CORE Metadata, citation and similar papers at core.ac.uk

Provided by Ghent University Academic Bibliography

Journal of Chromatography A 1624 (2020) 461231

Contents lists available at ScienceDirect

Journal of Chromatography A

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

Steroid profiling in urine of intact glucuronidated and sulfated using liquid chromatography-mass spectrometry

∗ Laurie De Wilde , Kris Roels, Pieter Van Renterghem, Peter Van Eenoo, Koen Deventer 1

Doping Control Laboratory (DoCoLab), Ghent University (UGent), Department Diagnostic Sciences, Technologiepark 30B, B-9052 Zwijnaarde, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Detection of endogenous anabolic androgenic steroids (EAAS) misuse is a major challenge in doping con- Received 26 November 2019 trol analysis. Currently, a number of endogenous steroids, which constitute the profile, are quanti-

Revised 6 May 2020 fied using gas chromatography (GC). With this methodology, only the sum of the free and glucuronidated Accepted 10 May 2020 steroids is measured together. A dilute-and-shoot LC-MS method, which is compliant with the quality Available online 23 May 2020 requirements for measuring EAAS established by the World Anti-Doping Agency (WADA), was devel- Keywords: oped and validated containing glucuronidated and sulfated steroids in order to gain some extra infor- Doping mation and to expand the existing steroid profile. The developed method is, to the best of our knowl- Urine edge, the first method to combine both steroid glucuronides and sulfates, which is compliant with the Steroid profile quality standards of the technical document on EAAS, established by WADA. The first advantage of this

Intact conjugates new steroid profile is the reduced sample preparation time, as it is a direct injection method of diluted Dilute-and-shoot urine. A second advantage is the ability of the used gradient to separate 5 α-androstane-3 α,17 β-diol- 3-glucuronide (5 ααβdiol3G), 5 α-androstane-3 α,17 β-diol-17-glucuronide (5 ααβdiol17G), 5 β-androstane- 3 α,17 β-diol-3-glucuronide (5 βαβdiol3G) and 5 β-androstane-3 α,17 β-diol-17-glucuronide (5 βαβdiol17G) allowing to gain specific information on these isomers, which cannot be accomplished in GC-MS screen- ing due to hydrolysis. This steroid profile also contains free , 5 α-androstane-3,17-dione and 5 β-androstane-3,17-dione as markers of degradation. In total, 17 compounds and 10 isotopically labelled internal standards are included in this method. © 2020 Elsevier B.V. All rights reserved.

1. Introduction trations of EAAS, which makes population limits insufficient to confirm whether an elevated concentration is a result of steroid Endogenous anabolic androgenic steroids (EAAS) are substances abuse. To circumvent this problem, the athlete’s steroidal pass- produced in the human body and are related to the metabolic port was introduced, as a part of the athlete’s biological pass- pathways of testosterone. In the clinical field, the measurement of port (ABP), which contains basal concentrations of EAAS and ra- steroids is very important for therapeutic drug monitoring, the di- tios among them, with the testosterone/ ratio (T/E) agnosis and monitoring of endocrine disorders and cancers (e.g., as the most important marker [ 8 , 9 , 11 , 16–21 ]. By using the adaptive breast, prostate and endometrium). Highly sensitive and specific model, individual reference limits can be established which allow methods are needed for an accurate diagnosis and the measure- the comparison of a measured value with results from previous ment of low concentrations of steroids. The most common matri- tests [ 8 , 9 , 11 , 16–19 , 21 ]. When suspicious results are found, a con- ces for steroid analysis are serum, plasma, urine and saliva [1–4] . firmation by gas chromatography-combustion-isotope ratio mass As the use of EAAS can enhance sports performances, it is pro- spectrometry (GC-C-IRMS) is performed to confirm the endoge- hibited by the World Anti-doping Agency (WADA) [5] . An abuse nous or exogenous origin of the steroids [ 9 , 11 , 13 , 15–17 , 19 , 20 , 22 ], with this class of doping substances is difficult to track down be- although eventually the ABP could also by itself be used to prose- cause these steroids are also formed endogenously [6–19] . Fur- cute a doping infraction. thermore, there is a large inter-individual variation in concen- The extensive metabolism of EAAS is divided into two phases. In phase I, oxidations and reductions occur to inactivate the com- pound and to facilitate its elimination. In phase II, the steroid is ∗ Corresponding author. coupled to glucuronic acid or a sulfate group by conjugation re- E-mail address: [email protected] (L. De Wilde). actions, making the steroid more polar, which also facilitates its 1 Doping Control Laboratory (DoCoLab), Ghent University, Department Diagnostic urinary excretion [ 7–9 , 14 , 15 , 17 , 22–27 ]. For example, 3 α-hydroxy Sciences, Technologiepark 30 B, B-9052 Zwijnaarde, Belgium.

https://doi.org/10.1016/j.chroma.2020.461231 0021-9673/© 2020 Elsevier B.V. All rights reserved. 2 L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231 steroids are conjugated with glucuronic acid, while 3 β-hydroxy sulfate, -d4 (AS-d4), epitestos- steroids are conjugated as sulfates [ 7 , 17 , 28 ]. Glucuronidation and terone glucuronide (EG), epitestosterone glucuronide-d3 (EG-d3), sulfatation at the 17 β-hydroxy group are well known for testos- epitestosterone sulfate (ES), epitestosterone sulfate-d3 (ES-d3), terone. The major metabolites of AAS are glucuronidated, however glucuronide (EtioG), etiocholanolone sulfate, some EAAS are also excreted as sulfates [ 7 , 14–18 , 22 , 25 ]. etiocholanolone sulfate-d5 (EtioS-d5), testosterone-d3 (T-d3), In the current steroid profile, constituted by concentrations of (TG), testosterone glucuronide-d3 (TG- EAAS and ratios among them, only glucuronidated steroids are in- d3), (TS) and testosterone sulfate-d3 (TS-d3) cluded, as T is mainly conjugated by UGT2B17 and mainly excreted were bought from NMI (Pymble, Australia). 5 α-androstane- in a glucuronidated form [ 11 , 13 , 16–18 , 20 , 22 , 24 , 29 ]. Hence, conven- 3,17-dione was purchased from Steraloids (Newport, R. I.), 5 β- tional sample preparation methods include an enzymatic hydrol- androstane-3,17-dione and testosterone from Sigma-Aldrich ysis step to release the aglycones, whereby information on the (Bornem, Belgium). sulfate (DHEAS) and free or sulfoconjugated fraction is lost. Subsequently, the samples dehydroepiandrosterone sulfate-d5 (DHEAS-d5) were bought from are preconcentrated and analysed by GC-MS after derivatisation Cerilliant (Round Rock, Texas). sulfate (EpiAS) was [ 8 , 9 , 13 , 15–18 , 20 , 22–25 , 27 ]. These steps can be time-consuming purchased from TRC (Toronto, Canada). and a source of variation, technical errors and inaccuracy [ 17 , 21 ]. As some people have a decreased glucuronidation of steroids, 2.2. Instruments and consequently a lower T/E because of a deletion/deletion (del/del) polymorphism in UGT2B17, it would be interesting to ex- The LC-MS/MS system consists of a Thermo Scientific Ultimate pand the steroid profile, to gain some information on free steroids 30 0 0, with a heated column compartment thermostated at 55 °C and to include the sulfoconjugated steroids [ 8 , 9 , 11 , 16 , 18 , 19 , 24 , 28 ]. and an XRS open autosampler, connected to a TSQ Vantage (both Previously, the group of Dehennin et al. suggested the ratio be- from Thermo Fisher Scientific). The separation of the compounds tween TG and the sum of EG and ES [ 9 , 28 ] and the ratio be- was performed on a Zorbax Eclipse Plus (C18 2.1 × 100 mm; 1.8 tween EG and ES [9] to distinguish between a physiologically and μm, Agilent Technologies). a pharmacologically high T/E, which was contradicted by the work of Borts et al. [27] . Another advantage of the addition of sulfate 2.2.1. LC gradient conjugates is the improvement of detection windows [ 11 , 19 , 24 , 30 ]. The mobile phases consisted of (A) water and (B) methanol Previous studies have shown the usefulness of epiandrosterone sul- both containing 0.01 % HCOOH/5 mM NH OOCH. A gradient elu- fate (EpiAS), which is known to be only excreted as a sulfoconju- 4 tion program started with 0 % B for 0.1 min after which it was gate [ 19 , 31–33 ]. Androsterone sulfate (AS) and etiocholanolone sul- immediately increased to 35 %. After 1 min of 35 % B, the gradient fate (EtioS) were also highlighted as promising markers after oral was further increased to 45 % B in 5 min, where it was kept for 5 testosterone undecanoate intake [ 8 , 9 ]. min and was then increased to 50 % in 19 min. For 4.5 min, the Unfortunately, inclusion of the sulfoconjugated steroids into the gradient increased to 90 % B, after which it increased to 100 % in existing GC-MS methods is difficult to achieve due to hydrolysis 0.5 min. This condition was kept for 2 min. The gradient decreased and extraction problems [ 6 , 22–24 , 27 ]. Additionally, hydrolysis of to starting conditions in 0.1 min and was kept for 0.9 min. the sulfonjugates yields the same aglycones as glucuronides and information on the phase II metabolites is lost. Previously, LC-MS has proven to be the ideal technique to anal- 2.2.2. MS parameters yse phase II metabolites of EAAS without the need to hydrolyse Detection of the steroids was carried out by electrospray ion- [ 6 , 8 , 9 , 11 , 12 , 14 , 15 , 18–20 , 23 , 25 , 27 , 34 ]. Some research groups anal- ization (ESI) in the selective reaction mode (SRM) and tuned S- ysed intact glucuronides [12] , intact sulfates [ 6 , 11 , 19 ] or a com- lens values were used. The capillary temperature was set to 350 °C. bination of both [ 8 , 9 , 15 , 18 , 25 , 27 ]. Except for the method of Pozo Sheath gas pressure, aux gas pressure and ion sweep gas pressure et al. [12] , these methods apply solid phase extraction (SPE) as were set to 45, 30 and 0.5 arbitrary units respectively. The spray sample preparation. Previous research in our laboratory has shown voltage for positive and negative polarity was set to 40 0 0 V and that even by using a dilute-and-shoot-LC-MS (DS-LC-MS) approach, 50 0 0 V, respectively. The collision gas pressure was 1.5 mTorr. steroid phase II metabolites can be detected [ 12 , 35 ]. Two transitions were found for each compound ( Table 1 ). Some The aim of the study was to develop and validate a method that compounds have the same mass and so the same transitions are can determine the steroid profile in urine at the same quality stan- applied for these compounds. Although it was difficult to separate dards as the currently used GC-MS methodology, while providing DHEAS and ES, transitions specific for ES were found and included additional information on the nature of the steroid conjugates and as well. allowing for the concurrent detection of free testosterone using LC- MS/MS. 2.3. Sample preparation

2. Materials and methods Steroid stripped urine was prepared by pouring urine of a child on a preconditioned XAD-2 column. Aliquots of 300 μl were pre- 2.1. Chemicals and reagents pared. Steroid stripped urine was spiked at six calibration lev- els using two different calibration mixes, which were prepared in Water and methanol (MeOH) were purchased from J. T. Baker methanol. The concentration of mix 1 was 10 times higher than (Deventer, The Netherlands). Formic acid (HCOOH) was ob- the lowest calibrator. The concentration of mix 2 was 5 times tained from Fisher Scientific (Geel, Belgium). Ammonium formate higher than the concentration of the fourth calibrator. 30, 60 and

(NH4 OOCH) was bought from Fisher Scientific (Loughborough, 300 μl of calibration mix 1 were added to calibrators 1,2 and 3 Leicestershire, UK). 5 α-androstane-3 α,17 β-diol-3-glucuronide respectively. 60, 120 and 240 μl of calibration mix 2 was added to (5 ααβdiol3G), 5 α-androstane-3 α,17 β-diol-17-glucuronide calibrator 4,5 and 6 respectively. The calibration mixes did not con- (5 ααβdiol17G), 5 α-androstane-3 α,17 β-diol-17-glucuronide-d4, tain 5 α-androstane-3,17-dione and 5 β-androstane-3,17-dione, be- 5 β-androstane-3 α,17 β-diol-3-glucuronide (5 βαβdiol3G), 5 β- cause there was no need for wide calibration ranges for these two androstane-3 α,17 β-diol-17-glucuronide (5 βαβdiol17G), andros- compounds. Therefore, 6, 15, 22.5, 30, 60 and 120 μl of a 5 α- terone glucuronide (AG), androsterone glucuronide-d4 (AG-d4), androstane-3,17-dione solution (1 μg/ml) were added to calibrators L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231 3

Table 1 Transitions and collision energies (CE) for every compound.

Compound Chemical formula Polarity Parent Product CE (eV)

∗ 5 ααβdiol3G C 25 H 40 O 8 - 467.2 84.9 40 74.9 40 ∗ 5 ααβdiol17G C 25 H 40 O 8 - 467.2 84.9 40 74.9 40

5 ααβdiol17G-d4 C 25 D 4 H 36 O 8 - 471.2 84.9 40 ∗ 5 α-androstane-3,17-dione C 19 H 28 O 2 + 289.2 213.2 16 271.2 8 ∗ 5 βαβdiol3G C 25 H 40 O 8 - 467.2 84.9 40 74.9 40 ∗ 5 βαβdiol17G C 25 H 40 O 8 - 467.2 84.9 40 74.9 40 ∗ 5 β-androstane-3,17-dione C 19 H 28 O 2 + 289.2 213.2 16 271.2 8 ∗ AG C 25 H 38 O 8 - 465.2 85.0 35 + 484.0 141.0 30

AG-d4 C 25 D 4 H 34 O 8 - 469.2 85.0 30 ∗ AS C 19 H 30 O 5 S - 369.2 96.9 35 80.0 60

AS-d4 C 19 D 4 H 26 O 5 S - 373.1 97.9 35 ∗ DHEAS C 19 H 28 O 5 S - 367.1 96.9 45 80.1 80

DHEAS-d5 C 19 D 5 H 23 O 2 - 372.1 97.8 45 ∗ EpiAS C 19 H 30 O 5 S - 369.2 96.9 35 80.0 60 ∗ EG C 25 H 36 O 8 - 463.2 85.0 35 + 465.3 289.4 19

EG-d3 C 25 D 3 H 33 O 8 - 466.2 85.0 25 ∗ ES C 19 H 28 O 5 S - 367.1 96.9 45 + 369.2 271.2 13

ES-d3 C 19 D 3 H 25 O 5 S - 370.1 97.9 40 ∗ EtioG C 25 H 38 O 8 - 465.2 85.0 35 + 484.0 141.0 30 ∗ EtioS C 19 H 30 O 5 S - 369.2 96.9 35 80.0 60

EtioS-d5 C 19 D 5 H 25 O 5 S - 374.1 80.0 35 ∗ T C 19 H 28 O 2 + 289.2 97.0 25 109.0 27

T-d3 C 19 D 3 H 25 O 2 + 292.2 97.0 21 ∗ TG C 25 H 36 O 8 + 465.3 289.4 19 108.9 39

TG-d3 C 25 D 3 H 33 O 8 + 468.0 97.0 30 ∗ TS C 19 H 28 O 5 S - 367.1 96.9 45 + 369.2 97.0 30

TS-d3 C 19 D 3 H 25 O 5 S - 370.1 98.0 50

∗ Ion transitions used for quantification.

1 to 6 respectively and 6, 12, 30, 60, 120 and 240 μl of a 5 β- 2.4.1. Linearity androstane-3,17-dione solution (500 ng/ml) was added to calibra- Three replicates of the calibration curve were prepared by tors 1 to 6 respectively. Table 2 presents the concentrations of the spiking steroid stripped urine at 6 concentration levels (cf. 2.3) steroid conjugates in every calibration sample. in compound-specific ranges ( Table 3 ) and evaluated using a 50 μl internal standard methanolic solution(5 ααβdiol17G-d4, weighted regression model (1/X). Linearity was evaluated at three

AG-d4, AS-d4, DHEAS-d5, EG-d3, ES-d3, EtioS-d5, T-d3, TG-d3; 1 levels, namely R², bias per level and the goodness-of-fit factor (g-  μg/ml) was added to an aliquot of 300 μl urine. The sample was ( )2 factor, g) calculated by the formula: g = %bias , with n being then centrifuged (5 min, 50 0 0 x g) and the supernatant was trans- n −1 ferred to a vial. 80 μl of sample was injected into the system. the number of calibrators. The maximum value for the goodness- of-fit factor depended on the concentration of the calibrators. If the concentration of more than half of the calibrators was below 100 ng/ml, the maximum g-factor value was 20. If the concentration of more than half of the calibrators was above 100 ng/ml, the allowed 2.4. Validation g-factor value was 10.

The steroids analysed with this method are endogenously pro- 2.4.2. Bias and precision duced and naturally found in urine samples. Therefore steroid Bias and precision were evaluated at the highest, the lowest stripped urine was chosen as a surrogate matrix to validate the and the fourth calibrator. To evaluate repeatability, the ratio of quantitative aspects of the method; i.e., preparation of the cali- the standard deviation and the nominal concentration was cal- bration curve, internal bias, repeatability and intermediate preci- culated. Intermediate precision was assessed by two analysts on sion. Selectivity, matrix effect, external bias and compliance with three separate days. External bias could only be evaluated for WADA’s stringent demands for quantitation were tested with real the glucuronidated steroids as there were no external reference urine samples. values for the sulfated steroids available. 25 external quality as- 4 L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231

Table 2 Concentrations of the steroid conjugates in every calibration sample.

Compound C1 C2 C3 C4 C5 C6

5 ααβdiol3G 5 10 50 200 400 800 ng/ml 5 ααβdiol17G 5 10 50 200 400 800 ng/ml 5 α-androstane-3,17-dione 20 50 75 100 200 400 ng/ml 5 βαβdiol3G 5 10 50 200 400 800 ng/ml 5 βαβdiol17G 5 10 50 200 400 800 ng/ml 5 β-androstane-3,17-dione 10 20 50 100 200 400 ng/ml AG 200 400 2000 3000 6000 12000 ng/ml AS 50 100 500 750 1500 3000 ng/ml DHEAS 20 40 200 2000 4000 8000 ng/ml EpiAS 15 30 150 250 500 1000 ng/ml EG 2 4 20 50 100 200 ng/ml ES 2 4 20 30 60 120 ng/ml EtioG 200 400 2000 3000 6000 12000 ng/ml EtioS 20 40 200 250 500 1000 ng/ml T 0.1 0.2 1 2.5 5 10 ng/ml TG 1 2 10 50 100 200 ng/ml TS 1 2 10 12.5 25 50 ng/ml

sessment scheme (EQAS) samples, organised by WADA and anal- result; Sw is theintermediate precision calculated by the following 2 2 ysed by all WADA–accredited doping control laboratories, were formula: S w = Bias % + RSD % and B ext is the external bias deter- analysed and z-scores were calculated using the following for- mined by the average bias in EQAS samples, which was calculated measured v alue − assigned v alue mula: z = . In this equation, the tar- t arget v alue for st and ard d e v iation with the measured concentrations of the conjugates expressed as get value for standard deviation depends on the compound. The concentrations of the free steroids. This was only calculated for allowed standard deviation is 25 % for the diols, 20 % for A, the parameters of the current steroid profile and at three differ- Etio, T and E and 15 % for the T/E ratio as defined by the In- ent levels, namely LOQ, QC1 and QC2. Measurement uncertainty, ternational Standard for Laboratories by WADA [36] . To be able described in WADA’s technical document on EAAS [38] and deci- to compare the EQAS results with the results of the devel- sion limits [39] , cannot exceed 30 % at the respective LOQ, 20 % oped method, the concentrations of the glucuronides were ex- for A and Etio or 25 % for the diols at 5 times the LOQ and 20 % pressed as concentrations of the free steroids whereby the con- for T and E when the concentration is greater than 5 ng/ml. centrations of 5 αdiol and 5 βdiol were the sums of their respec- tive glucuronide isomers (5 αdiol = 5 ααβdiol3G + 5 ααβdiol17G; 5 βdiol = 5 βαβdiol3G + 5 βαβdiol17G). 2.5. Comparison with routine GC method 2.4.3. Matrix effect

As the AAS in this method are naturally present in urine, de- 70 routine samples, previously analysed with the GC routine termination of the matrix effect was difficult. To circumvent this method [40], were analysed with the developed LC method. Re- problem, a previously used approach was applied [12] . Ten urines sults were compared using Bland-Altman plots, which express the

(specific gravity between 1.004 and 1.035; pH between 5.1 and 7.2) absolute difference between a sample measured on GC and on LC and a water sample were spiked with the internal standards. The plotted against the mean of the two measurements, and linear re- average area of the internal standards in the urine samples was gression plots. compared to the area in the water sample and the matrix effect was calculated with the formula: matrix effect (%) = ((urine/water area ratio) x 100) –100[37]. 2.6. Excretion study

2.4.4. Carry-over One capsule of testosterone undecanoate (TU) (Testocaps, 40 Carry-over was assessed by analysing a blank matrix sample af- mg) was administered orally to 6 healthy, male volunteers (age: ter an injection of the highest calibrator. 23-26 years, weight: 64-94 kg). Blank urine samples were collected for one week before administration in the morning, at noon and in 2.4.5. Quality control samples the evening. A blank sample was delivered just before administra- With each batch of samples, 2 quality control (QC) sampleswere tion as well. Post-administration samples were collected at 2 h, 4 analysed. These QCs existed of steroid stripped urine spiked at 2 h, 6 h, 8 h, 10 h, 12 h, 24 h, 30 h, and 36 h. From then on sam- different levels ( Table 3 ). The low QC (5LOQ) was spiked with 150 ples were collected in the morning, at noon and in the evening. All α μl of the first calibration mix, 18 μl of 5 -androstane-3,17-dione samples were stored frozen (-20 °C) until analysis. This study was β (1 μg/ml) and 21 μl of 5 -androstane-3,17-dione (500 ng/ml). The approved by the ethics committee of the Ghent University Hospital high QC was spiked with 90 μl of the second calibration mix, 45 μl (B67020064707). Each volunteer gave his written informed consent α β of 5 -androstane-3,17-dione (1 μg/ml) and 90 μl of 5 -androstane- prior to the study. 3,17-dione (500 ng/ml). Results were compared between the developed method and the routine GC method [40] by expressing the results of the LC method 2.4.6. Measurement uncertainty as concentrations of the free steroids. The concentrations of 5 αdiol Measurement uncertainty was determined with the use of in- and 5 βdiol were the sums of their respective glucuronide iso- termediate precision, QC and EQAS results, using the following for- mers. To compare urine samples longitudinally, concentrations of

( ) = 2 + ( )2 mula:uc y Sw u Bext . In this equation, uc (y) represents the studied steroids were corrected for dilution using the equation 1 . 020 − 1 the standard combined uncertainty associated with an individual C . = ∗ C . 1 020 Specific gra v ity −1 sample L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231 5

Table 3

Retention time (t R ), calibration range, R ², goodness-of-fit factor (g), bias, repeatability, intermediate precision, matrix effect (ME) ± SD. Concentrations are expressed as concentrations of the intact steroid conjugate.

Concentration Intermediate

t R Range QC1 QC2 level Repeatability precision ME Compound (min) (ng/ml) (ng/ml) (ng/ml) R ² g (ng/ml) Bias (%) (%) (%) IS (%)

5 ααβdiol3G 17.9 5-800 25 300 9.5 5 -4.2 4.0 9.7 5 ααβdiol17G-d4 -7.7 0.9980 ± 2.9 200 2.5 2.0 2.0 800 2.5 3.0 4.6 5 ααβdiol17G 18.9 5-800 25 300 7.2 5 -6.1 3.7 9.0 5 ααβdiol17G-d4 -7.7 0.9984 ± 2.9 200 2.2 2.5 2.5 800 6.1 3.3 5.4 5 α-androstane- 28.5 20-400 60 150 9.1 20 6.6 13.6 16.3 T-d3 -3.3 3,17-dione 0.9857 ± 4.9 100 -6.4 7.2 7.2 400 -2.3 9.4 10.0 5 βαβdiol3G 17.2 5-800 25 300 4.0 5 1.0 4.6 15.3 5 ααβdiol17G-d4 -7.7 0.9992 ± 2.9 200 0.4 1.8 1.8 800 1.4 2.9 4.6 5 βαβdiol17G 21.1 5-800 25 300 8.7 5 -9.8 4.4 7.9 5 ααβdiol17G-d4 -7.7 0.9982 ± 2.9 200 1.8 1.7 1.7 800 1.5 3.4 4.5 5 β-androstane- 31.7 10-400 35 150 9.0 50 -0.9 12.7 14.2 T-d3 -3.3 3,17-dione 0.9909 ± 4.9 200 -0.6 5.1 5.1 400 1.9 3.5 8.1 AG 22.1 200- 3.1 200 -5.7 1.7 9.5 AG-d4 -2.7 12000 1000 4500 0.9995 ± 2.2 3000 -5.9 1.7 3.8 12000 -2.8 3.4 5.0 AS 18.7 50- 250 5.3 50 5.9 1.4 10.7 AS-d4 3000 1125 0.9970 15.6 ± 9.9 750 10.4 1.2 1.2 3000 13.9 2.5 5.8 DHEAS 11.2 20- 100 3.8 20 3.1 1.4 11.3 DHEAS-d5 -2.8 8000 3000 0.9995 ± 6.6 2000 3.0 1.3 1.3 8000 5.8 2.5 5.1 EpiAS 13.0 15- 75 375 6.4 15 11.5 1.4 12.3 TS-d3 1000 0.9943 26.0 ± 10.7 250 8.6 1.1 1.1 1000 15.0 2.0 7.7 EG 16.2 2-120 10 75 5.3 2 9.1 3.3 14.6 EG-d3 0.9 0.9977 ± 3.0 50 5.4 1.8 1.8 200 11.6 2.8 7.1 ES 12.5 2-120 10 45 3.6 2 7.1 2.6 11.0 ES-d3 0.9991 40.7 ± 16.6 30 2.0 0.7 0.7 120 5.6 2.3 3.6 EtioG 19.5 200- 3.5 200 -2.8 1.6 9.6 AG-d4 -2.7 12000 1000 4500 0.9989 ± 2.2 ( continued on next page ) 6 L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231

Table 3 ( continued )

Compound t R Range QC1 QC2 R ² g Bias (%) IS ME (min) (ng/ml) (ng/ml) (ng/ml) Concentration Repeatability Intermediate (%) level (%) precision (ng/ml) (%)

3000 -6.8 1.6 1.6 12000 3.2 2.9 3.5 EtioS 17.6 20- 100 375 3.9 20 9.5 1.7 13.5 EtioS-d5 -0.4 1000 0.9985 ± 7.2 250 4.1 1.2 1.2 1000 6.3 2.0 4.2 T 21.8 0.1-10 0.5 3.75 0.1 47.3 17.8 37.0 T-d3 -3.3 0.9909 18.8 ± 4.9 2.5 10.5 4.7 4.7 10 2.9 2.8 7.9 TG 9.0 1-200 5 75 3.0 1 6.7 2.7 10.3 TG-d3 - 0.9996 12.7 ± 13.2 50 5.3 1.9 1.9 200 11.7 2.0 7.3 TS 10.0 1-50 5 5.2 1 -4.5 3.3 6.0 TS-d3 18.75 0.9986 26.0 ± 10.7 12.5 2.2 0.8 0.8 50 3.9 2.5 4.8

3. Results and discussion different isomers due to hydrolysis. Fig. 1 illustrates that, even at the highest calibration level, the diol glucuronides could be sepa- 3.1. Method development rated. As urine is injected directly onto the column, lifetime of the 3.1.1. Chromatography column was limited. The major issues observed were backpres- Separation of isomeric conjugates is a challenging task, espe- sure increasing quickly, retention time shifts and peak broadening. cially when they are present in different concentration ranges. Es- Eventually separation of the different compounds, especially the pecially separation between DHEAS and ES was a challenging task. diol glucuronides and EtioS/AS, was determinative. In summary, Therefore, different analytical columns were tested. Initially, an during this study, 5 columns were used (approximating about 3550 OmniSpher C18 column (100 × 2 mm, 3 μm), previously used by injections). Pozo et al. [12] , was investigated, but the desired separation was not achieved. An interesting column, the Zorbax Eclipse Plus C18 3.2. Sample preparation column (2.1 × 100 mm, 1.8 μm), was used in the work of Esquivel et al. [6] , allowing the separation of DHEAS and ES. In the first step of the development, turbulent flow online SPE As glucuronides and sulfates contain an acidic part, they can be was investigated as this had already been proven to be an effective ionized as [M-H] −. However, some glucuronides and sulfates can way of sample clean-up using no sample preparation [41] . + + + + also be ionized in positive mode as [M H] or [M NH4 ] . There- Previous work showed that phase II metabolites can be detected fore the addition of NH4 OOCH in the mobile phase is suitable for with sufficient sensitivity using a direct injection strategy, without the detection of these glucuronides and sulfates [ 12 , 23 , 25 , 26 , 34 ]. the need for sample clean-up [ 12 , 35 ]. Therefore, this approach was

Initially, 1 mM of NH4 OOCH was added to the mobile phase. applied. In order to improve peak shape, discussed in the work of Marcos The little sample preparation brings a couple of benefits. Next et al. [34] , the ionic strength of the mobile phase was increased to the reduced sample volume needed for analysis (300 μl), work- ± by adding 20 mM NH4 OOCH. Because of sensitivity problems, the load and sample preparation time ( 1 day for the routine GC concentration of NH4 OOCH was decreased to 5 mM, which gave a method versus adding internal standard, centrifuging and trans- good compromise between good peak shape and sensitivity and is ferring to a vial for the developed method), there are no param- in agreement with the findings of Esquivel et al. [6] . eters for hydrolysis and derivatisation that have to be checked as The gradient described in the work of Pozo et al. [12] was the intact conjugates can be analysed, avoiding technical errors first investigated, but the separation for the sulfates was not suf- described by Mareck et al. [21] . By omitting hydrolysis, LLE and ficient. The final gradient was based on the one used by Esquivel derivatisation, the developed method requires fewer reagents and et al. [6] but modifications are made since there are glucuronides is less prone to variation and human errors [17] . included in the method as well. Adjustments were needed for the separation of AG and EtioG, 5 α-androstane-3,17-dione and 3.3. Validation 5 β-androstane-3,17-dione and the diol glucuronides. Despite the fact that it is a long gradient, separation between 5 α- and 5 β- To establish calibration curves and test repeatability and inter- androstane-3,17-dione and all diol glucuronides was obtained. This nal bias, steroid stripped urine was used as a surrogate matrix. demonstrates another advantage of the developed method, as the This was preferred to other surrogate matrices, including synthetic GC-MS screening method is not able to distinguish between the urine, as best practice, as in the opinion of the authors it still rep- resents the best possible resemblance to real urine samples. It Is L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231 7

Fig. 1. Separation of the diol glucuronides at the highest calibration level on a new column.

clear that the stripping process also potentially removes other sub- ceeded 3 (unsatisfactory z-scores). The overall bias of these 25 stances present in the matrix. Therefore, for selectivity and exter- samples was 14.0 %, 16.0 %, 11.8 %, 8.8 %, 5.0 % and 10.2 % for 5 α- nal bias, real urine samples were used. The external bias was also diol, A, 5 β-diol, E, Etio and T respectively. included into the measurement uncertainty calculations. Finally, the cross validation/comparison of the results of this method with 3.3.3. Matrix effect the routine method, using GC-MS, and analysis of WADA EQAS If the matrix effect had been determined in steroid stripped samples also shows that the surrogate matrix was adequate. urine, the contribution of other endogenous steroids would not have been taken into account. Therefore, 10 real urine samples 3.3.1. Linearity were spiked with the deuterated internal standards to assess ma- A linear weighted regression model of 1/X was used for every trix effect [12] . As can be concluded from the results presented in compound except for 5 ααβdiol3G, 5 ααβdiol17G, 5 βαβdiol17G Table 3 , the matrix effect is small despite the fact that it is a di- and AG where a quadratic weighted regression model was used rect injection method. It might be that interferences are excluded due to the contribution of the natural abundances of the heavier due to the long gradient. The bigger standard deviations can be ex- isotopes in the target analyte [42] . As the g-factor is a more strin- plained by the fact that 10 different urine samples were used for gent criterion than bias per calibration level, occasionally outliers this experiment. had to be removed from the calibration curves with a maximum of 2 per calibration curve. In Table 3 , g-factors are presented for one 3.3.4. Carry-over of the intermediate precision batches. The g-factor is below 10 for Only for EtioG (0.01 %) and AG (0.02 %), carry-over could be ob- every compound except for T (18.8). As all calibrators are below served in the blank matrix sample analysed after the highest cali- 100 ng/ml, a maximum g-factor of 20 is allowed. brator. In every batch analysed, all calibration curves were compliant with the g-factor, R was higher than 0.98 and bias per calibrator 3.3.5. Measurement uncertainty level never exceeded 20 %. 18 intermediate precision measurements and 24 QC measure- ments (22 for 5 α-diol due to bad peaks) are included in the 3.3.2. Bias and precision calculation of measurement uncertainty. The results, tabulated in The results in Table 3 show that bias was the highest for free Table 4 , meet the criteria described in WADA’s technical document T at the lowest level (0.1 ng/ml) which is by far the lowest level on EAAS [38] . evaluated in this work. As free T is a marker of degradation [21] , the concentration only becomes important in case it exceeds the 3.4. Comparison with routine GC method threshold of 5 % of the total T concentration [38] . Quantification at such a low level of 0.1 ng/ml is therefore less important. The Bland-Altman plots and linear regression plots were chosen to bias of all other compounds never exceeded 20 %. Repeatability compare the developed LC method with the routine GC method and intermediate precision results have the same trend of being [40] for the glucuronidated steroids. The concentrations of these the highest for testosterone at the lowest level. steroids had to be expressed as the concentrations of the free The evaluation of 25 EQAS samples for the determination of steroids. Outliers were not removed. As can be seen from the re- the external bias generated 175 values. Only 6 out of 175 (3.4 %) sults shown in Fig. 2 , the mean differences approximate 0 (T: - z-scores were above 2 (questionable z-scores). Z-scores never ex- 1.6; E: -0.4; Etio: 43.3; 5 α-diol: -6.2; 5 β-diol: -19.5), except for 8 L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231

Fig. 2. Left side: Comparison between the developed LC method and the routine GC method represented by Bland-Altman plots. The yellow line represents the mean difference; the range in the orange lines contains 95 % of all values. Right side: Correlation between the routine GC method and the developed LC method.

Fig. 3. Steroid profile for one volunteer during an excretion study of oral testosterone undecanoate. The LC results are expressed as concentrations of the free steroids and are presented by dots, the GC results are presented by triangles. For the LC results, 5 αdiol = 5 ααβdiol3G + 5 ααβdiol17G; 5 βdiol = 5 βαβdiol3G + 5 βαβdiol17G.

A (-426.7). A Bland-Altman plot only gives information about the The results of the comparison with the routine GC method absolute difference between two measurements. When looking at [40] demonstrate that the developed LC method is fit for purpose. the higher concentrations of A, a bigger absolute difference can be seen while the relative difference is in line with the lower concen- 3.4.1. Measuring diol glucuronides separately trations. The linear regression plots show the correlation between The concentrations of the diol glucuronides were compared in the two methods. The slope of each regression curve approximates 81 samples. After performing a Wilcoxon Signed Rank Test, it can 1 (5 α-diol: 0.97; 5 β-diol: 1.11; A: 1.09; E: 0.97; Etio: 0.89 and T: be concluded that the concentration of 17-glucuronide is signifi- 1.03), meaning that there is a good correlation between both meth- cantly higher than the concentration of the 3-glucuronide for both ods. 5 α-diol glucuronide (58 out of 81, Z = -4.9; p = 7.6 E-7) and 5 β- L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231 9

Table 4 unteer. The pre-administration samples were used to calculate α β Measurement uncertainty for 5 -diolG, AG, 5 -diolG, EG, EtioG the 99 % reference limit (mean + 3x standard deviation (SD)). and TG at levels LOQ, QC1 and QC2. 5 βαβdiol17G/5 βαβdiol3G exceeded the individual threshold for 5

Uc LOQ (%) Uc QC1 (5LOQ) (%) Uc QC2 (%) volunteers. 5 α-diolG 17.3 18.8 17.2 In the research of Esquivel et al. [19] , a single dose of 120 mg AG 19.0 16.8 17.9 TU was administered. AS/ES and EpiAS/ES were found to be the β 5 -diolG 17.2 14.8 14.8 best candidates for the detection of oral T misuse and resulted EG 17.8 13.3 12.5 in longer detection times compared to the T/E ratio. In our study, EtioG 10.8 6.8 9.1

TG 18.2 12.9 11.8 mainly the results for volunteer 6 were similar to the ones of Es- quivel et al. [19] ( Fig. 5 ). This was less pronounced for the other volunteers.

4. Conclusion

An LC-method for the simultaneous detection of intact steroid glucuronides and sulfates was developed and validated, which is compliant with the quality criteria established by WADA for the reporting of analytical results of steroid profiling data. As all pa- rameters of the current steroid profile are included, the developed steroid profile screening method, in only 300 μl of urine, could be used as an alternative for the GC-MS based methods. The work- load is greatly reduced compared to the routine GC method as there is no sample preparation, no hydrolysis and no derivatisa- tion because of the dilute-and-shoot protocol. By analysing routine and EQAS samples, the method was found to be fit for purpose. A βαβ βαβ Fig. 4. 5 diol17G/5 diol3G after the oral administration of testosterone un- testosterone undecanoate administration study confirmed the im- decanoate. The orange line represents the 99 % reference limit (mean + 3x SD). portance of expanding the current steroid profile with steroid sul- fates.

diol glucuronide (72 out of 81, Z = -7.3; p = 2.6 E-13) in the ma- Declaration of Competing Interest jority of the samples. The authors declare that they have no known competing finan- 3.5. Excretion study cial interests or personal relationships that could have appeared to influence the work reported in this paper. In Fig. 3 , the steroid profile for one volunteer analysed with the developed method is presented together with his steroid profile analysed by the routine GC method [40] . In order to make a com- CRediT authorship contribution statement parison, concentrations of glucuronidated steroids were expressed as concentrations of the free steroids. As expected, the T/E ratio Laurie De Wilde: Conceptualization, Formal analysis, Investiga- reaches a maximum (T/E = 40) after 2 hours post-administration tion, Methodology, Project administration, Validation, Visualization, [8] . The concentrations of the metabolites also increase soon after Writing - original draft. Kris Roels: Methodology, Writing - review the administration. The same trend can be seen for both methods & editing. Pieter Van Renterghem: Conceptualization, Validation, and shows the similarity between the two methods. Writing - review & editing. Peter Van Eenoo: Conceptualization, Fig. 4 shows the 5 βαβdiol17G/5 βαβdiol3G ratio after the Resources, Writing - review & editing. Koen Deventer: Conceptu- oral administration of testosterone undecanoate for the same vol- alization, Methodology, Resources, Writing - review & editing.

Fig. 5. TG/EG and EpiAS/ES after the oral administration of TU for volunteer 6. The orange line represents the individual reference limit (mean + 3x SD). 10 L. De Wilde, K. Roels and P. Van Renterghem et al. / Journal of Chromatography A 1624 (2020) 461231

References [23] C. Gomez, A. Fabregat, Ó.J. Pozo, J. Marcos, J. Segura, R. Ventura, Analytical strategies based on mass spectrometric techniques for the study of steroid

[1] M.M. Kushnir, A.L. Rockwood, W.L. Roberts, B. Yue, J. Bergquist, A.W. Meikle, metabolism, TrAC - Trends Anal. Chem. 53 (2014) 106–116, doi:10.1016/j.trac.

Liquid chromatography tandem mass spectrometry for analysis of steroids in 2013.08.010.

clinical laboratories, Clin. Biochem 44 (2011) 77–88, doi: 10.1016/j.clinbiochem. [24] A.T. Cawley, R. Kazlauskas, G.J. Trout, A.V. George, Determination of urinary

2010.07.008 . metabolites using ion paired extraction, J. Chromatogr. B Anal.

[2] O. Nozaki, Steroid analysis for medical diagnosis, 935 (2001) 267–278. Technol. Biomed. Life Sci. 825 (2005) 1–10, doi:10.1016/j.jchromb.2005.01.042.

[3] B.E. Henderson , R.K. Ross , M.C. Pike , J.T. Casagrande , Endogenous Hormones as [25] F. Buiarelli, F. Coccioli, M. Merolle, B. Neri, A. Terracciano, Development of a

a Major Factor in Human Cancer, Cancer Res 42 (1982) 3232–3239 . liquid chromatography-tandem mass spectrometry method for the identifica-

[4] L. Konieczna, T. B aczek,˛ M. Belka, A. Fel, M. Markuszewski, W. Struck, tion of natural steroids and their conjugates in urine samples, Anal.

M. Markuszewski, R. Kaliszan, Steroid profiles as potential biomarkers in pa- Chim. Acta. 526 (2004) 113–120, doi:10.1016/j.aca.2004.09.068.

tients with urogenital tract cancer for diagnostic investigations analyzed by [26] A. Fabregat, O.J. Pozo, J. Marcos, J. Segura, R. Ventura, Use of LC-MS/MS for the

liquid chromatography coupled to mass spectrometry, J. Pharm. Biomed. Anal. open detection of steroid metabolites conjugated with glucuronic acid, Anal.

73 (2013) 108–115, doi: 10.1016/j.jpba.2012.03.016 . Chem 85 (2013) 5005–5014, doi:10.1021/ac4001749.

[5] The World Anti-Doping Agency, Standard Prohibited List, (2019) 1–9. https:// [27] D.J. Borts, L.D. Bowers, Direct measurement of urinary testosterone and

www.wada-ama.org/en/resources/science-medicine/prohibited- list- documents epitestosterone conjugates using high-performance liquid chromatogra-

(Accessed 5 February 2019). phy/tandem mass spectrometry, J. Mass Spectrom 35 (2000) 50–61, doi:10.   [6] A. Esquivel, É. Alechaga, N. Monfort, R. Ventura, Direct quantitation of endoge- 1002/(SICI)1096-9888(200001)35:1 50::AID-JMS912 3.0.CO;2-J.

nous steroid sulfates in human urine by liquid chromatography-electrospray [28] L. Dehennin, P. Lafarge, P. Dailly, D. Bailloux, J.P. Lafarge, Combined pro-

tandem mass spectrometry, Drug Test. Anal. (2018) 1–10, doi: 10.1002/dta.2413 . file of androgen glucuro- and sulfoconjugates in post-competition urine of

[7] W. Schänzer , Metabolism of anabolic androgenic steroids, Clin. Chem. 1020 sportsmen: A simple screening procedure using gas chromatography-mass

(1996) 1001–1020 . spectrometry, J. Chromatogr. B Biomed. Appl. 687 (1996) 85–91, doi:10.1016/

[8] F. Badoud, J. Boccard, C. Schweizer, F. Pralong, M. Saugy, N. Baume, Profiling of S0378-4347(96)00131-4.

steroid metabolites after transdermal and oral administration of testosterone [29] P. Martín-Escudero, J. Muñoz-Guerra, N. Del Prado, M.G. Canales, M.F. Ferrer,

by ultra-high pressure liquid chromatography coupled to quadrupole time-of- S. Vargas, A.B. Soldevilla, E. Serrano-Garde, F. Miguel-Tobal, M.M. de las Casas,

flight mass spectrometry, J. Steroid Biochem. Mol. Biol. 138 (2013) 222–235, C. Fernandez-Pérez, Impact of UGT2B17 gene deletion on the steroid profile of

doi: 10.1016/j.jsbmb.2013.05.018 . an athlete, Physiol. Rep. 3 (2015) 1–7, doi:10.14814/phy2.12645.

[9] J. Boccard, F. Badoud, E. Grata, S. Ouertani, M. Hanafi, G. Mazerolles, P. Lan- [30] W. Abushareeda, A. Vonaparti, K. Al Saad, M. Almansoori, M. Meloug, A. Saleh,

téri, J.L. Veuthey, M. Saugy, S. Rudaz, A steroidomic approach for biomarkers R. Aguilera, Y. Angelis, P.L. Horvatovich, A. Lommen, M. Alsayrafi, C. Geor-

discovery in doping control, Forensic Sci. Int. 213 (2011) 85–94, doi: 10.1016/j. gakopoulos, High resolution full scan liquid chromatography mass spectrom-

forsciint.2011.07.023 . etry comprehensive screening in sports antidoping urine analysis, J. Pharm.

[10] L. Dehennin, On the origin of physiologically high ratios of urinary testosterone Biomed. Anal. 151 (2018) 10–24, doi:10.1016/j.jpba.2017.12.025.

to epitestosterone: Consequences for reliable detection of testosterone admin- [31] T. Piper, W. Schänzer, M. Thevis, Genotype-dependent metabolism of exoge-

istration by male athletes, J. Endocrinol 142 (1994) 353–360, doi: 10.1677/joe.0. nous testosterone - new biomarkers result in prolonged detectability, Drug

1420353 . Test. Anal. (2016) 1163–1173, doi:10.1002/dta.2095.

[11] A. Esquivel, É. Alechaga, N. Monfort, S. Yang, Y. Xing, W. Moutian, R. Ventura, [32] T. Piper, M. Putz, W. Schänzer, V. Pop, M.D. McLeod, D.R. Uduwela, B.J. Steven-

Evaluation of sulfate metabolites as markers of intramuscular testosterone ad- son, M. Thevis, Epiandrosterone sulfate prolongs the detectability of testos-

ministration in Caucasian and Asian populations, Drug Test. Anal. (2019) 1218– terone, 4-, and misuse by means of car-

1230, doi: 10.1002/dta.2598 . bon isotope ratio mass spectrometry, Drug Test. Anal. 9 (2017) 1695–1703,

[12] O.J. Pozo, P. Van Eenoo, W. Van Thuyne, K. Deventer, F.T. Delbeke, Direct quan- doi:10.1002/dta.2291.

tification of steroid glucuronides in human urine by liquid chromatography- [33] T. Piper, U. Mareck, H. Geyer, U. Flenker, M. Thevis, P. Platen, W. Schanzer, De-

electrospray tandem mass spectrometry, J. Chromatogr. A. 1183 (2008) 108– termination of 13C/12C ratios of endogenous urinary steroids: method valida-

118, doi: 10.1016/j.chroma.2008.01.045 . tion, reference population and application to doping control purposes, Rapid

[13] P. Van Renterghem, P. Van Eenoo, H. Geyer, W. Schänzer, F.T. Delbeke, Ref- Commun. Mass Spectrom. 22 (2008) 2161–2175, doi:10.1002/rcm.

erence ranges for urinary concentrations and ratios of endogenous steroids, [34] J. Marcos, O.J. Pozo, Current LC–MS methods and procedures applied to the

which can be used as markers for steroid misuse, in a Caucasian population of identification of new steroid metabolites, J. Steroid Biochem. Mol. Biol. 162

athletes, Steroids 75 (2010) 154–163, doi: 10.1016/j.steroids.20 09.11.0 08 . (2016) 41–56, doi:10.1016/j.jsbmb.2015.12.012.

[14] F. Gosetti, E. Mazzucco, M.C. Gennaro, E. Marengo, Ultra high performance liq- [35] K. Deventer, O.J. Pozo, A.G. Verstraete, P. Van Eenoo, Dilute-and-shoot-liquid

uid chromatography tandem mass spectrometry determination and profiling chromatography-mass spectrometry for urine analysis in doping control and

of prohibited steroids in human biological matrices. A review, J. Chromatogr. B analytical toxicology, TrAC - Trends Anal. Chem. 55 (2014) 1–13, doi:10.1016/j.

Anal. Technol. Biomed. Life Sci. 927 (2013) 22–36, doi: 10.1016/j.jchromb.2012. trac.2013.10.012.

12.003 . [36] The World Anti-Doping Agency, International standard for labora-

[15] F. Badoud, E. Grata, J. Boccard, D. Guillarme, J.L. Veuthey, S. Rudaz, tories, (2016). https://www.wada-ama.org/en/resources/laboratories/

M. Saugy, Quantification of glucuronidated and sulfated steroids in human international-standard-for-laboratories-isl (Accessed 14 May 2019).

urine by ultra-high pressure liquid chromatography quadrupole time-of-flight [37] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for the assess-

mass spectrometry, Anal. Bioanal. Chem. 400 (2011) 503–516, doi: 10.1007/ ment of matrix effect in quantitative bioanalytical methods based on HPLC-

s00216- 011- 4779- 8 . MS/MS, Anal. Chem. 75 (2003) 3019–3030, doi:10.1021/ac020361s.

[16] P.E. Sottas, M. Saugy, C. Saudan, Endogenous Steroid Profiling in the Ath- [38] The World Anti-Doping Agency, Technical Document – TD2018EAAS Endoge-

lete Biological Passport, Endocrinol. Metab. Clin. North Am. 39 (2010) 59–73, nous Anabolic Androgenic Steroids WADA Technical Document – TD2018EAAS,

doi: 10.1016/j.ecl.20 09.11.0 03 . (2018) 1–17. https://www.wada-ama.org/en/resources/science-medicine/

[17] T. Kuuranne, M. Saugy, N. Baume, Confounding factors and genetic polymor- td2018eaas-0 (Accessed 6 June 2018).

phism in the evaluation of individual steroid profiling, Br. J. Sports Med. 48 [39] The World Anti-Doping Agency, WADA Technical Document – TD2019DL

(2014) 848–855, doi: 10.1136/bjsports- 2014- 093510 . Decision limits for the confirmatory quantification of threshold sub-

[18] J.J. Schulze, J.O. Thor ngren,¨ M. Garle, L. Ekstrom , A. Rane, Androgen sulfation stances - TD2019DL, (2019) 1–19. https://www.wada-ama.org/en/resources/ ¨ in healthy UDP-glucuronosyl transferase 2B17 enzyme-deficient men, J. Clin. science-medicine/td2019dl-version-2-0 (Accessed 5 July 2019).

Endocrinol. Metab. 96 (2011) 3440–3447, doi: 10.1210/jc.2011-0521 . [40] W. Van Gansbeke, M. Polet, F. Hooghe, C. Devos, P. Van Eenoo, Improved

[19] A. Esquivel, É. Alechaga, N. Monfort, R. Ventura, Sulfate metabolites improve sensitivity by use of gas chromatography—positive chemical ionization triple

retrospectivity after oral testosterone administration, Drug Test. Anal. (2018) quadrupole mass spectrometry for the analysis of drug related substances, J.

1–11, doi: 10.1002/dta.2529 . Chromatogr. B. 1001 (2015) 221–240, doi:10.1016/j.jchromb.2015.07.052.

[20] E. Strahm, I. Kohler, S. Rudaz, S. Martel, P.A. Carrupt, J.L. Veuthey, [41] L. De Wilde, K. Roels, M. Polet, P. Van Eenoo, K. Deventer, Identification and

M. Saugy, C. Saudan, Isolation and quantification by high-performance liq- confirmation of diuretics and masking agents in urine by turbulent flow online

uid chromatography-ion-trap mass spectrometry of androgen sulfoconjugates solid-phase extraction coupled with liquid chromatography-triple quadrupole

in human urine, J. Chromatogr. A. 1196–1197 (2008) 153–160, doi: 10.1016/j. mass spectrometry for doping control, J. Chromatogr. A. 1579 (2018) 31–40,

chroma.2008.04.066 . doi:10.1016/j.chroma.2018.10.032.

[21] U. Mareck, H. Geyer, G. Opfermann, M. Thevis, W. Sch, Factors influencing the [42] N. De Brabanter, W. Van Gansbeke, L. Geldof, P. Van Eenoo, An improved

steroid profile in doping control, J. Mass Spectrom 43 (2008) 877–891, doi: 10. gas chromatography screening method for doping substances using triple

1002/jms . quadrupole mass spectrometry, with an emphasis on quality assurance,

[22] R.L. Gomes, W. Meredith, C.E. Snape, M.A. Sephton, Analysis of conjugated Biomed. Chromatogr. 26 (2012) 1416–1435, doi:10.1002/bmc.2714. steroid : Deconjugation, derivatisation and associated issues, J. Pharm. Biomed. Anal. 49 (2009) 1133–1140, doi: 10.1016/j.jpba.2009.01.027 .