Analytica Chimica Acta 915 (2016) 102e110

Contents lists available at ScienceDirect

Analytica Chimica Acta

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

Fast and sensitive supercritical fluid chromatography e tandem mass

spectrometry multi-class screening method for the determination of
doping agents in urine
Lucie Nova
kova

a, Vincent Desfontaine b, Federico Ponzetto c, Raul Nicoli c,
Martial Saugy , Jean-Luc Veuthey b, Davy Guillarme b, *
c

a
Department of Analytical Chemistry, Faculty of Pharmacy, Charles University in Prague, Heyrovsk
eho 1203, 500 05 Hradec Kra
lov

e, Czech Republic
b
School of Pharmaceutical Sciences, University of Geneva, University of Lausanne, Boulevard d'Yvoy 20, 1211 Geneva 4, Switzerland
c
Swiss Laboratory for Doping Analyses, University Center of Legal Medicine Lausanne-Geneva, Centre Hospitalier Universitaire Vaudois and University of
Lausanne, Chemin des Croisettes 22, 1066 Epalinges, Switzerland

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

A high throughput UHPSFC-MS/MS

method was developed for
screening several classes of doping
agents in urine.

Supported liquid extraction in 48-
well plate format was successfully
applied to extract the doping agents
from urine samples.

Good extraction recoveries and
reasonable matrix effects were
observed for the whole set of doping
agents.

The method exhibited very low LODs,
below the Minimum Required Per-
formance Levels, for most of the 100
compounds.

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

Article history: This study shows the possibility offered by modern ultra-high performance supercritical fluid chroma-
Received 11 November 2015 tography combined with tandem mass spectrometry in doping control analysis. A high throughput
Received in revised form screening method was developed for 100 substances belonging to the challenging classes of anabolic
3 February 2016
agents, hormones and metabolic modulators, synthetic cannabinoids and glucocorticoids, which should
Accepted 7 February 2016
be detected at low concentrations in urine. To selectively extract these doping agents from urine, a
Available online 12 February 2016
supported liquid extraction procedure was implemented in a 48-well plate format. At the tested con-
centration levels ranging from 0.5 to 5 ng/mL, the recoveries were better than 70% for 48e68% of the
Keywords:
Doping agents
compounds and higher than 50% for 83e87% of the tested substances. Due to the numerous interferences
Urine related to isomers of steroids and ions produced by the loss of water in the electrospray source, the
Supercritical fluid chromatography choice of SFC separation conditions was very challenging. After careful optimization, a Diol stationary
Tandem mass spectrometry phase was employed. The total analysis time for the screening assay was only 8 min, and interferences as
Supported liquid extraction well as susceptibility to matrix effect (ME) were minimized. With the developed method, about 70% of
Steroids the compounds had relative ME within the range ±20%, at a concentration of 1 and 5 ng/mL. Finally,
limits of detection achieved with the above-described strategy including 5-fold preconcentration were

* Corresponding author.
E-mail address: davy.guillarme@unige.ch (D. Guillarme).

http://dx.doi.org/10.1016/j.aca.2016.02.010
0003-2670/© 2016 Elsevier B.V. All rights reserved.

kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110
below 0.1 ng/mL for the majority of the tested compounds. Therefore, LODs were systematically better
than the minimum required performance levels established by the World anti-doping agency, except for
very few metabolites.
1. Introduction
The list of prohibited substances published by the World anti-
doping agency (WADA) is yearly updated. Today, there are about
250 illicit compounds covering a wide range of physicochemical
properties that need to be tracked in the different specimens (urine
and blood) collected in-and-out of competition [1]. The declaration
of a presence or an absence of a doping agent is a multistep process
including an initial screening approach followed by a confirmation
procedure, if applicable. To tackle the huge chemical diversity of the
banned substances, various analytical methods are implemented in
anti-doping laboratories and among them, GCeMS(/MS) and LC-
MS/MS are considered as the methods of choice offering high
selectivity, sensitivity and fast turnaround. These methods focus on
the direct detection of prohibited substances, but also on the
determination of their major phase I and phase II metabolites, to
further improve the detection windows capabilities in the matrix of
interest [2]. To ensure homogeneous performance between labo-
ratories, Minimum Required Performance Levels (MRPLs) for
detection of non-threshold substances are established by WADA,
and correspond to the minimum concentration levels that should
be attained for routine analyses.
To date, supercritical fluid chromatography (SFC) has been
scarcely used for doping control purposes, despite some obvious
advantages of this separation technique [3]: i) the low fluid vis-
cosity and high diffusion coefficients under SFC conditions provide
excellent kinetic performance, ii) the organic solvent consumption
is limited compared to LC, despite the fact that a significant pro-
portion of organic modifier (up to 40%) may be used in modern
packed column SFC, iii) a wide range of compounds, from relatively
polar to highly non-polar, can be analyzed in SFC, with the same
mobile phase components (CO2 and methanol). Still, these benefits
were restricted by the poor quality of old-generation SFC in-
struments providing low sensitivity, unacceptable quantitative
performance and above all lack of reliability [4]. However, in the
last few years, providers of chromatographic instrumentation have
launched new generation of SFC systems which undertake the
above-mentioned shortcomings. These new SFC instruments are
described as UHPSFC which stands for ultra-high performance su-
percritical fluid chromatography [5]. UHPSFC systems offer an
improved compatibility with modern stationary phases, such as
columns packed with fully porous sub-2 mm particles as well as a
full compatibility with ESI-MS(/MS) devices thanks to commercial
interfaces [6]. Considering all the beneficial features of modern
UHPSFC-MS/MS instrumentation, a screening method was previ-
ously successfully developed in our laboratory for the determina-
tion of 110 doping agents in urine, including diuretics, b-blockers,
stimulants and narcotics, using a simple dilute-and-shoot proce-
dure [7,8].
Among the WADA list of banned substances, the determination
of the numerous anabolic agents, in particular androgenic steroids,
is particularly challenging. Indeed, these substances are excreted in
urine with very diverse concentrations, and MRPLs are extremely
demanding. In addition, there is a large number of isomers and
metabolites, among this particular class, which are difficult to
separate and identify in a satisfactory manner [9]. GCeMS(/MS)
103
© 2016 Elsevier B.V. All rights reserved.
remains the gold standard for analyzing these substances in anti-
doping laboratories, but often requires laborious liquideliquid
extraction procedures as well as time-consuming derivatization
steps prior to injection, to make the substances volatile and
improve their detectability. Surprisingly, SFC is known to be a
reference strategy for the analytical identification and character-
ization of various types of steroids [10e12], but has rarely been
applied so far for the determination of steroids and derivatives in
doping control analysis [11], and never in routine analyses.
The goal of this work was to evaluate the performance of a
combination of supported liquid extraction (SLE) and UHPSFC-MS/
MS analysis, for the high throughput screening of 100 substances
(parent compounds and phase I metabolites) in urine. All the
selected substances belong to the problematic classes of anabolic
androgenic steroids, hormones and metabolic modulators, syn-
thetic cannabinoids and glucocorticoids, for which there are a lot of
isobaric compounds that should be detected at very low concen-
tration levels in urine. The SLE recoveries, matrix effects and limits
of detection achieved in UHPSFC-MS/MS for the 100 substances will
be shown and critically discussed.
2. Experimental method
2.1. Reagents and analytes
All doping agents were kindly provided by the Swiss Laboratory
for Doping Analysis (Epalinges, Switzerland). The exhaustive list of
these target analytes is reported in Table S-1 and their structure in
Fig. S-1. Methanol (MeOH), ethanol (EtOH), isopropanol and
acetonitrile (ACN) of ULC/MS grade were provided by Biosolve
(Dieuze, France). Ammonium formate (AmF), ammonium acetate,
heptane extra dry 99%þ and methyl tert-butyl ether (MtBE) were
provided by SigmaeFluka (Buchs, Switzerland). Diethylether was
purchased from Acros Organics (Geel, Belgium). Pressurized gas
CO2 N48 (>99.998%) was purchased from Air Liquide (Dusseldorf,
Germany). Ultra-pure water was provided by a Milli-Q system from
Millipore (Bedford, MA, USA). b-glucuronidase from Escherichia coli
was purchased from Roche Diagnostics GmbH (Mannhein,
Germany).
2.2. Sample preparation of biological samples
A pool of blank urines was prepared by mixing 6 different urine
samples obtained from healthy volunteers. Each urine aliquot of
1 mL was first spiked with doping agent standards in ACN to obtain
5 different levels of concentration, namely 0.1, 0.5, 1, 5 and 10 ng/
mL. Then, they were extracted on Isolute SLEþ 48-well plates
(Biotage, Uppsala, Sweden). Urine was forced through the plate
sorbent using Biotage PRESSUREþ 96 positive pressure manifold at
3 psi. After a waiting time of 5 min, the elution was made by
percolating 3 mL of MtBE through the wells into a 48-well collec-
tion plate. The SLE wells were dried by operating positive pressure
again for few seconds. Then, the extracted samples in the collection
plate were evaporated until dryness using UNIVAPO Rotational
Vacuum Concentrator 150 ECH (Biolabo Scientific Instruments,
Ch^atel-Saint-Denis, Switzerland) at 1250 rpm, ambient
104
kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110
temperature and 30 Torr for 45 min. Then, they were reconstituted
in 200 mL of a mixture of water and ACN (1:1). After 5 min of
agitation, the contents of each well were finally transferred into
full-recovery vials (Waters, Milford, MA, USA) for injection into
UHPSFC-MS/MS.
The same procedure was also used to obtain post-spiked blank
urine extracts. After SLE extraction of blank urine and evaporation
until dryness, these samples were reconstituted in the mixture of
water and ACN spiked with the compounds of interest at different
concentrations. The results obtained with these samples were used
to calculate recoveries (RE) and to measure matrix effects (ME) for
each target compound: recoveries were calculated by dividing the
signal of pre-extraction spiked urine by the signal of post-
extraction spiked urine, whereas the matrix effect was calculated
making the ratio of signals of post-extraction spiked urine and
standards in solvent. Relative matrix effects (MErel) were calculated
as MErel ¼ ME-100.
2.3. UHPSFC-MS/MS instrumentation and method
All experiments were performed on a Waters Acquity UPC2
system equipped with a binary solvent delivery pump, an auto-
sampler which included a 10 mL loop for partial loop injection, a
column oven and a two-step (passive þ active) backpressure
regulator (BPR). The chromatographic system was coupled to a
Waters Xevo TQ-S triple quadrupole detector thanks to a dedicated
double-T splitter interface from Waters. Additional make-up sol-
vent was brought by a Waters Isocratic Solvent Manager (ISM)
pump. Pure MeOH was selected as the make-up solvent and
delivered at 0.3 mL/min. The hyphenation interface and splitter for
UHPSFC-MS/MS are detailed elsewhere [6].
Eight different columns were tested for the method develop-
ment, namely Waters Acquity UPC2 Torus Diol, 2-Picolylamine (2-
PIC), Diethylamine (DEA) and 1-Aminoanthracene (1-AA), Acquity
UPC2 BEH and BEH 2-Ethylpyridine (2-EP), Acquity UPC2 CSH
Fluoro-Phenyl (PFP) and Acquity UPC2 HSS C18 SB. All columns have
dimensions of 100  3.0 mm and particle sizes of 1.7 mm, except the
latter which had 1.8 mm particle sizes. When injecting urine sam-
ples, the selected column was protected with adequate pre-column
Acquity UPC2 Torus Diol VanGuard, 1.7 mm and 5  2.1 mm pur-
chased from Waters.
The mobile phase was a mixture of CO2 and 10 mM AmF dis-
solved in mixture of 98% MeOH and 2% water, as the organic
modifier. The gradient program lasted 6 min and consisted in a
1 min isocratic step at 2% organic modifier followed by a gradient
from 2 to 40% in 4 min and another 1 min isocratic step at 40%
organic modifier. The column was then reequilibrated for 2 min
prior to the next injection. The flow rate was 1.3 mL/min, column
temperature was fixed at 40  C, BPR pressure at 150 bar, and in-
jection volume was equal to 1 mL. A mixture of heptane and iso-
propanol (7:3) was used as the weak solvent wash and pure
isopropanol as the strong solvent wash. The volumes of weak and
strong wash solvents were 600 and 200 mL, respectively.
The Xevo TQ-S detector was exclusively operating in positive ESI
mode and the different parameters were finely tuned as follows to
attain the highest possible sensitivity: source temperature at
150  C, capillary voltage at 3.2 kV and ion source offset at 10 V.
Nitrogen was used as desolvation gas at 850 L/h and 550  C, cone
gas at 145 L/h and nebulizer gas at 7 bar. Argon was used as a
collision gas at 0.17 mL/min. The SRM (selected reaction moni-
toring) transitions, the cone voltages and collision energies for all
target analytes are reported in Table S-1. MassLynx 4.1 software was
used for instrument control and data acquisition. TargetLynx soft-
ware was used for data processing and peak integration.
3. Results and discussion
3.1. Optimization of the chromatographic conditions
Because SFC is known for being a complementary method to LC
and GC in terms of selectivity, this separation mode was evaluated
for the screening of doping agents. It has to be noted, though, that,
taking into account the high proportion of MeOH in the mobile
phase at the end of the gradient, it is very likely that the fluid is no
longer under a supercritical state but rather in a subcritical state.
This is not a drawback as there is continuity in the properties of
supercritical and subcritical fluids. One of the crucial step of
method development in SFC is the optimization of separation
conditions, to i) facilitate the identification of individual com-
pounds by separating isomers and interferences, ii) to ensure
elution of all compounds of interest during the gradient and iii) to
obtain narrow and symmetrical peaks.
The first step of the study was to define SRM transitions of in-
dividual analytes. This must have been performed using individual
compounds to define the transition correctly. As no positive criteria
have been set for screening methods, only one SRM transition was
selected for each compound. All the analytes were ionized using ESI
positive mode providing [MþH]þ ions for many tested analytes.
However, for steroid analysis, it is very common that protonated
molecules provide lower signal intensity than fragments obtained
by loss of one or two molecules of water. Therefore, [MþHeH2O]þ
or [MþHe2H2O]þ were monitored in many cases. Optimized SRM
transitions, cone voltages and collision energies were reported in
Table S-1 (supporting information). Dwell times were set-up
automatically using MassLynx software calculation and defined
peak width was 2.5 s.
High separation efficiency in SFC was obtained by using state-
of-the-art columns packed with sub-2 mm particles. To ensure the
highest possible selectivity, eight different sub-2 mm SFC stationary
phases including hybrid silica (BEH), hybrid silica modified with 2-
EP group, charged surface hybrid (CSH) phase modified with pen-
tafluorophenyl (PFP) group, non-endcapped silica based C18,
hybrid silica modified with diol group, hybrid silica modified with
2-picolylamine group (2-PIC), hybrid silica modified with 1-
aminoanthracene group (1-AA) and hybrid silica modified with
diethylamine group (DEA) were tested. The comparison of these
stationary phases was carried out under the generic mobile phase
conditions reported in Refs. [7,8]. A 4 min generic gradient elution
run from 2 to 40% of methanol with addition of ammonium formate
was used, with a 1 min isocratic step at the beginning and at the
end of the program. Such generic conditions allow covering a wide
range of analyte polarities, together with a reasonable analysis time
of 8 min including 2 min re-equilibration step. Three criteria were
selected for the choice of the best stationary phase: (1) number of
eluted compounds, (2) ability to separate isobaric compounds and
(3) peak shapes. As reported in Fig. 1, Torus Diol and Torus 2-PIC
columns clearly outperformed the other stationary phases in
terms of peak shape and selectivity. Among these two phases, the
Torus Diol was finally selected for further experiments since it had
slightly better peak shape than Torus 2-PIC. On the other hand,
Torus 1-AA was found less suitable for the analysis of selected
doping agents, due to very strong retention leading to non-elution
of several compounds. Also the C18 stationary phase provided
lower selectivity (many isobaric compounds were co-eluted) and
not enough symmetrical peaks. Very good distinction of isobaric
compounds and various isomeric species was observed on CSH PFP
stationary phase, however, at the cost of many distorted peaks (see
Fig. S-2). Remaining three stationary phases (BEH, Torus DEA and
BEH 2-EP) also provided interesting selectivity for the separation of
selected doping agents (Fig. S-2), but the best peak shapes in overall

kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110 105

Fig. 1. Selectivity charts for four of the tested columns. Each compound is plotted according to its retention time (X axis) and mass (Y axis) for the Diol, 2-PIC, 1-AA and HSS C18 SB
columns. The chromatograms of 2 compounds (30 OH-stanozolol in red and raloxifen in blue) are shown to highlight peak shape performance on each column. (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

comparison were obtained on Torus Diol.
Fig. 2 illustrates the complexity of steroid isomers separation,
which remained one of the critical parts of this screening approach.
In the first example (Fig. 2A), a typical case of isobaric compounds is
shown. Three compounds, i. e. mibolerone, methyl-1-testosterone
and metenolone have the same chemical formula C20H30O2 and
therefore the same molecular weight of 302.2. Due to their struc-
tural similarity, interferences in the SRM channel 303.2 > 187.2
were observed, and confirmed by the injections of individual
standard solutions. Thanks to the high resolution offered by
UHPSFC, an easy distinction of these isomers was possible. How-
ever, the presence of isobaric compounds was not the only issue in
method development. Indeed, most of the steroid structures pro-
vide a loss of one or two molecules of water in their fragmentation
pathway and these may also contribute to SRM channel of the other
compounds, such as in case of one metabolite of danabol, which
was monitored as [MþHeH2O]þ to have higher sensitivity. There-
fore, in the SRM channel of the metabolite of exemestane at
299.2 > 121.1, [MþHeH2O]þ ion of the metabolite of danabol was
also observed (Fig. 2B). Due to many interferences between isobaric
compounds and also [MþHeH2O]þ ions, the 100 compounds to be
analyzed were divided into two mixtures, as reported in Table S-1,
and specific acquisition windows centered on the retention time of
each compound were created in the SRM method. Otherwise, the
quantitative evaluation of the method, including calculation of re-
covery and matrix effects would not be possible. In addition, the
presence of interfering compounds in the generic screening step
may cause problems in the case of suspicious doping samples,
making the correct identification of doping agent more challenging.
This problem may be addressed with higher success rate during the
confirmation step. In fact, for an unequivocal identification of the
suspected substance and related metabolites, experimental condi-
tions for the confirmation procedure could be further optimized
using SFC-MS, including the use of second column chemistry with
complementary selectivity (e.g. 2-PIC chemistry), chromatographic
gradients with improved peak capacity and additional SRM tran-
sitions. Furthermore, especially in the case of steroids, the use of an
orthogonal technique such as GC-EI-MS/MS could also be envis-
aged for confirmation purposes, providing more informative MS/
MS spectra through an additional derivatization procedure of the
compounds of interest. For confirmation, at least two SRM transi-
tions should be monitored, as required by WADA. Then, unequiv-
ocal identification of compounds should be obtained by comparing
chromatographic retention times (±1% or ±0.1 min) and relative ion
intensities of the two monitored transitions in real samples with
those of certified reference material and/or administration studies
[24].
After having selected the best stationary phase, fine method
tuning was performed. Based on previous findings [7,8], the influ-
ence of additive on peak shape, separation selectivity and especially
MS response was monitored. According to the provider, the Torus
columns are reported to provide good peak shapes even without
additives in the mobile phase. Various types of mobile phase
components were tested, including i) pure methanol, ii) methanol
with 2% of water, iii) methanol þ10 mM ammonium acetate, iv)
methanol þ10 mM ammonium formate and finally v)
methanol þ10 mM ammonium formate þ2% water. Symmetrical
peak shapes were observed for most of the investigated analytes
that are neutral or weakly acidic, when using methanol or meth-
anol with 2% water. However, using these organic modifiers, MS
106
kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110

Fig. 2. UHPSFC-MS/MS chromatograms for SRM transitions of metenolone (blue) and metabolite of exemestane (red). Conditions: column Torus Diol (100 mm  3.0 mm, 1.7 mm),
mobile phase A: CO2, mobile phase B: methanol with 10 mM ammonium formate þ 2% water, 2% B for 1 min then gradient from 2 to 40% in 4 min and 40% B for 1 min, flow rate:
1.3 mL/min, temperature: 40  C, backpressure: 150 bar, make-up pump: methanol at 0.3 mL/min, injection volume: 1 mL. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

sensitivity was systematically decreased, often by more than 70%
compared to the reference, which was methanol þ10 mM ammo-
nium formate. The performance of the two buffers, ammonium
formate and acetate in methanol, was very similar for our selected
group of compounds. In agreement with our previous findings [7,8],
the MS signal was systematically increased when 2% water was
added to the mobile phase containing methanol þ ammonium
formate. The increase in signal was about 30e70% in most cases,
reaching even more than 200% for a few substances. Therefore,
water was added to the mobile phase. Finally, some experiments
were also performed to verify the influence of the make-up solvent
on the MS response using methanol, ethanol and isopropanol in
presence or absence of 2 and 5% of water. While ethanol and iso-
propanol decreased the MS signal compared to the reference
response of pure methanol, the addition of water did not change
the response significantly or lead to some slight decrease in the
case of 2% of water addition. Therefore, pure methanol was used as
make-up solvent.
3.2. Optimization of the sample preparation step
Using optimized conditions, developed UHPSFC-MS/MS method
was tested with standard solutions to determine its linearity and
sensitivity. The method demonstrated very good sensitivity and
linearity (data on standard solutions not shown), which was a good
starting point for further development of sample preparation pro-
cedure in urine. For this purpose, SLE was selected as a simplified
variant of LLE. SLE columns contain specially processed wide-pore
diatomaceous earth, a chemically inert matrix, which only acts as
a holder for the sample [13]. After loading the sample, an
immiscible organic solvent is passed through the column to extract
the desired components, typically 2e3 times of the sample volume.
The residue is subsequently evaporated to dryness, reconstituted
and injected onto the chromatographic system. Due to the low
polarity of most of the tested compounds, SLE is a very interesting
strategy providing urine preconcentration and sample clean-up
(elimination of polar interfering compounds present in urine
samples which do not participate in non-polar organic solvents
partitioning). A 48 well-plate format was selected in this research
to ensure high sample throughput, which is required in routine
doping control analysis. Even if this study was carried out using
only spiked urines, selected compounds are mainly excreted as
glucuronides in real urine samples. Therefore, preliminary extrac-
tion experiments have been performed to evaluate a possible in-
fluence of the enzymatic hydrolysis step on SLE procedure. For
these tests, the pool of blank urines (pH determined at 7) was
spiked with selected substances and was then subjected to hy-
drolysis with b-glucuronidase from E. coli for 1 h at 50  C, followed
by SLE procedure as described above. No significant difference in
peak area was observed on the chromatograms after hydrolysis
compared to non-hydrolysed spiked samples (data not shown);
therefore this step was omitted for practical reasons in the
continuation of this work. Finally, performance of SLE was also
evaluated with two of the most widely used solvents in the
extraction of steroids, namely methyl tert-butyl ether (MtBE) and
diethylether (DEE). MtBE was selected since the overall recoveries
were comparable or even better when using MtBE vs. DEE.

kova
L. Nova
3.3. Evaluation of extraction recoveries
Urines spiked with mixture 1 or mixture 2 (see Table S-1) were
tested at five concentration levels with the final protocol to obtain
extraction recovery values. The summary of results when using
MtBE as an extraction solvent is shown in Table S-2 (supporting
information) for all compounds. The suitability of the SLE proce-
dure for the sample treatment is shown in Fig. 3. For most of the
compounds, the extraction recovery was >70% (48%, 59% and 68% of
compounds at selected concentration levels of 0.5, 1 and 5 ng/ml,
respectively), while only nine compounds provided recovery lower
than 50% at low concentration levels equal to 0.5 and 1 ng/ml. Such
recoveries are perfectly adapted for the routine screening analysis
of selected doping agents, provided that MRPLs of all compounds
are reached. The capability of the method to reach MRPLs is further
influenced by the amount of matrix effects, besides method pre-
cision, sensitivity and recovery. The lowest recoveries were sys-
tematically observed for zilpaterol (~5%), due to its high polarity
and tamoxifen (~20%). However, this issue was compensated by a
good response of both compounds in UHPSFC-MS/MS and thus
sufficient method sensitivity. The MS responses of the investigated
compounds were indeed quite different depending on their
physico-chemical properties, as illustrated at 5 ng/ml. At this con-
centration level, there were still three compounds (metabolites of
norethisterone, norbolethone and turinabol) which provided very
low signal in UHPSFC-MS/MS and thus cannot be detected, while
six other compounds cannot be quantified anymore due to the
saturation of the detector. This is typically the case of synthetic
cannabinoids, which provide very good response in ESI positive
mode. Method precision was tested by performing quadruplicate
extractions of urine at all five concentration levels. Calculated RSD
values were in most cases lower than 10e15%, with the large ma-
jority well below 10%, which is considered to be excellent for the
extraction of small compounds from biological material, without
the use of internal standards in the case of MS detection (see
Table S-3). Worse results (RSD > 20%) were only observed for a few
compounds including zilpaterol, toremifene, triamcinolone and
Fig. 3. Recovery [%] of doping agents using SLE procedure for the extraction of substances from urine at concentration levels of 0.5, 1.0 and 5.0 ng/ml. Individual values of recoveries
for each compound are shown in Table S-2 of the supporting information.

et al. / Analytica Chimica Acta 915 (2016) 102e110 107
metabolite 1 of budesonide.
3.4. Evaluation of matrix effects
Matrix effects (ME) are sometimes an important issue of LC-MS
analytical methods, as they can negatively influence method ac-
curacy, precision and sensitivity. Due to the co-eluting compounds
from the matrix (or other sources), the signal response may be
altered leading to ion suppression or enhancement [14,15]. There-
fore, it is important to evaluate the amount of ME when developing
new bioanalytical methods [16]. While the phenomenon of ME has
been widely explored in LC-MS bioanalytical methods [15,17e19],
only few reports deal with the determination of matrix effects in
SFC-MS [8,20,21]. SFC might be a powerful strategy to reduce the
amount of ME [8,20], due to its very different separation mecha-
nism mostly based on H-bond and dipoleedipole interactions.
Experimentally, blank urine samples were extracted using the final
protocol and then spiked with both standard mixtures (see Table S-
1) at five concentration levels to obtain matrix effect values for each
compound. The calculation of ME was based on the post-extraction
addition approach described elsewhere [22]. Since relative matrix
effects (MErel) were calculated, positive values implicitly indicated
signal enhancement, while negative values corresponded to signal
suppression. The amount of ME observed in our study was very low,
when using SLE as sample preparation step and UHPSFC-MS/MS for
the analysis. The results of MErel for all analytes at five concentra-
tion levels are shown in Table S-2 (supporting information), while
Fig. 4 shows the comparison of the amount of ME at three inter-
mediate concentration levels (0.5, 1 and 5 ng/ml) as a function of
retention times. For most of the compounds, typically 60e70% of
tested doping agents, matrix effects were negligible (MErel < þ20%
and > 20%) using SLE as a sample preparation procedure. As
shown, slightly more positive MEs were observed for 23% of com-
pounds at 0.5 ng/ml, for 17% of compounds at 1.0 ng/ml and for 12%
of compounds at 5.0 ng/ml, while negative MEs were observed for
7e10% of compounds at all three concentration levels. For most of
the analytes, ME remained reasonable, except for the metabolite 2
108
kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110
Fig. 4. Relative matrix effects [%] observed in UHPSFC-MS/MS method after SLE procedure as a function of retention times at three concentration levels: 0.5 ng/ml (blue diamonds),
1 ng/ml (red circles) and 5 ng/ml (grey triangles). All the experimental data were now shown, to keep the clarity of the figure, but Table S-2 provides an exhaustive list of relative
matrix effects at all concentrations levels. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
of prostanozol, demonstrating matrix effects of 206% at 0.5 ng/ml,
179% at 1 ng/ml and 113% at 5 ng/ml. Therefore, these values were
not included in Fig. 4 and are only shown in Table S-2 (supporting
information). Bolasterone, 30 OH-stanozolol and beclomethasone
were also importantly influenced by ME (MErel > þ50% or < 50%).
Similarly to Fig. 3, at the highest concentration level of 5 ng/ml, six
compounds could not be evaluated quantitatively, due to the de-
tector saturation. Also, in the case of ME, it is important to have
consistent values through the measurements of different samples
[16], therefore, RSD values were calculated for the individual con-
centration levels, and obtained RSD values were always better than
10%, except for the metabolites of budesonide, triamcinolone and
toremifene (see Table S-3).
3.5. Method sensitivity and comparison with MRPLs
Obviously, the most important aspect of the developed
screening method is to reach MRPL required by WADA. According
to dedicated technical document [23], substances should be
unambiguously detected at 100% MRPL value, while the estimated
LOD (limit of detection) should not be higher than 50% of MRPL.
Only if these conditions are met, the method may be applicable for
routine doping screening analysis. In Fig. 5, the MRPLs of 100 tested
substances were compared with the achievable sensitivities esti-
mated for target analytes spiked in urine samples. This figure
confirms that the developed UHPSFC-MS/MS method using high
throughput SLE as a sample preparation (including a 5-fold pre-
concentration factor) is applicable for doping screening analysis
of the selected classes of substances. The 50% MRPLs were
comfortably reached for almost all of the 100 doping agents, since
the achieved LODs were in average between 5 and 50-fold lower
than the MRPLs. Because LODs were excellent, the detection time
window of these doping agents can be extended. As shown in Fig. 5,
there were only three exceptions, namely the metabolites of nor-
ethisterone, norbolethone and turinabol. These compounds pro-
vided low response in UHPSFC-MS/MS, thus higher
preconcentration during the SLE step would be needed to attain the
requested MRPLs. However, this does not have to be necessarily
done, as other metabolites (e.g. norbolethone met. 2) or the parent
compound itself (e.g turinabol) were monitored with sufficient
sensitivity. Regarding norethisterone, additional metabolites
should be included in this screening method to improve detect-
ability of this compound in routine analyses.
4. Conclusion
In the field of doping control analysis, GCeMS(/MS) and
LCeMS(/MS) are the gold standard techniques for screening and
confirmatory analysis of substances prohibited in sport practice.
The objective of this work was to assess the potential of UHPSFC-
MS/MS, as an alternative to the above-mentioned analytical stra-
tegies, for the determination of 100 compounds in urine, including
parent substances and their main phase I metabolites. These ana-
lytes belong to the difficult classes of anabolic agents, hormone and
metabolic modulators, synthetic cannabinoids and glucocorticoids.
As a sample preparation procedure, SLE was found to be particu-
larly well suited for these relatively lipophilic substances and pro-
vided a high throughput extraction using 48-well plate format.
Depending on the tested concentration levels (0.5, 1.0 and 5.0 ng/
mL), recoveries were better than 70% for 48e68% of the target
compounds and higher than 50% for 83e87% of the tested sub-
stances. The chromatographic conditions were finely optimized,
considering the presence of numerous interferences related to
steroid isomers and ions produced by the loss of water in the ESI
source. Various stationary phase chemistries were tested and, a
Torus diol 100  3 mm, 1.7 mm column was finally selected. Using a
generic gradient from 2 to 40% MeOH in presence of 10 mM
ammonium formate þ2% water, the elution of all the substances
was possible in only 8 min, with narrow peaks and limited in-
terferences issues (high chromatographic selectivity). The suscep-
tibility of the developed method towards ME was also evaluated at
different concentration levels, ranging from 0.1 to 10 ng/mL. The
relative ME was reasonable, as 60e70% of the tested substances
have relative ME within the range ±20%. In addition, there were

kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110 109

Fig. 5. Concentration levels providing sufficient sensitivity and precision (blue lines) versus required MRPLs (red lines) for individual doping agents of mixture 1 (A) and mixture 2
(B). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

generally slightly more signal enhancement rather than ion sup-
pression. Finally, achieved LODs were compared to MRPLs fixed by
WADA for all these classes of compounds. Thanks to the 5-fold SLE
preconcentration step, and the high sensitivity provided by modern
UHPSFC-MS/MS device, LODs were systematically below the
MRPLs, except for one metabolite of each of these substances:
norethisterone, norbolethone and turinabol.
In order to conclude, the developed SLE-UHPSFC-MS/MS
method provided excellent performance in terms of turnaround,
selectivity and sensitivity. It could potentially be employed in
doping control laboratories for screening purposes in the near
future, as a replacement strategy to the most widely used LCeMS/
MS and GCeMS/MS platforms. Moreover, despite that the hydro-
lysis step should be maintained, the rapid sample preparation
developed in this study using automated SLE could be of great value
for routine analysis avoiding tedious manipulations during LLE,
evaporation, reconstitution and derivatization steps required for
GCeMS/MS analysis.
Acknowledgements
The authors gratefully acknowledge Waters, and particularly
Marcus Winkler, Stephane Canarelli and Joel Fricker for the kind
opportunity to use the Waters Acquity UPC2 and Xevo TQ-S in-
struments in their demo-lab in Eschborn, Germany. This project has
been carried out with the support of the World Anti-Doping Agency
(14A21RN).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.aca.2016.02.010.
References
[1] M. Thevis, T. Kuurane, H. Geyer, W. Schanzer, Annual banned-substance re-
view: analytical approaches in human sports drug testing, Drug Test. Anal. 7
(2015) 1e20.
[2] F. Badoud, D. Guillarme, J. Boccard, E. Grata, M. Saugy, S. Rudaz, J.-L. Veuthey,
Analytical aspects in doping control: challenges and perspectives, Forensic Sci.
Int. 213 (2011) 49e61.
[3] E. Lesellier, C. West, The many faces of packed column supercritical fluid
110
kova
L. Nova
et al. / Analytica Chimica Acta 915 (2016) 102e110
chromatography e a critical review, J. Chromatogr. A 1382 (2015) 2e46.
[4] V. Desfontaine, D. Guillarme, E. Francotte, L. Nov

, Supercritical fluid
akova
chromatography in pharmaceutical analysis, J. Pharm. Biomed. Anal. 113
(2015) 56e71.
[5] A. Grand-Guillaume Perrenoud, J.-L. Veuthey, D. Guillarme, The use of col-
umns packed with sub-2 mm particles in supercritical fluid chromatography,
Trends Anal. Chem. 63 (2014) 44e54.
[6] A. Grand-Guillaume Perrenoud, J.-L. Veuthey, D. Guillarme, Coupling state-of-
the-art supercritical fluid chromatography and mass spectrometry: From
hyphenation interface optimization to high-sensitivity analysis of pharma-
ceutical compounds, J. Chromatogr. A 1339 (2014) 174e184.
[7] L. Nova
kova
, A. Grand-Guillaume Perrenoud, R. Nicoli, M. Saugy, J.-L. Veuthey,
D. Guillarme, Ultra high performance supercritical fluid chromatography
coupled with tandem mass spectrometry for screening of doping agents. I:
Investigation of mobile phase and MS conditions, Anal. Chim. Acta 853 (2015)
637e646.
[8] L. Nova
kova
, A. Grand-Guillaume Perrenoud, R. Nicoli, M. Saugy, J.-L. Veuthey,
D. Guillarme, Ultra high performance supercritical fluid chromatography
coupled with tandem mass spectrometry for screening of doping agents. II:
analysis of biological samples, Anal. Chim. Acta 853 (2015) 647e659.
[9] W. Schanzer, M. Thevis, Human sports drug testing by mass spectrometry,
Mass Spec. Rev. (2015), http://dx.doi.org/10.1002/mas.21479.
[10] L. Nova
kov

a, P. Chocholous, P. Solich, Ultra-fast separation of estrogen ste-
roids using subcritical fluid chromatography on sub-2-micron particles,
Talanta 121 (2014) 178e186.
[11] M. Doue, G. Dervilly-Pinel, K. Pouponneau, F. Monteau, B. Le Bizec, Analysis of
glucuronide and sulfate steroids in urine by ultra-high-performance super-
critical-fluid chromatography hyphenated tandem mass spectrometry, Anal.
Bioanal. Chem. 407 (2015) 4473e4484.
[12] X. Xu, J.M. Roman, T.D. Veenstra, J. Van Anda, R.G. Ziegler, H.J. Issaq, Analysis
of fifteen estrogen metabolites using packed column supercritical fluid
chromatography-mass spectrometry, Anal. Chem. 78 (2006) 1553e1558.
[13] L. Nova
kov
a, H. Vl
, A review of current trends and advances in modern
ckova
bio-analytical methods: chromatography and sample preparation, Anal. Chim.
Acta 656 (2009) 8e35.
[14] L. Tang, P. Kebarle, Dependence of ion intensity in electrospray mass spec-
trometry on the concentration of the analytes in the electrosprayed solution,
Anal. Chem. 65 (1993) 3654e3668.
[15] A. Cappiello, G. Famiglini, P. Palma, E. Pierini, V. Termopoli, H. Trufelli,
Overcoming matrix effects in liquid chromatography-mass spectrometry,
Anal. Chem. 80 (2008) 9343e9348.
[16] Committee for medicinal products for human use, Guideline on Bioanalytical
Method Validation, EMA, London, UK, 2011.
[17] R. Dams, M.A. Huestis, W.E. Lambert, C.M. Murphy, Matrix effect in bio-
analysis of illicit drugs with LC-MS/MS: influence of ionization type, sample
preparation, and biofluid, J. Am. Soc. Mass Spectrom. 14 (2003) 1290e1294.
[18] E. Chambers, D.M. Wagrowski-Diehl, Z. Lu, J.R. Mazzeo, Systematic and
comprehensive strategy for reducing matrix effects in LC/MS/MS analyses,
J. Chromatogr. B 852 (2007) 22e34.
[19] A. Van Eckhaut, K. Lanckmans, S. Sarre, I. Smolders, Y. Michotte, Validation of
bioanalytical LC-MS/MS assays: evaluation of matrix effects, J. Chromatogr. B
877 (2009) 2198e2207.
[20] T. Berg, L. Kaur, A. Risnes, S.M. Havig, R. Karinen, Determination of a selection
of synthetic cannabinoids and metabolites in urine by UHPSFC-MS/MS and by
UHPLC-MS/MS, Drug Test. Anal. (2015), http://dx.doi.org/10.1002/dta.1844 (in
press).
[21] Y. Hsieh, F. Li, C.J. Duncan, Supercritical fluid chromatography and high-
performance liquid chromatography/tandem mass spectrometric methods
for the determination of cytarabine in mouse plasma, Anal. Chem. 79 (2007)
3856e3861.
[22] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for the
assessment of matrix effect in quantitative bioanalytical methods based on
HPLC-MS/MS, Anal. Chem. 75 (2003) 3019e3030.
[23] TD2015MRPL, World Anti-doping Agency, Montreal, 2015. https://wada-
main-prod.s3.amazonaws.com/resources/files/wada_td2015mrpl_minimum_
required_perf_levels_en.pdf (Last accessed on October 2015).
[24] TD2015IDCRL, World Anti-doping Agency, Montreal, 2015. https://wada-
main-prod.s3.amazonaws.com/resources/files/wada_td2015idcr_minimum_
criteria_chromato-mass_spectro_conf_en.pdf (Last accessed on January 2016).