An automated sample preparation for detection of 72 doping-related substances

Transcription

1 Research article Drug Testing Received: 30 April 2013 Revised: 8 August 2013 Accepted: 13 August 2013 Published online in Wiley Online Library ( DOI /dta.1538 An automated sample preparation for detection of 72 doping-related substances Darío Cuervo,* Pablo Díaz-Rodríguez and Jesús Muñoz-Guerra Automation of sample preparation procedures in a doping control laboratory is of great interest due to the large number of samples that have to be analyzed, especially in large events where a high throughput protocol is required to process samples over 24 h. The automation of such protocols requires specific equipment capable of carrying out the diverse mechanical tasks required for accomplishing these analytical methodologies, which include pipetting, shaking, heating, or crimping. An automated sample preparation procedure for the determination of doping-related substances by gas chromatography mass spectrometry (GC-MS) and gas chromatography-tandem mass spectrometry (GC-MS/MS) analysis, including enzymatic hydrolysis, liquid-phase extraction and derivatization steps, was developed by using an automated liquid handling system. This paper presents a description of the equipment, together with the validation data for 72 doping-related compounds including extraction efficiency, evaluation of carry-over, interferences, and robustness. Validation was approached as a comparison between the results obtained using the manual protocol and the transferred automated one. The described methodology can be applied for sample preparation in routine anti-doping analysis with high sample throughput and suitable performance. Copyright 2013 John Wiley & Sons, Ltd. Keywords: doping control analysis; urine analysis; automated sample preparation; liquid-liquid extraction Introduction Sample preparation in a doping control laboratory is laborious work due to both the large number of samples to be treated and the variety of compounds to be monitored. The diverse structural and chemical characteristics of the analytes included on the List of Prohibited Substances published annually by the World Anti-Doping Agency (WADA), [1] entails different sample preparation procedures in order to purify, concentrate and adequate the samples for instrumental analysis by specific techniques, including gas chromatography mass spectrometry (GC-MS), gas chromatography-tandem mass spectrometry (GC- MS/MS) or liquid chromatography-tandem mass spectrometry (LC-MS/MS). The automation of these sample preparation procedures is of great interest considering the savings in terms of time, security inside the laboratory and, eventually, the quality of the results obtained. Automation of sample preparation protocols needs specific equipment capable of carrying out the diverse mechanical tasks required for accomplishing the different analytical methodologies involved. [2,3] Several studies have been undertaken in the past by this laboratory [4,5] and others [6 11] regarding the automation of drug analysis in urine or blood using different platforms. Given that the detection critteria for the different compounds included on the WADA List of Prohibited Substances and Methods in Sport are more and more challenging, particularly in terms of sensitivity, new solutions for automation need to be developed and investigated. Our laboratory has recently acquired an automated pipetting system for liquid-liquid extraction from Zinsser Analytic GmbH (Frankfurt, Germany), [12] with further capabilities including shaking, heating, or crimping. In brief, the system was specifically designed for treating up to 96 samples in a sample preparation protocol which includes several heating and extraction steps. As a first step, we focused on the automation of the method implemented in our laboratory for the determination of anabolic agents, narcotics, anti-estrogenic substances, cannabinoids, diuretics, and stimulants in urine by GC-MS and GC-MS/MS analysis, the sample preparation procedure of which includes enzymatic hydrolysis, liquid-liquid extraction, and derivatization steps. The aim was to implement an automated sample preparation protocol able to carry out the whole scope of the method without loss of performance. The paper discusses the validation of the automated sample preparation method, approached as a comparison between both protocols. Validation data including extraction efficiency and evaluation of interferences and carry over effects are also presented. Experimental Chemicals, reagents, and materials All solvents and reagents used were of analytical grade. Tertbutyl methyl ether (TBME) and methanol were from Merck (Darmstadt, Germany). Potassium dihydrogen phosphate (KH 2 PO 4 ), sodium hydrogen carbonate (NaHCO 3 )andpotassium carbonate (K 2 CO 3 ) were from Scharlab (Sentmenat, Spain). Disodium hydrogen phosphate dihydrate (Na 2 HPO 4 2 H 2 0) was from Panreac (Castellar del Vallès, Spain). N-Methyl- N-(trimethylsilyl) trifluoroacetamide (MSTFA) was from * Correspondence to: Darío Cuervo, PhD, Laboratorio de Control del Dopaje, Agencia Española de Protección de la Salud en el Deporte, Ministerio de Educación, Cultura y Deporte, Gobierno de España, c/ Pintor El Greco s/n, 28040, Madrid, Spain. dcuervo@aea.gob.es Laboratorio de Control del Dopaje, Agencia Española de Protección de la Salud en el Deporte, Ministerio de Educación, Cultura y Deporte, Gobierno de España, c/ Pintor El Greco s/n, 28040, Madrid, Spain 1 Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd.

2 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra 2 Macheray-Nagel (Düren, Germany). 1,4-Dithioerythritol and ammonium iodide were from Sigma-Aldrich (St. Louis, MO, USA). Β-Glucuronidase from E. coli K12 was from Roche Diagnostics GmbH (Mannheim, Germany). Internal standard (ISTD) and standards used for process control: methyltestosterone was from Sigma-Aldrich (St Louis, MO, USA), androsterone glucuronide-d 4 and etiocholanolone-d 5 were from NMI (Pymble, Australia), and timolol was from USP (Rockville, MD, USA). Commercial solid standards: the systematic names of each compound are listed in Table 1. Nandrolone M1, nandrolone M2, stanozolol M1, stanozolol M2, 4OH-testosterone, 5α and 5βmethyltestosterone, metandienone M1, M2, M3 and M4, bolasterone M1, boldenone M1, calusterone, carphedon, cyclofenil M2, clostebol M1, danazol M2, drostanolone M1, oxandrolone M1, trenbolone M1, fluoxymesterone M1 and M2, furazabol M1, letrozole M1, mesterolone M1, metenolone M1, methyldienolone, methyl-1- testosterone, 6-monoacetylmorphine, norboletone M1 and M2, noretandrolone M1 and M2, dehydrochlormethyltestosterone M1, tamoxifen M1, zeranol and zeranol M1 were from NMI (Pymble, Australia). Tibolone M1 and M2, and bromantane M1 were from Atlanchim Pharma (Nantes, France). Tibolone M3 and zilpaterol were from TRC (Toronto, Canada). Androstatrienedione, clenbuterol and danazol were from Sigma-Aldrich (St Louis, MO, USA). Aminoglutethimide was from Alltech (State College, PA, USA). Bolasterone, boldenone, danazol M1, estradienedione and mibolerone were from Steraloids (Newport, RI, USA). Buprenorfine and parahydroxyamphetamine were from USP (Rockville, MD, USA). Spironolactone was from European Pharmacopoeia (Strasbourg, France). Estradienedione M1 was from AK Scientific (Union City, CA, USA). Furazabol was supplied free of charge by a pharmaceutical company. Metasterone M1 was obtained from the World Association of Anti-Doping Scientists (WAADS). Commercial standards in solution: ampoules of codeine (1.0 mg/ml), fluoxymesterone (1.0 mg/ml), hydromorphone (1.0 mg/ml), mesterolone (1.0 mg/ml), metenolone (1.0 mg/ml), morphine (1.0 mg/ml), buprenorphine M1 (1.0 mg/ml), oxycodone (1.0 mg/ml), oxymorphone (1.0 mg/ml), pentazocine (1.0 mg/ml) and THC-9-COOH (0.1 mg/ml) were obtained from Cerilliant (Round Rock, TX, USA). Ampoules of hydrocodone (1.0 mg/ml), oxandrolone (1.0 mg/ml) and oxymesterone (1.0 mg/ml) were from Alltech (State College, PA, USA). Preparation of solutions The phosphate buffer was prepared by mixing g of KH 2 PO 4 and g of Na 2 HPO 4 2H 2 0 in 200 ml of purified water. The carbonate buffer was prepared by mixing 20 g of NaHCO 3 and 40 g of K 2 CO 3 in 200 ml of purified water. The silylating reagent was prepared by mixing 120 mg of 1,4-dithioerythritol and 60 mg of NH 4 I with 30 ml of MSTFA. A 100 ml solution of the internal standard and the standards used for process control (ISTD mixture), containing 5 μg/ml of methyltestosterone, 5 μg/ml of timolol, 20 μg/ml of androsterone glucuronide-d 4 and 20 μg/ml of etiocholanolone-d 5 was prepared from methanolic solutions of 1 mg/ml for methyltestosterone and timolol and by dissolving 2 mg of the standards of androsterone glucuronide-d 4 and etiocholanolone-d 5. For fluoxymesterone, hydrocodone, mesterolone, buprenorphine M1, oxandrolone, oxymesterone and pentazocine 100 μg/ml solutions were prepared by diluting 1 ml of the 1.0 mg/ml standard solutions in 10 ml of methanol. For metenolone a 50 μg/ml solution was prepared by diluting 0.5 ml of the 1.0 mg/ml standard solution in 10 ml of methanol. For 6-monoacethylmorphine, a 1000 μg/ml solution was prepared by dissolving 1.0 mg of standard in 1 ml of methanol. Standards of codeine, hydromorphone, morphine and oxymorphone were used as acquired (solutions of 1.0 mg/ml). The standardofthc-9-coohwasalsousedasacquired(asolutionof 100 μg/ml). For the rest of the substances, 100 μg/ml solutions were prepared by dissolving 1.0 mg of standards in 10 ml of methanol. Two separate standard stock solutions containing the compounds listed in Table 1 were prepared from those solutions. Standard stock solution B contained all the standards in concentrations 20 times those indicated in Table 1. Standard stock solution A contained all the standards listed except clenbuterol and norboletone M1, in concentrations 20 times those indicated in Table 1. Clenbuterol and norboletone M1 were not included in the standard stock solution A due to the fact that these two substances were not included in the scope of the method when the sensitivity experiments with nonconductive disposable tips were undertaken. Both solutions were used during the validation experiments for recovery calculations. Control samples Negative and positive urine samples (2 ml) available in our laboratory, as well as blank samples of purified water (2 ml), were used for all the optimization and validation experiments. Negative samples consisted of urine donations from the laboratory staff, checked for adequate ph (between 5.5 and 6.5), density (between and 1.020), free of interferences and of any trace of substances included in this study. Positive samples consisted of negative ones spiked with the methanolic solutions of standards at the concentrations depicted in Table 1. These concentrations are consistent with the minimum required performance levels (MRPL) established by WADA in the technical document TD2013MRPL [13], and represent the concentration of a prohibited substance or any of its metabolites or markers that accredited laboratories must be able to routinely detect and identify. Manual sample preparation In routine work, the compounds listed in Table 1 can be extracted from urine samples following a manual three-step procedure which includes enzymatic hydrolysis, liquid-liquid extraction, and derivatization, as depicted in Figure 1. [14] The ph of the samples is brought to 7 by previous adjustment (if necessary) and subsequent addition of the phosphate buffer (100 μl). After the addition of 100 μl of ISTD mixture and 50 μl ofβ-glucuronidase enzyme, the samples are incubated for 1 h at 55 C, in order to cleave the glucuronic linkage of most of the analytes with the glucuronic acid. After cooling, samples are brought to ph 11 by addition of the carbonate buffer (300 μl) and 5 ml of TBME are added. After shaking and centrifugation, the upper organic layer is separated to clean tubes after freezing of the aqueous phase and evaporated to dryness under a stream of nitrogen. 50 μl of silylating reagent are added and the samples are transferred into GC vials after incubation at 65 C for 30 min. Description of the automated sample preparation system An automated liquid-liquid extraction system specifically designed for the needs of our laboratory, with additional shaking, heating, drying and crimping capabilities, was acquired from Zinsser Analytics (Frankfurt, Germany). [12] The equipment consists of a wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)

3 Automated sample preparation for detection of doping-related substances Drug Testing Table 1. List of substances analysed and concentrations in the positive urine sample. Substances analyzed by GC-MS/MS are shown in italics in the compound column. All other analytes were determined by GC-MS CATEGORY IN WADA PROHIBITED LIST COMPOUND SYSTEMATIC NAME PROHIBITED COMPOUND CONCENTRATION IN URINE SAMPLES (PPB) S1. Anabolic agents 4-OH-Testosterone Androst-4-en-4,17β-diol-3-one 4-OH-Testosterone 12 5α-Methyltestosterone 17α-methyl-5α-androstan-3α,17β-diol Mestanolone, Methyltestosterone, 2 Methyl-1-testosterone, Oxymetolone 5β-Methyltestosterone 17α-methyl-5β-androstan-3α,17β-diol Metandienone, Methandriol, 2 Methyltestosterone Bolasterone 7α-17α-dimethylandrost-4-en-17β-ol-3-one Bolasterone 5 Bolasterone M1 7α,17α-dimethyl-5β-androstan-3α,17β-diol 5 Boldenone Androst-1,4-dien-17β-ol-3-one Boldenone 5 Boldenone M1 5β-androst-1-en-17β-ol-3-one 5 Calusterone 7β,17α-dimetilandrost-4-en-17β-ol-3-one Calusterone 5 Clenbuterol 1-(4-amino-3,5-dichlorophenyl)-2-(tert-butylamino)ethanol Clenbuterol 0.2 Clostebol M1 4-chloro-androst-4-en-3α-ol-17-one Clostebol 5 Danazol 17α-ethynyl-androst-4-en-17β-ol-(2,3-d)-isoxazole Danazol 5 Danazol M1 (Ethisterone) 17α-ethynyl-androst-4-en-17β-ol-3-one 5 Danazol M2 17α-ethynyl-2α-hydroxymethyl-androst-4-en-17β-ol-3-one 5 Dehydrochlormethyltestosterone M1 4-chloro-17α-methylandrost-1,4-dien-6β,17β-diol-3-one Dehydrochlormethyltestosterone 2 Drostanolone M1 2α-methyl-5α-androst-3α-ol-17-one Drostanolone 5 Estradienedione Estra-4,9-diene-3,17-dione Estradienedione 5 Estradienedione M1 (9(10)- Dehydronandrolone) Estra-4,9-diene-17β-ol-3-one 5 Fluoxymesterone 9α-fluoro-17α-methylandrost-4-en-11β,17β-diol-3-one Fluoxymesterone 5 Fluoxymesterone M1 9α-fluoro-17α-methyl-androst-4-en-3α,6β,11β,17β-tetrol 5 Fluoxymesterone M2 9α-fluoro-11β-ol-18-nor-17,17-dimethylandrost-4,13-dien-3-one 5 Furazabol 17α-methyl-5α-androstan-17β-ol-[2,3-c]-furazan Furazabol 5 Furazabol M1 17α-methyl-5α-androsta-16ε,17β-diol-[2,3-c]-furazan 5 Mesterolone 1α-methyl-5α-androstan-17β-ol-3-one Mesterolone 5 Mesterolone M1 1α-methyl-5α-androstan-3α,17β-diol 5 Metandienone M1 (Epimetendiol) 17β-methyl-5β-androst-1-en-3α,17α-diol Metandienone 2 Metandienone M2 (6-OH-Dianabol) 17α-methylandrosta-1,4-dien-6β,17β-diol-3-one 2 Metandienone M3 18-nor-17,17-dimethyl-5β-androst-1,13-dien-3α-ol 10 Metandienone M4 (17-Epimetandienone) 17β-metil-androst-1,4-dien-17α-ol-3-ona 2 Metasterone M1 (3-OH-Metasterone) 2α,17α-dimethyl-5α-androst-3α,17β-diol Metasterone 5 Metenolone 1-methyl-5α-androst-1-en-17β-ol-3-one Metenolone 5 Metenolone M1 1-methylen-5α-androstan-3α-ol-17-one 5 Methyldienolone 17α-methylestra-4,9-dien-17β-ol-3-one Methyldienolone 5 Methyl-1-testosterone 17α-methyl-5α-androst-1-en-17β-ol-3-one Methyl-1-testosterone 5 Mibolerone 7α,17α-dimethylestr-4-en-17β-ol-3-one Mibolerone 5 Nandrolone M1 (19-Norandrosterone) 5α-estran-3α-ol-17-one Nandrolone 2 (Continues) 3 Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/dta

4 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra 4 Table 1. (Continued) CATEGORY IN WADA PROHIBITED LIST COMPOUND SYSTEMATIC NAME PROHIBITED COMPOUND CONCENTRATION IN URINE SAMPLES (PPB) S4. Hormone and metabolic modulators S5. Diuretics and other masking agents Nandrolone M2 (19-Noretiocholanolone) 5β-estran-3α-ol-17-one 5 Norboletone M1 18-methyl-5α-estr-17α-ethyl-3α,17β-diol Norboletone 5 Norboletone M2 18-methyl-5β-estr-17α-ethyl-3α,17β-diol 5 Norethandrolone M1 17α-ethyl-5β-estran-3α-17β-diol Norethandrolone 10 Norethandrolone M2 17α-ethyl-5α-estran-3α-17β-diol 5 Oxandrolone 17α-methyl-2-oxa-5α-androstan-17β-ol-3-one Oxandrolone 5 Oxandrolone M1 (Epioxandrolone) 17β-methyl-2-oxa-5α-androstan-17α-ol-3-one 5 Oxymesterone 17α-methylandrost-4-en-4,17β-diol-3-one Oxymesterone 5 Stanozolol M1 3 -hydroxy-17α-methyl-5α-androst-17β-ol-[3,2-c]pyrazol Stanozolol 2 Stanozolol M2 17α-methyl-5α-androst-4β,17β-diol-[3,2-c]pirazol 2 Tibolone M1 17α-ethynyl-7α-methyl-estr-5(10)-en-3β,17β-diol Tibolone 5 Tibolone M2 17α-ethynyl-7α-methyl-estr-5(10)-en-3α,17β-diol 5 Tibolone M3 17α-ethynyl-7α-methyl-estr-4-en-17β-ol-3-one 5 Trenbolone M1 (Epitrenbolone) Estra-4,9,11-trien-17α-ol-3-one Trenbolone 5 Zeranol (3S,7R)-7,14,16-trihydroxy-3-methyl-3,4,5,6,7,8,9,10,11, Zeranol 5 12-decahydro-1H-2-benzoxacyclotetradecin-1-one Zeranol M1 (Taleranol) (3S,7S)-7,14,16-trihydroxy-3-methyl-3,4,5,6,7,8,9,10,11, 12-decahydro-1H-2-benzoxacyclotetradecin-1-one Zilpaterol Trans-4,5,6,7-tetrahydro-7-hydroxy-6- (isopropylamino)imidazo[4,5,1-jk][1]benzazepin-2(1h)-one Zilpaterol 5 Aminoglutethimide 3-(4-aminophenyl)-3-ethyl-piperidine-2,6-dione Aminoglutethimide 20 Androstatrienedione Androsta-1,4,6-triene-3,17-dione Androstatrienedione 50 Letrozole M1 Bis(4-cyanophenyl)methanol Letrozole 20 Tamoxifen M1 Z-2-[4-(1-phenyl-2-(3-hydroxy-4-methoxyphenyl)-1- Tamoxifen 20 butenyl)phenoxy]-n,n-dimethylethanamine Spironolactone 7α-acetylthio-3-oxo-17α-pregn-4-ene-21,17-carbolactone Spironolactone 200 S6. Stimulants Bromantane M1 (6-OH-Bromantane) 2-(4-bromophenylamine)-adamantan-6-ol Bromantane 100 Carphedon 2-(2-oxo-4-phenylpyrrolidin-1-yl)acetamide Carphedon 100 Parahydroxyamphetamine 4-(2-aminopropyl)phenol Parahydroxyamphetamine 100 S7. Narcotics 6-Monoacetylmorphine 3-hydroxy-6-acetyl-(5α,6α)-7,8-didehydro-4,5-epoxy- Diacetylmorphine (Heroine) methylmorphinan Buprenorfine 9α-cyclopropylmethyl-4,5-epoxy-6,14-ethano-3-hydroxy-6- Buprenorfine 5 methoxymorphinan-7-yl]-3,3-dimethylbutan-2-ol Buprenorfine M1 (Norbuprenorfine) 4,5-epoxy-6,14-ethano-3-hydroxy-6-methoxymorphinan-7-yl]- 5 3,3-dimethylbutan-2-ol Morphine (5α,6α)-7,8-didehydro- 4,5-epoxy-17-methylmorphinan-3,6-diol Morphine 50 Oxycodone 4,5α-epoxy-14-hydroxy-3-methoxy-17-methylmorphinan-6-one Oxycodone 50 Oxymorphone 4,5α-epoxy-3,14-dihydroxy- 17-methylmorphinan-6-one Oxymorphone wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)

5 Automated sample preparation for detection of doping-related substances Drug Testing Table 1. (Continued) COMPOUND SYSTEMATIC NAME PROHIBITED COMPOUND CONCENTRATION IN URINE SAMPLES (PPB) CATEGORY IN WADA PROHIBITED LIST Pentazocine 2-dimethylallyl-5,9-dimethyl-2 -hydroxybenzomorphan Pentazocine 50 S8. Cannabinoids THC-9-COOH 1-hydroxy-6,6-dimethyl-3-pentyl-6a,7,8,10atetrahydrobenzo[c]chromene-9-carboxylic Tetrahydrocannabinol 15 acid Codeine 50 Codeine (5α,6α)-7,8-didehydro-4,5-epoxy-3-methoxy-17- methylmorphinan-6-ol Other non-prohibited compounds Hydrocodone 4,5α-epoxy-3-methoxy-17-methylmorphinan-6-one Hydrocodone 50 Figure 1. Flow chart of the manual sample preparation workbench divided into different zones where the analytical tasks are carried out (Figure 2). The layout integrates several modules among which a mobile gripper transports, picks up and drops off the different racks and tools, and four pipetting probes dispense and transfer solvents and reagents among the different areas. These probes self-load and discharge disposable tips thus avoiding potential cross-contamination. Briefly, the system was designed for treating up to 96 samples in a sample preparation protocol which could include several heating and extraction steps, to which end a module for heating, a module for shaking and/or heating, four racks for 24 vials of 8 ml, two racks for 48 GC vials (1.8 ml), two racks for 48 GC vial caps, three racks for a total of 288 disposable tips, a reagent rack, a drying manifold, two solvent reservoirs, a crimper and a gripper tool for GC vials and caps were included in the design. All operations and methods are programmed and launched via the software Zinsser WinLissy (version 7.0.7). The entire system is located under a large fume hood in order to avoid dissemination of vapours. The 8 ml vials, where samples are initially placed, were from Zinsser Analytics (Frankfurt, Germany). The 1.8 ml high recovery GC vials were from Agilent Technologies (Palo Alto, CA, USA). The carbonate buffer is placed in a 40-ml brown glass vial on the reagent rack. The silylating reagent is also located on the reagent rack in an 8-ml glass vial protected from ambient light and covered with a septum sealed with a screw-top. The septum allows the disposable tips to 5 Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/dta

6 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra 6 Disposable tip racks Reagent rack Heating plate Heating/Vortexing plate Solvent reservoirs Racks for 8 ml vials enter and self-closes during each reagent loading. The pipetting module is able to work with two different types of 1100 μl disposable tips for the addition and transfer of reagents and solvents: conductive disposable tips (from Zinsser Analytics (Frankfurt, Germany)) and non-conductive ones (from Molecular BioProducts (San Diego, CA, USA)). Conductive tips allow working with detection of interphase by conductivity: after shaking of the urine samples with TBME and waiting for phase separation, the pipetting probes loaded with conductive tips can detect the interphase position individually for each sample by closure of an electric circuit, and transfer the organic layer from the 8-ml vials to the GC vials. On the other hand, when non-conductive disposable tips are used, the volume of water phase in the samples must be indicated in the software, which calculates the position of the interphase in order to aspirate and transfer the organic layer. In this case, the same volume must be necessarily used in all samples, since there is not an individual detection of interphase. Both kinds of disposable tips were evaluated in this study in order to assess possible performance differences. Automated sample preparation Drying manifold Parking area Cap racks Racks for GC vials Crimper Gripper tool for GC vials and caps Figure 2. The Zinsser Lissy GXL automated sample preparation system The automation of the process described above and depicted in Figure 1 required specific previous studies in order to adapt the protocol to the characteristics of the equipment, in particular regarding the liquid-liquid extraction step. The shaking/heating module of the system consists of a high speed vortexer, in contrast to the lineal shaking used routinely in the manual sample preparation. Thus the speed and time of shaking had to be carefully optimized in order to enhance the recoveries of the analytes. On the other hand, the extraction is carried out in the equipment by the pipetting probes which transfer the organic phase from the 8-ml vials to 1.8 ml high recovery GC vials after shaking. The organic solvent is then eliminated in the shaking/heating module by means of the drying manifold, through which a light nitrogen gas stream is distributed. As a consequence, just a little more than 1 ml of solvent could be used in each extraction step (otherwise GC vials would overflow), and therefore several consecutive extractions had to be envisaged in order to accomplish adequate recoveries. Finally, in the manual sample preparation the separation of organic and water phase is achieved after freezing of the latter and decanting of the TBME phase to clean tubes. In contrast, phase separation in the equipment is carried out leaving the samples to stand for the necessary time. In short, time and speed of shaking, waiting time for phase separation, number of extractions, quantity of organic solvent and drying time had to be studied and selected in advance before the validation experiments were undertaken. The general protocol followed for the previous studies, validation experiments and current routine work was as follows. Adjustment of ph when necessary, together with addition of the phosphate buffer (100 μl), the ISTD mixture (100 μl) and the β-glucuronidase enzyme (50 μl), was carried out manually in the test tubes containing the 2-ml samples. Tubes were then vortexed, transferred to the 8 ml vials and placed on the racks of the equipment. The automated procedure was then initiated by the enzymatic hydrolysis reaction, which was carried out in the heating module under the same temperature and time conditions as the manual preparation (55 C, 60 min). Following this, the rack containing the samples was transferred to the initial position and maintained for 10 min in order to allow the samples to cool to room temperature. The extraction process, the distinctive features of which were explained in the previous paragraphs, was systematically studied and optimized as described in the Results and Discussion section. The derivatization reaction was carried out under the same conditions as the manual preparation (65 C, 30 min). In the automated procedure the addition of 50 μl of derivatization agent to each sample was carried out directly by the pipetting probes in the GC vials, which were subsequently capped by means of the gripper tool and crimper. The rack was finally transferred to the shaking/heating module where samples were incubated with slight agitation. For the investigation of the process parameters and validation experiments just one rack of 8-ml vials (up to 24 samples) needed to be used, and all the operations described in the previous paragraph were accomplished consecutively. In contrast, when more than one rack is used (meaning the processing of 25 to 96 samples), which is often the case in routine analysis, the different steps of the sample preparation method are carried out alternately for each rack (i.e. hydrolysis of samples on rack 1, hydrolysis of samples on rack 2 during extraction of those on rack 1, extraction of rack 2 during hydrolysis of rack 3, etc.). GC-MS analysis Equipment All the samples were analysed by mass spectrometry coupled to GC. The instrumental methods of detection used in this study were previously established in our laboratory and are not discussed here. The compounds studied are included in the scope of procedures where analysis is based either on GC-MS (SIM mode) or on GC-MS/MS (MRM mode). Due to this, some of the compounds included in this work were determined by a triple quadrupole mass spectrometer and the rest by a single quadrupole spectrometer. Both systems were from Agilent Technologies (Palo Alto, CA, USA). The single quadrupole GC-MS system was a 6890N chromatograph with a 7683 Series injector/autosampler, coupled to a 5973 mass analyzer operating in electron ionization (EI) mode. wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)

7 Automated sample preparation for detection of doping-related substances Drug Testing The triple quadrupole GC-MS/MS system was a 7890 chromatograph with a 7693 injector/autosampler, combined with a 7000 triple quad mass analyzer. Data analysis was carried out with the Chemstation (D ) and MassHunter Workstation (B.05.00) software from Agilent Technologies. Chromatographic conditions Separations were performed using Agilent J&W ScientificHP-Ultra1 capillary columns. For GC-MS the dimensions were 30 m length, 0.25 mm I.D. and 0.25 mm film thickness. For GC-MS/MS the dimensions were 17 m length, 0.25 mm I.D. and 0.25 mm film thickness. Split, straight liners with glass wool, non-deactivated (Part No ) from Agilent Technologies were used for both systems. MS detection In all mass spectrometric measurements, the injector port, transfer line, quadrupoles and ion source temperature were set at 280, 280, 150, and 230 C, respectively. In the GC-MS system the EI mode was used at low resolution with ionization energy of 70 ev. All analyses were performed in selected ion monitoring (SIM) mode for the GC-MS system and in multiple reaction monitoring (MRM) mode for the GC-MS/MS system. Substances analysed by GC-MS/MS are shown in italics in Table 1. The rest of the analytes were determined by GC-MS. For each compound a minimum of two ions/transitions were selected in order to monitor its presence. Validation of each experiment Several analytical criteria are used routinely in our laboratory to ensure that hydrolysis, extraction and derivatization steps, as well as instrumental analysis, have been successfully accomplished. These control parameters, based on relations among the different substances included in the ISTD mixture, were maintained and monitored during this study in order to validate that each experiment was performed correctly. The following ratios were controlled: The relationship between areas of the main ion of androsterone-d 4 and ethiocholanolone-d 5 as proof of a correct enzymatic hydrolysis reaction (in the GC-MS analysis). The ratio between areas of the main ion of methyltestosterone (ISTD) and timolol as evidence for the correct extraction procedure (in the GC-MS analysis). The relationship between areas of the selected ion of androsterone bis-tms/androsterone mono-tms and etiocholanolone bis-tms/etiocholanolone mono-tms for the control of the derivatization reaction (in the GC-MS analysis). Retention time and area (GC-MS analysis) or height (GC-MS/MS analysis) of the main ion/transition of methyltestosterone as proof of a correct injection in the chromatographic system. Validation of the automated sample preparation procedure The validation was approached as a comparison between the results obtained with the manual sample preparation method and the automated one; therefore the instrumental chromatographic analysis procedure was not changed. The following parameters were evaluated: Interferences Four 2-ml negative urine samples were analysed (five replicate experiments) according to the automated protocol finally selected. Analytical results were then compared to those obtained in the manual sample preparation, in order to check if there were any interfering peaks in the chromatograms, coming from the equipment or the materials used, that could avoid or complicate the determination of any substance. Extraction efficiency Recoveries were determined by extracting pairs of 2-ml distilled water samples. One of them was spiked prior to the preparation procedure at four times the concentrations indicated in Table 1: 200 μl of methanolic standard stock solution A (for validation with non-conductive tips) or B (for conductive ones) were added to an 8-mL vial, dried under a light stream of N 2, and then 2 ml of distilled water were added. Another 2-ml distilled water sample was also prepared together with a GC vial spiked similarly and dried, indicating full recovery. Samples were then extracted by using the automated sample preparation method finally selected. The same protocol was concurrently followed with two 2-ml blank distilled water samples by using the manual sample preparation method, in order to compare automated/manual extraction efficiency for each substance. This protocol was carried out five times. Extraction efficiency was calculated as mean percentages of the full-recovery samples. Accomplishment of minimum concentration levels To check whether the automation allows ensuring the performance to fulfil the requirements of the WADA technical document TD2013MRPL, [13] 2-ml positive urine samples (negative samples spiked at the concentrations shown in Table 1) were extracted in triplicate according to the selected procedure. Analytical results were then compared to those obtained in the manual sample preparation. Additionally, signal-to-noise (S/N) ratios for each ion/transition of each compound and sample were calculated by using the root mean square (RMS) algorithm. Splash contamination and carry-over In order to check possible contamination between samples (particularly during the drying step) and carry-over effects, a batch of 48 water samples (2 ml) was extracted following the automated procedure. Seven of the samples were spiked at twenty times the concentration indicated in Table 1 and placed in different positions (external and central) on the racks. The absence of traces of any substance in the non-spiked samples was then checked. Results and discussion Optimization of parameters of automated sample preparation As depicted above, the automation of the manual sample preparation protocol described in the previous section required the investigation of several process parameters related to the extraction step, in order to adapt the method to the special features of the system and enhance the analytical results for all the substances included in the study. Shaking speed and time The type of agitation of the shaking/heating module included in the system (vortex) is different from that used in the manual method (linear). So the speed and time of shaking had to be optimized at first. Obviously, the faster the agitation, the more effective the recoveries will be. However after the first experiments it was immediately 7 Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/dta

8 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra evident that excessively strong shaking in terms of speed or time led to wide emulsified interphases, which were inconsistent with adequate phase separations. Several experiments were undertaken with negative urine samples of a range of densities at different agitation speeds, covering from 600 to 900 rpm and times from 5 to 30 min. A shaking speed of 750 rpm for 15 min was selected, given that these values represented the highest velocity during a convenient time for which no emulsions were observed in the interphases. Waiting time for phase separation Several experiments showed that when the interphase is emulsified due to an excessive agitation speed or time, long waiting times did not help to eliminate the emulsion. So 5 min were selected as the waiting time after agitation. Volume of sample indicated in the software (for non-conductive tips) When non-conductive disposable tips are used, the position of the interphase is calculated by the software according to the sample volume indicated at the launch of the method. The total sample volume before the extraction step was 2.85 ml after addition of ISTD mixture, enzyme, and buffers to the initial volume of the urine. After testing several values, the sample volume which must be indicated in the software was 3.00 ml, otherwise undesired aspirations of water phase during the extractions could take place. Number of extractions 8 While 5 ml of TBME are used in the manual sample preparation method, a series of tests showed that no more than 1.3 ml of solvent should be used in an extraction step in the automated system, in order to prevent solvent overflow from the GC vials. So, to attain the recoveries of the manual sample preparation and avoid loss of performance, several extractions must be carried out. The necessary number of extractions was then studied by monitoring the area of the main ion of the ISTD (methyltestosterone), and comparing it with that obtained with the manual method. Several experiments with different numbers of extractions showed that at least three of them (a total of 3.9 ml of TBME) were necessary in order to have similar signal intensities. So for the automated sample preparation method, three extractions were selected, each one including addition of TBME, agitation, time for phase separation, and transfer and elimination of organic solvent. Drying time The time selected should be the minimum one that guarantees the full removal of organic solvent for every sample and every extraction. Seven minutes at 45 C were selected. Table 2. Parameters optimized in the automated sample preparation prior to the validation studies PARAMETER VALUE Volume of sample indicated in the software 3.0 ml (for non-conductive tips) Shaking speed 750 rpm Shaking time 15 min Waiting time for phase separation 5 min Volume of organic solvent for each extraction 1.3 ml Number of extractions 3 Drying time 7 min Figure 3. Flow chart of the automated sample preparation. Automated steps are inside the grey square. Steps that required optimization prior to the validation study are shown in bold letters A summary of the values selected after the optimization studies is shown in Table 2. The rest of the conditions were maintained as in the manual sample preparation, as shown in Figure 3 which depicts the complete flow chart of the automated method. Validation Study of interferences During the initial experiments, which were carried out with conductive tips, from the outset an important interference was noticed that affected the detection of 19-norandrosterone, a metabolite of nandrolone. Both species elute at almost the same retention time and have in common both of the ions monitored for 19- norandrosterone (m/z and 420.3). As can be seen in Figure 4, when non-conductive tips are used, this interference is much less intense, and so 19-norandrosterone can be determinedcorrectly.wewereabletoclarifythatthenon-conductive tips do not produce this interfering product, so its presence on the chromatograms comes from other parts of the equipment. The interference signal is also present as a little signal when samples are treated manually (Figure 4). Several analyses were carried out with the purpose of discovering the structure of this species, but we have not been wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)

9 Automated sample preparation for detection of doping-related substances Drug Testing Conductive tips Non conductive tips Negative Positive (autom. method) Positive (manual method) when recoveries between the validations carried out with conductive versus non-conductive tips are compared, those obtained with non-conductive tips were slightly better for most of the compounds. Recovery values ranged widely due to the different nature of the substances analysed. For steroids recoveries were between 70 and 100% for most of them. Narcotics ranged between 60 and 100% except for morphine and hydromorphone, with recovery ranging from 19 to 33% in manual and automated methods. Anti-estrogenic substances ranged from 60 to 90%. Analytes belonging to the other anabolic agents, category which gathers substances of diverse chemical structures, presented recoveries ranging from 30 40% (zilpaterol) to % (clenbuterol, tibolone M1, M2, and M3, zeranol and taleranol). The stimulants studied in this research (bromantane and its metabolite 6-OH-bromantane, carphedon and parahydroxyamphetamine) have also very different structures and therefore chemical properties. In this way, recoveries for bromantane and 6-OH-bromantane were quite good (74 91%), while those observed for carphedon and parahydroxyamphetamine were low (8 42%). The diuretic spironolactone had recoveries of 72 80% and the metabolite of cannabis THC-9-COOH showed values ranging from 79 to 87%. Figure 4. Interference in the window of 19-norandrosterone. When using conductive tips its intensity does not allow correct detection of this analyte able to elucidate its exact nature to date. Apparently, the interference comes from a degradation product of a polymer of common use in analytical material, and which is also part of the formulation of the conductivetips. On the other hand, no further interfering peaks were observed that could affect the performance of the method for the rest of the analytes included. So regarding selectivity, non-conductive tips must be used to perform the whole scope of the method. The required experiments of validation for the inclusion of 19-norandrosterone in the GC-MS/MS instrumental method are currently being developed in our laboratory. Due to the better selectivity of the MRM technique (GC-MS/MS) versus the SIM one (GC-MS), the interference signal should be eliminated and both types of tips may be used in the near future, in terms of selectivity. Extraction efficiency The recovery values, standard deviations and relative standard deviations (in percentage) are shown in Table 3 for the automated sample preparation with both types of disposable tips as well as the manual sample preparation. Overall, the average recoveries were slightly inferior in the automated sample preparation method versus the manual for most of the analytes. This is probably a consequence of the lesser quantity of TBME used in the automated method (3.9 ml versus 5 ml). Further experiments showed that a supplementary step of extraction would equal the results in terms of recoveries. Nevertheless, values were good enough to maintain the three extractions, which implies a saving in terms of quantity of solvent needed, quantity of disposable tips required and time taken for performing the whole method. On the other hand, Accomplishment of minimum concentration levels Analytical results for each compound were compared between automated and manual protocols in samples spiked at the concentrations depicted in Table 1. Additionally, S/N ratios were calculated (RMS algorithm) for every compound and every ion/transition, being higher than three in all cases (data not shown). Overall, successful determination was achieved for all the analytes included in the study after the automated sample preparation, except for nandrolone M1 by using conductive tips. Based on the similar results obtained for the overall scope of the method by manual and automated sample preparations, the limits of detection assigned were maintained as regards those obtained individually for each compound in their validation studies. Splash contamination and carry-over An experiment to check contamination between samples or carry-over effects was conducted according to the protocol described above. No trace of any compound was detected in any of the non-spiked samples, demonstrating the absence of cross contamination or carry-over issues. Routine applicability The automated sample preparation protocol has been used daily in our laboratory for several weeks in 24 to 96 routine sample batches. No mechanical or computer failures were observed during the experiments that could affect the performance of the preparations, so all of them were successfully performed in times ranging from 5 h for 24 samples to 17 h for 96 samples. Examples of routine results for several compounds of spiked blank, negative and positive samples, extracted with the manual and automated method, are shown in Figure 5. 9 Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/dta

10 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra Table 3. Recoveries, standard deviations and relative standard deviations (percentage) for the automated and manual sample preparation COMPOUND DETECTION OF INTERPHASE BY CONDUCTIVITY DETECTION OF INTERPHASE BY HEIGHT MANUAL PREPARATION AV REC STD DEV RSD AV REC STD DEV RSD AV REC STD DEV RSD 10 4-OH-Testosterone α-Methyltestosterone β-Methyltestosterone Monoacetylmorphine Aminoglutethimide Androstatrienedione Bolasterone Bolasterone M Boldenone Boldenone M Bromantane M1 (6-OH-Bromantane) Buprenorphine Buprenorphine M1 (Norbuprenorphine) Calusterone Carphedon Clenbuterol a Clostebol M Codeine Cyclofenil M Danazol Danazol M1 (Ethisterone) Danazol M Dehydrochlormethyltestosterone M Drostanolone M Estradienedione Estradienedione M1 (9(10)-Dehydronandrolone) Fluoxymesterone Fluoxymesterone M Fluoxymesterone M Furazabol Furazabol M Hydrocodone Hydromorphone Letrozole M Mesterolone Mesterolone M Metandienone M1 (Epimetendiol) Metandienone M2 (6-OH-Dianabol) Metandienone M Metandienone M4 (17-Epimetandienone) Metasterone M1 (3-OH-Metasterone) Metenolone Metenolone M Methyldienolone Methyl-1-testosterone Mibolerone Morphine Nandrolone M1 (19-Norandrosterone) b Nandrolone M2 (19-Noretiocholanolone) Norboletone M1 a Norboletone M Norethandrolone M Norethandrolone M Oxandrolone (Continues) wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)

11 11 Automated sample preparation for detection of doping-related substances Drug Testing Table 3. (Continued) COMPOUND DETECTION OF INTERPHASE BY CONDUCTIVITY DETECTION OF INTERPHASE BY HEIGHT MANUAL PREPARATION AV REC STD DEV RSD AV REC STD DEV RSD AV REC STD DEV RSD Oxandrolone M1 (Epioxandrolone) Oxycodone Oxymesterone Oxymorphone Parahydroxyamphetamine Pentazocine Spironolactone Stanozolol M Stanozolol M Tamoxifen M THC-9-COOH Tibolone M Tibolone M Tibolone M Trenbolone M1 (Epitrenbolone) Zeranol Zeranol M1 (Taleranol) Zilpaterol a Recoveries for clenbuterol and norboletone M1 were not calculated in the sample preparation with conductive tips and manual sample preparation, due to the fact that these substances were included in the scope of the procedure after the finalisation of this part of the validation. b 19-norandrosterone (nandrolone M1) in the sample preparation with conductive tips could not be determined due to the presence of interference (see the study of interferences section in Results and Discussion). SUBSTANCES MANUAL PREP. AUTOMATED PREPARATION Positive Blank 1 Blank 2 Positive Negative 1 Negative 2 Negative 3 Negative 4 ISTD Stanozolol M1 Metandienone M1 Oxandrolone M1 Furazabol Figure 5. Examples of chromatograms of a positive sample prepared manually and spiked blanks, positive and negative samples prepared using the automated method Drug Test. Analysis (2013) Copyright 2013 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/dta

12 Drug Testing D. Cuervo, P. Díaz-Rodríguez and J. Muñoz-Guerra Conclusion A fully automated sample preparation method for the analysis of 72 substances including anabolics, anti-estrogenics, narcotics, diuretics, cannabinoids, and stimulants was developed by using an automated liquid handling system specifically designed for the laboratory. The method allows preparation of up to 96 urine samples in one simple experiment with minimal manual preparation prior to the launch of the automated method. The recoveries and concentration levels detected with the automated sample preparation method were similar to those obtained with the manual method and fulfilled the requirements of WADA 13 for all the substances involved. The competence of the automated protocol was also tested in terms of interferences, contamination and carry-over effects. By using one of the disposable tips checked (non-conductive ones), the full scope of the compounds including in the method can be successfully determined. As a result, the described method is at present suitable for routine analyses and is being applied daily in our laboratory with high sample throughput. After the initial investment, the implementation of the protocol would allow an anti-doping laboratory to benefit from a fully automated sample preparation method that can coexist with the traditional manual one without loss of performance. The automation minimizes human errors while analysts are released from hazardous exposure to reagents and time-consuming tasks, thus being more available for additional duties carried out in the laboratories. References [1] World Anti-Doping Agency. The 2013 List of Prohibited Substances and Methods. Available at: Anti-Doping_Program/WADP-Prohibited-list/2013/WADA-Prohibited- List-2013-EN.pdf [29 April 2013]. [2] M.D. Luque de Castro, J.L. Luque García, Automation of Sample Preparation, in Comprehensive Analytical Chemistry, (Eds: Z. Mester, R. Sturgeon), Elsevier, Amsterdam, 2003, 41, [3] H.L. Lord, E.A. Pfannkoch, Sample Preparation Automation for GC Injection, in Comprehensive Sampling and Sample Preparation, (Ed: J. Pawliszyn), Elsevier, Amsterdam, 2012, 2, [4] C. Soriano, J. Muñoz-Guerra, D. Carreras, C. Rodríguez, A.F. Rodríguez, R. Cortés. Automated analysis of drugs in urine. J. Chromatogr. B 1996, 687, 183. [5] E. Haber, J.A. Muñoz-Guerra, C. Soriano, D. Carreras, C. Rodríguez, F.A. Rodríguez. Automated sample preparation and gas chromatography mass spectrometry analysis of urinary androgenic anabolic steroids. J. Chromatogr. B 2001, 755, 17. [6] R.W. Taylor, S.D. Le. Robotic method for the analysis of cocaine and benzoylecgonine in urine. J. Anal. Toxicol. 1991, 15, 276. [7] B. Houlihan. Robotic extraction of cocaine and benzoylecgonine by solid phase chemistry. Adv. Lab. Autom. Rob. 1991, 7, 583. [8] D.M. Steinberg, L.J. Sokoll, K.C. Bowles, J.H. Nichols, R. Roberts, S.K. Schultheis, C.M. O Donnell. Clinical evaluation of Toxi Prep: A semiautomated solid-phase extraction system for screening of drugs in urine. Clin. Chem. 1997, 43, [9] D. Vuckovic, E. Cudjoe, D. Hein, J. Pawliszyn. Automation of solidphase microextraction in high-throughput format and applications to drug analysis. Anal. Chem. 2008, 80, [10] W. Xie, W.M. Mullett, C.M. Miller-Stein, J. Pawliszyn. Automation of in-tip solid-phase microextraction in 96-well format for the determination of a model drug compound in human plasma by liquid chromatography with tandem mass spectrometric detection. J. Chromatogr. B 2009, 877, 415. [11] H. Kataoka. Recent developments and applications of microextraction techniques in drug analysis. Anal. Bioanal. Chem. 2010, 396, 339. [12] Zinsser Analytic Website. A description of the characteristics of this type of systems. Available at: [09 September 2013] [13] WADA. Technical document TD2013MRPL: Minimum required performance levels for detection and identification of non-threshold substances. Available at: Anti-Doping_Program/WADP-IS-Laboratories/Technical_Documents/ WADA-TD2013MRPL-Minimum-Required-Performance-Levels-v EN.pdf [29 April 2013]. [14] R. Massé, C. Ayotte, R. Dugal. Studies on anabolic steroids: I. Integrated methodological approach to the gas chromatographic-mass spectrometric analysis of anabolic steroid metabolites in urine. J. Chromatogr. B 1989, 489, wileyonlinelibrary.com/journal/dta Copyright 2013 John Wiley & Sons, Ltd. Drug Test. Analysis (2013)