Analytica Chimica Acta 1042 (2018) 52e59

Contents lists available at ScienceDirect

Analytica Chimica Acta

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

Development and validation of an open screening method for doping

substances in urine by gas chromatography quadrupole time-of-flight
mass spectrometry
Michael Polet*, Wim Van Gansbeke, Peter Van Eenoo
Ghent University, Department of Clinical Chemistry, Microbiology and Immunology, Doping Control Laboratory, Technologiepark 30 B, B-9052, Zwijnaarde,
Belgium

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

An open and complete GC-MS high

resolution screening method.

Target analyte list no longer required
due to open screening.

Expanded list of compounds in the
gas chromatography screening
method.

The high resolution screening
method is compliant with WADA
regulations.

Equal or lower LLODs for 80.2% of the
compounds in comparison with CI
GC-QQQ-MS.

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

Article history: In anti-doping, a high number of classes of substances are prohibited and laboratories need to detect
Received 17 July 2018 these at low urinary concentrations. Traditionally, testing is done using complimentary liquid
Received in revised form chromatography mass spectrometry and gas chromatography mass spectrometry. High resolution
22 August 2018
mass spectrometric acquisition has some important advantages over triple quadrupole instruments
Accepted 26 August 2018
(e.g., open screening due to full scan high resolution data acquisition with retrospectivity,
Available online 28 August 2018
compatibility with libraries and a straightforward and effortless addition and validation of new
compounds in the future). Doping samples can be stored for 10 years and retrospective data analysis
Keywords:
Time-of-flight
can be used to re-evaluate previously acquired data (e.g., searching for prohibited (designer) sub-
Gas chromatography stances that were unknown at the initial moment of analysis). During the past decade, these ad-
High resolution vantages have led to the wide-scale transfer of liquid chromatography triple quadrupole mass
Doping spectrometry screening to liquid chromatography high resolution mass spectrometry screening for
Open screening doping control purposes. Up to now, for gas chromatography a similar switch to high resolution
Drugs of abuse screening has not yet occurred, because so far no method has been developed that combines suf-
ficient sensitivity with wide-scale drug detection.
In this work, the current gas chromatography triple quadrupole mass spectrometry screening
method for human doping control purposes was successfully converted into an equivalent and
complete gas chromatography high resolution acquisition screening method. This new screening
method on a gas chromatography quadrupole time-of-flight mass spectrometer has been developed
and validated. The method is compliant with the World Anti-Doping Agency requirements and allows
the detection of 294 target compounds (and 14 internal standards), including diuretics, stimulants,

* Corresponding author.
E-mail address: michael.polet@UGent.be (M. Polet).

https://doi.org/10.1016/j.aca.2018.08.050
0003-2670/© 2018 Elsevier B.V. All rights reserved.
 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59
narcotics, beta-2-agonists, beta-blockers, hormone modulators, anabolic agents and the quantification
of 14 endogenous steroids in a single fast run (14.1 min).
1. Introduction
The World Anti-Doping Agency (WADA) defines which sub-
stances are regarded as prohibited in sports and these compounds
cover a wide range of chemically and pharmacologically different
classes [1]. The lowest concentrations that WADA accredited labo-
ratories are required to detect on a day-to-day basis is set by WADA
as the minimum required performance limit (MRPL) [2]. Nowadays,
efficient doping control laboratories use one gas chromatography
mass spectrometry (GC-MS) method and one liquid chromatography
MS (LC-MS) method for their initial testing procedure [3e5]. These
two screening methods are complementary and together they cover
all low to medium molecular weight drugs of abuse that need to be
detected according to the WADA regulations [1]. For example, some
compounds are difficult or impossible to detect by using LC-MS (e.g.,
low ionization efficiency for oxymesterone) and these need to be
included in the GC-MS method [6]. Other compounds are easy on LC-
MS, but challenging on GC-MS (e.g., tetrahydrogestrinone) [7]. For
GC-MS screening purposes, prohibited substances can be divided
into two categories: exogenous substances that need to be detected
qualitatively (i.e., as sensitive as possible) and endogenous steroids
that need to be quantified as accurate as feasible within a very wide
concentration range [2,8]. Both categories require unique and con-
flicting method characteristics and a performant screening method
demands an optimal compromise (e.g., high sensitivity versus high
capacity, high chromatographic resolution versus short analysis
time). For example, a higher injection volume may lower the limit of
detection (LOD), but will simultaneously deteriorate the quantifica-
tion of endogenous steroid isomers (e.g., etiochanolone and
androsterone, two essential endogenous steroid epimers with high
urinary concentrations) due to coelution. As such, the GC-MS
screening method should include as many prohibited substances
as possible, have a large linear range for the quantitative substances,
sufficient chromatographic resolution (for accurate quantification of
isomers) and a short analysis time. It has to be sensitive enough
(especially for the exogenous compounds) and as cost-effective as
feasible. GC triple quadrupole MS (GC-QQQ-MS) is capable of
meeting these demands and nowadays GC-QQQ-MS instruments
have become the standard GC-MS screening tool, in parallel with
many other analytical fields, due to their high sensitivity and excel-
lent linear range [9e16].
Up to now, high resolution technology in combination with GC
has not yet been used as a complete GC-MS screening method for
doping control analyses. To the best of our knowledge, there are only
two papers that describe the use of GC high resolution as screening
tool in the doping control field [17,18]. The first paper screens with a
Waters GC orthogonal acceleration time-of-flight MS and the second
paper presents a screening method using conventional EI on a Agi-
lent 7200 quadrupole time-of-flight MS (QTOF-MS). Both screenings
are primarily for exogenous steroids and a substantial part of the
compounds that need to be covered for a full screening method are
not included. Up to now, the available GC high resolution in-
struments were not sensitive enough for a number of compounds
and/or had inadequate linearity to comply with all the required
WADA criteria, meaning that the current GC-QQQ-MS screening
methods could not be replaced by an equivalent method on GC high
resolution MS. However, high resolution technology has some
53
© 2018 Elsevier B.V. All rights reserved.
important advantages in comparison with GC-QQQ-MS (e.g., open
screening due to full scan high resolution data acquisition with ret-
rospectivity, compatibility with libraries and a straightforward and
effortless addition and validation of new compounds in the future).
During the past decade, these advantages have led to the wide-scale
transfer of LC-QQQ-MS screening methods to a LC high resolution MS
screening [3,19e23]. Up to now, for GC-MS a similar switch to high
resolution screening has not yet occurred.
This paper describes the first successful conversion of the cur-
rent GC-QQQ-MS screening method for human doping control
purposes into an equivalent and complete GC high resolution
acquisition screening method on a new generation GC-QTOF-MS
instrument. The developed and validated method includes 280
qualitative compounds and 14 quantitative compounds for analysis
of human doping control samples. Due to the full scan high reso-
lution acquisition, a large amount of compounds can be added to
the method, making it a very versatile approach that can easily be
transferred and applied in other analytical fields such as environ-
ment, food safety, toxicology and horse doping [24,25]. This open
screening method does not rely on a targeted analyte list, allowing
the identification of new compounds more quickly, followed by
retrospective data analysis without having to reanalyze samples. In
addition, the performance of the method is not restricted by the
amount of compounds that is monitored, as is the case for GC-QQQ-
MS screening.
2. Experimental section
2.1. Reagents and reference standards
Ethanethiol (reagent grade) was bought from Acros (Geel,
Belgium). N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA)
was purchased from Chem. Fabrik Karl Bucher (Waldstedt, Ger-
many); ammonium iodide (NH4I) and dodecane (anhydrous) were
obtained from Sigma Aldrich (Bornem, Belgium). Acetonitrile (LC-
MS grade) and methyl tert-butyl ether (HPLC grade) were from
Biosolve (Valkenswaard, The Netherlands). b-Glucuronidase from
E. coli K12 was obtained from Roche Diagnostics GmbH (Mannheim,
Germany). Sodium hydrogen carbonate (NaHCO3) was from Fisher
Scientific (Loughborough, UK). Potassium carbonate (K2CO3),
disodium hydrogen phosphate (Na2HPO4), sodium dihydrogen
phosphate (NaH2PO4), sodium acetate (NaOAc) and sodium sulfate
(Na2SO4) were obtained from Merck (Darmstadt, Germany). XAD-2
was from Serva (Heidelberg, Germany).
Phosphate buffer (pH 7) was prepared by dissolving 7.1 g of
Na2HPO4$2H2O and 1.4 g of NaH2PO4$H2O in 100 mL of deionized
water. Carbonate buffer (pH 9.5) was prepared by dissolving 135 g
of K2CO3 and 111 g of NaHCO3 in 900 mL of double-distilled water.
The origin of all the reference standards is described in detail in our
previous work [3,26].
The internal standard (IS) consisted of d3-Testosterone glucu-
ronide (10 mg mL1), d4-Epitestosterone glucuronide (5 mg
mL1), d4-Androsterone glucuronide (50 mg mL1), d5-
Etiocholanolone (50 mg mL1), d3-Dihydrotestosterone glucuro-
nide (2 mg mL1), d6-Dehydroepiandrosterone (10 mg mL1), d3-5a-
Androstane-3a,17b-diol (10 mg mL1), d5-5b-Androstane-3a,17b-
diol (10 mg mL1), 17a-Methyltestosteron (5 mg mL1), 19-
54 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59
Norandrosteron-d4 glucuronide (2 mg mL1), 19-
Noretiocholanolone-d4 (2 mg mL1), Nalorphine (2 mg mL1),
Sibutramine (2 mg mL1) and Salbutamol-d3 (10 mg mL1).
2.2. Sample preparation
For the analysis of urine samples, the following procedure was
applied: to 500 mL of urine, 25 mL of IS, 1 mL phosphate buffer (0.1 M;
pH ¼ 7) and 25 mL of b-glucuronidase from E. coli were added. The
samples were incubated for 1 h at 56  C. Afterwards, aqueous
NaHCO3/K2CO3 solution (pH ¼ 9.5, 1 mL) and methyl tert-butyl ether
(5 mL) were added. Extraction was performed by rolling the test
tubes during 20 min. The organic layer was transferred to tubes and
evaporated to dryness under nitrogen. Then, the residue was
reconstituted in 150 mL acetone, transferred to vials and evaporated
to dryness under nitrogen. To the dried residues, 40 mL of derivati-
zation mixture was added and the samples were incubated at 80  C
for 30 min. The derivatization mixture consisted of 2 mL of MSTFA/
ethanethiol/NH4I (500:4:2, v/v/w) and 1 mL of dodecane.
2.3. Method and instrumentation
An Agilent 7250 GC-QTOF-MS (Agilent Technologies, Palo Alto,
USA) with back flush system was used and 0.8 mL was injected. The
GC was equipped with a 15 m HP ultra 1 column (200 mm I.D.,
0.11 mm df) (Agilent Technologies, Palo Alto, USA). The column was
divided into two parts: 5 m from the inlet to the back flush unit and
10 m from the back flush unit to the detector. The back flush system
was installed to extend the lifetime of the column by removing
pollutants during the final stage of the GC run that accumulated
during the run on the first column (5 m). These pollutants are flush
out towards the inlet (and not towards the MS ion source). This also
minimizes maintenance cost as the 5 m column simultaneously
functions as a precolumn that can easily be replaced (instead of a
complete 15 m column). The oven temperature was programmed
from 110  C (0.1 min) to 125  C at 75  C min1, then to 165  C at
25  C min1, then to 185  C at 40  C min1, then to 210  C at 3.5  C
min1, then to 240  C at 15  C min1 and finally to 325  C (0.9 min)
at 75  C min1. Afterwards, the back flush was performed for
0.5 min at 325  C during which the first part of the column (5 m)
was back flushed for 25 void volumes. The total GC run time was
14.1 min. Helium was used as carrier gas at a constant flow rate of
1 mL min1. Analyses were performed with low energy electron
impact (LE-EI) at 18 eV. The instrument operated in full scan mode
from m/z 50 to 750. During 7 small time intervals the instrument
measured both in full scan mode and MS/MS mode by continuous
switching. Table 1 lists the 7 time intervals and the corresponding
precursor ions that were selected in the Q of the QTOF for MS/MS.
Nitrogen was used as collision gas at a flow rate of 1 mL min1.
Chromatographic and mass spectrometric details of all analytes can
Table 1
Time intervals during which the GC-QTOF-MS operates in both full scan (m/z 50e750) and MS/MS mode by
continuous switching.
time interval (min)
10.6e10. 95
11.05e11.40
11.40e11.70
11.75e12.01
12.01e12.25
12.55e12.73
12.73e12.95
be found in supplementary material S2.
2.4. Method validation
2.4.1. Qualitative
The method validation of the qualitative compounds was per-
formed on 10 randomly chosen blank urine samples covering a
normal range of pH, specific gravity and turbidity for anti-doping
samples. Validation was performed by spiking these urine sam-
ples with reference mixtures at MRPL and ½ MRPL to check WADA
MRPL compliance. These experiments were performed on different
days and with multiple operators to evaluate the robustness of the
method under routine operating conditions as well. These experi-
ments provided data concerning the repeatability and reproduc-
ibility. A compound was regarded as validated if the substance
could be detected in all 10 urine samples at the 1/2 MRPL level (S/
N > 3). Selectivity and specificity were tested by analysis of 10 blank
urine samples. The lowest LOD (LLOD) is defined as the lowest level
that can be detected in an ideal situation (i.e., urine sample con-
taining very few interferences). LLOD is determined by analysis of
one urine sample with a low specific gravity spiked with reference
mixtures at 2, 1, ½, 1/5, 1/10, 1/25 and 1/100 of the MRPL.
2.4.2. Quantitative
Six point calibration curves (Table 2) were made by spiking
steroid stripped urine. Steroid stripped urine was prepared by
pouring negative urine from an infant on a preconditioned XAD-2
column. Removal of the naturally present androgenic steroids
was necessary for correct quantification. Each validation batch
consisted of 3 calibration curves and 9 additional spiked samples at
the lowest (n ¼ 3), midrange (n ¼ 3) and highest (n ¼ 3) calibration
levels. In total, three validation batches were analysed, each by a
different analyst and on a different day. Unweighted least squares
regression was used to set up the calibration curves. Accuracy,
expressed as bias, was defined as the relative difference between
the theoretical value and the measured concentration. Repeat-
ability and reproducibility were calculated as the relative standard
deviation (RSD) and expressed as percentages. Acceptable toler-
ances (%) for precision were calculated from 2/3
RSDmax ¼ 2(10.5logC) [27]. Tolerances for the accuracy, expressed as
bias, were set at maximum of 15% [28].
To assess external bias, matrix matched urine samples from the
WADA external quality assessment scheme (EQAS) were also ana-
lysed and z-scores were calculated.
3. Results and discussion
3.1. Conversion of CI GC-QQQ-MS to LE-EI GC-QTOF-MS screening
The 7250 GC-QTOF-MS has an optimized source configurations,
MS/MS mode: precursor ion (m/z) and collision energy (eV)
446 (25)
339 (25)
442 (25)
446 (30)
460 (45)
506 (45)
428 (30)
520 (45)
640 (35)
490 (30)
 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59 55

Table 2
Target substances for quantitative analysis.

Substance Internal standard Calibration levels in Coefficient of RSD and corresponding 2/3RSDmax between Bias at LOQ (%)
urine (ng mL1) determination (r2) brackets at LOQ (%) (n ¼ 18) (n ¼ 18)

Testosterone d3-Testosterone 1-3-10-30-100-400 0.9995 5.8 (30.2) 12.6

glucuronide
Epitestosterone d4-Epitestosterone 1-3-10-30-100-400 0.9997 8.0 (30.2) 3.4
glucuronide
Androsterone d4-Androsterone 24-72-240-720-2400- 0.9993 5.1 (18.7) 9.3
glucuronide 9600
Etiocholanolone d5-Etiocholanolone 24-72-240-720-2400- 0.9987 6.4 (18.7) 3.9
9600
Dihydrotestosterone d3-Dihydrotestosterone 0.5-1.5-5-15-50-200 0.9986 6.8 (33.5) 14.1
glucuronide
Dehydroepiandrosterone d6- 2-6-20-60-200-800 0.9993 5.2 (27.2) 13.9
Dehydroepiandrosterone
4-androstene-3,17-dione d3-Dihydrotestosterone 0.5-1.5-5-15-50-200 0.9988 4.7 (33.5) 11.9
glucuronide
5a-androstane-3a,17b- d3-5a-Androstane- 2-6-20-60-200-800 0.9997 7.0 (27.2) 5.3
diol 3a,17b-diol
5b-androstane-3a,17b- d5-5b-Androstane- 2-6-20-60-200-800 0.9995 4.5 (27.2) 13.6
diol 3a,17b-diol
5a-androstane-3,17- d5-5b-Androstane- 0.5-1.5-5-15-50-200 0.9987 16.0 (33.5) 12.8
dione 3a,17b-diol
5b-androstane-3,17- d5-5b-Androstane- 0.5-1.5-5-15-50-200 0.9986 9.3 (33.5) 5.7
dione 3a,17b-diol
6aOH-androstenedion d3-Dihydrotestosterone 0.25-0.75-2.5-7.5-25- 0.9988 16.1 (37.2) 13.9
glucuronide 100
4OH-androstenedion d3-Dihydrotestosterone 0.25-0.75-2.5-7.5-25- 0.9986 15.5 (37.2) 14.7
glucuronide 100
5b-pregnanediol d5-5b-Androstane- 2-6-20-60-200-800 0.9971 9.5 (27.2) 5.6
3a,17b-diol

lenses, magnets and tuning for LE-EI (12e20 eV) measurements.
LE-EI allows a softer ionization than conventional EI at 70 eV. Use of
LE-EI therefore leads to a higher relative abundance of the high m/z
ions (i.e., ions with high diagnostic value), analogous to chemical
ionization (CI). For a compound such as 6b-OH-4-
Chlordehydromethyltestosteron for example, LE-EI at 18 eV leads
to a lower LOD (factor of 2) in comparison with conventional EI at
70 eV (Fig. 1). As a consequence, LE-EI at 18 eV was used for the LE-
EI GC-QTOF-MS screening method.
Most doping control laboratories use a conventional EI GC-QQQ-
MS approach for their GC screening [26]. Since 2014, our laboratory
uses CI in combination with GC-QQQ-MS because this leads to
higher sensitivities for more than 90% of the compounds in com-
parison with EI GC-QQQ-MS [4]. However, CI does also require more
maintenance and instrument intervention. To convert our current
CI GC-QQQ-MS doping control screening method into an equivalent
and complete GC high resolution acquisition screening method on
LE-EI GC-QTOF-MS, a small number of chromatographic changes
were necessary. First, the 12 m HP-1MS (250 mm I.D. x 0.25 mm df)
column was replaced with a 15 m HP ultra 1 (200 mm I.D. x 0.11 mm

Fig. 1. GC-QTOF-MS analysis using LE-EI (18 eV) and conventional EI (70 eV) of a urine sample spiked at different levels of 6b-OH-4-Chlordehydromethyltestosteron.
56 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59
Fig. 2. Mass accuracy measurements (ppm) of epimetenediol during 48 h: first 100 injections without calibration runs, last 100 injections with calibration runs every 5 injections.
Average mass deviation and 2 x standard deviation in full and dotted line respectively.
df) column. Both columns have an identic stationary phase. The
extension in length is compensated by the reduction of the internal
diameter and film thickness, leading to comparable behaviour on
both the CI GC-QQQ-MS screening method and the new LE-EI GC-
QTOF-MS screening method. Secondly, on the LE-EI GC-QTOF-MS
0.8 mL out of 40 mL of sample is injected whereas with the CI GC-
QQQ-MS method, an injection volume of 2 mL out of 60 mL of sam-
ple was used.
One of the most desirable attributes of a high resolution mass
analyser is its mass accuracy. To improve the mass accuracy of the
instrument, it is advisable to execute an external calibration run at
regular intervals. This automatic external calibration run is pro-
vided and incorporated in the operation software of the instru-
ment and does not jeopardize the high throughput characteristics
of the approach as it only requires 1 min and is executed during GC
cool down time. Accuracy measurements during approximately
48 h are presented in Fig. 2 and supplementary material S1 for
17a-methyltestosterone (m/z 446.3031 and 301.1982) and epi-
metenediol (m/z 143.0887, 216.1873, 325.2686 and 448.3193).
Before the first injection, a calibration run was performed on the
instrument. During the subsequent 99 injections (±24 h) no cali-
bration runs were executed. Starting from injection 101 until in-
jection 200, a calibration run was performed every 5 injections.
The maximum observed deviations between the measured mass
and the theoretical mass during the first 100 injections were 4.9,
3.3, 2.8, 2.3, 4.2 and 4.2 ppm for m/z 446.3031, 301.1982, 143.0887,
216.1873, 325.2686 and 448.3193 respectively. When calibration
runs are performed at regular intervals (i.e., injection 101 until
200), the maximum observed deviations drop to 4.0, 2.7, 1.4, 1.4,
3.6 and 3.1 for m/z 446.3031, 301.1982, 143.0887, 216.1873,
325.2686 and 448.3193 respectively. The standard deviation
(STDEV) of the mass deviations remain roughly the same (Fig. 2
and supplementary material S1). Without regular calibration
runs, the STDEVs range from 0.46 until 0.62 ppm. With regular
calibration runs, the STDEVs range from 0.50 until 0.65 ppm. In
general, the mass analyser has a tendency to deviate towards a
higher mass than the exact mass and the deviation becomes larger
for ions with a higher m/z (Fig. 2 and supplementary material S1).
When very high amounts of compound reach the mass analyser
(i.e., samples with very high concentrations), the mass analyser
can get saturated leading to a mass shift towards the higher
masses (data not shown). This phenomenon has to be taken into
account. If the mass window is set too narrow during the pro-
cessing of the samples, there is a risk of not detecting substances
that are present in very high concentration due to this mass shift.
Therefore, a conservative mass window of 20 ppm was used for
the processing of the compounds. This mass window takes into
account the required selectivity for the respective compound and
the expected possible maximum concentration of that compound
in urine.
3.2. Validation
3.2.1. Qualitative
An overview of all the compounds that are included in the LE-EI
GC-QTOF-MS method can be found in supplementary material S2.
All retention times were compliant with WADA's identification
criteria [2]. The LLOD defines the lowest level that can be detected
in an ideal situation. As most prohibited substances have a zero-
tolerance, anti-doping laboratories need to be able to detect sub-
stances at the lowest possible concentration. This concentration is
not only dependent on the method and the target substance, but
also on the sample matrix. In a field where results are highly
challenged in court, the LLOD of a method needs to reflect the true
LLOD as much as possible, as otherwise lawyers will claim that it is
impossible to detect a substance below the validated LOD. In
practice, this can occur as matrix is an important determining
factor. Nevertheless, it is important that the LLOD reflects as much
as possible a best case scenario. The LLODs of the CI GC-QQQ-MS
and the LE-EI GC-QTOF-MS screening method are listed in sup-
plementary material S2. For 14 out of 68 stimulants, the LE-EI GC-
QTOF-MS has higher LLODs than the CI GC-QQQ-MS and for 7 out of
20 narcotics the LE-EI GC-QTOF-MS has lower LLODs than the CI
GC-QQQ-MS. However, for both these classes, the LLOD is not of
utmost importance, as WADA has recommended not to report
concentrations below ½ MRPL. However, for steroids it is important
to have LLODs which are as low as possible, as lower LODs will
mean a longer detection window after administration of a sub-
stance due to decreasing concentrations in function of time. The LE-
EI GC-QTOF-MS has equal or lower LLODs for 150 out of 187 (80.2%)
of the tested compounds than the CI GC-QQQ-MS screening
method, while previously it has been shown that CI GC-QQQ-MS
has lower LODs than EI GC-QQQ-MS [4].
The high resolution open screening allows retrospectivity and
compatibility with libraries. In addition, a large amount of com-
pounds can be added to the method, without effecting other
simultaneously detected substances. In contrast, the number of
compounds that can be added in a GC-QQQ-MS method is
restricted because it is a targeted approach with a limited number
of transitions that can be included. In QQQ-MS, the number of
monitored transitions will therefore affect the sensitivity of the
method. Here, a total of 280 compounds was validated for quali-
tative detection, most of which in full scan mode. Only 8 com-
pounds required the use of the MS/MS mode to comply with
WADA's stringent criteria. All other compounds had sufficient
sensitivity and selectivity by using the full scan high resolution
mode. The list of compounds that is included in the LE-EI GC-QTOF-
MS screening method is considerably larger than the list of sub-
stances that is included in current CI GC-QQQ-MS screening
methods [4,26,29]. For example, diuretics, a class of doping sub-
stances that is not covered in the CI GC-QQQ-MS screening method,
 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59
is included in the LE-EI GC-QTOF-MS screening. In the CI GC-QQQ-
MS method, these compounds could not be included due to the
limited number of transitions that can be monitored without
effecting the sensitivity of other compounds, illustrating the benefit
of an open screening method. The LE-EI GC-QTOF method is also
capable of screening for HIF stabilizer and SARMS (e.g., Molidustat,
andarine, ostarine, LGD-4033). These are relatively new classes of
doping substances that can be difficult to detect on LC-MS. How-
ever, they are important doping substances. For example, SARMS
have anabolic properties, in parallel with anabolic steroids, but can
be bought legally and cheap, making them easily accessible for
athletes.
Fig. 3 shows the LE-EI GC-QTOF-MS chromatograms of some
compounds that can be challenging if one uses a conventional EI
GC-QQQ-MS screening. The LE-EI GC-QTOF-MS screening does not
have any issues to reach the WADA MRPLs for these compounds.
Fig. 4 visualizes the LE-EI GC-QTOF-MS analyses of a random urine
sample spiked with long term metabolites (LTM) at different levels.
These LTM are metabolites of anabolic steroids that are excreted in
(very) low concentrations in urine, but that have very long detec-
tion times. Therefore, they are of great importance for doping
control analyses as the extended detection times resulted in a
4e80-fold increase of adverse analytical findings for anabolic ste-
roids [30].
3.2.2. Quantitative
For most compounds a zero tolerance policy applies in doping
control. According to WADA regulations, quantitative data is
required for a defined set of endogenous steroids [8]. For this
purpose, the method was validated quantitatively for 14 endoge-
nous steroids, including the two minor metabolites 4OH-androst-
4-ene-3,17-dione (formestane) and 6aOH-androst-4-ene-3,17-dion
(Table 2) [31,32]. Six point calibration curves were made by spiking
steroid stripped urine. The concentration ranges differ for the
different compounds according to their naturally occurring con-
centrations. The low-, mid- and high-level standards were analysed
Fig. 3. LE-EI GC-QTOF-MS analysis of a urine sample spiked at the MRPL.
57
in 6 replicates. This procedure was repeated in triplicate (three
different analysts at three different points in time). All STDEVs were
acceptable as they were lower than 2/3 of the maximum residual
standard deviation (RSDmax ¼ 2(10.5logC)) as defined by Horwitz
[27]. The bias at each of these points was below 15% [28]. The limit
of quantification (LOQ) is defined as the lowest calibration point of
the corresponding calibration curve. All calibration curves had a
coefficient of determination r2 > 0.98.
Within WADAs EQAS urine samples are frequently sent to
accredited laboratories. Fitness for purpose of the method was
assessed by reanalysis of 17 EQAS samples, each in triplicate, and
the mean of measured values were compared with the reported
consensus values. For 110 out of 119 values, the z-scores were lower
than 1, indicating satisfactory accuracy of the method (supple-
mentary material S3). Only 2 out of 119 measured values had a z-
score higher than 2 (i.e., sample 1 Epitestosterone z-score ¼ 2.15,
sample 4 Testosterone z-score ¼ 2.43) which is acceptable as sta-
tistically you would only expect 95% of the values to be within a z-
score of 2.
A comparison between the CI GC-QQQ-MS screening method
and the LE-EI GC-QTOF-MS screening method is shown in Fig. 5 and
supplementary material S4. On CI GC-QQQ-MS, the endogenous
steroid concentrations of 100 urine samples analysed over 5
batches was determined. Subsequently, the same samples were
additionally analysed on a second CI GC-QQQ-MS instrument and
on the LE-EI GC-QTOF-MS to verify that the bias and the spread on
the bias is comparable between the CI GC-QQQ-MS analyses and
the analyses on the LE-EI GC-QTOF-MS screening method. The re-
sults indicate an excellent correlation (r2) and comparability (slope
between 0.926 and 1.093) between both techniques.
3.3. Application to excretion/routine samples
The suitability of the developed method was proven by ana-
lysing 105 urine samples previously declared positive for doping
substances and authentic samples from controlled administration
58 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59

Fig. 4. LE-EI GC-QTOF-MS analysis of a negative urine sample and a urine sample spiked with LTMs at different levels (MRPL ¼ 5 ng mL1). Each column corresponds to the above
mentioned concentration. The circles indicate the lowest detectable concentration for the corresponding LTM.

Fig. 5. Quantification of endogenous steroids in 100 urines samples determined with the CI GC-QQQ-MS screening method and the LE-EI GC-QTOF-MS screening method: cor-
relation between CI GC-QQQ-MS and LE-EI GC-QTOF-MS and visualization of the CI GC-QQQ-MS vs CI GC-QQQ-MS bias and the CI GC-QQQ-MS vs LE-EI GC-QTOF-MS bias.

studies. All positive cases were detected and no false negative
samples were identified. In addition, 241 blank urine samples were
analysed and no false positive samples were identified.
4. Conclusions
A comprehensive high resolution open screening method on LE-
EI GC-QTOF-MS has been developed and validated for the analysis
of 294 compounds (and 14 internal standards) in urine. All detec-
tion limits were compliant with the required WADA MRPLs. The LE-
EI GC-QTOF-MS has equal or lower LLODs for 80.2% of the tested
compounds compared to the CI GC-QQQ-MS screening method.
The method allows the replacement and transition of the cur-
rent GC-QQQ-MS screening method for human urine doping con-
trol purposes into an equivalent and complete GC high resolution
acquisition screening method. The high resolution open screening
allows retrospectivity and compatibility with libraries and, in
addition, a large amount of compounds can be added to the
method. On top of that, the full scan high resolution acquisition
makes it a versatile method and approach that can easily be
transferred and applied in other related analytical fields. An open
screening method does not rely on a targeted analyte list, allowing
the identification of new compounds more quickly, followed by
retrospective data analysis without having to reanalyze samples.
 M. Polet et al. / Analytica Chimica Acta 1042 (2018) 52e59
The performance of the method is not restricted by the amount
of compounds that is monitored, as is the case for GC-QQQ-MS
screening. As such, the list of compounds that are enclosed in the
LE-EI GC-QTOF-MS screening method has been expanded consid-
erably in comparison with the CI GC-QQQ-MS screening method. It
also includes HIF, SARMS and diuretics. These are doping classes
that are not covered in the current CI GC-QQQ-MS screening
method.
Funding
This work was supported by the Partnership for Clean Compe-
tition (grant number 2018R1000323G).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.aca.2018.08.050.
References
[1] WADA, The 2018 Prohibited List International Standard, World Anti-Doping
Agency, 2018.
[2] WADA, Technical Document Minimum Required Performance Limits for
Detection of Prohibited Substances: TD2018MRPL, World Anti-Doping
Agency, 2018.

n, K. Deventer, K. Roels, P. Van Eenoo, Development and validation of
[3] A.J. Giro
an open screening method for diuretics, stimulants and selected compounds
in human urine by UHPLC-HRMS for doping control, Anal. Chim. Acta 721
(2012) 137e146, https://doi.org/10.1016/j.aca.2012.02.002.
[4] W. Van Gansbeke, M. Polet, F. Hooghe, C. Devos, P. Van Eenoo, Improved
sensitivity by use of gas chromatographydpositive chemical ionization triple
quadrupole mass spectrometry for the analysis of drug related substances,
J. Chromatogr. B 1001 (2015) 221e240, https://doi.org/10.1016/
j.jchromb.2015.07.052.
[5] M. Mazzarino, X. de la Torre, F. Botre , A screening method for the simulta-
neous detection of glucocorticoids, diuretics, stimulants, anti-oestrogens,
beta-adrenergic drugs and anabolic steroids in human urine by LC-ESI-MS/
MS, Anal. Bioanal. Chem. 392 (2008) 681e698, https://doi.org/10.1007/
s00216-008-2292-5.
[6] M. Polet, W. Van Gansbeke, L. Geldof, K. Deventer, P. Van Eenoo, Identification
and characterization of novel long-term metabolites of oxymesterone and
mesterolone in human urine by application of selected reaction monitoring
GC-CI-MS/MS, Drug Test. Anal. (2017), https://doi.org/10.1002/dta.2183.
[7] D.H. Catlin, M.H. Sekera, B.D. Ahrens, B. Starcevic, Y.-C. Chang, C.K. Hatton,
Tetrahydrogestrinone: discovery, synthesis, and detection in urine, Rapid
Commun. Mass Spectrom. 18 (2004), https://doi.org/10.1002/rcm.1495,
1245e1049.
[8] WADA, Technical Document Endogenous Anabolic Androgenic Steroids:
TD2016EAAS, World Anti-Doping Agency, 2016.
[9] Z. He, Y. Wang, L. Wang, Y. Peng, W. Wang, X. Liu, Determination of 255
pesticides in edible vegetable oils using QuEChERS method and gas chroma-
tography tandem mass spectrometry, Anal. Bioanal. Chem. 409 (2017)
1017e1030, https://doi.org/10.1007/s00216-016-0016-9.
[10] K. Kalachova, T. Cajka, C. Sandy, J. Hajslova, J. Pulkrabova, High throughput
sample preparation in combination with gas chromatography coupled to
triple quadrupole tandem mass spectrometry (GC-MS/MS): a smart procedure
for (ultra)trace analysis of brominated flame retardants in fish, Talanta 105
(2013) 109e116, https://doi.org/10.1016/j.talanta.2012.11.073.
[11] M. Monteleone, A. Naccarato, G. Sindona, A. Tagarelli, A reliable and simple
method for the assay of neuroendocrine tumor markers in human urine by
solid-phase microextraction-gas chromatography-triple quadrupole mass
spectrometry, Anal. Chim. Acta 759 (2013) 66e73, https://doi.org/10.1016/
j.aca.2012.11.017.
[12] A. Thomas, C. Widmer, G. Hopfgartner, C. Staub, Fast gas chromatography and
negative-ion chemical ionization tandem mass spectrometry for forensic
analysis of cannabinoids in whole blood, J. Pharmaceut. Biomed. Anal. 45
(2007) 495e503, https://doi.org/10.1016/j.jpba.2007.08.019.
[13] L. Truta, A.L. Castro, S. Tarelho, P. Costa, M.G.F. Sales, H.M. Teixeira, Antide-
pressants detection and quantification in whole blood samples by GC-MS/MS,
for forensic purposes, J. Pharmaceut. Biomed. Anal. 128 (2016) 496e503,
https://doi.org/10.1016/j.jpba.2016.06.027.
[14] K. Banerjee, S. Utture, S. Dasgupta, C. Kandaswamy, S. Pradhan, S. Kulkarni,
59
P. Adsule, Multiresidue determination of 375 organic contaminants including
pesticides, polychlorinated biphenyls and polyaromatic hydrocarbons in fruits
and vegetables by gas chromatography-triple quadrupole mass spectrometry
with introduction of semi-quantification, J. Chromatogr. A 1270 (2012)
283e295, https://doi.org/10.1016/j.chroma.2012.10.066.
[15] M. Polet, W. Van Gansbeke, P. Van Eenoo, K. Deventer, Efficient approach for
the detection and identification of new androgenic metabolites by applying
SRM GC-CI-MS/MS: a methandienone case study, J. Mass Spectrom. 51 (2016)
524e534, https://doi.org/10.1002/jms.3781.
[16] M. Polet, L. De Wilde, P. Van Renterghem, W. Van Gansbeke, P. Van Eenoo,
Potential of saliva steroid profiling for the detection of endogenous steroid
abuse: reference thresholds for oral fluid steroid concentrations and ratios,
Anal. Chim. Acta 999 (2018) 1e12, https://doi.org/10.1016/j.aca.2017.11.015.
[17] C.G. Georgakopoulos, A. Vonaparti, M. Stamou, P. Kiousi, E. Lyris, Y.S. Angelis,
G. Tsoupras, B. Wuest, M.W.F. Nielen, I. Panderi, M. Koupparis, Preventive
doping control analysis: liquid and gas chromatography time-of-flight mass
spectrometry for detection of designer steroids, Rapid Commun. Mass Spec-
trom. 21 (2007) 2439e2446, https://doi.org/10.1002/rcm.3103.
[18] W. Abushareeda, E. Lyris, S. Kraiem, A. Al Wahaibi, S. Alyazidi, N. Dbes,
A. Lommen, M. Nielen, P.L. Horvatovich, M. Alsayrafi, C. Georgakopoulos, Gas
chromatographic quadrupole time-of-flight full scan high resolution mass
spectrometric screening of human urine in antidoping analysis, J. Chromatogr.
B 1063 (2017) 74e83, https://doi.org/10.1016/j.jchromb.2017.08.019.
[19] K. Deventer, O.J. Pozo, A.G. Verstraete, P. Van Eenoo, Dilute-and-shoot-liquid
chromatography-mass spectrometry for urine analysis in doping control and
analytical toxicology, TrAC Trends Anal. Chem. (Reference Ed.) 55 (2014)
1e13, https://doi.org/10.1016/j.trac.2013.10.012.
[20] A. Musenga, D.A. Cowan, Use of ultra-high pressure liquid chromatography
coupled to high resolution mass spectrometry for fast screening in high
throughput doping control, J. Chromatogr. A 1288 (2013) 82e95, https://
doi.org/10.1016/j.chroma.2013.03.006.
[21] S. Odoardi, V. Valentini, N. De Giovanni, V.L. Pascali, S. Strano-Rossi, High-
throughput screening for drugs of abuse and pharmaceutical drugs in hair by
liquid-chromatography-high resolution mass spectrometry (LC-HRMS),
Microchem. J. 133 (2017) 302e310, https://doi.org/10.1016/
j.microc.2017.03.050.
[22] W. Abushareeda, A. Vonaparti, K. Al Saad, M. Almansoori, M. Meloug, A. Saleh,
R. Aguilera, Y. Angelis, P.L. Horvatovich, A. Lommen, M. Alsayrafi,
C. Georgakopoulos, High resolution full scan liquid chromatography mass
spectrometry comprehensive screening in sports antidoping urine analysis,
J. Pharmaceut. Biomed. Anal. 151 (2018) 10e24, https://doi.org/10.1016/
j.jpba.2017.12.025.
[23] I. Ojanper€a, M. Kolmonen, A. Pelander, Current use of high-resolution mass
spectrometry in drug screening relevant to clinical and forensic toxicology
and doping control, Anal. Bioanal. Chem. 403 (2012) 1203e1220, https://
doi.org/10.1007/s00216-012-5726-z.
[24] H.H. Maurer, M.R. Meyer, High-resolution mass spectrometry in toxicology:
current status and future perspectives, Arch. Toxicol. 90 (2016) 2161e2172,
https://doi.org/10.1007/s00204-016-1764-1.
[25] F. Herna
ndez, T. Portole

s, E. Pitarch, F.J. Lo

pez, Gas chromatography coupled to
high-resolution time-of-flight mass spectrometry to analyze trace-level
organic compounds in the environment, food safety and toxicology, TrAC
Trends Anal. Chem. (Reference Ed.) 30 (2011) 388e400, https://doi.org/
10.1016/j.trac.2010.11.007.
[26] P. Van Eenoo, W. Van Gansbeke, N. De Brabanter, K. Deventer, F.T. Delbeke,
A fast, comprehensive screening method for doping agents in urine by gas
chromatography-triple quadrupole mass spectrometry, J. Chromatogr. A 1218
(2011) 3306e3316, https://doi.org/10.1016/j.chroma.2010.09.082.
[27] W. Verwaal, M. van Bavel, A. Boot, J. Bravenboer, F. de Goei, C. Maas, A. van der
Putten, Valideren van (fysisch-) chemische en (fysisch-) mechanische meth-
oden op het niveau van de wettelijke eis, Ware(n) Chem. 26 (1996) 106.
[28] F. Bressolle, M. Bromet-Petit, M. Audran, Validation of liquid chromatographic
and gas chromatographic methods. Applications to pharmacokinetics,
J. Chromatogr. B Biomed. Appl. 686 (1996) 3e10. http://www.ncbi.nlm.nih.
gov/pubmed/8953186. (Accessed 25 September 2014).
[29] N. De Brabanter, W. Van Gansbeke, L. Geldof, P. Van Eenoo, An improved gas
chromatography screening method for doping substances using triple quad-
rupole mass spectrometry, with an emphasis on quality assurance, Biomed.
Chromatogr. 26 (2012) 1416e1435, https://doi.org/10.1002/bmc.2714.
[30] H. Geyer, W. Scha €nzer, M. Thevis, Anabolic agents : recent strategies for their
detection and protection from inadvertent doping, Br. J. Sports Med. 48 (2014)
820e826, https://doi.org/10.1136/bjsports-2014-093526.
[31] M. Polet, P. Van Renterghem, W. Van Gansbeke, P. Van Eenoo, Studies on the
minor metabolite 6a-hydroxy-androstenedione for doping control purposes
and its contribution to the steroid profile, Drug Test. Anal. 6 (2014) 978e984,
https://doi.org/10.1002/dta.1618.
[32] M. Polet, P. Van Renterghem, W. Van Gansbeke, P. Van Eenoo, Profiling of
urinary formestane and confirmation by isotope ratio mass spectrometry,
Steroids 78 (2013), https://doi.org/10.1016/j.steroids.2013.07.011.