 2a. Also, commonly used antiepileptic drugs such as valproic acid, phenobarbital, lamotrigine, pregabalin, topiramate, levetiracetam,
106
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
Fig. 3. Control charts at MVL (a), MV1 (b), MV2 (c) and MV3 (d) levels. Dashed lines correspond to + 3 as the upper control limit (UCL), and − 3 as the lower control limit (LCL).
carbamazepine, as well as paracetamol and amoxicillin did not affect signal at LLOQ level, since no interference was detected. 3.1.2. Calibration curve The current bioanalytical method was fully validated according to FDA and EMA guidelines. Five analytical runs were carried out, each one containing one calibration curve with seven non-zero standards, as presented in Table 1. As for the weighting factor, 1/x2 was chosen as the best one, compared to other weighting schemes and unweighted linear regression, resulting in the least sum of percentage relative error (%RE) per linear regression. Calibration standards data fulfilled the acceptance preset criteria, while linearity was excellent with regression coefficient R2 greater than 0.991 for the examined range of 0.500–50.0 g/mL. Moreover, ANOVA revealed no statistically significant lack-of-fit at the 5% significance level (p = 0.988). Chromatogram of the lowest calibration level, which is considered to be the LLOQ, is depicted in Fig. 2b. 3.1.3. Accuracy and precision Both within-run and between-run data for accuracy (expressed as %recovery) and precision (expressed as % relative standard Table 1 Statistical parameters of the calibration equations. Run
1 2 3 4 5
Regression equations
y = 0.0000155 + 0.168x y = −0.00343 + 0.167x y = −0.0249 + 0.166x y = 0.0230 + 0.308x y = 0.0369 + 0.308x
R2
0.997 0.992 0.995 0.991 0.991
SD Slope
Intercept
0.0043 0.0067 0.0053 0.0130 0.0146
0.0049 0.0077 0.0061 0.0149 0.0160
deviation, %RSD), are presented in Table 2. Acceptance criteria were formed as follow: for accuracy, mean value must not exceed ±15% of the nominal value for each concentration level (±20% at MVL level), and for precision, % RSD must be within 15% (20% at MVL level). In each case, results were within the predicted range. Furthermore, as shown in the control charts (Fig. 3), MV values fall inside the region set by the upper control limit (UCL) and lower control limit (LCL), defined as ± 3. No data drift was observed.
3.1.4. Carryover Evaluation of carryover by injecting blank solvent solution after calibrator of the highest concentration revealed that no carryover was observed, since no signal of both analyte or internal standard was detected.
Table 2 Within-run and between-run accuracy and precision results. MV sample (g/mL)
Within-run accuracya (%recovery)
Between-run accuracyb (%recovery)
Within-run precisionc (%RSD)
Between-run precisionb (%RSD)
MVL (0.500) MV1 (1.50) MV2 (7.50) MV3 (35.0)
105 87.7 92.9 94.6
97.2 96.0 97.5 95.9
15 6.6 7.7 9.6
15 10 9.4 9.5
a (n = 5), expressed as 100 × (mean calculated concentration)/(nominal concentration). b Values obtained from all 5 runs (n = 25). c (n = 5).
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108 Table 3 Recovery and matrix effect results.
Vigabatrin Internal standard *
Concentration (g/mL)
Recovery* (%)
Matrix effect* (%)
1.50 7.50 35.0 0.500
70.1 78.0 76.8 74.8
97.2 102.9 95.5 100.3
Mean values, n = 3.
3.1.5. Extraction recovery and matrix effect results Mean values of % extraction recovery for vigabtrin and IS are presented in Table 3. Similar extraction behaviour was observed for the three different concentration levels. In the same table, mean results for % matrix effect are given. Results confirm lack of any impact of endogenous plasma compounds on vigabatrin analysis. This outcome was expected, considering (i) the minor plasma volume (∼5 L), (ii) the LLE procedure that led to very clean extracts and iii) the high volume of reconstitution solution (500 L). 3.1.6. Stability results There was no available data in literature regarding potential stability problems of vigabatrin under normal storage or handling conditions, which is expected considering its chemical structure. All tests showed variations
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Vigabatrin in dried plasma spots: Validation of a novel LC–MS/MS method and application to clinical practice Nad¯a Kostic´ a , Yannis Dotsikas b , Nebojˇsa Jovic´ c , Galina Stevanovic´ c , And¯elija Malenovic´ a,∗ , Mirjana Medenica d a
University of Belgrade, Faculty of Pharmacy, Department of Drug Analysis, Vojvode Stepe 450, Belgrade, Serbia University of Athens, School of Pharmacy, Department of Pharmaceutical Chemistry, Panepistimioupoli Zografou, Athens, Greece c University of Belgrade, Faculty of Medicine, Clinic of Neurology and Psychiatry for Children and Youth, Dr Suboti´ca 6a, Belgrade, Serbia d University of Belgrade, Faculty of Pharmacy, Department of Physical Chemistry and Instrumental Methods, Vojvode Stepe 450, Belgrade, Serbia b
a r t i c l e
i n f o
Article history: Received 5 March 2014 Accepted 17 May 2014 Available online 27 May 2014 Keywords: Vigabatrin Chloroformates Dried plasma spots Mass spectrometry
a b s t r a c t This paper presents a LC–MS/MS method for the determination of antiepileptic drug vigabatrin in dried plasma spots (DPS). Due to its zwitterionic chemical structure, a pre-column derivatization procedure was performed, aiming to yield enhanced ionization efficiency and improved chromatographic behaviour. Propyl chloroformate, in the presence of propanol, was selected as the best derivatization reagent, providing a strong signal along with reasonable run time. A relatively novel sample collection technique, DPS, was utilized, offering easy sample handling and analysis, using a sample in micro amount (∼5 L). Derivatized vigabatrin and its internal standard, 4-aminocyclohexanecarboxylic acid, were extracted by liquid-liquid extraction (LLE) and determined in positive ion mode by applying two SRM transitions per analyte. A Zorbax Eclipse XDB-C8 column (150 × 4.6 mm, 5 m particle size) maintained at 30 ◦ C, was utilized with running mobile phase composed of acetonitrile: 0.15% formic acid (85:15, v/v). Flow rate was 550 L/min and total run time 4.5 min. The assay exhibited excellent linearity over the concentration range of 0.500–50.0 g/mL, which is suitable for the determination of vigabatrin level after per os administration in children and youths with epilepsy, who were on vigabatrin therapy, with or without co-medication. Specificity, accuracy, precision, recovery, matrix-effect and stability were also estimated and assessed within acceptance criteria. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Vigabatrin (4-amino-5-hexenoic acid) is an antiepileptic drug of newer generation, indicated predominantly as adjunctive therapy for patients with refractory complex partial seizures, as well as monotherapy for pediatric patients with infantile spasms [1]. It is also considered as a first-choice drug in the treatment of West syndrome, particularly in cases associated with tuberous sclerosis [2]. The suggested mechanism of vigabatrin action is irreversible inhibition of ␥-amino butyric acid (GABA) transaminase, the enzyme responsible for the metabolic degradation of the GABA. As a result, level of GABA in the central nervous system is increased [3,4]. Although vigabatrin is produced as a racemic mixture, it was shown that only S-(+) enantiomer is pharmacologically active [1]. Pharmacokinetic properties of the active enantiomer are not affected by the presence of R-(−) vigabatrin, since chiral inversion
∗ Corresponding author. Tel.: +381 11 3951 333; fax: +381 11 3972 840. ´ E-mail address: [email protected] (A. Malenovic). http://dx.doi.org/10.1016/j.jchromb.2014.05.037 1570-0232/© 2014 Elsevier B.V. All rights reserved.
in vivo is not observed [5]. Vigabatrin is eliminated in urine without any significant metabolic reaction, and therefore there is no significant metabolite which should be determined in biological fluids. A measurement of vigabatrin concentration is very important, especially because of visual impairment, associated with field defects, the most severe adverse affect observed during vigabatrin therapy. Therefore, a suitable analytical protocol for vigabatrin determination could be very useful in therapeutic drug monitoring, including evaluation of toxicity risks, as well. Literature survey revealed a certain number of articles dealing with different methodologies applied for the quantification of vigabatrin alone, or in combination with other antiepileptic drugs. However, regarding mass spectrometric (MS) methods, several articles applying gas chromatography (GC) and liquid chromatography (LC) were described. Pharmacokinetics of the individual enantiomers of vigabatrin was studied mainly by using stereospecific GC–MS protocols [6–8]. The simultaneous determination of vigabatrin and gabapentin in serum was obtained by a GC–MS method [9]. For the direct quantification of vigabatrin in plasma, one LC–MS/MS protocol was applied [10]. For the
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
analysis of vigabatrin with pregabalin and gabapentin [11], as well as in the combination with 21 other antiepileptic drugs [12], two UPLC–MS/MS protocols were described. Vigabatrin, pregabalin and gabapentin were determined in human serum using liquid chromatography of hydrophilic interactions with MS detection [13]. All previous LC–MS/MS protocols were characterized by sample preparation based on protein precipitation and lack of derivatization reaction. This combination led to a relatively short preparation time, however it is also responsible for not so clean extracts and unsatisfactory retention in RP chromatographic systems. DPS present a relatively novel alternative sample collection technique referring to the collection and preservation of plasma samples on filter paper. When combined with conventional plasma samples, DPS offer certain advantages regarding sample storage, transfer, stability and avoidance of contamination. In combination with LC–MS/MS, DPS provide an attractive approach for quantitative determinations in bioanalysis and thus the number of published methods is increasing [14–20]. Although dried blood spots (DBS) constitute the majority of applications of Dried Matrix Spots, especially when neonates are participants, DPS seem to have a promising future, due to two major advantages: (i) lack of hematocrit effect, the main concern in DBS analysis, and (ii) pharmacokinetic data mainly refer to plasma. The aim of this study was to develop the first LC–MS/MS method for the determination of vigabatrin in DPS. Due to its zwitterionic structure and coexistence of both positive and negative charge, which greatly affects ionization efficiency, a derivatization reaction based on alkyl chloroformate was performed [21]. It was recently shown that the length of alkyl moiety, originating from derivatization reagents, affects both MS signal and retention time in C8 column [22]. The selection of propyl chloroformate/n-propanol among all examined combinations (ranging from methyl to butyl chloroformate and from methanol to n-butanol) seems to provide the best compromise between signal enhancement and vigabatrin retention time/peak shape. Novel method was validated in accordance to guidelines for bioanalytical method validation [23,24] and its suitability was proven by analyzing samples obtained from epileptic children and youths. 2. Materials and methods 2.1. Chemicals and reagents Reference standards of vigabatrin and internal standard 4aminocyclohexanecarboxylic acid were obtained from British Pharmacopeia Commission Laboratory (Teddington, UK) and Alfa Aesar (Chembiotin, Athens, Greece), respectively. Acetonitrile (MS grade), n-propanol (HPLC grade), propyl chloroformate, pyridine, chloroform, ethyl acetate and formic acid were purchased from Sigma-Aldrich (St. Louis, USA). Hydrochloric acid was obtained from Lach-Ner (Neratovice, Czech Republic), while sodium hydroxide and n-hexane were acquired from J.T. Baker (Deventer, Holland) and LGC Promochem GmbH (Wesel, Germany), respectively. Whatman 903 filter papers were purchased from GE Healthcare (NJ, USA). Aqueous solutions were prepared with de-ionized and doubledistilled water (Resistivity >18 M) from Simplicity 185 (Millipore, Billerica, USA). Plasma samples were obtained from Clinic of Neurology and Psychiatry for Children and Youth in Belgrade. 2.2. Instrumentation Sample preparation was carried out by using Vortex—Genie 2T (Scientific Industries, Inc., New York, USA), Reacti—Vap III evaporation unit (Thermo Fisher Scientific Inc., San Jose, USA) and EBA 20 Hettich zentrifugen (DJB Labcare Ltd, England). Disks from DPS
103
were obtained by a 5 mm diameter non-commercial puncher. A TSQ Quantum Access MAX triple quadrupole mass spectrometer (Thermo Fisher Scientific Inc., San Jose, USA), equipped with electrospray ionization source (ESI), was utilized for tandem mass spectrometric detection. The chromatographic analysis was performed using Accela Thermo Scientific system, consisted of Accela pump and autosampler. Optimal conditions for the mass spectrometric and chromatographic parameters were defined by implementation of a step-by-step optimization strategy, as described in our previous paper [25]: position D for probe mount, 2.0 for micrometer settings, 0.05 m/z for scan width, 0.01 s for scan time, 4500 V for spray voltage, 50 and 10 arbitrary units for sheath and auxiliary gas pressure, respectively, 290 ◦ C for capillary temperature, 350 ◦ C for vaporizer temperature and 1.0 mTorr for collision pressure. The analytes were detected by monitoring the precursor → product ion transitions, using selected reaction monitoring (SRM) scan mode. The selected m/z transitions with corresponding collision energy values were 257.95 → 113.04 (15 eV) and 257.95 → 155.05 (7 eV) for derivatized vigabatrin and 272.03 → 127.01 (17 eV) and 272.03 → 212.00 (7 eV) for derivatized IS. All runs were performed using Zorbax Eclipse XDB-C8 column (150 × 4.6 mm, 5 m particle size) maintained at 30 ◦ C. The injection volume was 20 L with total run time of 4.5 min and running mobile phase was acetonitrile: 0.15% formic acid (85:15, v/v) with flow rate of 550 L/min. Data acquisition was performed using Xcalibur 1.2 software (Thermo Fisher Scientific Inc., San Jose, USA). 2.3. Sample preparation 2.3.1. Preparation of calibration standards and quality control/method validation samples Stock solutions of vigabatrin and internal standard (IS) were prepared by dissolving accurately weighted reference substances into 0.1 mol/L HCl to obtain a concentration of 500 g/mL and 100 g/mL, respectively. Working solutions were prepared by serial dilution of vigabatrin stock solution in 0.1 mol/L HCl with the following concentration levels: 500, 200, 100, 50.0, 20.0, 10.0 and 5.00 g/mL, while working solution for IS was prepared at concentration of 0.500 g/mL in 0.1 mol/L HCl. Another stock solution of vigabatrin (500 g/mL) was prepared after separate weighting for the preparation of quality control/method validation (QC/MV) working solutions in four levels: 350, 75.0, 15.0 and 5.00 g/mL in 0.1 mol/L HCl. Calibration curve consisted of seven non-zero vigabatrin standards prepared by a 10-fold dilution of respective working solutions in drug-free human plasma in 1.5 mL eppendorf tubes with final concentrations: 50.0, 20.0, 10.0, 5.00, 2.00, 1.00 and 0.500 g/mL. Final concentrations for QC/MV samples after 10fold dilution in drug-free plasma were 35.0 g/mL (QC3 /MV3 ), 7.50 g/mL (QC2 /MV2 ), 1.50 g/mL (QC1 /MV1 ) and 0.500 g/mL (QCL /MVL ). QC samples were used as the criterion of accepting or rejecting analytical runs, while MV samples were used for calculation of assay accuracy and precision. 2.3.2. Preparation of dried plasma spots DPS samples were prepared by spiking specific amount of standard/QC/unknown plasma sample onto filter paper. 40 L of each sample were carefully added to the center of the preprinted circle of labelled cards. In order to provide accurate and reliable results, spreading of plasma should be within the boundaries of the printed circle. Cards were allowed to dry at room temperature for 2–3 h, protected from light and humidity. Then they were packed into a plastic bag containing desiccant and stored in refrigerator till analysis.
104
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
2.4.2. Calibration curve A linear regression model (y = ax + b) was evaluated, where x is the concentration of vigabatrin and y corresponds to the areas ratios of derivatized vigabatrin to derivatized IS. Due to heteroscedasticity, a suitable weighting factor had to be estimated, based on the least sum of percentage relative error (%RE) per linear regression.
Fig. 1. Structures of underivatized (left) and derivatized (right) vigabatrin (a) and IS (b).
Plasma volume, extracted from each spot after punching, can be calculated using the following equation:
Vdisk
2
d 2 d/2 A = disk Vspot = Vspot = Vspot 2 Aspot D D/2
where V stands for volume, A for area, d represents the extracted plasma spot disc diameter and D the dried plasma spot diameter. Taking into account that d = 5 mm and D = 14 mm, the calculated plasma volume corresponding to each punched disc was ∼5 L. 2.3.3. Patients’ samples The current study was in accordance with Declaration of Helsinki and approved by Ethical Committee of the Clinic of Neurology and Psychiatry for Children and Youth in Belgrade. Signed informed consent was obtained by patients and/or parents or caregivers if appropriate. Patients having epilepsy or West syndrome received commercially available vigabatrin tablets/oral solution with or without co-medication. DPS from patients’ samples were prepared after collection of 500 L blood into tubes containing EDTA-K3 as anticoagulant. Blood was centrifuged for 5 min at 4000 rpm (1538 g) and then plasma supernatant was spread on card, creating four spots per filter paper. 2.3.4. DPS extraction and derivatization procedure A 5 mm diameter DPS disc from the center of the spot was punched out in 1.5 mL eppendorf tubes, followed by addition of 100 L of working IS solution. Extraction of analytes took place while tubes were vortex-mixed for 10 min. Derivatization reaction was performed by adding 150 L NaOH (0.33 mol/L), 80 L of n-propanol in pyridine solution (77:23, v/v) and 50 L propyl chloroformate reagent solution (mixture of propyl chloroformate, chloroform and n-hexane in ratio 17.4:71.6:11.0, v/v). LLE of newly formed derivatives was performed with 500 L of ethyl acetate. After vortex-mixing, all solutions were left to relax for 10 min and then 450 L of the upper layer were transferred to new vials and evaporated to dryness under nitrogen. Dry extracts were reconstituted with 500 L of acetonitrile and placed into autosampler at 10 ◦ C for LC–MS/MS analysis. Structures of underivatized and derivatized vigabatrin and IS are presented in Fig. 1. 2.4. Validation procedure 2.4.1. Selectivity Selectivity was tested with blank samples obtained from six different sources of plasma. Each blank sample was tested for interferences caused by potentially co-administered drugs, such as valproic acid, phenobarbital, lamotrigine, pregabalin, topiramate, levetiracetam, carbamazepine, as well as commonly prescribed drugs like paracetamol and amoxicillin. Selectivity was tested at 0.500 g/mL (LLOQ level).
2.4.3. Accuracy and precision Within-run and between-run accuracy and precision were estimated by analysis of five sets of MV samples at four levels during five different days. Accuracy was expressed as % recovery, while precision was expressed as % relative standard deviation (%RSD) at each MV level. 2.4.4. Carryover Carryover evaluation was estimated during analytical run by injecting blank solvent solution after calibrator of the highest concentration. 2.4.5. Extraction recovery Extraction recovery from DPS was calculated by comparing the peak area values of extracted and non-extracted MV samples, at three levels. Non-extracted samples were treated by applying derivatization reaction in 5 L vigabatrin solutions, prepared in 0.1 mol/L HCl and reconstituted with blank extract. Recovery test samples with concentrations corresponding to MV1 , MV2 and MV3 levels were analyzed in triplicate. 2.4.6. Matrix effect Matrix effect phenomenon was quantitatively determined at concentrations corresponding to MV1 , MV2 and MV3 levels. Two sets of vigabatrin samples in 0.1 mol/L HCl (5 L) underwent derivatization and extraction procedures in triplicate, as previously described. Samples of the first set (SOL) were reconstituted with pure acetonitrile, while the other samples were reconstituted with extracts originating from blank DPS (ME). %Matrix effect, per each investigated concentration level, is expressed as ME/SOL area ratio values multiplied by 100. 2.4.7. Stability studies Stability experiments were performed as part of the method validation protocol, in order to evaluate short-term and long-term stability, working solution and stock solution stability, as well as the stability of derivatized drug in autosampler. In each case, sample stability was validated by calculating the mean variation in percent. Autosampler, short-term and long-term stability were investigated by using two sample types, one with low to medium concentration of 2.00 g/mL (SL ) and one with medium to high of 40.0 g/mL (SH ). During stability testing all samples were analyzed in triplicate. Short-term stability was estimated by comparing fresh DPS with DPS standing at room temperature for 4 h. To evaluate long-term stability, 4 months old DPS, kept in refrigerator at 2–8 ◦ C, were analyzed versus freshly prepared vigabatrin DPS. As for autosampler stability, samples analyzed at the beginning of an analytical run were reanalyzed after 15 h, considering the case of analysis of 200 samples, which is the maximum number that the autosampler can accommodate during one sequence. Stock and working solutions stability (stored at 2–8 ◦ C) were estimated for the respective periods (3 months and 2 weeks). Old solutions diluted with 0.1 mol/L HCl in order to reach concentrations of 1 g/mL and 0.5 g/mL for stock and working stability testing, respectively were analyzed versus freshly prepared solutions.
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
105
Fig. 2. Representative SRM chromatograms of vigabatrin (top) and IS (bottom) obtained from a blank (a), LLOQ (b) and patient (c) sample.
2.4.8. Incurred sample reanalysis (ISR) The analytical method was also tested with ISR. All 12 samples were reanalyzed and the ISR results were compared with the original data, according to the following formula: (repeat sample − original sample) × 100/mean value. At least 67% of the samples should be within ±20%. 2.4.9. Dillution test QC sample in concentration of 90 g/mL in plasma was prepared and 40 L of this solution was placed on the filter paper. Sample preparation procedure was performed as already described and sample was reconstituted with 500 L of acetonitrile and additional 1000 L of acetonitrile used to reconstitute zero samples. Five replicates were prepared and tested as QC samples with three times dilution from nominal value, at concentration of 30 g/mL. Deviation from theoretical concentration, as well as %RSD should not exceed 15%.
resulted in increased signal and improved chromatographic behaviour in C8 columns [22]. Using just one disc from DPS, corresponding to ∼5 L of plasma, was proven to be adequate for achieving the desired lower limit of quantification (LLOQ) value of 0.500 g/mL. Derivatized vigabatrin and IS were detected in positive ESI mode, while quantitative analysis was performed via 2 SRM transitions per analyte. In the case of vigabatrin, parent molecular ion at m/z 257.95 underwent fragmentation to form product ions at m/z 113.04 and 155.05. Regarding IS, molecular ion [M + H]+ at m/z 272.03 was fragmented generating two strongest product ions at m/z 127.01 and 212.00. The optimized chromatographic conditions provided sharp, symmetric and Gaussian peaks for both analyte and IS. The retention times were 3.54 min and 3.76 min for derivatized vigabatrin and IS, respectively.
3. Results and discussion
3.1. Validation results
Mass spectrometry, in combination with dried matrix spots for sample collection, is obtaining increased popularity in modern bioanalysis. Inherent MS/MS sensitivity along with enhanced plasma concentrations in real samples, enable application of dried spots for the quantitation of vigabatrin. Implementation of derivatization reaction employing propyl chloroformate/n-propanol combination
3.1.1. Selectivity No interference was observed in the retention times of both analytes when analyzing blank DPS from various sources. Chromatogram of blank plasma sample is presented in Fig. 2a. Also, commonly used antiepileptic drugs such as valproic acid, phenobarbital, lamotrigine, pregabalin, topiramate, levetiracetam,
106
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108
Fig. 3. Control charts at MVL (a), MV1 (b), MV2 (c) and MV3 (d) levels. Dashed lines correspond to + 3 as the upper control limit (UCL), and − 3 as the lower control limit (LCL).
carbamazepine, as well as paracetamol and amoxicillin did not affect signal at LLOQ level, since no interference was detected. 3.1.2. Calibration curve The current bioanalytical method was fully validated according to FDA and EMA guidelines. Five analytical runs were carried out, each one containing one calibration curve with seven non-zero standards, as presented in Table 1. As for the weighting factor, 1/x2 was chosen as the best one, compared to other weighting schemes and unweighted linear regression, resulting in the least sum of percentage relative error (%RE) per linear regression. Calibration standards data fulfilled the acceptance preset criteria, while linearity was excellent with regression coefficient R2 greater than 0.991 for the examined range of 0.500–50.0 g/mL. Moreover, ANOVA revealed no statistically significant lack-of-fit at the 5% significance level (p = 0.988). Chromatogram of the lowest calibration level, which is considered to be the LLOQ, is depicted in Fig. 2b. 3.1.3. Accuracy and precision Both within-run and between-run data for accuracy (expressed as %recovery) and precision (expressed as % relative standard Table 1 Statistical parameters of the calibration equations. Run
1 2 3 4 5
Regression equations
y = 0.0000155 + 0.168x y = −0.00343 + 0.167x y = −0.0249 + 0.166x y = 0.0230 + 0.308x y = 0.0369 + 0.308x
R2
0.997 0.992 0.995 0.991 0.991
SD Slope
Intercept
0.0043 0.0067 0.0053 0.0130 0.0146
0.0049 0.0077 0.0061 0.0149 0.0160
deviation, %RSD), are presented in Table 2. Acceptance criteria were formed as follow: for accuracy, mean value must not exceed ±15% of the nominal value for each concentration level (±20% at MVL level), and for precision, % RSD must be within 15% (20% at MVL level). In each case, results were within the predicted range. Furthermore, as shown in the control charts (Fig. 3), MV values fall inside the region set by the upper control limit (UCL) and lower control limit (LCL), defined as ± 3. No data drift was observed.
3.1.4. Carryover Evaluation of carryover by injecting blank solvent solution after calibrator of the highest concentration revealed that no carryover was observed, since no signal of both analyte or internal standard was detected.
Table 2 Within-run and between-run accuracy and precision results. MV sample (g/mL)
Within-run accuracya (%recovery)
Between-run accuracyb (%recovery)
Within-run precisionc (%RSD)
Between-run precisionb (%RSD)
MVL (0.500) MV1 (1.50) MV2 (7.50) MV3 (35.0)
105 87.7 92.9 94.6
97.2 96.0 97.5 95.9
15 6.6 7.7 9.6
15 10 9.4 9.5
a (n = 5), expressed as 100 × (mean calculated concentration)/(nominal concentration). b Values obtained from all 5 runs (n = 25). c (n = 5).
N. Kosti´c et al. / J. Chromatogr. B 962 (2014) 102–108 Table 3 Recovery and matrix effect results.
Vigabatrin Internal standard *
Concentration (g/mL)
Recovery* (%)
Matrix effect* (%)
1.50 7.50 35.0 0.500
70.1 78.0 76.8 74.8
97.2 102.9 95.5 100.3
Mean values, n = 3.
3.1.5. Extraction recovery and matrix effect results Mean values of % extraction recovery for vigabtrin and IS are presented in Table 3. Similar extraction behaviour was observed for the three different concentration levels. In the same table, mean results for % matrix effect are given. Results confirm lack of any impact of endogenous plasma compounds on vigabatrin analysis. This outcome was expected, considering (i) the minor plasma volume (∼5 L), (ii) the LLE procedure that led to very clean extracts and iii) the high volume of reconstitution solution (500 L). 3.1.6. Stability results There was no available data in literature regarding potential stability problems of vigabatrin under normal storage or handling conditions, which is expected considering its chemical structure. All tests showed variations
