Abstract
Pregabalin is a gabapentinoid approved for the treatment of general anxiety disorder, neuropathic pain and as adjunctive therapy for focal seizures in patients with epilepsy. In addition, there are a number of conditions for which pregabalin is prescribed off-label. Along with the widespread use there are a significant number of reports describing the misuse of pregabalin over the last decade. Over time, it became clear that pregabalin should become part of routine testing in toxicology laboratories. The aim of this paper was to present validation of a LC-MS/MS method for the quantification of pregabalin in plasma of acutely poisoned patients. Simple sample preparation step and rapid chromatographic separation shortened the overall analysis time, which was the goal of method development. The presence of pregabalin was confirmed with three ion transitions, ensuring high selectivity of the validated method. The statistical data obtained showed good precision and accuracy over a wide concentration range. No endogenous or other interference was detected, and there was no matrix effect influence with this method. The LC-MS/MS method was applied to quantify pregabalin in plasma samples of patients admitted to the emergency department due to a possible acute pregabalin overdose. Different concentrations were found, and we report, to the best of our knowledge, the highest plasma concentration of pregabalin in the plasma of a patient with acute poisoning. In conclusion, we developed a fast and simple LC-MS/MS method for reliable determination of pregabalin and demonstrated the developed method was suitable for routine use in clinical toxicology setting.
Introduction
Pregabalin (PG) is psychoactive drug that belongs to the gabapentinoids, a class of drugs synthesized as structural analogs of γ-aminobutyric acid (GABA) (Fig. 1). Gabapentinoids however, do not bind to GABA receptors, but exert their effects primarily by inhibiting voltage-gated calcium channels. Following oral administration PG shows an extensive and rapid absorption rate, independent of the ingested dose. Oral bioavailability is over 90% and peak plasma concentrations occur within an hour of dosing. The metabolic profile of PG involves minimal biotransformation and the drug is excreted in urine almost completely unchanged. The elimination half-life of PG in healthy adults is 5–7 h. Beside approved indications (managing of general anxiety disorder, neuropathic pain and adjunctive therapy for focal seizures in patients with epilepsy), pregabalin is increasingly used as off-label drug in the treatment of various conditions such as bipolar disorder, restless legs syndrome, sleep disorders, alcohol withdrawal syndrome, etc. [1, 2]. The therapeutic plasma range is 2–8 mg L−1, but steady state plasma concentrations were up to 14.2 mg L−1 in patients maintained on daily dose of 600 mg [3]. Toxicologically relevant concentrations are several times higher than therapeutic ones [4–6]. The highest reported pregabalin concentration was 66.5 mg L−1, measured in the serum of patient, after a single oral PG poisoning [7].
Structure of GABA and GABA analogs
Citation: Acta Chromatographica 36, 2; 10.1556/1326.2023.01104
The prescription and consumption of PG has been constantly increasing since its first release on market in 2004 [8]. Along with the widespread use, there has been a worrying increase in cases of PG misuse. In the last decade a number of reports of PG acute poisoning cases have been published, but data on concentrations in overdose cases are still lacking [9–12]. The National Poison Control Center (NPCC) of the Military Medical Academy in Belgrade is a reference facility in Serbia for the treatment of patients with acute poisoning. The number of pregabalin overdose cases in NPCC has increased dramatically over the past five years, as have the number of requests for laboratory testing to determine drug concentration.
Various bioanalytical methods for the determination of PG can be found in the literature. Liquid chromatography coupled with a tandem mass spectrometer (LC-MS/MS) is preferable method due to its selectivity, accuracy, precision, and robustness [13–15]. Some of the published methods require a derivatization step during sample preparation, which complicates the process and increases the overall time of analysis [16]. Most of the published methods are linear near the therapeutic range, while only a few of them have an upper limit of quantification (ULOQ) greater than 20 mg L−1 [17, 18]. Therefore, the aim of this study was to develop and validate bioanalytical method suitable for use in clinical toxicology laboratories. We validated a rapid, precise, accurate and highly selective method with a simple sample preparation step that can measure a wide range of therapeutic and toxic PG concentrations in human plasma. The method was applied for quantification of PG in plasma samples of patients, admitted to the NPCC emergency department, with suspected pregabalin overdose.
Experimental
Chemical and reagents
Pregabalin (Mylan Laboratories Ltd, India) was used as the reference standard. Gabapentin (Belupo Pharmaceuticals & Cosmetics Inc, Croatia) was used as the internal standard (IS). Chemicals used for preparation of samples, stock solutions and mobile phase (water, acetonitrile, methanol and acetic acid) were LC/MS grade purity and obtained from Merck, Darmstadt, Germany. Blank human plasma was obtained from healthy donors in accordance with hospital ethical approval.
Analytical instrumentation
The analysis was performed using a ACQUITY UPLC H-Class System coupled with a XEVO TQD triple quadruple mass spectrometer (Waters Corp., Milford, MA, USA). The mass spectrometer was equipped with an electrospray ionization (ESI) source for mass analysis and detection. Analyst MassLynx-Intellistart software was used for detector optimization and data acquisition.
Liquid chromatographic conditions
Chromatographic separation was achieved on a Waters Acquity UPLC® BEH C18, 1.7 µm, 2.1 × 100 mm column. The column and sample temperature were maintained at 50 °C and 4 °C, respectively. Mobile phase A consisted of 0.5% acetic acid in water; mobile phase B consisted of 0.1% acetic acid in acetonitrile. Isocratic elution was applied at a constant flow rate of 0.40 mL min−1 (A: B = 65:35 v/v) and total run time for each sample was 2.5 min. Injection volume was 1 μL.
Mass spectrometric conditions
The mass spectrometer was operated using ESI source in positive ion mode. Optimization of mass spectrometer was done by injecting PG and gabapentin solution (1 mg L−1) directly into the mass spectrometer. Detection and quantification was performed with multiple reactions monitoring (MRM) mode. Three mass transitions for PG and two for IS were monitored in this method. One transition was used for quantification (quantifier) and the others for confirmation of detected substances (qualifiers). The optimization and MS/MS method parameters are listed in Table 1.
The optimization and MS-MS method parameters
| Compound | Ion transition | Parent ion (m/z) | Cone voltage (V) | Daughter ion (m/z) | Collision energy (V) |
| Pregabalin | MRM 1 | 160.09 | 28 | 54.96 | 22 |
| MRM 2 (quantifier) | 160.09 | 28 | 97.02 | 14 | |
| MRM 3 | 160.09 | 24 | 142.00 | 10 | |
| Gabapentin (IS) | MRM 1 | 172.09 | 26 | 54.95 | 22 |
| MRM 2 (quantifier) | 172.09 | 34 | 136.95 | 14 |
The source block temperature was 150 °C and the capillary voltage 1.30 kV. The desolvation gas (nitrogen) was heated to 500 °C and the flow was set to 1000 L h−1. The cone gas (nitrogen) was delivered at a flow rate of 20 L h−1.
Stock solutions, calibration standards and quality control samples
Two stock solutions of PG (1 mg mL−1) and one stock solution of IS (1 mg mL−1) were prepared by dissolving 10.0 mg (±0.1 mg) of standard substances in 10 mL of methanol. One stock solution of PG was used to prepare working standard solutions for calibration plasma samples (Cs) and the other stock solution for the preparation of working standards for Quality Control samples (QCs). Working solutions were prepared freshly on each day of analysis as serial dilutions in appropriate ratio. The IS was prepared from a stock solution in methanol, at a concentration of 50 mg L−1.
For the calibration curve eight concentration levels of Cs were prepared, after diluting the appropriate working solutions of PG with human EDTA plasma, so as to obtain final concentrations: 1 mg L−1, 5 mg L−1, 10 mg L−1, 20 mg L−1, 35 mg L−1, 50 mg L−1, 75 mg L−1 and 100 mg L−1. Plasma sample with PG concentration of 0.05 mg L−1 was prepared to be analyzed as a Limit of detection (LoD). Zero sample (plasma sample with IS) was prepared to confirm that no interferences were present at the retention time of PG.
Quality control samples were prepared in three concentration levels by diluting the appropriate working solutions with human EDTA plasma, so that the following concentrations were obtained: Low–2.5 mg L−1 (QCL); Medium–40 mg L−1 (QCM); High–80 mg L−1 (QCH) as and 150 mg L−1 to test the sample dilution effect (QCDE).
Blood samples for obtaining “pool” plasma were collected from healthy volunteers in accordance with ethical approval of hospital.
Sample preparation
All samples (Cs, QCs and patient samples) were prepared in the same manner according to the following procedure: 200 µL of plasma sample, 40 µL of IS and 800 µL of cold acetonitrile were mixed in the plastic tube and shaken on vortex mixer for 30 s. After centrifugation at 8,935 rpm for 10 min, 200 µL of supernatant was diluted in 800 µL of mobile phase A. Final solution was filtered and 1 µL was injected into LC-MS/MS for analysis.
Method validation
The method was validated for selectivity, carry over, linearity, precision and accuracy, dilution effect, matrix effect and stability according to European medicines agency (EMA) guidelines for validation of bioanalytical methods [19].
Selectivity was evaluated by analyzing chromatograms of the blank plasma samples from 6 different sources. Additionally, potential interferences were tested in the presence of drugs and substances that are most often co-ingested with PG in acute overdose cases [12]. In this purpose methanol solutions of various substances (diazepam, clonazepam, bromazepam, lorazepam, valproic acid, carbamazepine, lamotrigine, tramadol, methadone, morphine, codeine, 6-monoacetylmorphine, amphetamine, cocaine, benzoylecgonine, paracetamol, olanzapine, mirtazapine, sertraline) were injected into LC-MS/MS system. Produced interferences were considered insignificant if the signal was less than 20% for PG and less than 5% for IS, compared with signal obtained for lower limit of quantification sample (LLOQ).
The linearity, accuracy and precision were evaluated using 8 calibrators and three levels of quality control samples (QC). Calibration curves were constructed with 8 calibration samples using a 1/x2 weighted linear regression of the peak area ratios (PG/IS) vs the plasma concentration. The lower limit of quantification (LLOQ) was defined as the lowest concentration of the calibration curve, while the LoD was set on defined concentration level and at least two out of three MRM transitions were positive. Calibration curves were considered acceptable when the correlation coefficient (r2) was greater than 0.99 and if the bias of the calculated concentrations ranged within ±15% of the nominal concentrations and ±20% for the LLOQ. Accuracy and precision (intra- and inter-day) were considered acceptable if the values obtained for recovery and RSD were ±15% for all QC levels and ±20% for the LLOQ.
Carry over effect (signal remaining from the previously analyzed sample) was tested throughout the analytical sequence, by processing the blank sample after the calibration standard of the highest concentration (ULOQ) and after sample predicted for dilution effect testing (QCDE). The carry-over is considered to be acceptable if the peak area in blank sample injected after high concentration samples was less than 20% and 5% of peak area observed in LLOQ sample at retention time of PG and IS, respectively.
Dilution effect was tested for a sample with the concentration of PG above the range of calibration curve (QCDE–150 mg L−1). Sample was prepared in a duplicate and then diluted using pool plasma in ratio 1:1 in order to adjust concentration to the linearity range of the method. Testing was performed in 5 reruns.
Matrix effect was estimated at two QC levels (low and high) in terms of IS-normalized matrix factor (ISNMF) using peak response ratio (PG/IS) in the presence and absence of matrix. It was considered there is no significant matrix effect if the % RSD of the obtained ISNMF was less than 15%.
Two QC levels (low and high) were used for stability testing of samples. All samples were analyzed in triplicate under various storage conditions. Short-term stability was estimated by comparison of freshly prepared samples with ones kept at ambient temperature for 4 h. Stability in autosampler was estimated by analyzing samples at the beginning and 24 h after storage in autosampler at 10 °C. Stability during frosting (−20 °C) and thawing was estimated after three cycles with the time range between cycles 12–24 h. Long term stability was tested after storage of QC samples for one month at −20 °C. Samples were considered stable under defined storage conditions if the mean concentration of stability samples was in the range 85–115% of the nominal concentration. The stability of the stock solutions of PG and IS was evaluated by testing solutions after two months at 4–8 °C.
Method application
The validated method was applied for the quantification of PG in plasma samples of patients admitted to the NPPC emergency department, with suspected PG overdose. Blood samples were delivered to the toxicology laboratory of NPCC in sterile polypropylene vacuum tubes, 3.5 mL volume, with EDTA as an anticoagulant. Immediately after receipt blood was centrifuged, plasma was separated and samples were processed using validated method. Testing was done with approval of Ethical board of Military Medical Academy, Belgrade, Republic of Serbia.
Results and discussion
Method development
During method development, different extraction procedures were tested to optimize the sample preparation. First, we tried solid-phase extraction with hydrophilic lipophilic balance (HLB) extraction cartridges, however a significant matrix effect and low extraction recovery were observed for pregabalin. Furthermore, protein precipitation was tested with different organic solvents, and the best results in terms of extraction recovery, matrix effect and peak shape obtained with cold acetonitrile. The use of deuterated internal standards is the best option in analytics, however these compounds are expensive and often unavailable to laboratories. We chose gabapentin as an internal standard because it is structurally similar and shares some common features with pregabalin (e.g., physicochemical properties, extraction recovery value). The chromatographic separation was performed on a Waters Acquity UPLC® BEH C18, 1.7 µm, 2.1 × 100 mm column. A mobile phase of acetic acid in water combined with methanol and acetonitrile in various combinations was evaluated. An isocratic elution with mobile phase consisted of 0.5% acetic acid in water and 0.1% acetic acid in acetonitrile gave the best results in terms of retention time and peak shape. Operation at higher column temperature reduces the mobile phase viscosity and allows higher flow rates without significant loss in efficiency and increase in column back pressure. The column temperature was set at 50 °C which provided better peak shape and shorter retention time of both PG and IS. These optimized extraction procedures and chromatographic conditions resulted in a shortened overall analysis time with accurate and precise detection of pregabalin in plasma samples, which met the requirements of urgent toxicological analysis.
Selectivity, linearity, lower limit of quantification and limit of detection
Representative chromatograms of blank sample, LoD, LLOQ and patient sample are shown in Fig. 2 (a, b) and Fig. 3 (a, b). Retention times for PG and IS were 0.67 and 0.65 min, respectively. As shown in blank sample chromatogram, no detectable endogenous interferences in PG and IS retention times were found. Analysis of chromatograms of drugs and substances commonly associated with misuse of PG showed no interference at specific retention times. The method showed linearity over the tested range of PG concentrations (1–100 mg L−1) which was confirmed with high correlation factor (r2 = 0.9965). The mean accuracy and precision for back calculated concentrations of each point from calibration curve met all acceptance criteria. The typical equation of the calibration curve was y = 0.021x−0.0039, where y represents the peak area ratio (PG/IS) and x the nominal plasma concentrations. The lowest calibrator sample that fulfilled validation acceptance criteria was taken as the LLOQ. The sample spiked with PG in concentration of 0.05 mg L−1 was positive for all three MRM transitions and this concentration level was accepted as the LoD of validated method.
Representative chromatograms of: a) blank plasma b) spiked plasma at LoD level
Citation: Acta Chromatographica 36, 2; 10.1556/1326.2023.01104
Representative chromatograms of: a) spiked plasma at LLOQ level b) patient plasma sample
Citation: Acta Chromatographica 36, 2; 10.1556/1326.2023.01104
Accuracy, precision, matrix, carry over effect and dilution integrity
The data for accuracy, intra- and inter-day precision remained within acceptance criteria and data are shown in Table 2. Obtained results from matrix effect testing suggest there is no significant matrix effect on the ionization of PG and IS under this method conditions. Analyzing peak area at retention times of PG and IS in blank sample that is processed after the ULOQ and QCDE showed insignificant carry over effect which is in accordance with validation criteria. The concentration of the diluted sample was within 90–110% of that of the QCDE sample.
Intra-day and inter-day accuracy and precision
| QC sample | Accuracy (Recovery %) | Precision (% RSD) | ||
| Intra-day | Inter-day | Intra-day | Inter-day | |
| LLOQ (1 mg L−1) | 107.8 | 107.4 | 11.99 | 13.84 |
| QCL (2.5 mg L−1) | 101.2 | 100.72 | 5.96 | 7.04 |
| QCM (40 mg L−1) | 99.39 | 97.62 | 1.77 | 1.97 |
| QCH (80 mg L−1) | 98.68 | 98.78 | 1.24 | 1.47 |
Stability
The results presented in Table 3 indicate that PG was stable in human plasma samples under defined conditions of short-term, autosampler, three freeze–thaw cycles, and long-term stability tests. The stock solutions of PG and IS were found to be stable at refrigerator temperature (4–8 °C) for at least two months.
Stability of PG in plasma samples; results are expressed as variation from nominal concentration
| Short-term stability | Autosampler stability | Freeze-thaw stability | Long-term stability | |
| QCL | 2.4 | 1.8 | 3.2 | 4.8 |
| QCH | 0.6 | 0.9 | 1.1 | 2.9 |
Method application
During the study period, validated method was successfully applied to 95 plasma samples of patients admitted to the emergency department of NPCC due to potential PG overdose. Pregabalin was tested positive in 85 cases (89%) and plasma PG concentrations in quantified samples ranged between 1.05 and 76.16 mg L−1, with median value of 5.77 mg L−1. Figure 4 illustrates distribution and number of PG positive cases in different concentration ranges.
Concentration ranges of pregabalin positive cases
Citation: Acta Chromatographica 36, 2; 10.1556/1326.2023.01104
Discussion of validated method
There are numerous reports of pregabalin use and misuse in the literature, but data on pregabalin concentrations in acute poisoning are mostly limited to case reports [4, 5, 7–12]. In this work, we presented the validation of LC-MS/MS method for the quantification of pregabalin in human plasma from patients with suspected overdose of this drug.
We chose protein precipitation as the most convenient method for sample preparation. This method is widely used for the preparation of biological fluids in drug analysis because it is fast, easy to perform and has advantages regarding costs when compared to the other sample preparation techniques [20]. Our method requires a small amount of plasma (200 µL), which is very important in routine laboratory practice, since pregabalin is rarely misused alone in cases of acute poisoning and additional drug analysis is required. Additionally, the small sample amount and minimal injection volume in LC-MS/MS system enable longer use of analytical columns.
By choosing the specified column, mobile phase composition and applying isocratic elution, we achieved good peak resolution and a very short run time. The time of analytical run was 2.5 min, while retention time for PG and IS were 0.67 and 0.65 min, respectively. The total analysis time (sample preparation and analytical run) was less than 20 min, making this method ideal for urgent analysis.
Although tandem mass spectrometry is a specific method, there is still a possibility of false positive results when only one MRM transition is monitored for identification [21]. Since PG is often co-ingested with other drugs and substances, we confirmed its presence with three MRM transitions, to minimize the possibility of false positive results. In addition, we analyzed methanol solutions of various substances that could be potential source of interferences, and no detectable signal was found in the retention times of PG and IS, which makes this method highly selective. Furthermore, no endogenous interferences were observed with this method which is confirmed by the analysis of blank samples (Fig. 2), while filtering and dilution of the prepared sample before injection into the column resulted in a negligible matrix effect.
Compared to other published methods, this method showed precision, accuracy and linearity over a wider range of concentrations (1–100 mg L−1), which was confirmed by a high correlation factor (r2 > 0.99) [13–18]. With this calibration range, no additional dilution steps are required, which are necessary for methods with a narrower measurement range. These features make this method useful for toxicology laboratories where both therapeutic and toxic concentrations are expected. The concentration chosen for LLOQ met all validation criteria and fully satisfy the purpose of the validated method. There was no need to reach a more sensitive point for LLOQ since lower concentrations are not of interest in acute poisoning cases. The LoD was set to a specific concentration value rather than theoretically calculating it and the advantage of this approach is that objective data are used [22]. In our opinion, this approach is more appropriate for toxicology testing. Stability was evaluated under various conditions and the data showed that the samples can be stored at refrigerator temperature if the analysis cannot be performed immediately, however freezing the samples (−20 °C) is recommended for a longer period of time.
We applied this method to quantify PG in 95 plasma samples from patients admitted to the NPCC emergency department. Eighty-nine percent of the cases tested positive for PG, while in 11% of the cases, the concentrations were below LoD and it was concluded that some other reason was the cause of hospitalization. One of analyzed samples was obtained from a 45-year-old female patient presented to the NPCC emergency department approximately 3.5 h after ingestion and heteroanamnestic data indicated that she has been abusing large doses of pregabalin for a longer period of time. This patient was found to have a pregabalin plasma concentration of 76.16 mg L−1, which is, to the best of our knowledge, the highest reported concentration in the literature to date. Pregabalin plasma concentrations were determined three times during the patient's hospitalization: on admission, 12 and 36 h after admission as shown in Fig. 5.
Pregabalin plasma level during the hospitalization
Citation: Acta Chromatographica 36, 2; 10.1556/1326.2023.01104
The application of this method has shown, when applied in routine clinical practice, it can be very helpful to all clinicians treating acutely poisoned patients, especially in situations where the cause of the poisoning is unknown. The type and concentration of the ingested substance represent valuable information that can determine further direction of diagnostics and treatment of patients presenting to the emergency department. The use of the developed method in clinical toxicology laboratories can be of scientific importance as well, since there is not enough data on the concentration of pregabalin in acute poisoning cases.
Conclusion
The increasing use and misuse of pregabalin in the last decade makes this drug appear more frequently as a cause of acute poisoning and indicate this drug should be included in toxicological screening and routine laboratory testing. A rapid and reliable method for quantification of pregabalin with a simple sample preparation step was validated. The method was successfully applied for quantification of PG in acutely poisoned patients. A wide range of concentrations was observed during the study period and we report the highest measured pregabalin concentration in plasma of an acutely poisoned patient. The statistical data demonstrated that the validated method is precise, accurate, highly selective and linear over a wide concentration range making it suitable for use in clinical toxicology laboratories.
Acknowledgments
This work was supported by University of Defence, Republic of Serbia (Award Number: MF VMA 06/22–24).
References
- 2.
Evoy, K. E.; Peckham, A. M.; Covvey, J. R.; Tidgewell, K. J. J. Clin. Pharmacol. 2021, 61(Suppl. 2), S89–S99.
- 4.↑
Slocum, G. W.; Schult, R. F.; Gorodetsky, R. M.; Wiegand, T. J.; Kamali, M.; Acquisto, N. M. Toxicol. Commun. 2018, 2(1), 19–20.
- 6.
Elliott, S. P.; Burke, T.; Smith, C. J. Forensic Sci. 2017, 62(1), 169–173.
- 7.↑
Wood, D. M.; Berry, D. J.; Glover, G.; Eastwood, J.; Dargan, I. P. J. Med. Toxicol. 2010, 6(4), 435–437.
- 8.↑
Cairns, R.; Schaffer, A. L.; Ryan, N.; Pearson, P. A.; Buckley, N. A. Addiction 2019, 114(6), 1026–1034.
- 9.↑
Daly, C.; Griffin, E.; Ashcroft, D. M.; Webb, R. T.; Perry, I. J.; Arensman, E. Clin. Drug Investig. 2018, 38(4), 373–380.
- 10.
Dufayet, L.; Care, W.; Deheul, S.; Laborde-Casterot, H.; Nisse, P.; Langrand, J.; Vodovar, D. Clin. Toxicol. 2021, 59(11), 1027–1030.
- 11.
Rietjens, S. J.; Sikma, M. A.; Hunault, C. C.; De Lange, D. W.; Hondenbrink, L. Br. J. Clin. Pharmacol. 2022 Mar, 88(3), 1288–1297.
- 12.↑
Isoardi, K. Z.; Polkinghorne, G.; Harris, K.; Isbister, G. K. Br. J. Clin. Pharmacol. 2020 Dec, 86(12), 2435–2440.
- 13.↑
Dwivedi, J.; Namdev, K. K.; Chilkoti, D. C.; Verma, S.; Sharma, S. Ther. Drug Monit. 2018, 40(5), 610–619.
- 14.
Vaidya, V. V.; Yetal, S. M.; Roy, S. M. N.; Gomes, N. A.; Joshi, S. S. Chromatographia 2007, 66(11–12), 925–928.
- 15.
Shah, G. R.; Ghosh, C.; Thaker, B. T. J. Adv. Pharm. Technol. Res. 2010 Jul, 1(3), 354–357.
- 16.↑
Martinc, B.; Roškar, R.; Grabnar, I.; Vovk, T. J. Chromatogr. B Analyt Technol. Biomed. Life Sci. 2014, 962, 82–88.
- 17.↑
De La Vega, H.; Fox, K.; Pardi, J.; Santiago-Tirado, W.; Cooper, G. J. Anal. Toxicol. 2019, 43(9), 696–702.
- 18.↑
Nahar, L.; Smith, A.; Patel, R.; Andrews, R.; Paterson, S. J. Anal. Toxicol. 2017, 41(5), 441–450.
- 19.↑
European Medicines Agency Guideline on bioanalytical method validation, 2011 Jul. http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2011/08/WC500109686.pdf (accessed September, 2022).
