Gas chromatography/chemical ionization triple quadrupole mass spectrometry analysis of anabolic steroids: Ionization and collision-induced dissociation behavior

Article abstract of DOI:10.1002/rcm.7472

Rationale The detection of new anabolic steroid metabolites and new designer steroids is a challenging task in doping analysis. Switching from electron ionization gas chromatography triple quadrupole mass spectrometry (GC/EI-MS/MS) to chemical ionization (CI) has proven to be an efficient way to increase the sensitivity of GC/MS/MS analyses and facilitate the detection of anabolic steroids. CI also extends the possibilities of GC/MS/MS analyses as the molecular ion is retained in its protonated form due to the softer ionization. In EI it can be difficult to find previously unknown but expected metabolites due to the low abundance or absence of the molecular ion and the extensive (and to a large extent unpredictable) fragmentation. The main aim of this work was to study the CI and collision-induced dissociation (CID) behavior of a large number of anabolic androgenic steroids (AAS) as their trimethylsilyl derivatives in order to determine correlations between structures and CID fragmentation. Clarification of these correlations is needed for the elucidation of structures of unknown steroids and new metabolites. Methods The ionization and CID behavior of 65 AAS have been studied using GC/CI-MS/MS with ammonia as the reagent gas. Glucuronidated AAS reference standards were first hydrolyzed to obtain their free forms. Afterwards, all the standards were derivatized to their trimethylsilyl forms. Full scan and product ion scan analyses were used to examine the ionization and CID behavior. Results Full scan and product ion scan analyses revealed clear correlations between AAS structure and the obtained mass spectra. These correlations were confirmed by analysis of multiple hydroxylated, methylated, chlorinated and deuterated analogs. Conclusions AAS have been divided into three groups according to their ionization behavior and into seven groups according to their CID behavior. Correlations between fragmentation and structure were revealed and fragmentation pathways were postulated.

Full text of DOI:10.1002/rcm.7472

Research Article
Received: 9 September 2015
Revised: 19 October 2015
Accepted: 26 November 2015
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
(wileyonlinelibrary.com) DOI: 10.1002/rcm.7472
Gas chromatography/chemical ionization triple quadrupole mass
spectrometry analysis of anabolic steroids: ionization and
collision-induced dissociation behavior
*
Michael Polet , Wim Van Gansbeke, Peter Van Eenoo and Koen Deventer
Department of Clinical Chemistry, Microbiology and Immunology, Doping Control Laboratory, Ghent University, Technologiepark
30 B, B-9052, Zwijnaarde, Belgium
RATIONALE: The detection of new anabolic steroid metabolites and new designer steroids is a challenging task in
doping analysis. Switching from electron ionization gas chromatography triple quadrupole mass spectrometry
(GC/EI-MS/MS) to chemical ionization (CI) has proven to be an efficient way to increase the sensitivity of
GC/MS/MS analyses and facilitate the detection of anabolic steroids. CI also extends the possibilities of GC/MS/MS
analyses as the molecular ion is retained in its protonated form due to the softer ionization. In EI it can be difficult
to find previously unknown but expected metabolites due to the low abundance or absence of the molecular ion
and the extensive (and to a large extent unpredictable) fragmentation. The main aim of this work was to study
the CI and collision-induced dissociation (CID) behavior of a large number of anabolic androgenic steroids (AAS)
as their trimethylsilyl derivatives in order to determine correlations between structures and CID fragmentation.
Clarification of these correlations is needed for the elucidation of structures of unknown steroids and
new metabolites.
METHODS: The ionization and CID behavior of 65 AAS have been studied using GC/CI-MS/MS with ammonia as the
reagent gas. Glucuronidated AAS reference standards were first hydrolyzed to obtain their free forms. Afterwards, all the
standards were derivatized to their trimethylsilyl forms. Full scan and product ion scan analyses were used to examine
the ionization and CID behavior.
RESULTS: Full scan and product ion scan analyses revealed clear correlations between AAS structure and the obtained
mass spectra. These correlations were confirmed by analysis of multiple hydroxylated, methylated, chlorinated and
deuterated analogs.
CONCLUSIONS: AAS have been divided into three groups according to their ionization behavior and into seven groups
according to their CID behavior. Correlations between fragmentation and structure were revealed and fragmentation
pathways were postulated. Copyright © 2016 John Wiley & Sons, Ltd.
Due to the performance-enhancing properties of anabolic
androgenic steroids (AAS), the World Anti-Doping
Association (WADA) has banned their use but, according to
the annual report of WADA, steroids are still very popular
amongst athletes and are responsible for half of all adverse
analytical findings.[1,2]
Nowadays, analyses for the detection of AAS are based
on gas chromatography/mass spectrometry (GC/MS
(MS)) or liquid chromatography/mass spectrometry
(LC/MS/(MS)).[3–10] GC/MS/MS methods are becoming
increasingly popular due to a substantial improvement
in selectivity and sensitivity.[11,12] As in most analytical
applications and fields, electron ionization (EI) in
combination with GC/MS is predominantly used in the
anti-doping world.[5,13–16]
Chemical ionization (CI) is a soft ionization technique and
results, in comparison with EI, in fewer fragments with
higher intensities, ideal for selection as precursor ions
for MS/MS.[17,18] GC/CI-MS/MS also opens up new
possibilities in the search for new metabolites. In EI it can
be difficult to find previously unknown but expected
metabolites due to the low abundance (or absence) of the
molecular ion and the extensive (and to
a
large
extent unpredictable) fragmentation. Searching for new
metabolites by selection of predicted transitions for
expected metabolites is much more straightforward in
LC/MS/MS for example, due to the soft ionization
techniques typically employed in that technique.[19–21]
Interpretation of the CI product ion spectra of AAS in
GC/CI-MS/MS and the study of their collision-induced
dissociation (CID) pathways are therefore important in
the search for new metabolites. In addition, knowledge of
their ionization and CID behavior is essential for
understanding product ion specificity in AAS detection
analyses and for the mass spectrometric characterization
of new designer steroids.
* Correspondence to: M. Polet, Department of Clinical Chemistry,
Microbiology and Immunology, Doping Control Laboratory,
Ghent University, Technologiepark 30 B, B-9052 Zwijnaarde,
Belgium.
E-mail: michael.polet@ugent.be
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.

M. Polet et al.
EXPERIMENTAL
(Cologne, Germany). Methylclostebol (MeCLO), 6αOH-
ADION, 7βOH-DHEA and mibolerone (MIB) were from
Steraloids (Newport, RI, USA).
Reagents and reference standards
N-Methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA) was
purchased from Chem. Fabrik Karl Bucher (Waldstedt,
Germany), and ammonium iodide (NH4I), 2-mercaptoethanol
and dodecane were obtained from Sigma Aldrich (Bornem,
Belgium). Acetonitrile and methyl tert-butyl ether were from
Biosolve (Valkenswaard, The Netherlands). β-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) and
sodium sulphate (Na2SO4) were obtained from Merck
(Darmstadt, Germany).
Sample preparation and derivatization
Glucuronidated AAS standards were first hydrolyzed to the
free steroid forms by adding 1 mL phosphate buffer and
50 μL of β-glucuronidase from E. coli. The samples were
incubated for 1.5 h at 56 °C in test tubes. Then 1 mL of
NaHCO3/K2CO3 solution in water (pH 9.5) and 5 mL of
methyl tert-butyl ether were added. Extraction was
performed by rolling the test tubes for 20 min. The organic
layer was transferred and evaporated to dryness under
nitrogen.
Prior to analysis, the free steroid form of the reference
standards was converted by derivatization into its
trimethylsilylmethyl derivative. To the dried residues, 15 μL
of acetonitrile was added, followed by derivatization for
30 min at 80 °C with 75 μL MSTFA/ethanethiol/NH4I
(500:4:2), and afterwards 10 μL dodecane was added.
The phosphate buffer (pH 7) was prepared by dissolving
7.1 g Na2HPO4.2H2O and 1.4 g NaH2PO4.H2O in 100 mL
water. The carbonate buffer (pH 9.5) was prepared by
dissolving 135 g K2CO3 and 111 g NaHCO3 in 900 mL
double-distilled water.
Testosterone (T), epitestosterone (epiT), androsterone (A),
etiocholanolone (Et), 11β-hydroxyandrosterone (11βOH-A),
4-androstene-3,17-dione (ADION), 5α-androstene-3,17-dione
(5α-ADION), 5β-androstene-3,17-dione (5β-ADION), 17β-diol
(ααβ-diol), 5β-androstane-3α, 17β-diol (βαβ-diol), 5β-pregnan-
3α,20α-diol (PD), formestane (F), ethisterone (ETHI), super-
drol (SUPER) and d6-dehydroepiandrosterone (DHEA-d6)
were obtained from Sigma Aldrich. D3-Epitestosterone
glucuronide (epiT-d3), d4-epitestosterone glucuronide
(epiT-d4), d3-dihydrotestosterone glucuronide (DHT-d3),
d4-androsterone glucuronide (A-d4), 1-androstenedione
(1-Adiol), 19-norandrosterone (NorA), 19-noretiocholanolone
(19-NorEt), 1-testosterone (1-T), 2α-hydroxymethylethisterone
(2OHMe-ETHI), 4-OH-testosterone (4OH-T), 4-OH-nandrolone
(4OH-NAN), nandrolone (NAN), 9α-fluoro-11β-hydroxy-17,17-
Instrumentation
The Agilent GC 7890 gas chromatograph (Agilent
Technologies, Palo Alto, CA, USA) employed in this study
was equipped with a programmed temperature evaporation
(PTV) injector (CIS4C, Gerstel GmbH, Mülheim a/d Ruhr,
Germany). For sample preparation and introduction, a
Gerstel MPS2 autosampler was utilized.
The system was coupled to an Agilent 7000B triple
quadrupole mass spectrometer. For the high-resolution
experiments, an Agilent 7200B gas chromatograph quadrupole
time-of-flight (QTOF) instrument was used (mass resolution
>12500, mass accuracy <5 ppm). Nitrogen was used as the
collision gas at a flow rate of 1.5 mL/min and helium was
purged into the collision cell at a flow rate of 2.25mL/min to
eliminate the effect of metastable He ions in the mass
spectrometer. The CI experiments were performed with
ammonia as a reagent gas using a CI gas flow set at 20%
(standard configuration of the Agilent system).
dimethyl-18-norandrosta-4,13-diene-3-one
(18norfluoxyMES),
17α-methyl-1-1testosterone (1-MeT), 17-methyl-5α-androstan-3,17-
diol (αMeTmet), 17-methyl-5β-androstan-3,17-diol (βMeTmet),
epimethandienone (epiDIAN), 6β-hydroxy-methandienone
(6βOH-DIAN), 4-chloro-3α-hydroxy-androst-4-en-17-one (CLOmet),
3α-hydroxy-2α-methyl-5α-androstan-17-one (DROSmet), 3α-
hydroxy-1α-methyl-5α-androstan-17-one (MESTmet), 3α-hydroxy-
1-methylen-5α-androstan-17-one (METHmet), β-androstan-7β,
The gas chromatograph was equipped with a 12m × 250 μm
I.D., 0.25 μm df HP1-MS column (Agilent Technologies). The
column temperature was programmed as follows: initial
temperature of 90 °C (0.25 min), ramped at 35°C/min to
193 °C, then at 2.6 °C/min to 214 °C, at 15 °C/min to 250 °C
and finally at 75 °C/min to 315 °C (1.1 min). Helium was
17α-dimethyl-3α,17β-diol
(CALmet),
17α-methyl-1,4-
androstdien-11α,17β-diol-3-one (Fbmet), 17α-ethylestrane-
3α,17β-diol (NorETHAmet), norboletone (NorBOLE),
oral-turinabol (ORAL-T) and epimethendiol (EpiMdiol) were
from NMI (Pymble, Australia). Dehydroepiandrosterone
(DHEA) was from Serva (Heidelberg, Germany) and
dihydrotestosterone (DHT) from Piette International
Laboratories (Drogenbos, Belgium). Methenolone (METH)
was a kind gift from the Drug Control Centre of King’s
College London (London, UK), bolasterone (BOLA) from
Upjohn (Kalamazoo, MI, USA) and 3α-hydroxytibolone
(TIBmet) and 17α-methyltestosterone (MeT) were a kind gift
from Akzo Nobel (Oss, The Netherlands). Norclostebol
(NorCLO) and stenbolone (STEN) were from TRC (Toronto,
Canada). Boldenone (B), 5β-androst-1-en-17β-ol-3-one (Bmet)
and oxymesterone (OxyMES) were purchased from the
Institute für Biochemie of the Deutsche Sporthochschule
used as the carrier gas at
a constant flow rate of
0.96 mL/min for 12.271 min that was then increased from
1.5 mL/min to 4 mL/min in order to flush out the column.
1.5 μL of sample was injected pulsed splitless at an injector
temperature of 200 °C (and increased to 340 °C at 12 °C/s)
and a pressure of 16 psi.
The gas chromatographs and mass detectors were
controlled with GC/MS MassHunter software (version B.07,
Agilent Technologies, Waldbronn, Germany), while the
Gerstel equipment was controlled by Maestro software
(version 1.4, Gerstel).
The optimization of the instrument settings and
ionization parameters was described in our previous
work.[17]
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522

GC/CI-MS/MS analysis of anabolic steroids
RESULTS AND DISCUSSION
equilibrium towards the enol form. AAS without keto
functions will rapidly lose one or two TMSOH molecules
resulting in abundant [M + H–90]+ (or [M + H–180]+) ions
due to the presence of less stable derivatized hydroxyl
groups.
In general, the CI behavior of AAS can be divided into
three groups. In Fig. 1 a representative full scan spectrum of
a compound of each group is given.
Ionization behavior
Derivatization by silylation is common practice in the doping
control field as it shows excellent results for a wide range of
doping substances. Silylation under acidic conditions
(MSTFA/ethanethiol/NH4I) is under thermodynamic control
and is therefore determined by the stability of the intermediate
enol and end product, meaning that trimethylsilylation of,
for example 3-keto-4-androstene, leads to the 3,5-dienolTMS
derivative.[22]
As described elsewhere, ammonia is used instead of
methane as the reagent gas for all our analyses.[17] The proton
affinity of ammonia is higher than that of methane, causing
less fragmentation and generating a larger proportion of the
[M + H]+ ion. Due to the high proton affinity of the TMS ether,
neutral loss of TMSOH ([M + H–90]+) and TMS ([M + H–72]+)
will also be frequently observed in the full scan spectra. AAS
containing a carbonyl are more likely to have an abundant
[M + H]+ ion and formation of the [M + H–90]+ fragment ion
is more difficult. A carbonyl increases the stability and avoids
neutral loss of TMSOH due to the keto-enol equilibrium that
is formed when a keto function is derivatized (Fig. 4).
Formation of the TMS derivative pushes the keto-enol
Group A: AAS that contain one keto function
These AAS are characterized by a relative abundance of
[M + H]+ between 50 and 100% (100% in most cases)
and a relative abundance of [M + H–90]+ ≥ 5%.
Group B: AAS that contain two keto functions (one at C3 and one at C17)
These AAS are characterized by a relative abundance of
[M + H]+ between 50 and 100% (100% in most cases)
and a relative abundance of [M + H–90]+ < 5%.
Group C: AAS that do not contain a keto function
These AAS are characterized by a relative abundance of
[M + H]+ < 20%.
Figure 1. Positive CI full scan spectra of the trimethylsilylated forms of (A) ethisterone (group A), (B) formestane (group B), and
(C) 1-adiol (group C).
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

M. Polet et al.
Table 1. AAS structural information
A ring
B ring
C ring
D ring
Group 1
2OHMe-ETHI
4OH-T
6OH-ADION
18norfluoxyMES
ADION
BOLA
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
Δ
4 3CO 2CH2OH
4 3CO 4OH
4 3CO
17OH 17C ≡ CH
17OH
6OH
17CO
4 3CO
9F
11OH nor18CH3 13C = 14
C
17CH3 17CH3
17CO
4 3CO
4 3CO
7CH3
17OH 17CH3
17OH
epiT
4 3CO
16,16,17
epiT-d3
epiT-d4
ETHI
4 3CO
17OH D
17OH D
4 3CO D1
4 3CO
16,16,17
17OH 17C ≡ CH
17CO
F
4 3CO 4OH
4 3CO 4Cl
4 3CO
MeCLO
MeT
17OH 17CH3
17OH 17CH3
17OH 17CH3
17OH
OxyMES
T
T-d8
4 3CO 4OH
4 3CO
4 3CO D2,2,4
D6,6
17OH D
16,16,17
Group 2
4OH-NAN
NAN
Δ
Δ
Δ
Δ
Δ
4 3CO 4OH nor19CH3
4 3CO nor19CH3
17CO
17CO
MIB
4 3CO nor19CH3
7CH3
6OH
17OH 17CH3
NorBOLE
NorCLO
Group 3
6βOH-DIAN
B
BDION
DIAN
epiDIAN
Fbmet
ORAL-T
4 3CO nor19CH3
17OH 17CH2CH3 18CH2CH3
17OH
4 3CO 4Cl nor19CH3
Δ
Δ
Δ
Δ
Δ
Δ
Δ
1,4 3CO
1,4 3CO
1,4 3CO
1,4 3CO
1,4 3CO
1,4 3CO
1,4 3CO 4Cl
17OH 17CH3
17OH
17CO
17OH 17CH3
17OH 17CH3
17OH 17CH3
17OH 17CH3
11OH
Group 4
1-MeT
1-T
Δ
Δ
Δ
Δ
Δ
1 3CO
17OH 17CH3
17OH
1 3CO
Bmet
1 3CO
17OH
METH
STEN
1 3CO 1CH3
1 2CH3 3CO
17OH
17OH
Group 5
5α-ADION
5β-ADION
DHT
DHT-d3
MEST
3CO 5α-H
3CO 5β-H
3CO 5α-H
3CO 5α-H
1CH3 3CO 5α-H
3CO 5α-H
1CH3 3CO
17CO
17CO
17OH
17OH D16,16,17
17OH
MESTA
SUPER
17OH 17CH3
17OH 17CH3
Group 6
7βOH-DHEA
11βOH-A
A
3OH
Δ
5 7OH
17CO
17CO
17CO
17CO
17CO
17CO
3OH 5α-H
3OH 5α-H
11OH
A-d4
3OH 5α-H D2,2,4,4
4
CLOmet
DHEA
Δ
4 3OH Cl
3OH
Δ5
(Continues)
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522

GC/CI-MS/MS analysis of anabolic steroids
Table 1. (Continued)
A ring
B ring
5Δ D5
C ring
D ring
DHEA-d6
DROSmet
Et
3OH D2,2,3,4,4
17CO
17CO
17CO
17CO
17CO
17CO
17CO
2CH3 3OH 5α-H
3OH 5β-H
METHmet
MESTmet
NorA
1C = CH2 3OH 5α-H
1CH3 3OH 5α-H
3OH 5α-H nor19CH3
3OH 5β-H nor19CH3
NorEt
Group 7
1-Adiol
Δ
1 3OH
17OH
17OH
ααβ-diol
αMeTmet
βαβ-diol
βMeTmet
CALmet
EES
3OH 5α-H
3OH 5α-H
3OH 5β-H
3OH 5β-H
3OH
17OH 17CH3
17OH
17OH 17CH3
7CH3
17OH 17CH3
Δ4 nor19CH3
17OH 17CH2CH3
17OH 17CH3
EpiMdiol
NorBOLEmet
NorETHAmet
PD
Δ
1 3OH
3OH nor19CH3
3OH nor19CH3
3OH
17OH 17CH2CH3 18CH2CH3
17OH 17CH2CH3
17CHOHCH3
17OH 17C ≡ CH
TIBmet
3OH
5C = 10 C 7CH3
Collision-induced dissociation behavior
compounds (Table 2). As expected, subsequent GC/QTOF
analysis of epiT resulted in a measured mass of m/z 169.1047
(C9H17OSi exact mass m/z 169.1043; mass error 2.27ppm).
Likewise, T showed the product ion at m/z 181 whereas in
T-d8 it shifted to m/z 184, indicating that C2 and C4 are
included in this fragment, but not C6. In 4OH-T the ion shifted
to m/z 269 (=181 + 88) and in MeCLO it shifted to m/z 215
(=181 + 34) confirming the presence of C4 in the fragment.
Similar observations were made for other compounds (Table 2).
GC/QTOF analysis of epiTresulted in a measured mass of m/z
181.1048 (C10H17OSi exact mass m/z 181.1043; mass error
2.68 ppm). The only exception was BOLA. In this case the ion
shifted to m/z 195 and C6 appears to be part of the fragment,
meaning that the fragmentation pathway described in Fig. 5 is
more likely. The occurrence of a methyl group at C7 seems to
shift the position of the bond cleavage. This was also confirmed
by GC/QTOF analysis of BOLA (measured mass m/z 195.1202;
C11H19OSi exact mass m/z 195.1200; mass error 1.19 ppm).
For AAS that contain a hydroxyl function in the corresponding
fragment, we can see a double transition. In 4OH-T, for example,
ions at m/z 257 (=fragment without loss of TMSOH) and m/z
169 (=same fragment with loss of TMSOH) are both present.
The same observation is made for m/z 269 and 181.
The CID behavior of 65 AAS, metabolites and deuterated
AAS was studied. The structures of the different AAS are
given in Table 1 and the obtained characteristic transitions
in Table 2. [M + H]+ was chosen as the precursor ion except
for those AAS that did not contain a keto function. For those
compounds [M + H–90]+ or [M + H–180]+ was the most
abundant ion and it was therefore selected as the precursor
ion. Based on our fragmentation study, AAS can be
subdivided into seven groups (Figs. 2 and 3).
Group 1: 3-keto-4-ene-anabolic steroids
Sixteen compounds belonged to this group, and all showed
similar fragmentation behavior. They were characterized by
abundant product ions at m/z 169 and 181. Figure 4 illustrates
the proposed fragmentation pathways corresponding to the
formation of these ions.
The pathway for m/z 169 is identical to that for the formation
of the ion at m/z 97 in the LC/ESI-MS/MS CID of AAS.[23]
Support for the proposed formation pathway of these product
ions from the A-ring was obtained by analysis of hydroxylated,
methylated, chlorinated and deuterated analogs. T showed the
product ion at m/z 169 whereas in T-d8 it shifted to m/z 172,
indicating that C2 and C4 are included in this fragment, but
not C6. In 4OH-T and OxyMES (hydroxylated A-ring) the ion
shifted to m/z 257 (=169+ 88) and in MeCLO (chlorinated
A-ring) it shifted to m/z 203 (=169 + 34) confirming the presence
of C4 in the fragment. Similar observations were made for other
Group 2: 19-nor-3-keto-4-ene-anabolic steroids
Five compounds belonged to this group. They were
characterized by abundant product ions at m/z 155 and 181.
The proposed fragmentation pathway for the formation of
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

M. Polet et al.
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522

GC/CI-MS/MS analysis of anabolic steroids
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

M. Polet et al.
Figure 2. CID spectra (30 eV) of the [M + H]+ ion of MeT (group 1), NorBOLE (group 2), B (group 3), and Bmet (group 4) in their
trimethylsilylated forms.
m/z 155 is the same as for group 1 (Fig. 4(A), but without the
inclusion of C19). In LC/ESI-MS/MS, in contrast to
CI-MS/MS, the 19-nor-analogs did not undergo this
fragmentation pathway.[23] Figure 5 describes the proposed
fragmentation pattern corresponding to the formation of
m/z 181. The 19-nor-3-keto-4-ene-anabolic steroids constitute
a different group in CI-MS/MS CID behavior because the
product ion at m/z 181 has a different origin from that in group
1 due to an alternative dissociation pathway. In contrast to the
pathway depicted in Fig. 4(B), the C5–C6 bond is not cleaved
in the initial formation stage.
to m/z 189 (=155+ 34), indicating that C4 is present in the
fragment. For MIB the ion remained at m/z 155, confirming
that C7 is not included in this fragment. Basically, the
product ion at m/z 155 is formed by the same fragmentation
pathway as the ion at m/z 169 in group 1. Confirmation by
GC/QTOF analysis of NAN resulted in a measured mass
of m/z 155.0889 (C8H15OSi exact mass m/z 155.0887; mass
error 1.51 ppm).
In 4OH-NAN the m/z 181 product ion shifted to m/z 269
(=181 + 88) and in NorCLO it shifted to m/z 215 (=181 + 34)
indicating that C4 is present in the fragment. For MIB the
product ion remained at m/z 181, confirming that C7 is not
included in this fragment. GC/QTOF analysis of NAN
resulted in a measured mass of m/z 181.1044 (C10H17OSi
exact mass m/z 181.1043; mass error 0.45 ppm).
Support for the fragmentation mechanisms was found by the
analysis of 4OH-NAN, NorCLO and MIB. In 4OH-NAN
(hydroxylated A-ring) the m/z 155 product ion shifted to m/z
243 (=155+ 88) and in NorCLO (chlorinated A-ring) it shifted
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522

GC/CI-MS/MS analysis of anabolic steroids
Figure 3. CID spectra (30 eV) of the [M + H]+ ion of MESTA (group 5) and NorEt (group 6) and the CID spectra of the
[M + H–180]+ ion of PD and the [M + H–90]+ ion of 1-Adiol (group 7) in their trimethylsilylated forms.
Group 3: 3-keto-1,4-diene-anabolic steroids
Support for the proposed fragmentation mechanism was
found by the analysis of ORAL-T and 6bOH-DIAN. In
ORAL-T (chlorinated A-ring) this product ion shifted to m/z
227 (=193 + 34), demonstrating that C4 is part of the fragment,
and in 6bOH-DIAN to m/z 281 (=193 + 88), indicating that C6
is also present in the fragment. The m/z 193 ion also occurs in
6bOH-DIAN (=same fragment with loss of TMSOH at
position C6). Confirmation of the ion by GC/QTOF analysis
of B resulted in a measured mass of m/z 193.1046 (C11H17OSi
exact mass m/z 193.1043; mass error 1.47 ppm).
Seven compounds belonged to this group and they were all
characterized by an abundant product ion at m/z 193. The
pathway for the formation of m/z 193 is very similar to
that for the formation of m/z 121 in the LC/ESI-MS/MS
of 3-keto-1,4-diene-anabolic steroids (Fig. 6).[24,25] The
mechanism is initiated by protonation of the oxygen of the
TMS ether. A keto-enol tautomerization occurs and a
conjugated π-electron system of the 3-keto-1,4-diene function
is formed. This is followed by the migration of the C8
hydrogen to C9 and fission of the linkage between C9 and
C10. The temporary charge at C8 triggers cleavage of
the C6–C7 bond, resulting in the formation of the ion at
m/z 193. This structure can be stabilized by charge
delocalization to a tropylium cation.[26]
Group 4: 3-keto-1-ene-anabolic steroids
Five compounds belonged to this group and the characteristic
abundant product ions were m/z 187 and 201. For this group
of compounds, however, there was not enough data available
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

M. Polet et al.
Figure 4. Proposed fragmentation pathways for the formation of
product ions m/z 169 (A) and 181 (B) (group 1).
Figure 5. Proposed fragmentation pathway for the formation of
product ion m/z 181 (group 2).
Figure 6. Proposed fragmentation pathway for the formation of
product ion m/z 193 (group 3).
to determine the fragmentation pathways corresponding to
these transitions. GC/QTOF analysis of Bmet yielded measured
masses of m/z 187.1489 and 201.1640, corresponding to a
C14H19 structure (exact mass m/z 187.1481; mass error
4.15 ppm) and a C15H21 structure (exact mass m/z 201.1638;
mass error 1.11 ppm), respectively.
originated from the D-ring as all three were present in
DHT-d3. The product ion obtained for MEST (methylated
A-ring) also indicates that C1 is included in m/z 143 and 159
but not in m/z 187. For m/z 159 and 187 there was not enough
data available to determine the corresponding fragmentation
pathways. The pathway for the formation of m/z 143 could
be clarified and is depicted in Fig. 7. The mechanism is
initiated by protonation of the oxygen of the TMS ether,
inducing a keto-enol tautomerization. This is followed by
simultaneous fission of the C1–C10 and C4–C5 bonds, and
subsequent formation of the m/z 143 ion. This structure can
be stabilized by electron delocalization and tautomerization.
Group 5: unconjugated-3-keto-anabolic steroids
Seven compounds belonged to this group and the
characteristic abundant product ions were m/z 143, 159 and
187. We can conclude that none of these three transitions
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522

GC/CI-MS/MS analysis of anabolic steroids
Analysis of MESTmet and METHmet provides evidence that
C1 is part of these fragments and analysis of DHEA and
DHEA-d6 shows that C5 and C6 are also present.
Group 7: anabolic steroids without keto functions
Twelve compounds belonged to this group. Due to the
absence of a carbonyl in the steroid structure, this group
was also characterized by extensive fragmentation leading
to numerous product ions. The most abundant product ions
were m/z 81, 93/95 and 105/107/109. Other frequently
occurring ions were m/z 119/121, 133/135, 147/149,
159/161 and 175. Subsequent product ions differ by 14 m/z
units, which means that subsequent fragments differ by the
absence of a methyl group. This basically indicates that the
complete steroid structure is being shattered, generating
fragments which differ by a methyl group from each other.
Figure 7. Proposed fragmentation pathway for the formation
of product ion m/z 143 (group 5).
Support for this suggested mechanism was obtained by
analysis of the methylated analogs (SUPER and MEST) and
a
deuterated analog (DHT-d3). The introduction of a
methyl group at C1 shifts the product ion at m/z 143 to 157
(=143 + 14) indicating that C1 remains in the fragment. The
m/z 143 ion also occurs in SUPER (=same fragment but
lacking the methyl group). Introduction of deuterium atoms
in the D-ring yielded an unchanged product ion at m/z 143.
GC/QTOF analysis of DHT-d3 yielded measured masses of
m/z 143.0899 (C7H15OSi exact mass m/z 143.0887; mass error
8.67 ppm), 159.1178 (C12H15 exact mass m/z 159.1168; mass
error 6.15 ppm) and 187.1484 (C14H19 exact mass m/z
187.1481; mass error 1.47 ppm).
General phenomena
Based on the data that we have gathered, we have divided
the AAS into seven groups depending on the correlation
between structure and CID behavior. Progressing from group
1 to 7, more extensive fragmentation occurs with less specific
product ions and as a result compounds can have characteristic
product ions from multiple groups. Compounds with an alkyl
group at the hydroxylated C17 position, for example, have
transitions from both groups 6 and 7. Secondly, one has to keep
in mind the structure of the steroid after derivatization. The
presence of a π-bond in the A-ring, for example, leads to
transitions of group 1 to 4 because of the structural similarity
after derivatization. 1-Adiol (1-ene), for example, belongs to
group 7, but after derivatization it has the structural
characteristics of a compound of group 4, leading to the product
ions at m/z 187 and 201 (characteristic of group 4, Fig. 3).
[M + H–90]+ and [M + H–180]+ are also important and frequently
occurring product ions across all groups. [M + H–116]+ is an
important and characteristic ion for compounds with a C17
hydroxyl function. When the C17 position is also methylated,
this ion shifts to [M + H–130]+; and when C17 is ethylated, it
shifts to [M + H–144]+ (Table 2). The proposed fragmentation
pathway for the neutral loss of 116 Da is depicted in Fig. 8(B).
Combinations obviously also occur; for example, MeT has
a product ion at [M + H–220]+ (simultaneous loss of 130
and 90 Da (=TMSOH)) and NorBOLE has a product ion
at [M + H–234]+ (simultaneous loss of 144 and 90 Da
(=TMSOH)) (Fig. 2).
Group 6: 3-hydroxy-17-keto-anabolic steroids
Thirteen compounds belonged to this group. Due to the
absence of a carbonyl at the C3 position this group was
characterized by more extensive fragmentation. The
frequently occurring product ions were m/z 145, 147, 159,
161, 199 and 213. Figure 8(A) shows the proposed structures
for m/z 199 and 213 but there was not enough data available
to elucidate the fragmentation pathway. Support for these
suggested structures was obtained by analysis of analogs. In
NorEt the m/z 199 ion shifts to 185 and m/z 213 shifts to
199, indicating that C19 is included in both fragments.
Likewise, in DROSmet, m/z 199 shifts to 213 and m/z 213
shifts to 227 due to the presence of a methyl group at C2.
CONCLUSIONS
The CID behavior of 65 trimethylsilylated AAS was studied
and correlations between fragmentation and structure were
revealed. The presence of certain product ions can be indicative
of a specific moiety or structure. When a group (methyl, ethyl,
hydroxyl, halogen, deuteron,...) is incorporated or removed
from the steroid structure, the m/z values of these product ions
can change substantially, but in general the fragmentation
mechanism remains the same. These data will be helpful in
the elucidation of the structures of unknown steroids and in
the search for new metabolites.
Figure 8. Proposed structures for the formation of product
ions m/z 199 and 213 (group 6) (A) and the proposed
fragmentation pathway for the neutral loss of 116 Da (B).
Rapid Commun. Mass Spectrom. 2016, 30, 511–522
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm

M. Polet et al.
of oxymetholone in human urine. J. Steroid Biochem. Mol.
Biol. 1992, 42, 229.
REFERENCES
[1] World Anti-doping Agency. The 2015 Prohibited List
International Standard, 2015. Available: https://wada-
main-prod.s3.amazonaws.com/resources/files/wada-2015-
prohibited-list-en.pdf (accessed 23.07.2015).
[2] World Anti-doping Agency. 2014 Anti-Doping Testing
Figures Report, 2015. Available: https://wada-main-prod.
s3.amazonaws.com/wada_2014_anti-doping-testing-figures_
full-report_en.pdf (accessed 09.08.2015).
[3] W. Schänzer, M. Donike. Metabolism of anabolic steroids in
man: synthesis and use of reference substances for
identification of anabolic steroid metabolites. Anal. Chim.
Acta 1993, 275, 23.
[4] C. Ayotte, D. Goudreault, A. Charlebois. Testing for natural
and synthetic anabolic agents in human urine. J. Chromatogr.
B 1996, 687, 3.
[15] W. Schänzer, H. Geyer, M. Donike. Metabolism of
metandienone in man: identification and synthesis of
conjugated excreted urinary metabolites, determination
of excretion rates and gas chromatographic-mass spectrometric
identification of bis-hydroxylated metabolites. J. Steroid
Biochem. Mol. Biol. 1991, 38, 441.
[16] D. de Boer, E. G. de Jong, R. A. Maes, J. M. van Rossum. The
methyl-5 alpha-dihydrotestosterones mesterolone and
drostanolone; gas chromatographic/mass spectrometric
characterization of the urinary metabolites. J. Steroid
Biochem. Mol. Biol. 1992, 42, 411.
[17] W. Van Gansbeke, M. Polet, F. Hooghe, C. Devos, P. Van Eenoo.
Improved sensitivity by use of gas chromatography-positive
chemical ionization triple quadrupole mass spectrometry for
the analysis of drug related substances. J. Chromatogr. B
2015, 1001, 221.
[18] M. Raro, T. Portolés, J. V. Sancho, E. Pitarch, F. Hernández,
J. Marcos, R. Ventura, C. Gómes, J. Segura, O. J. Pozo. Mass
spectrometric behavior of anabolic androgenic steroids
using gas chromatography coupled to atmospheric pressure
chemical ionization source. Part I: ionization. J. Mass
Spectrom. 2014, 49, 509.
[19] M. Thevis, H. Geyer, U. Mareck, W. Schänzer. Screening for
unknown synthetic steroids in human urine by liquid
chromatography-tandem mass spectrometry. J. Mass
Spectrom. 2005, 40, 955.
[20] 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 metabolism. Trends Anal.
Chem. 2014, 53, 106.
[21] O. J. Pozo, K. Deventer, P. Van Eenoo, F. T. Delbeke. Efficient
approach for the comprehensive detection of unknown
anabolic steroids and metabolites in human urine by liquid
chromatography-electrospray-tandem mass spectrometry.
Anal Chem 2008, 80, 1709.
[22] D. H. Van de Kerkhof, J. W. Zwikker, V. Kaats-Richters,
K. Bouwman, R. D. van Ooijen, D. de Boer, R. Maes.
Trimethylsilylation of 3-keto-androst-4-ene steroids: thermo-
dynamic versus kinetic control of enolization. Recent Adv.
Doping Anal. 2000, 8, 129.
[5] M. Saugy, C. Cardis, N. Robinson, C. Schweizer. Test
methods: anabolics, Best Practice
Endocrinol. Metab. 2000, 14, 111.
& Research. Clin.
[6] A. Leinonen, T. Kuuranne, T. Kotiaho, R. Kostiainen.
Screening of free 17-alkyl-substituted anabolic steroids in
human urine by liquid chromatography-electrospray
ionization tandem mass spectrometry. Steroids 2004, 69, 101.
[7] M. Mazzarino, F. Botrè. A fast liquid chromatographic/
mass spectrometric screening method for the simultaneous
detection of synthetic glucocorticoids, some stimulants,
anti-oestrogen drugs and synthetic anabolic steroids. Rapid
Commun. Mass Spectrom. 2006, 20, 3465.
[8] M. Thevis, G. Fußhöller, H. Geyer, G. Rodchenkov,
U. Mareck, G. Sigmund, A. Koch, A. Thomas, W. Schänzer.
Detection of stanozolol and its major metabolites in human
urine by liquid chromatography-tandem mass spectrometry.
Chromatographia 2006, 64, 441.
[9] K. Deventer, P. Van Eenoo, F. T. Delbeke. Screening for
anabolic steroids in doping analysis by liquid
chromatography/electrospray ion trap mass spectrometry.
Biomed. Chromatogr. 2006, 20, 429.
[10] O. J. Pozo, P. Van Eenoo, K. Deventer, F. T. Delbeke.
Development and validation of a qualitative screening
method for the detection of exogenous anabolic steroids in
urine by liquid chromatography-tandem mass spectrometry.
Anal. Bioanal. Chem. 2007, 389, 1209.
[11] M. A. Delgadillo, L. Garrostas, O. J. Pozo, R. Ventura, B. Velasco,
J. Segura, J. Marcos. Sensitive and robust method for anabolic
agents in human urine by gas chromatography-triple
quadrupole mass spectrometry. J. Chromatogr. B 2012, 897, 85.
[12] 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 2011,
1218, 3306.
[13] M. Thevis, T. Kuuranne, H. Geyer, W. Schänzer. Annual
banned-substance review: analytical approaches in human
sports drug testing. Drug Test. Anal. 2014, 6, 164.
[14] H. Bi, P. Du, R. Massé. Studies on anabolic steroids – 8.
GC/MS characterization of unusual seco acidic metabolites
[23] O. J. Pozo, P. Van Eenoo, K. Deventer, S. Grimalt,
J. V. Sancho, F. Hernández, F. T. Delbeke. Collision-induced
dissociation of 3-keto anabolic steroids and related compounds
after electrospray ionization. Considerations for structural
elucidation. Rapid Commun. Mass Spectrom. 2008, 22, 4009.
[24] M. Thevis, W. Schänzer. Mass spectrometric analysis of
androstan-17beta-ol-3-one and androstadiene-17beta-ol-3-
one isomers. J Am Soc Mass Spectrom 2005, 16, 1660.
[25] F. Guan, L. R. Soma, Y. Luo, C. E. Uboh, S. Peterman.
Collision-induced dissociation pathways of anabolic steroids
by electrospray ionization tandem mass spectrometry. J. Am.
Soc. Mass Spectrom. 2006, 17, 477.
[26] C. Lifshitz. Tropylium ion formation from toluene: Solution
of an old problem in organic mass spectrometry. Accounts
Chem. Res. 1994, 27, 138.
wileyonlinelibrary.com/journal/rcm
Copyright © 2016 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2016, 30, 511–522