Introduction
The analytical methods for screening and confirmatory test are essential for identifying substances of abuse which conducts the credibility of testimony. The preliminary test used to identify the targets from urine samples over their threshold values is known as screening test.1 The confirmatory test recognized as secondary test is performed by contrasting the positive result from screening test.1–4
In forensic analysis, immunoassays are used for drug screening because of their high sensitivity and ease of use. However, limitations such as constantly changing new psychoactive substances (NPS) and the inability to distinguish between false negative/false positive results limit their advantages over NPS analysis.5–7 In contrast, chromatographic techniques such as gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-tandem mass spectrometry (LC-MS/MS) are efficient tools for screening or confirmatory tests and are widely used in forensic science and clinical toxicology applications for their excellent sensitivity and specificity.
The adoption of analytical methods that are sensitive, specific, and conclusive is essential for the testing of substances of abuse, while the methods applied meet strict criteria for sample integrity, legal documentation, validation of targets identified and confirmation of the results.1 It should demonstrate equivalent effectiveness in target identification no matter what method is used in confirmatory test. In this context, statistical measures are an efficient tool to assess the compatibility from analytical methods. When assessing the degree of agreement/disagreement (independence/difference) between assessors of binary variables (positive/negative or yes/no), Cohen's Kappa Coefficient and McNemar’s test are two metrics used most often.8–12
With the widespread emergence of NPS in recent decades, phenethylamines (or PEAs for short) have become one of the groups abused most which PEAs, along with other stimulants, make up the majority of stimulants. By 2023, PEA and other stimulants account for the majority of the reported NPS, with 398 items in total.17 Physical and psychological effects associated with PEAs include high blood pressure, hyperthermia, hallucinations, agitation, aggression, dissociation, attention deficit hyperactivity disorder, liver and kidney failure, and overdose can lead to severe poisoning and even death.18
Materials and methods
Reagents
Seventy-six standards and four internal standards (IS) were obtained from six vendors. The detailed information is shown in Table 1. Among all, the standards obtained from GreenChem were commission-synthesized products associated with Taiwan Food and Drug Administration (TFDA) and have been identically assessed by NMR, HRMS, and FT-IR, whereas the purity for all items are above 98%. Ammonium hydroxide and methanol were analytical grade purchased from J.T. Baker (Phillipsburg, NJ, USA). Ethyl acetate and isopropanol were analytical grade purchased from Avantor (Phillipsburg, NJ, USA). Sodium hydroxide, dipotassium phosphate, monopotassium phosphate, and glacial acetic acid were analytical grade purchased from E. Merck (Darmstadt, Germany). The derivatization agents heptafluorobutyric acid (HFBA) and pentafluoropropionic anhydride (PFPA) were analytical grade and purchased from Sigma-Aldrich (St. Louis, MO, USA). Normal human urine, certified drug negative containing 0.01% sodium azide, was purchased from UTAK Laboratories, Inc. (Santa Clarita, CA, USA). The solid-phase extraction (SPE) cartridge Bond Elut SPEC DAU cartridge (15 mg, 3 mL) was purchased from Agilent Technologies (Santa Clara, CA, USA). A total of 20 authentic urine samples were provided by 6 different analytical laboratories authorized by local law enforcement agencies of Taiwan. The authentic urine samples were stored at -20ºC and acclimated to room temperature before analysis. The sampling of urine specimens in this study followed the regulations of the Ministry of Health and Welfare, Taiwan.
Table 1. MRM parameters for 76 targets and 4 IS of PEAs by GC/MS
a. GreenChem Corporation (Taichung, Taiwan)
b. Cayman Chemical (Ann Arbor, MI, USA)
c. LGC Limited (Teddington, UK)
d. Sigma-Aldrich Corporation (St. Louis, MO, USA)
e. Cerilliant Corporation (Austin, TX, USA), methanolic solutions (1 mg/mL)
*Quantifier ion
Instrumentation
The GC/MS analysis was performed on an Agilent 8890 gas chromatograph (GC) system (Agilent Technologies, Santa Clara, CA, USA) equipped with an Agilent 7693A autosampler and connected to an Agilent 5977C mass selective detector (MSD). The chromatographic analysis was performed using an Agilent HP-5MS capillary column (30 m × 0.25 mm i.d., 0.25 μm). Helium was used as carrier gas at a flow rate of 0.9 mL/min. The target PEAs were divided into two groups (see Table 1) and the parameters of GC/MS analysis are as follows. Group I analytes: temperature program of GC started at the initial temperature of 120ºC and held for 0.5 min, increased to 180ºC at 15ºC/min, increased to 200ºC at 3ºC/min, and then increased to 285ºC at 30ºC/min and hold for 3 min, whereas the total run time was within 17 min; the injector was set at 260ºC as the inlet temperature under splitless mode with injection volume of 2 μL; the ion source was set at 230º C and interface temperature of 280ºC under electron ionization (EI) mode at 70 eV. Group II analytes: temperature program of GC started at the initial temperature of 100o C and held for 1 min, increased to 265ºC at 10ºC/min, and then increased to 300ºC at 20ºC/min and hold for 5 min, whereas the total run time was within 24.3 min; the injector was set at 285ºC as the inlet temperature under splitless mode with injection volume of 1.5 μL; the ion source was set at 230ºC and interface temperature of 300ºC under EI mode at 70 eV. The data acquisition was carried out under total ion monitoring mode in the scanning range of 30 to 550 m/z to determine the retention time and characteristic mass fragments for each analyte so that the selected ion monitoring (SIM) mode could be applied for detecting the targets in the sample. The instrumentation as well the preparation of standard solutions and urine samples for LC-MS/MS analysis were in accordance with those published in a previous study and are presented as Supplementary Material 1 (SM 1).19
Preparation of standard solutions
The 76 PEA standards and 4 IS for GC/MS analysis were separately dissolved in methanol to the concentration of 1.0 mg/mL as stock solutions, whereas the working solutions were prepared by diluting the stock solutions with methanol to the concentration of 10 μg/mL. Before use, all stock and working solutions were stored at -20ºC and allowed to acclimate to room temperature before use.
Pretreatment of urine samples
The pretreatment of the urine samples was carried out using a Bond Elut SPEC DAU SPE cartridge on a Biotage Rapid Trace+ SPE Workstation system (Vimpelgatan, Switzerland). The sample solution was prepared by mixing 2 mL of the homogenized urine sample, 20 μL of the IS solution (equivalent 100 ng/mL), and 1 mL of 0.1 M phosphate buffer solution (pH 6). The setting of automatic SPE is as follows: Condition, methanol, 1 mL, 12 mL/min, to waste; condition, water, 1 mL, 12 mL/min, to waste; load, sample solution, 2 mL, 1.8 mL/min, to waste; purge-cannula, water, 5 mL, 42 mL/min, to cannula; rinse, 0.1 M acetic acid, 1 mL, 3 mL/min, to waste; purge-cannula, water, 5 mL, 42 mL/min, to cannula; rinse, methanol, 1 mL, 3 mL/min, to waste; dry, nitrogen gas, 2 bar, 4 min; collect, elution buffer (dichloromethane/isopropanol/25% ammonia, 80:20:2, v/v/v), 2 mL, 4.8 mL/min, to fraction; purge-cannula, methanol, 5 mL, 42 mL/min, to cannula. After the extraction, collect the eluent and transfer to a micro-reaction vial. Evaporate the eluent to dryness by gently flushing with a steam of nitrogen gas at 40ºC and keep the residue. The residue was then derivatized by the following procedure: dissolve the residue with 50 μL of HFBA (or PFPA) and 50 μL of ethyl acetate. Cap the micro-reaction vial, vortex the residue solution to homogeneity, and react at 90ºC for 15 min. Cool the solution to room temperature and evaporate with nitrogen gas at 40ºC until dry. Dissolve the derivatized residue with 100 μL of ethyl acetate and then apply to GC/MS analysis. Drug-free urine (DFU) consisted of urine but without spiking any target analyte was used as the blank matrix and negative control.
The carryover was assessed by injecting DFU (n = 3) immediately after analysis of the urine samples spiked with the concentration of the highest calibrator (800/1000/2000 ng/mL) and the response ratio of residual to LLOQ was then calculated. The limit of carryover is 20% which is considered ignorable on target analysis. To evaluate the selectivity, different DFU samples (n = 20) were analyzed and no interfering peaks appeared during the target analysis i.e. the absence of evident interfering signals from the matrix at retention times nearby the characteristic ions.
Validation of GC/MS method
In the practice of forensic analysis, the validation is assessed to ensure the reliability and feasibility of the proposed method. In this study, the validation was performed in accordance with the guidelines “Working Group for Forensic Toxicology Standard Practices (SWGTOX) for Method Validation in Forensic Toxicology” and “Guidance for the Validation of Analytical Methodology and Calibration of Equipment used for Testing of Illicit Drugs in Seized Materials and Biological Specimens”.20,21 The methods were validated in terms of carryover, selectivity, linearity, sensitivity, extraction recovery, accuracy, and precision. To assess the linearity, the standard solutions of 76 PEAs (n = 3) were analyzed at six concentration levels ranging from 50 to 2000 ng/mL (IS of 100 ng/mL included). The calibration curve was plotted from the peak area ratio of standard/IS versus the concentration of standard using the least-square method, and the acceptable correlation coefficient (r) value was greater than 0.995. Quantification of target analytes was determined using a calibration curve based on the IS method within an established linear range. The acceptable ranges for qualitative and quantitative determination were as follows: relative ion ratio > 50%, RSD ± 20%; relative ion ratio 20–50%, RSD ± 25%; relative ion ratio 10–20%, RSD ± 30%; relative ion ratio ≤ 10%, RSD ± 50%. The acceptable retention time deviation of target was ± 0.2 min. Sensitivity was evaluated in terms of the LOD and LLOQ (lower limit of qualification) defined by the estimated signal-to-noise ratios (S/N) of 3 and 10, respectively. Six replicates (n = 6) were used for each analyte. The quality control (QC) was performed on analyte spiked urine samples. The SPE recovery, intra-day and inter-day precision (% CV) and accuracy (%) were evaluated in triplicate over five different runs at three concentration levels from low to high (100–1000 ng/mL). The acceptable range of recovery, precision and accuracy are 80%–120% and ± 20%, respectively. To extract the detailed explanation above, a flowchart was drawn to express it as shown in Figure 1.
Figure 1. Flowchart for the development of the proposed methods.
Experimental
Determination of target analyte
Twenty authentic urine samples were analyzed applying the presented GC/MS methods. The established LC-MS/MS is also applied as a comparison. The cutoffs for positivity of target analyte were based on the legal threshold or lower limit of quantification (LLOQ) of each item. Specifically, the legal thresholds for five items in Taiwan are as follows: 500 ng/mL for amphetamine; 500 ng/mL for methamphetamine while amphetamine is detected of above 100 ng/mL; 500 ng/mL for MDMA. Meanwhile, it is identified as MDMA positive when MDMA and MDA are detected simultaneously with individual amount less than 500 ng/mL while the total amount is above 500 ng/mL; 500 ng/mL for MDA; 500 ng/mL for MDEA. The overall identification of a sample is defined as positive if one or more target analytes are detected above their cutoff concentration. Conversely, a negative is defined as long as the target is not detected or is below the cutoff concentration. The above identification must meet the acceptance criteria for relative ion intensity (see Table S1 in supplemental material). Furthermore, Cohen's kappa coefficient and McNemar’s test were evaluated on the analysis results obtained by both methods to examine the consistency and difference between the results.
Statistical analysis
The consistency and difference of diagnostic tests (binary outcome, positive/negative or +/–) for urinalysis of PEAs in authentic samples applying GC/MS and LC-MS/MS was evaluated by the Cohen’s kappa coefficient (κ) and McNemar’s test (chi-square x2) with 2 × 2 contingency tables. The data a and d in the matrix represent the number of agreements (positive samples of target analyte), whereas b and c represent the number of disagreements (negative samples of target analyte). The data n represents the total number of agreements and disagreements (i.e. a + b + c + d). The definition of κ is as follows:
\(\begin{align}\kappa=\frac{p_{o}-p_{e}}{1-p_{e}}\end{align}\)
where po is the relative observed agreement among raters and pe is the hypothetical probability of chance agreement among raters, described as follows:
\(\begin{align}p_{o}=\frac{a+d}{n}\end{align}\)
\(\begin{align}\begin{aligned} p_{e} & =\frac{(a+b)(a+c)}{n} \times \frac{(c+d)(b+d)}{n} \\ & =\frac{(a+b)(a+c)+(c+d)(b+d)}{n^{2}}\end{aligned}\end{align}\)
The x2 of McNemar’s test is conducted as follows:
\(\begin{align}\chi^{2}=\frac{(b-c)^{2}}{b+c}\end{align}\)
Meanwhile, if the total number n is less than 40, the Yates continuity correction is applied for compensating the deviations from the theoretical probability distribution.22 The corrected x2 is defined as follows:
\(\begin{align}\chi^{2}=\frac{(|b-c|-1)^{2}}{b+c}\end{align}\)
The McNemar’s test was assessed by the online calculator.23
Results and discussion
Method development
When analyzing drugs and substances of abuse, the selectivity of GC/MS method influences the identification of target analytes. Therefore, various measures have been taken to improve the performance of chromatographic analysis, including derivatization, complexation, and ion-pair formation. Of the reactions listed above, the derivatization is the most commonly used for GC/MS analysis.24 The derivatization process was introduced in this study to improve the detection and resolution of target PEA analysis. In the pretest, various derivatization reagents, including N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), HFBA, N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA), PFPA, and trifluoroacetic anhydride (TFAA) were tested. The scheme chart of derivatization of PEAs was expressed as Figure 2. Among the derivatization agents tested, HFBA and PFPA were observed to have the main effect on improving GC/MS detectability in the analysis of target PEA (data not shown). The target PEAs were divided into two groups according to the adapted derivatization agents. The derivatized monitoring ions were collected as shown in Table 1, and the total ion chromatogram (TIC) of PEAs in urine spiked at 500 ng/mL under SIM mode is shown in Figure 3. The chromatographic separation was achieved within 17 and 23 minutes in separate runs. From Figure 3, it is observed that the SIM mode not only eliminates interference from the urine matrix, but also demonstrate processing efficiency in multi-target analysis by monitoring designated ions. Based on the above advantages, SIM mode has emerged as an efficient technique to analyze multiple PEA analogues especially NPS with the same frame structure, by monitoring fragment ions in a specific detection window with fixed retention times. This technique allows the identification and quantification of PEA-type substances with high accuracy and specificity. Good separation of the various analogues was achieved by applying appropriate SIM program settings determined by retention time. The cross contribution was further investigated through subsequent method validation to ensure limited impact on the analysis and sufficient specificity for quantification of analytes with the same mass pattern.
Figure 2. Scheme chart of derivatization applying HFBA and PFPA for two groups of PEAs. (A) Group I, take 25H-NBOMe as example; (B) Group II, take 25T4-NBOMe as example.
Figure 3. Total ion chromatograms (TIC) of blank urine, 76 targets, and 4 IS at 500 ng/mL in urine, ordered by item number of analytes from Table 1: (A) HFBA-derivatized blank urine under scanning and SIM modes; (B) and (C), HFBA-derivatized Group I PEAs under SIM mode; (D) PFPA-derivatized blank urine under scanning and SIM modes; (E) PFPA-derivatized Group II PEAs under SIM mode.
Method validation
In forensic analysis, the reliability and credibility of an analytical process influence court decisions; therefore, the method validation is indispensable. The results showed that the carryovers for 76 PEAs were all less than 0.01% (data not shown), and no interference was seen in the subsequent analysis of the samples. In the evaluation of selectivity, no obvious impact from the above elements have been found and indicated the proposed methods as a selective technique for analyzing PEAs. Subsequently, the linearity was assessed through setting up calibration curves with the concentration range of 25 to 2000 ng/mL, based on the correlation coefficient, r. The r values of the calibration curve for all analytes was above 0.995, and the determined LOD and LLOQ were 10–200 ng/mL and 25–200 ng/mL, respectively (see Table 2, alone with LC-MS/MS result of Table S2 in supplemental material). In particular, the targets lisdexamfetamine, 2C-T-2, 2C-T-4, 2C-T-7, 25T2-NBOMe, 25T4-NBOMe, and 25T7-NBOMe showed higher LOD and LLOQ which can be attributed to their low volatility. The obtained SPE extraction recovery was 80.2%–111.6% for all analytes, corresponding to the criteria of 80%–120%. To assess accuracy and precision, intra-day and inter-day experiments were performed using triplicate samples at three concentration levels. The ion ratio was over 50%, and the accuracy ranged from -12.5% to 18.9%. These values are within ±20% according to the respective criteria and are satisfactory. The results of extraction recovery, precision, and accuracy are shown as Table S3 in supplemental material.
Table 2. Linearity, LOD, and LLOQ of 76 targets and 4 IS of PEAs for GC/MSa
Application
The goal of this study was to establish a set of derivatized GC/MS methods that could be used to analyze authentic urine samples and identify 76 PEAs. The current methods were applied to the analysis of authentic urine samples that were supplied by the accredited local analytical laboratories. A total of 20 samples were examined, and the results of urinalysis for targets above the LLOQs are shown in Table 3. Thirteen out of 20 samples tested positive for five target PEAs including N,N-DMA, PMMA, 4-CA, amphetamine, and methamphetamine. The detected concentration levels ranged from 94.5 ng/mL to 124.0 µg/mL.
Table 3. Targets detected above LLOQ from authentic urine samples applying GC/MS
Meanwhile, the authentic samples were also analyzed using the previously established LC-MS/MS method, to ensure whether there was equivalence between the two techniques in PEA monitoring.19 The findings of the statistical assessments are shown as Table 4 (comparative result as Table S4 in supplementary material). In the LC-MS/MS analysis, a perfect agreement (kappa coefficient value = 1) was observed when the dilution rates of urine samples were 50× for N,N-DMA, PMMA, and 4-CA, and 500× for amphetamine and methamphetamine, indicating a consistent result. Insufficient dilution rate (50×) of urine samples (in terms of amphetamine and methamphetamine) could result in invalid LC-MS/MS identification, making it impossible to reduce the matrix effect to meet the criterion for ion ratio of qualifiers, and hence a dilution rate of 500x was introduced to ensure consistency of the results.
Table 4. Statistical analysis for target determination applying LC-MS/MS and GC/MS methods
The bulk of the analytical methods for identifying multiple PEAs in urine was established using LC-MS/MS.25−27 In contrast, the GC/MS is generally considered to be less advantageous than the LC-MS/MS in terms of sensitivity, selectivity, and adaptability. Nonetheless, this study established the effective GC/MS methods with different derivatization process in identifying multiple PEAs. The current methods were suggested as an effective tool in screening/confirmatory testing as LC-MS/MS for identifying multiple PEAs, according to the result from statistical assess.
Conclusions
This study presents a set of derivatized GC/MS methods for detection of 76 target PEAs in urine. The proposed methods were applied to the analysis of authentic urine samples, of which 13 samples were detected positive for six identified targets in total out of 20 samples. The results of the urinalysis were statistically analyzed and compared with those obtained using the LC-MS/MS method that was previously established. The statistical results suggested that GC/MS was as efficient as LC-MS/MS for PEA urinalysis. Furthermore, this study provided a systematic technique for preliminary or confirmatory testing of PEAs, and the methods can be used to identify drug components in urine in the future.
Acknowledgments
This work was supported by the Taiwan Food and Drug Administration, Ministry of Health and Welfare, R.O.C.
References
- Lum, G.; Mushlin, B. Lab. Med., 2004, 35, 368, DOI: https://doi.org/10.1309/QHJCKA4235EGPEGF
- Akhgari, M.; Bahmanabadi, L.; Iravani, F. S. S.; Jokar, F. Toxicol. Anal. et Clin. 2021, 33, 109, DOI: https://doi.org/10.1016/j.toxac.2020.10.089
- Graziano, S.; Anzillotti, L.; Mannocchi, G.; Pichini, S.; Busardo, F. P. J. Pharm. Biomed. Anal. 2019, 163, 170, DOI: https://doi.org/10.1016/j.jpba.2018.10.011
- Favretto, D.; Pascali, J. P.; Tagliaro, F. J. Chromatogr. A 2013, 1287, 84, DOI: https://doi.org/10.1016/j.chroma.2012.12.049
- Staack, R. F.; Fritschi, G.; Maurer, H. H. J. Chromatogr. B 2002, 773, 35, DOI: https://doi.org/10.1016/S1570-0232(01)00619-5
- Kraemer, T.; Wennig, R.; Maurer, H. H. J. Anal. Toxicol. 2001, 25, 333, DOI: https://doi.org/10.1093/jat/25.5.333.
- Hawks, R. L.; Chiang, C. N. Urine testing for drugs of abuse. Research monograph series 73. National Institute on Drug Abuse. U.S. Department of Health and Human Services 1986. https://archives.nida.nih.gov/sites/default/files/monograph73.pdf
- Casati, S.; Angeli, I.; Ravelli, A.; Del Fabbro, M.; Minoli, M.; Orioli, M. Forensic Sci. Int. 2019, 304, 109951, DOI: https://doi.org/10.1016/j.forsciint.2019.109951
- Warrens, M. J. J. Psychol. Psychother. 2015, 5, 1000197, DOI: https://doi.org/10.4172/2161-0487.1000197
- Feuerman, M.; Miller, A. R.; J. Eval. Clin. Pract. 2008, 14, 930, DOI: https://doi.org/10.1111/j.1365-2753.2008.00984.x
- Pembury Smith, M. Q. R.; Ruxton, G. D. Behav. Ecol. Sociobiol. 2020, 74, 133, DOI: https://doi.org/10.1007/s00265-020-02916-y
- ElMehy, A.; Wood, D.; El Madah, E.; Hassan, N.; Saad, K.; Dargan, P. Asia Pac. J. Med. Toxicol. 2022, 11, 4, DOI: https://doi.org/10.22038/apjmt.2022.19912
- Landis, J. R.; Koch, G. G. Biometrics 1977, 33, 159, DOI: https://doi.org/10.2307/2529310
- Ti, L.; Tobias, S.; Lysyshyn, M.;, Laing, R.; Nosova, E.; Choi, J. C.; Arredondo, J.; McCrae, K.; Tupper, K.; Wood, E. Drug Alcohol Depend. 2020, 212, 108006, DOI: https://doi.org/10.1016/j.drugalcdep.2020.108006
- Nwala, G. C.; Ibeneme, C. A.; Ojinnaka, N. C. Int. j. sci. res. 2021, 10, 930, DOI: https://www.ijsr.net/getabstract.php?paperid=SR21311103801 103801
- Gjerde, H.; Gjersing, L.; Furuhaugen, H.; BrettevilleJensen, A. L. Subst. Use Misuse 2019, 1337, DOI: https://doi.org/10.1080/10826084.2019.1580295
- UNODC Early Warning Advisory on NPS - Summary Dashboard, assessed May 2024. Available at:? https://www.unodc.org/LSS/Page/NPS/DataVisualisations
- Mercolini, L. Chapter 20 - New Psychoactive Substances: An Overview, In A. Dasgupta (2nd Ed.), Critical Issues in Alcohol and Drugs of Abuse Testing, Academic Press, 2019, 247. DOI: https://doi.org/10.1016/B978-0-12-815607-0.00020-4
- Fan, S. Y.; Zang, C. Z.; Shih, P. H.; Ko, Y. C.; Hsu, Y. H.; Lin, M. C. Tseng, S. H.; Wang, D. Y. Forensic Sci. Int. 2021, 325, 110884, DOI: https://doi.org/10.1016/j.forsciint.2021.110884
- Scientific Working Group for Forensic Toxicology. J. Anal. Toxicol. 2013, 37, 452, DOI: https://doi.org/10.1093/jat/bkt054
- Laboratory and Scientific Section of the United Nations Office on Drugs and Crime, Guidance for the Validation of Analytical Methodology and Calibration of Equipment used for Testing of Illicit Drugs in Seized Materials and Biological Specimens, United Nations Office on Drugs and Crime 2009. Available at: https://www.unodc.org/documents/scientific/validation_E.pdf
- Giannini, E. H.; Design, measurement, and analysis of clinical investigations. In Textbook of Pediatric Rheumatology 2005, 142, DOI: https://doi.org/10.1016/B978-1-4160-0246-8.50012-7
- Omnicalculator online https://www.omnicalculator.com/statistics/mcnemars-test
- He, Y. Liquid-based microextraction techniques for environmental analysis. In Comprehensive Sampling and Sample Preparation 2012, 835, DOI: https://doi.org/10.1016/B978-0-12-381373-2.00116-2
- Wagmann, L.; Maurer, H. H. Exp. Pharmacol. 2018, 252, DOI: https://doi.org/10.1007/164_2017_83
- Simao, A. Y.; Antunes, M.; Marques, H.; Rosado, T.; Soares, S.; Goncalves, J.; Mario, B.; Maristela, A.; Gallardo, E. Bioanalysis 2020, 12, DOI: https://doi.org/10.4155/bio-2020-0148
- Di Trana, A.; Mannocchi, G.; Pirani, F.; La Maida, N.; Gottardi, M.; Pichini, S.; Busardo, F. P. J. Anal. Toxicol. 2020, 44, DOI: https://doi.org/10.1093/jat/bkaa103