lable at ScienceDirect

Analytica Chimica Acta 964 (2017) 123e133

Contents lists avai

Analytica Chimica Acta

journal homepage: www.elsevier .com/locate/aca

Coupling of gas chromatography and electrospray ionization highresolution mass spectrometry for the analysis of anabolic steroids astrimethylsilyl derivatives in human urine

Eunju Cha a, 1, Eun Sook Jeong a, 1, Sangwon Cha b, Jaeick Lee a, *

a Doping Control Center, Korea Institute of Science and Technology, Hwarangno 14-gil 5, Seongbuk-gu, Seoul 136-791, South Koreab Department of Chemistry, Hankuk University of Foreign Studies, Yongin, 449-791, South Korea

h i g h l i g h t s

* Corresponding author.E-mail address: jaeicklee@kist.re.kr (J. Lee).

1 Both authors contributed equally to this work.

http://dx.doi.org/10.1016/j.aca.2017.01.0580003-2670/© 2017 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� A GC-ESI/HRMS was developed andoptimized to enhance the ionizationefficiency.

� Key parameters were investigatedand GC-ESI/HRMS was applied tosteroids analysis.

� Regardless of analyte phase orderivatization, steroids showedsimilar ionization profile.

� GC-ESI/HRMS with TMS derivativesshowed narrow peak width and goodsensitivity.

� This method has potential as novelionization tool for steroids analysis.

a r t i c l e i n f o

Article history:Received 9 August 2016Received in revised form7 January 2017Accepted 13 January 2017Available online 1 February 2017

Keywords:Gas chromatographyElectrospray ionizationAnabolic steroidsTrimethylsilylation

a b s t r a c t

In this study, gas chromatography (GC) was interfaced with high resolution mass spectrometry (HRMS)with electrospray ionization source (ESI) and the relevant parameters were investigated to enhance theionization efficiency. In GC-ESI, the distances (x-, y- and z) and angle between the ESI needle, GC capillarycolumn and MS orifice were set to 7 (x-distance), 4 (y-distance), and 1 mm (z-distance). The ESI spraysolvent, acid modifier and nebulizer gas flow were methanol, 0.1% formic acid and 5 arbitrary units,respectively. Based on these results, analytical conditions for GC-ESI/HRMS were established. In partic-ular, the results of spray solvent flow indicated a concentration-dependent mechanism (peak dilutioneffect), and other parameters also greatly influenced the ionization performance. The developed GC-ESI/HRMS was then applied to the analysis of anabolic steroids as trimethylsilyl (TMS) derivatives in humanurine to demonstrate its application. The ionization profiles of TMS-derivatized steroids were investi-gated and compared with those of underivatized steroids obtained from gas chromatography-electrospray ionization/mass spectrometry (GC-ESI/MS) and liquid chromatography-electrospray ioni-zation/mass spectrometry (LC-ESI/MS). The steroids exhibited ionization profiles based on their struc-tural characteristics, regardless of the analyte phase or derivatization. Groups I and II with conjugated orunconjugated keto functional groups at C3 generated the [MþH]þ and [MþH-TMS]þ ions, respectively.On the other hand, Groups III and IV gave rise to the characteristic fragment ions [MþH-TMS-H2O]þ and[MþH-2TMS-H2O]

þ, corresponding to loss of a neutral TMS$H2O moiety from the protonated molecularion by in-source dissociation. To the best of our knowledge, this is the first study to successfully ionizeand analyze steroids as TMS derivatives using ESI coupled with GC. The present system has enabled the

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133124

ionization of TMS derivatives under ESI conditions and this method has potential as a novel ionizationtool. It is also useful for the simultaneous analysis of steroids as TMS derivatives.

© 2017 Elsevier B.V. All rights reserved.

1. Introduction

Anabolic steroids are synthetic analogues of the male hormonetestosterone and these substances promote muscle growth.Anabolic steroids are widely used to enhance athletic performancein sports. In doping control, the abuse of anabolic steroids is bannedby the World Anti-Doping Agency [1], and anabolic steroids are themost frequently detected class of prohibited substances. Moreover,the number of illegal steroids is constantly increasing. Thus, thedetection of steroids is a continuous challenge in doping controlanalysis.

Anabolic steroids are organic compounds with four rings ar-ranged in a specific configuration. The structures and molecularweights of steroids are very similar and endogenous steroids arepresent in human biological matrices. For these reasons, the chro-matographic resolution is important and essential for steroidsanalysis. Analytical methods for the analysis of anabolic steroidshave been developed on the basis of gas chromatography-electronionization/mass spectrometry (GC-EI/MS). GC-EI/MS methods arewell-established and widely used for steroids because of their highchromatographic resolving power and the wide ionization range ofthe EI ionization source [2,3]. Although GC-EI/MS methods arepowerful and have some advantages for steroids analysis, the EIionization can cause extensive fragmentation owing to the ener-getic ionization process with excessive ionization energies of 70 eV,resulting in the absence of the molecular ion.

Recently, various analytical methods based on liquidchromatography-electrospray ionization/tandem mass spectrom-etry (LC-ESI/MS/MS) have been developed and applied to steroidsanalysis [4e6]. LC-ESI/MS/MS is suitable for determining the mo-lecular ion owing to its softer ionization by ESI. Unfortunately,however, LC-ESI/MS/MS faces serious problems for steroid analysis.This method provides poor chromatographic separation, comparedwith GC-EI/MS. Therefore, GC-ESI/MS combining the advantages ofGC-MS and LC-MS has potential for steroid analysis. Such gaschromatography-electrospray ionization/high resolution massspectrometry (GC-ESI/HRMS) can provide efficient chromato-graphic separation by GC with a capillary column, and it can lead toenhanced selectivity and sensitivity by ESI/HRMS. When consid-ering these points, GC-ESI/HRMS seems to be an attractive instru-ment. However, GC-ESI/HRMS is not commercially available to date.A few publications regarding the ionization of gaseous analytes byESI have been reported [7e9]. Among these, Kostiainen et al. re-ported a simple method for coupling GC to ESI/MS and this is theonly publication of a practical GC-ESI/MS instrument [9]. Theyinvestigated and proposed the x, y, and z distances between the ESIneedle, the orifice, and the GC column, as well as the effect of theESI solvent on the ionization efficiency. Furthermore, the sensitivityof GC-ESI/MS, GC-atmospheric-pressure chemical ionization(APCI)/MS, and GC-EI/MS were investigated, and the mass spectraproduced by these instruments were compared. As a result, theysuccessfully ionized and analyzed several analytes using their GC-ESI/MS system. Nevertheless, various studies of the availabilityand potential of GC-ESI/MS systems as alternative analytical toolsare not sufficient to make progress in this field.

Based on previous reports [7e9], the aims of this study are (1) todevelop and establish more sensitive and practical GC-ESI/HRMS

instrumentation, and (2) to subsequently investigate and evaluateGC-ESI/HRMS as a novel ionization tool for the simultaneous anal-ysis of anabolic steroids as trimethylsilyl (TMS) derivatives. For thispurpose, a self-made resistor-heated column transfer line was pre-pared, and the optimum distances between the ESI needle, orificeand GC column outlet were established. Furthermore, the ESI sol-vent, acid modifier, flow rate of ESI solvent, and flow rate of nebu-lizing gas were investigated to improve the sensitivity. Based on theresult, ionization profiles of gas-phase steroids as TMS derivativeswere investigated using GC-ESI/HRMS, and the results werecompared with gas-phase steroids and liquid-phase underivatizedsteroids under ESI conditions. Finally, the present GC-ESI/HRMSmethod was successfully applied to the simultaneous qualitativeanalysis of anabolic steroids as TMS derivatives in human urine.

2. Materials and methods

2.1. Standards and reagents

Anabolic steroids and chemicals were purchased from Akzo-nobel (BM Arnhem, Netherlands), NMI (Pymble, Australia), Ster-aloids (Newport, RI, USA), Sigma (St. Louis, MO, USA), and NARL(Pymble, Australia). b-Glucuronidase was purchased from Roche(Mannheim, Germany). SeRDoLIT® PAD-1 was purchased fromServa Electrophoresis GmbH (Heidelberg, Germany). The HPLCgrademethanol (MeOH), acetonitrile (ACN), isopropyl alcohol (IPA),diethyl ether and ethyl acetate were purchased from Burdick &Jackson (Ulsan, Korea). Analytical grade formic acid, acetic acid,sodium phosphate and potassium carbonate were purchased fromSigma (St. Louis, MO, USA). Deionized water (DW) was generatedfrom an in-house water purification system (Milli-Q, Bedford, MA,USA). For derivatization, N-methyl-N-(trimethylsilyl) tri-fluoroacetamide (MSTFA), ammonium iodide (NH4I), and dithioer-ythritol (DTE) were purchased from Sigma (St. Louis, MO, USA).

2.2. Sample preparation

Twelve anabolic steroids were used for the establishment ofoperating parameters, and these compounds are summarized inTable 1. To compare their ionization profiles according to analytephase, underivatized steroids were dried and reconstituted usingdiethyl ether and 45% formic acid in MeOH for GC-ESI/HRMS andLC-ESI/MS/MS, respectively. For analysis of TMS steroids de-rivatives, 79 steroids and deuterium labeled internal standard suchas 16,16,17-d3-testosterone were spiked in human urine at10 ng mL�1 and 20 ng mL�1, respectively, and then pretreated fordetermination. Targeted steroids were classified into four differentcategories [10,11]: conjugated- or unconjugated-keto functionalgroups at C3 (Groups I and II) and conjugated- or unconjugated-hydroxyl functional groups at C3 (Groups III and IV). Samplepreparation was based on our previous report [11]. Urine (2 mL)samples were loaded onto a PAD-1 column that had been washedwith 2 mL of DW. After loading, the columnwas washed with 2 mLof DW and eluted with 4 mL of MeOH. The eluent was dried byevaporation in a rotary evaporator. The dry residue was incubatedwith b-glucuronidase (50 mL) in 1mL of phosphate buffer (pH 7.2) at55�C for 1 h. The pH of the reaction mixture was adjusted with

Table 1The classified steroids for optimization using GC-ESI/HRMS.

Structure Compound Abbr. M.W Structure Compound Abbr. M.W

Group IConjugated keto functional group at C3

(C17 keto þ C17 mono or other C multi hydroxyl function)

Group IIUnconjugated keto functional group at C3 (C17 mono or other C multi hydroxyl function)

1-Androstenedione 1-AND 286 Mestanolone MESTA 304

Bolasterone BOLAS 316 Methasterone METHAS 318

6b-OH-Fluoxymesterone FLUO-M3 352 Oxandrolone OXNAD 306

Group IIIConjugated hydroxyl functional group at C3

(C17 keto þ C17 mono or other C multi hydroxyl function)

Group IVUnconjugated hydroxyl functional group at C3(C17 keto þ C17 mono or other C multi hydroxyl function)

5a-Androst-1-ene-3a-ol-17-one 1-TE-M 288 1a-Methyl-5a-andro-stan-3a-ol-17-one MESTE-M1 304

Bolandiol BOLAD 276 13b,17a-Diethyl-5b-gonane-3a,17b-diol

NORB-M2 320

3a-OH-Tibolone TIBO-M2 314 17a-Hydroxyethyl-5b-estrane-3a,17b-diol

NORE-M2 322

Fig. 1. Instrumentation for the developed GC-ESI/HRMS as novel ionizations tool (top): system overview (left), installation of heated column transfer line towards the MS (middle)and positions for MS orifice, ESI probe and end-tip of interface (right). Configuration of the developed GC-ESI/HRMS (bottom): tube fitting (a), GC column (b), glass fiber (c), coppertubing (d), heating cord with power supply (e), temperature sensor with temperature controller (f), pedestal with micrometric screw (g), gas phase analyte (h), and micro syringepump (i).

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133 125

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133126

potassium carbonate (5%, 0.7 mL, pH 11.7) and the mixture wasextracted with 5 mL of diethyl ether. After centrifugation (1530 g,5 min), the organic layer was removed and evaporated under ni-trogen at 50�C. Extracted analytes were dried and converted toTMSderivatives. Steroids were dried in the vacuum desiccators for atleast 20 min. The dried residue was derivatized at 60�C for 20 minby adding 50 mL of MSTFA/NH4I/DTE (500:4:2,v/w/w). After deriv-atization, 2 mL of the sample was injected into the GC-ESI/HRMS orGC-ESI/MS/MS [11].

To determine the limit of detection (LOD), five different urinesamples were prepared at concentrations ranging from 0.02 to200 ng mL�1, analyzed simultaneously with blank urine to deter-mine LOD. The blank urine was a pooled human urine sample(approximately 300 people). The LOD was defined as the lowest

Fig. 2. Summary of the key parameters of GC-ESI/HRMS.

Fig. 3. Results for the establishment of GC-ESI parameters using FLUO-M3: x, y, and z-distaand nebulizer gas flow (F).

concentration of analyte that could be detected in all of the spikedsamples with a S/N ratio �3.

2.3. Instrumentation

GC-ESI/HRMS or GC-ESI/MS/MS experiments were performedon an Agilent 6890N GC (Agilent, Palo Alto, CA) equipped with aself-made heated column transfer line that was heated by a pro-grammable temperature controller TC200P with a temperaturesensor (Misung Scientific Co., Seoul, Korea). The GC column was anUltra-1® capillary column (95%methylpolysiloxane,17m� 0.2mm,0.11 mm; J&W Scientific, Agilent Technologies Inc.). The oven tem-perature was programmed to increase by 10�Cmin�1 from 180�C to290�C. The flow rate of the carrier gas (He, 99.9999%) was1.0 mL min�1 at constant pressure. The injector and GC columntransfer line were maintained at 280�C. Samples (2.0 mL) wereinjected in the splitless mode with a purge time delay of 0.8 min.The mass spectrometers were linear quadruple ion trap-Orbitrap(LTQ Orbitrap, Thermo Finnigan, San jose, USA) and triple quad-rupole (TSQ Quantum discovery Max, Thermo Finnigan, San Jose,CA, USA) equipped with an ESI source. Data acquisition was underthe control of the Finnigan Xcalibur software version 2.0 (ThermoFinnigan, San Jose, CA, USA). For ionization profile, data were ac-quired in the 100 to 650mass ranges at a resolving power of 15,000(full width at half maximum), and the obtained data was extractedwith a mass error tolerance of 10 ppm using LTQ Orbitrap. Fordetermination of LOD, data were acquired in the positive modeusing the selected reactionmonitoringmode of GC-ESI/MS/MS [11].

The ESI source was used in the positive mode with the followingsetting: spray voltage, 4.2 kV; capillary voltage, 35 V; capillarytemperature, 270�C; tube lens offset voltage, 120 V; nebulizer gaspressure, 5 arbitrary units (abr). The ESI solvent was sprayed using a

nces (A)e(C), ESI spray solvent and acid modifier (D), flow rate of ESI spray solvent (E),

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133 127

Harvard Apparatus microsyringe pump (Holliston, MA) with a flowrate of 0.1 mL min�1.

LC-ESI/MS/MS experiments were performed using an LC-20 ADultra-fast liquid chromatography system (Shimadzu, Tokyo, Japan)interfaced to a TSQ Quantum Ultra Mass Spectrometer (ThermoFinnigan, San Jose, CA, USA) with a heated ESI (H-ESI) ion source.The optimized H-ESI source parameters were as follows: sheath gaspressure, 60 abr; auxiliary gas flow, 30 abr; ion sweep gas pressure,2 abr; spray voltage, 4.2 kV; capillary temperature, 320�C. The massspectrometer was operated in the positive mode using full-scanmode. LC separation was performed on a Gemini-NX C18 column(100 mm � 2.0 mm, 3 mm; Phenomenex, Torrance, CA, USA) at aflow rate of 300 mL min�1. The mobile phase was composed of 0.1%formic acid in DW (solvent A) and 0.1% formic acid in MeOH (sol-vent B) and was varied linearly by changing the percentage ofsolvent B: 0 min, 45%; 12.50 min, 75%; 13.00 min, 98%; 18.00 min,98%; 18.01 min, 45%; 20.00 min, 45%. Data processing was per-formed using Finnigan Xcalibur version 2.07 (Thermo Finnigan, SanJose, CA, USA). In addition, detailed analytical instruments andconditions for GC-EI/MS, GC-EI/MS/MS and LC-AgþCIS/MS/MSweredescripted in our previous studies [10e12].

3. Results and discussion

3.1. Investigation of GC-ESI parameters

Fig.1 illustrates the detailed composition of the GC-ESI interface.

Fig. 4. Comparison of ionization profiles between analyte phases with or without TMS derivaunconjugated keto functional groups (A and B) and conjugated/unconjugated hydroxyl fun

For the establishment of a robust system, maintenance of high thetemperature of the transfer line was important. In detail, coppertubing was used as transfer line because it has a good thermalconductivity. The copper tubingwas tightly coiled to supply thehightemperature using a heating cord connected to a power supply. Formonitoring the high temperature as 280�C at the GC-ESI interface, asensor locatedon theheating cord and connected to a controllerwasused. The high temperature of GC-ESI interface was used to mini-mize the condensation, adsorption and desorption effects.

Subsequently, interface was wrapped using glass fiber to mini-mize changes in temperature. To ensure good reproducibility andrepeatability, the position of the GC-ESI interface was adjustedusing a micrometric screw. The developed GC-ESI interface wascombined with the GC using a tube fitting, and the GC column waspassed through the heated column transfer line. After passing theinterface, the end tip of the GC columnwas placed into the ESI/MS.In this system, the length of the GC column end tip from the heatedcolumn transfer line is of great importance to the generations ofstable signals and peak shapes. When a stable high temperaturewas not supplied andmaintained, analytes were afflicted by seriouspeak broadening owing to column diffusion. Under these undesir-able conditions, steroids as gas-phase analytes could not passthrough the interface because they were adsorbed onto the GCcolumn. Therefore, the length of the column outlet from the coppertubing was set to be approximately 1 mm. Column outlet lengthabove 1 mm gave rise to the excessive column diffusion oradsorption of analytes because the high temperature could not be

tization using GC-ESI/HRMS and LC-ESI/MS/MS classified into four groups: conjugated/ctional groups (C and D).

Table 2Characteristic ionization profile of steroids as TMS derivatives using GC-ESI/HRMS (m/z, relative abundance, %).

Compound Abbr. Characteristic ionization profile

nTMS

[MþnTMSþH]þ Loss of 1TMS Loss of1TMS�H2O

Loss ofnTMS

Loss ofnTMS�H2O

Loss ofnTMS�2H2O

Others

Group I Conjugated keto functional group at C3I-1 Keto-substituted at C17 (Conjugated C3 keto þ C17 keto function)1-Androstenedione 1-AND 2 431 (100) 359 (20) 341 (4)4-Hydroxyandrostenedione 4-OH TE-M 3 519 (100) 447 (24)6-OXO-Androstenedione 6-OXO 3 517 (100) 445 (24) 410 (10)4-Androstene-6a-ol-3,17-

dione6-OXO-M 3 519 (100) 447 (62) 429 (5) 375 ([Mþ(n-2)

TMSþH]þ,12), 357 ([Mþ(n-1)TMS-H2OþH]þ,2)

19-Norandrostenedione 19-NOR 2 417 (100) 345 (42) 273 (4)Androstatrienedione ATD 2 427 (12) 355 (100) 283 (2)Boldione BOLDI 2 429 (100) 357 (62) 285 (10)I-2 Mono-hydroxyl substituted at C17 (Conjugated C3 keto þ C17 mono-hydroxyl function)1-Testosterone 1-TE 2 433 (100) 361 (52) 231 (53)Bolasterone BOLAS 2 461 (100) 389 (30) 371 (52) 299 (14)Boldenone BOLD 2 431 (100) 359 (50) 341 (20) 269 (5) 429 (76), 417 (66), 416 (50),

414 (36), 345 (19), 343 (16),339 (17)

5b-Androst-1-en-17b-ol-3-one

BOLD-M 2 433 (100) 361 (84) 271 (4)

Calusterone CALUS 2 461 (100) 389 (30) 371 (48) 299 (12) 410 (38)Ethisterone DANA-M 2 457 (100) 385 (40) 367 (36) 295 (4) 410 (18)Dehydrochlormethyl-

testosteroneDHMT 2 479 (62) 407 (100) 389 (76) 317 (72)

Gestrinone GES 2 453 (100) 381 (10) 363 (24) 309 (8)Metenolone METE 2 447 (100) 375 (52)Methandienone METH 2 445 (100) 373 (60) 355 (36) 283 (32) 315 (14), 341 (20)Methyl-1-testosterone ME-1-TE 2 447 (100) 375 (42) 357 (60) 285 (9)Methyldienolone METLD 2 431 (100) 359 (30) 341 (66) 269 (18) 371 (10)Methylnortestosterone MENORTE 2 433 (100) 361 (36) 343 (42) 271 (8)Methyltestosterone MET 2 447 (100) 375 (44) 357 (62) 285 (10)Mibolerone MIBO 2 447 (100) 375 (28) 357 (42) 285 (10)Nandrolone NAND 2 419 (100) 347 (38) 329 (4)Norclostebol NORC 2 453 (100) 381 (88) 309 (2) 291 (4) 397 (10)Norethandrolone NORE 2 447 (100) 375 (30) 357 (40) 285 (12)Tetrahydrogestrinone THG 2 457 (100) 367 (60) 295 (4) 283 (12), 311 (6), 337 (4)Tibolone TIBO 2 457 (100) 385 (24) 367 (30)4d-Tibolone TIBO-M3 2 457 (100) 385 (36) 367 (38) 295 (4)Trenbolone TREN 2 415 (100) 343 (4) 325 (5) 371 (10), 355 (6)17a-Trenbolone TREN-M 2 415 (100) 343 (5) 325 (17)Stenbolone STEN 2 447 (100) 375 (24)I-3 Multi-hydroxyl substituted(Conjugated C3 keto þ C17 multi-hydroxyl function)4-Hydroxytestosterone 4-OH-TE 3 521 (100) 449 (18) 433 (5)Fluoxymesterone FLUO 3 553 (100) 481 (34) 463 (20) 391 ([Mþ(n-2)TMS-

H2OþH]þ, 6)6b-OH-Fluoxymesterone FLUO-M3 4 641 (100) 569 (16) 551 (14)2-Hydroxymethyl-17a-

methyl-androstadiene-11a,17b-diol-3-one

FOR-M 4 635 (70) 563 (20) 545 (34) 577 (14), 529 (74), 531(100), 482 (90), 473 (46),459 (40), 455 (26), 445 (60),371 (82)

Oxabolone OXAB 3 507 (100) 435 (10) 417 (2)Oxymesterone OXYM 3 535 (100) 463 (18) 445 (32) 373 ([Mþ(n-2)TMS-

2H2OþH]þ,2)6b-OH-4-Chloro-

dehydromehty-ltestosterone

DHMT-M 3 495 (60) 405 (42) 333 (24) 315 (62) 423 ([Mþ(n-2)TMSþH]þ,100), 281 (80),289 (32)

6b-OH-Methandienone METH-M2 3 533 (46) 461 (100) 371 (32) 353 ([Mþ(n-2)TMS-2H2OþH]þ, 10), 517 (54),445 (82), 429 (32), 387 (42)

I-4 Not substituted9a-Fluoro-17,17-dimethyl-

18-nor-androsta-4,13-diene-11b-ol -3-one

FLUO-M2 2 463 (100) 391 (24)

17,17-dimethyl-18-norandrosta-1,4,13-triene-3-one

METH-M3 1 355 (100) 283 (34) 337 ([MþnTMS-H2O]þ, 20),339 (40), 229 (40), 325 (24)

Group II Unconjugated keto functional group at C3 (Unconjugated C3 keto)II-1 Mono-hydroxyl substituted at C17 (Unconjugated C3 keto þ C17 mono-hydroxyl function)Mestanolone MESTA 2 449 (40) 377 (100) 359 (48) 287 (34) 269 (10)Methasterone METHAS 2 463 (70) 391 (100) 373 (76) 301 (35) 283 (10)Oxandrolone OXAND 2 451 (4) 379 (100) 361 (8) 289 (44)17-Epioxandrolone OXAND-M 2 451 (2) 379 (100) 289 (83) 396 (14), 447 (10)Oxymetholone OXYMTH 3 549 (100) 477 (5) 459 (34) 565 (5)

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133128

Table 2 (continued )

Compound Abbr. Characteristic ionization profile

nTMS

[MþnTMSþH]þ Loss of 1TMS Loss of1TMS�H2O

Loss ofnTMS

Loss ofnTMS�H2O

Loss ofnTMS�2H2O

Others

Group III Conjugated hydroxy1 functional group at C3 (Conjugated C3 hydroxyl)III-1 Keto-substituted at C17 (Conjugated C3 hydroxyl þ C17 keto function)5a-Androst-1-ene-3a-ol-

17-one1-TE-M 2 433 (34) 361 (10) 343 (98) 271 (60) 253 (100)

4-Chloro-4-androsten-3a-ol-17-one

CLOS-M 2 467 (40) 377 (45) 305 (18) 287 (100)

III-2 Mono-hydroxyl substituted at C17 (Conjugated C3 hydroxyl þ C17 mono-hydroxyl function)Bolandiol BOLAD 2 331 (72) 259 (28) 241 (100) 215 (18)Epimetendiol METH-M1 2 269 (100) 213 (8), 201 (13)3b-OH-Tibolone TIBO-M1(b) 2 459 (35) 387 (2) 369 (65) 297 (50) 279 (100) 229 (25)3a-OH-Tibolone TIBO-M2(a) 2 459 (16) 369 (54) 297 (18) 279 (100) 229 (24)III-3 Multi-hydroxyl substituted (Conjugated C3 hydroxyl þ C17 multi-hydroxyl function)9a-Fluoro-17a-methyl-4-

androsten-3a,6b,11b,17b-tetra-ol

FLOU-M1 4 553 (85) 481 ([Mþ(n-2)TMS-H2OþH]þ, 100), 391([Mþ(n-3) TMS-2H2O þH]þ,10)

Group IV Unconjugated hydroxy1 functional group at C3 (Conjugated C3 hydroxyl)IV-1 Keto-substituted at C17 (Unconjugated C3 hydroxyl þ C17 keto function)19-Norandrosterone 19-NA 2 421 (2) 349 (32) 331 (10) 277 (18) 259 (100) 241 (56)19-Noretiocholanolone 19-NE 2 421 (10) 349 (64) 331 (4) 277 (38) 259 (100) 241 (64) 366 (15)2a-Methyl-5a-androstan-

3a-ol-17-oneDROS-M 2 449 (6) 377 (10) 359 (4) 305 (8) 287 (100) 269 (38)

1a-Methyl-5a-androstan-3a-ol-17-one

MESTE-M1 2 449 (50) 377 (8) 359 (54) 287 (48) 269 (100)

1-Methylene-5a-androstan-3a-ol-17-one

METE-M 2 447 (14) 375 (14) 357 (6) 303 (30) 285 (100) 267 (24)

IV-2 Mono-hydroxyl substituted at C17 (Unconjugated C3 hydroxyl þ C17 mono-hydroxyl function)7a-17a-Dimethyl-5b-

androstane-3a,17b-diolBOLAS-M 2 375 (6) 303 (5) 285 (100)

7b-17a-Dimethyl-5b-androstane-3a,17b-diol

CALUS-M 2 303 (3) 285 (100)

1a-Methyl-5a-androstane-3a,17b-diol

MESTE-M2 2 451 (2) 361 (20) 289 (18) 271 (100)

Methandriol METHD 2 359 (20) 287 (6) 269 (100)17a-Methyl-5a-

androstane-3a,17b-diolMET-M1 2 271 (100)

17a-Methyl-5b-androstane-3a,17b-diol

MET-M2 2 361 (2) 271 (100)

13b,17a-Diethyl-5a-gonane-3a,17b -diol

NORB-M1 2 285 (100)

13b,17a-Diethyl-5b-gonane-3a,17b-diol

NORB-M2 2 375 (8) 285 (100)

17a-Ethyl-5b-estrane-3a,17b-diol

NORE-M1 2 289 (8) 271 (100)

IV-3 Multi-hydroxyl substituted (Unconjugated C3 hydroxyl þ C17 multi-hydroxyl function)17a-Hydroxyethyl-5b-

estrane-3a, 17b-diolNORE-M2 3 287 (100) 269 ([Mþ(n-3)TMS-

3H2OþH]þ, 26), 243 (14),Group V Non-steroid structure groupClenbuterol CLEN 2 421 (98) 349 (4) 331 (100) 275 (21)Danazol DANA 2 482 (100)Ethylestrenol ETHLE 1 271 (100) 147 (28), 175 (15), 201 (10),

287 (16),Stanozolol STAN 2 473 (100) 401 (44) 383 (26) 311 (10) 489 (4)30-OH-Stanozolol STAN-M1 3 561 (100) 489 (6) 471 (22)16b-OH-Stanozolol STAN-M2 3 561 (100) 489 (84) 471 (36) 399 ([Mþ(n-2)TMS-

H2OþH]þ,12), 381 ([Mþ(n-2)TMS-2H2OþH]þ,20)

4b-OH-Stanazolol STAN-M3 3 561 (100) 471 (32) 381 ([Mþ(n-2)TMS-2H2OþH]þ,38), 399([Mþ(n-2)TMS-H2OþH]þ,8)

3-OH-Prostanozol 30-OH-PROS 3 545 (100) 473 (12) 401 ([Mþ(n-2)TMSþH]þ,2)a-Zearalanol ZERA 4 611 (16) 539 (100) 521 (30) 449 ([Mþ(n-2)TMS-

H2OþH]þ,92), 431 ([Mþ(n-2)TMS-2H2OþH]þ,84), 377([Mþ(n-3)TMS-H2OþH]þ,4)

b-Zearalanol ZERA-M 4 611 (5) 539 (96) 521 (25) 449 ([Mþ(n-2)TMS-H2OþH]þ,100), 431([Mþ(n-2)TMS-2H2OþH]þ,98),

16b-OH-Furazabol FUR-M 2 491 (100) 419 (24) 401 (56) 329 (34) 311 (50)

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133 129

Fig. 5. Representative chromatograms of 79 steroid-spiked urine samples at 10 ngmL�1 collected using GC-ESI/HRMSwith TMS derivatization. Abbreviations: 1-Androstenedione (1-AND), 4-hydroxyandrostenedione (4-OH TE-M), 6-OXO-androstenedione (6-OXO), 4-androstene-6a-ol-3,17-dione (6-OXO-M),19-norandrostenedione (19-NOR), androstatrienedione(ATD), boldione (BOLDI), 1-testosterone (1-TE), bolasterone (BOLAS), boldenone (BOLD), 5b-androst-1-en-17b-ol-3-one (BOLD-M), calusterone (CALUS), ethisterone (DANA-M),dehydrochlormethyl-testosterone (DHMT), gestrinone (GES), metenolone (METE), methandienone (METH), methyl-1-testosterone (ME-1-TE), methyldienolone (METLD), methyl-nortestosterone (MENORTE),methyltestosterone (MET),mibolerone (MIBO), nandrolone (NAND), norclostebol (NORC), norethandrolone (NORE), tetrahydrogestrinone (THG), tibolone(TIBO), 4d-tibolone (TIBO-M3), trenbolone (TREN),17a-trenbolone (TREN-M), stenbolone (STEN), 4-hydroxytestosterone (4-OH-TE), fluoxymesterone (FLUO), 6b-OH-fluoxymesterone(FLUO-M3), 2-hydroxymethyl-17a-methyl-androstadiene-11a,17b-diol-3-one (FOR-M), oxabolone (OXAB), oxymesterone (OXYM), 6b-OH-4-chloro-dehydromehtyltestosterone(DHMT-M), 6b-OH-methandienone (METH-M2), 9a-fluoro-17,17-dimethyl-18-norandrosta-4,13-diene-11b-ol-3-one (FLUO-M2), 17,17-dimethyl-18-norandrosta-1,4,13-triene-3-one(METH-M3), mestanolone (MESTA), methasterone (METHAS), oxandrolone (OXAND), 17-epioxand-rolone (OXAND-M), oxymetholone (OXYMTH), 5a-androst-1-ene-3a-ol-17-one(1-TE-M), 4-chloro-4-androsten-3a-ol-17-one (CLOS-M), bolandiol (BOLAD), epimetendiol (METH-M1), 3b-OH-tibolone (TIBO-M1), 3a-OH-tibolone (TIBO-M2), 9a-fluoro-17a-methyl-4-andro-sten-3a,6b,11b,17b-tetra-ol (FLOU-M1), 19-norandrosterone (19-NA), 19-noretiocholanolone (19-NE), 2a-methyl-5a-androstan-3a-ol-17-one (DROS-M), 1a-methyl-5a-androstan-3a-ol-17-one (MESTE-M1), 1-methylene-5a-androstan-3a-ol-17-one (METE-M), 7a-17a-dimethyl-5b-androstane-3a,17b-diol (BOLAS-M), 7b-17a-dimethyl-5b-andro-stane-3a,17b-diol (CALUS-M), 1a-methyl-5a-androstane-3a,17b-diol (MESTE-M2), methandriol (METHD), 17a-methyl-5a-androstane-3a,17b-dio (MET-M1), 17a-methyl-5b-andros-tane-3a,17b-diol (MET-M2), 13b,17a-diethyl-5a-gonane-3a,17b-diol (NORB-M1), 13b,17a-diethyl-5b-gonane-3a,17b-diol (NORB-M2), 17a-ethyl-5b-estrane-3a,17b-diol (NORE-M1),17a-hydroxyethyl-5b-estrane-3a,17b-diol (NORE-M2), clenbuterol (CLEN), danazol (DANA), ethylestrenol (ETHLE), stanozolol (STAN), 30-OH-stanozolol (STAN-M1),16b-OH-stanozolol(STAN-M2), 4b-OH-stanazolol (STAN-M3), 3-OH-prostanozol (30-OH-PROS), a-zearalanol (ZERA), b-zearalanol (ZERA-M), 16b-OH-furazabol (FUR-M).

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133 131

maintained; therefore, peaks arising from the steroids were notsuccessfully detected. The sensitivity of the GC-ESI/HRMS wasrepeatedly monitored using quality control samples to evaluate therobustness of the system. Such control experiments were per-formed before the steroid analysis.

Subsequently, the GC-ESI parameters were evaluated for theestablishment of a system. The GC-ESI parameters could not bechecked for cross-links owing to the numerous possible combina-tions of parameters; therefore, an evaluation process using GC-ESIparameters was effectively designed, in this study. First, the dis-tances and position between the MS orifice, ESI needle, and GCcolumn outlet were critical factors in the establishment of GC-ESI.The x-, y-, and z-distances were investigated, and the conditionsfor the other parameters were as follows: ESI solvent and modifier,MeOH containing 0.1% formic acid; flow rate of ESI solvent,0.1 mL min�1; and nebulizer gas flow, 5 abr. As a result, the x-, y-,and z-distances were set to approximately 7, 4, and 1 mm using 12steroids, respectively (Fig. 2). As shown in Fig. 3(A)e(C), longer x-,y-, and z-distances resulted in a decrease of the signal intensity andidentical results were observed for 12 steroids. These resultsrevealed that the short distances gave rise to better ionizationefficiency.

Parameters such as the nebulizing gas flow, ESI spray solvent/acidmodifier and flow rate of ESI spray solvent were investigated toenhance the ionization efficiency. First, several ESI spray solventsand acid modifiers were evaluated. As shown in Fig. 3(D), severalconditions, including 0.1% acetic acid or 0.1% formic acid wereinvestigated. The conditions for the other parameters were as fol-lows: The x-, y-, and z-distances, 7, 4, and 1 mm, respectively; ESIsolvent, MeOH; flow rate of ESI solvent, 0.1 mL min�1; and nebulizergas flow, 5 abr.When comparing the twoMeOH systems containing0.1% acetic acid or 0.1% formic acid, the MeOH containing 0.1%formic acid showed better ionization efficiency. As the acid modi-fier, 0.1% formic acid can help the ionization under GC-ESIconditions.

Several solvents containing 0.1% formic acid, such as MeOH, 50%MeOH, DW, IPA and ACN were compared. The conditions for theother parameters were as follows: the x-, y-, and z-distances, 7, 4,and 1mm, respectively; ESI solvent andmodifier, MeOH containing0.1% formic acid; and flow rate of ESI solvent, 0.1 mL min�1; nebu-lizer gas flow, 5 abr. The results of these experiments show thatprotic solvents provide better ionization efficiency than aproticsolvents. Among the tested protic solvents, the MeOH systemcontaining 0.1% formic acid showed the best ionization efficiency.Thus, it was used as the optimal ESI spray solvent and acidmodifier.

The flow rate of the ESI spray solvent was investigated in therange of 0.1 mL min�1 to 50 mL min�1. The conditions for the otherparameters were as follows: the x-, y-, and z-distances, 7, 4, and1 mm, respectively; ESI solvent and modifier, MeOH containing0.1% formic acid; nebulizer gas flow, 5 abr. As illustrated in Fig. 3(E),a flow rate of 0.1 mL min�1 generated the best signal intensity andthe increasing of flow rate resulted in a decrease in intensity. It iswell-known that ESI involves a concentration-dependent mecha-nism (peak dilution effect). This result implies that the present GC-ESI may undergo the same concentration-dependentmechanism asLC-ESI. Consequently, the flow rate was set to 0.1 mL min�1. Thenebulizing gas is important for the stable generation of droplets inthe ionization process. As shown in Fig. 3(D), the effect of nebu-lizing gas flow rate was investigated. The conditions for the otherparameters were as follow; the x-, y-, and z-distances, 7, 4, and1 mm, respectively; ESI solvent and modifier, MeOH containing0.1% formic acid; and flow rate of ESI solvent, 0.1 mL min�1. Whenthe nebulizing gas was not used or a lower gas flow rate was used,the best results were obtained for 12 target steroids. However, thelowgas flow gave rise to the serious fluctuations of peak intensities.

Consequently, the nebulizing gas flow rate was set to 5 abr.

3.2. Comparison of the ionization profile of liquid-phase and gas-phase steroids, and gas-phase TMS-steroids under ESI conditions

In this study, the gas-phase analytes eluted from GC were suc-cessfully protonated by the present GC-ESI/HRMS method. Basedon this result, GC-ESI/HRMS was compared with LC-ESI/MS toinvestigate the ionization profiles resulting from the different an-alyte phases. For this purpose, the comparison of full-scan spectrabetween LC-ESI and GC-ESI was performed using underivatizedsteroids. The characteristic ions of underivatized steroids aresummarized in Fig. 4. In Fig. 4(A) and (B), conjugated or unconju-gated keto functional groups at C3 (Groups I and II) were mainlyionized as the [MþH]þ ion, regardless of whether a gas- or liquid-phase analyte was used. The [MþH]þ ion was detected as themost abundant precursor ion, and the only difference originatingfrom the analyte phase was the presence of sodium adduct ions,[MþNa]þ. In Fig. 4(C) and (D), conjugated or unconjugated hydroxylfunctional groups at C3 (Groups III and IV) were mainly ionized as[MþH-H2O]þ or [MþH-2H2O]þ in both GC-ESI and LC-ESI.

As a result, both gas- and liquid-phase underivatized analytesshowed ionization profiles that were dependent on the structuralcharacteristics of the analyte, regardless of the analyte phase. Therelationship between ionization profile and structure in LC-ESI/MSis already well-known and has been discussed in previous reports[10]. The steroids containing conjugated keto functional groups atC3 showed good proton affinity and stability and generated the[MþH]þ ion as themost abundant precursor ion. On the other hand,the steroids containing conjugated- or unconjugated hydroxylfunctional groups at C3 mainly generated [MþH-H2O]þ or [MþH-2H2O]þ ions owing to poor stability [10,13]. Consequently, theionization profiles of gas-phase underivatized steroids were similarto the results obtained from liquid-phase analytes. To the best ofour knowledge, this is the first study to investigate and compare theionization profiles between analyte phases using ESI.

In a previous GC-ESI report [9], several compounds such aspyridine, 2-aminopyridine andN-vinyl-pyrrolidonewere ionized tothe [MþH]þ ion without derivatization. Furthermore, limits ofdetection, linearity, and repeatability with target compounds wereevaluated for GC-ESI/MS, GC-APCI/MS, and GC-EI/MS [9]. In gen-eral, derivatization, such as silylation, acetylation and alkylationand esterification is a suitable strategy for GC-based analysisbecause active protons such as hydroxyl groups can result in poorchromatographic properties such as peak tailing for broadening onGC column [14]. Chemical modification with TMS is the most usedderivatization, and it is known to improve sensitivity and selectivity[15]. In the present study, gas-phase steroids as TMS derivativescould be ionized under the GC-ESI/HRMS. Ionization profiles of gas-phase steroids as TMS derivatives were investigated using GC-ESI/HRMS. The obtained results were compared with those of gas-phase underivatized steroids to evaluate the effect of TMS underESI conditions. The characteristic ions of TMS derivatives aresummarized in Table 2 and Fig. 4. As shown in Fig. 4(A) and (B),conjugated or unconjugated keto functional groups at C3 (Groups Iand II) generated the [MþH]þ and [MþH-TMS]þ ions, respectively.Group I reliably generated the [MþH]þ ion as the most abundantion because of the good stability of the TMS analyte due to conju-gation. Group II generated [MþH-TMS]þ as the dominant ion via in-source dissociation, and [MþH]þ was obtained at lower abundancein Group II than in Group I. On the other hand, conjugated or un-conjugated hydroxyl functional groups at C3 (Groups III and IV)generated [MþH-2TMS-2H2O]þ as the most abundant ion. Asshown in Fig. 4(C) and (D), Groups III and IV were influenced bystructural characteristics such as poor stability. Their ionization

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133132

profiles were similar to the profile results for underivatized steroidsunder GC-ESI conditions. Groups I and II created the characteristicfragment ion [MþH-TMS]þ via in-source dissociation, corre-sponding to the loss of a neutral TMS moiety from the protonatedmolecular ion (even electron cation, EEþ). Groups III and IV gaverise to the characteristic fragment ions [MþH-TMS-H2O]þ and[MþH-2TMS-H2O]þ corresponding to the loss of a neutral TMS$H2Omoiety from the protonated molecular ions by in-source dissocia-tion. The keto functional groups at C3 (Groups I and II) lead torearrangement of double bonds when converted to TMS de-rivatives. The formation of a ketoeenol equilibriumwould increasestability and avoid the loss of the TMS$H2O moiety [16]. Pozo et al.investigated the mass spectrometric behavior of steroids using GCcoupled to an APCI source. Our ionization profiles including in-source dissociation patterns of steroids as TMS derivatives underGC-ESI, are consistent with those reported by Pozo et al. [16].Therefore, in the present study, the relationship between TMS de-rivatives, ionization profiles and dissociation was successfully

Table 3Concentration range, linearity (r2), repeatability and limit of detection (LOD) for the dev

Analyte (Abbr.) GC-ESI/MS/MS LODs (ng mL�

Conc.range(ng/mL)

Linearity(r2)

Repeatability(%RSDa)

GC-ESI/MS/M[11]

Group I1-AND 5e200 0.9864 18.31 5.0BOLAS 2e200 0.9641 20.97 2.0FLUO-M3 5e200 0.9922 16.24 5.0

Group IIMESTA 5e200 0.9977 19.23 5.0METHAS 10e200 0.9961 17.08 10.0OXAND 5e200 0.9935 21.03 5.0

Group IIIBOLAD 5e200 0.9960 16.80 5.0TIBO-M1 10e200 0.9742 9.66 10.0FLUO-M1 1e200 0.9997 10.05 1.0

Group IVMESTE-M1 5e200 0.9973 17.97 5.0NORE-M2 0.5e200 0.9955 22.14 0.519-NE 0.5e200 0.9992 13.89 0.5

a Relative standard deviation (%RSD, n ¼ 5).

Fig. 6. Averages and standard deviations for peak width at half-maximum height (Wh), ardeviation (SD, n ¼ 5), b; relative standard deviation (%RSD, n ¼ 5).

demonstrated using GC-ESI/HRMS.

3.3. Analysis of steroid TMS derivatives in human urine

Seventy-nine steroids were spiked at 10 ng mL�1 in humanurine. After sample preparation and TMS derivatization, the resul-tant samples were analyzed using GC-ESI/HRMS. Steroids as TMSderivatives were ionized and determined without significant in-terferences [11]. A representative chromatogram is shown in Fig. 5.Steroids as TMS derivatives in urine samples were protonated as[MþH]þ. In addition, fragment ions arising from the loss of Si(CH3)3or Si(CH3)3$H2O moieties from the protonated molecular ion wereobtained such as [MþH-TMS]þ, [MþH-TMS-H2O]þ, [MþH-2TMS-H2O]þ, and [MþH-2TMS-2H2O]þ. The targeted steroids wereextracted using the dominant or predominant ion, and they wereperfectly identifiable in human urine. As shown in Fig. 5, narrowpeak widths and good signal to noise ratios were obtained withoutinterference from the human urine matrix. In detail, averages and

eloped GC-ESI system.

1, S/N � 3) [11]

S GC-EI/MS[11]

GC-EI/MS/MS[11]

LC-ESI/MS/MS[10,11]

LC-AgþCIS/MS/MS[10,12]

2.0 0.1 1.0 1.01.0 2.0 0.05 0.02510.0 0.5 0.2 0.01

2.0 0.2 2.0 1.02.0 0.2 20.0 1.0200 20.0 1.0 0.25

1.0 0.5 0.5 0.051.0 0.2 10.0 0.055.0 2.0 0.5 0.01

2.0 0.2 2.0 0.252.0 2.0 0.5 1.01.0 0.1 2.0 0.2

ea and height based on the representative chromatogram at 10 ng mL�1. a; standard

Fig. 7. The robustness for GC-ESI/MS with standard deviation (SD, s) of peak areausing d3-testosterone.

E. Cha et al. / Analytica Chimica Acta 964 (2017) 123e133 133

standard deviation (SD) for peak width at half-maximum height(Wh), area and height at 10 ng mL�1 using CAUS and METE wereshown in Fig. 6. As shown in Fig. 6, Wh for CALUS and METE wereobserved as 5.796 s and 5.676 s, respectively. In addition, re-peatabilities for Wh, area and height were evaluated as relativestandard deviation (%RSD). Results showed 4.886 to 9.197 for Wh,2.837 to 10.25 for area and 8.174 to 14.61 for height. These resultsmean that GC-ESI/MS method provides the consistent and reliabledata.

This result means that for the GC-ESI/HRMSmethod, including acapillary column and performing TMS derivatization bothcontribute to the improvement of selectivity and sensitivity.Therefore, GC-ESI/HRMS with TMS derivatives is suitable forsimultaneous analysis of steroids as a novel ionization andanalytical tool.

3.4. Performance evaluations of GC-ESI

The LOD, concentration range, linearity and repeatability of thepresent GC-ESI/MS method was evaluated to compare the presentGC-ESI/MSmethodwith conventional methods. GC-ESI/MSmethodprovided reliable results and the results were summarized inTable 3. Linearity (r2) and repeatability (%RSD) were observed in therange of 0.9641 to 0.9992 and 9.66 to 22.14, respectively.

The LODs obtained using various ionization methods werecompared in Table 3 [11]. In addition, robustness of GC-ESI methodwas performed with standard deviation (SD, s), and it was evalu-ated as peak areas of d3-testosterone based on the results of 180sample analysis. As shown Fig. 7, the all most data were obtained inthe range of ±2s, hence it means that GC-ESI/MS method is stableand robust. Consequently, the present GC-ESI/MS provided satis-factory results for LOD, S/N, linearity, repeatability and robustness.We have demonstrated that the present GC-ESI/MS is reliablemethod, and established the possibility as an alternative for anal-ysis of steroids, in this study.

4. Conclusions

GC-ESI/MS was developed to combine the advantages of GC andESI using a self-made resistor-heated column transfer line. Steroids

as with or without TMS derivatization were protonated under ESIconditions; this is the first study to successfully ionize and analyzesteroids as TMS derivatives using GC-ESI/MS. Moreover, funda-mental and useful data were obtained using GC-ESI/MS, and ioni-zation profiles from analytes phases or TMS derivatives werecompared and discussed based on their structure.When comparingthe results obtained from GC-ESI/MS and LC-ESI/MS, steroidsshowed similar ionization profiles based on their structural char-acteristics, regardless of the analyte phase or derivatization. Inaddition, the developed GC-ESI/MS method was applied to theanalysis of anabolic steroids as TMS derivatives. The applicability ofour newly developed method to doping analysis been demon-strated. Consequently, the present GC-ESI/MS method has enabledthe ionization of TMS derivatives under ESI conditions; and haspotential as a novel ionization tool. These results suggest that thepresent method is suitable for the simultaneous analysis of TMSderivatives in human urine.

References

[1] World Anti-Doping Agency, Prohibited List, 2016. Available at: https://www.wada-ama/org/en/resources/science-medicine/prohibited-list [1 January2016].

[2] J. Maros, J.A. Pascual, X. de la Torre, J. Segura, Fast screening of anabolic ste-roids and other banned doping substances in human urine by gas chroma-tography/tandem mass spectrometry, J. Mass Spectrom. 37 (2002)1059e1073.

[3] A.I. Revelsky, A.S. Samokhin, E.D. Virus, G.M. Rodchenkov, I.A. Revelsky, Highsensitive analysis of steroids in doping control using gas chromatography/time-of-flight mass-spectrometry, Drug Test. Anal. 3 (2011) 263e267.

[4] O.J. Pozo, P. Van Eenoo, K. Deventer, F.T. Delbeke, Ionization of anabolic ste-roids by adduct formation in liquid chromatography electrospray massspectrometry, J. Mass Spectrom. 42 (2007) 497e516.

[5] G. Van Poucke, M. Van De Velde, C. Van Peteghem, Combination of liquidchromatography/tandem mass spectrometry and gas chromatography/massspectrometry for the detection of 21 anabolic steroid residues in bovine urine,J. Mass Spectrom. 40 (2005) 731e738.

[6] B.W. Jeon, H.H. Yoo, E.S. Jeong, H.J. Kim, C. Jin, D.H. Kim, J. Lee, LC-ESI/MS/MSmethod for rapid screening and confirmation of 44 exogenous anabolic ste-roids in human urine, Anal. Bioanal. Chem. 401 (2011) 1353e1363.

[7] C.Y. Lee, J. Shiea, Gas chromatography connected to multiple channel elec-trospray ionization mass spectrometry for the detection of volatile organiccompounds, Anal. Chem. 70 (1998) 2757e2761.

[8] H.J. Hsu, T.L. Kuo, S.H. Wu, J.N. Oung, J. Shiea, Characterization of syntheticpolymers by electrospray-assisted pyrolysis ionization-mass spectrometry,Anal. Chem. 77 (2005) 7744e7749.

[9] N. Brenner, M. Haapala, K. Vuorensola, R. Kostiainen, Simple coupling of gaschromatography to electrospray ionization mass spectrometry, Anal. Chem.80 (2008) 8334e8339.

[10] E. Cha, S. Kim, H.W. Kim, K.M. Lee, H.J. Kim, O.S. Kwon, J. Lee, Relationshipsbetween structure, ionization profile and sensitivity of exogenous anabolicsteroids under electrospray ionization and analysis in human urine usingliquid chromatography-tandem mass spectrometry, Biomed. Chromatogr. 30(2016) 555e565.

[11] E. Cha, S. Kim, H.J. Kim, K.M. Lee, K.H. Kim, O.S. Kwon, J. Lee, Sensitivity of GC-EI/MS, GC-EI/MS/MS, LC-ESI/MS/MS, LC-AgþCIS/MS/MS, and GC-ESI/MS/MSfor analysis of anabolic steroids in doping control, Drug Test. Anal. 7 (2015)1040e1049.

[12] S. Kim, E. Cha, K.M. Lee, H.J. Kim, O.S. Kwon, J. Lee, Simultaneous ionizationand analysis of 84 anabolic androgenic steroids in human urine using liquidchromatography-silver ion coordination ionspray/triple-quadrupole massspectrometry, Drug Test. Anal. 6 (2014) 1174e1185.

[13] Y.C. Ma, H.Y. Kim, Determination of steroids by liquid chromatographyemassspectrometry, J. Am. Soc. Mass Spectrom. 8 (1997) 1010e1020.

[14] K. Blau, J.M. Halket, Derivatives for Gas Chromatography-mass Spectrometry,second ed., Wiley, Chichester, 1993.

[15] E.M. Brun, R. Puchades, �A. Maquieira, Analytical methods for anti-dopingcontrol in sport: anabolic steroids with 4,9,11-triene structure in urine,TrAC Trends Anal. Chem. 30 (2011) 771e783.

[16] M. Raro, T. Portol�es, J.V. Sancho, E. Pitarch, F. Hern�andez, J. Marcos, R. Ventura,C. G�omez, J. Segura, O.J. Pozo, Mass spectrometric behavior of anabolicandrogenic steroids using gas chromatography coupled to atmosphericpressure chemical ionization source. Part I: ionization, J. Mass. Spectrom. 49(2014) 509e521.