Journal of Analytical Toxicology 2014;38:637 –644

doi:10.1093/jat/bku098 Advance Access publication August 7, 2014 Article

Screening for Anabolic Steroids in Urine of Forensic Cases Using Fully Automated
Solid Phase Extraction and LC –MS-MS
David W. Andersen* and Kristian Linnet
Department of Forensic Chemistry, University of Copenhagen, Frederik Femte vej 11, DK-2100 Copenhagen, Denmark

*Author to whom correspondence should be addressed. Email: [email protected]

A screening method for 18 frequently measured exogenous anabolic The number of published LC – MS methods for screening for
steroids and the testosterone/epitestosterone (T/E) ratio in forensic AAS has steadily increased over the last decade. A challenge
cases has been developed and validated. The method involves a fully when using LC – MS-MS is the lack of ionization of some com-
automated sample preparation including enzyme treatment, addition pounds. Atmospheric pressure photo ionization, atmospheric
of internal standards and solid phase extraction followed by analysis pressure chemical ionization and electrospray ionization (ESI)
by liquid chromatography –tandem mass spectrometry (LC –MS-MS) have all been used in the analysis of AAS. An early comparison be-
using electrospray ionization with adduct formation for two com- tween the three ionization sources suggested that ESI was the
pounds. Urine samples from 580 forensic cases were analyzed to best technique (11) although the chemical properties of the an-
determine the T/E ratio and occurrence of exogenous anabolic ste- alytes should determine the choice of the source of ionization
roids. Extraction recoveries ranged from 77 to 95%, matrix effects (12). Formation of adducts has been applied to compounds
from 48 to 78%, overall process efficiencies from 40 to 54% and that ionize poorly, thereby facilitating the use of LC – ESI –
the lower limit of identification ranged from 2 to 40 ng/mL. In the MS-MS analysis in the detection of AAS (13–15). Several manual
580 urine samples analyzed from routine forensic cases, 17 (2.9%) and semi-automated online sample preparation procedures using
were found positive for one or more anabolic steroids. Only seven dif- solid phase extraction (SPE) have been developed for analyzing
ferent steroids including testosterone were found in the material, sug- AAS in human, horse and bovine urine samples (16 –19).
gesting that only a small number of common steroids are likely to Findings of AAS in police cases in Sweden from 1999 to 2009
occur in a forensic context. The steroids were often in high concen- showed that the mean concentrations of the investigated
trations (>100 ng/mL), and a combination of steroids and/or other AAS ranged from 135 to 2,080 ng/mL (n ¼ 6,362) with a maxi-
drugs of abuse were seen in the majority of cases. The method pre- mum concentration in a single sample over 30,000 ng/mL
sented serves as a fast and automated screening procedure, proving (19-norandrosterone) (20). A high rate of positive samples was
the suitability of LC– MS-MS for analyzing anabolic steroids. seen (33.5%), but this was due to the fact that samples only
were analyzed for AAS when police suspected an abuse. The
most frequent motive for using AAS concerns the anabolic ef-
fects, including the aspect of obtaining a better looking body
Introduction and performing better in sports (21). Among other self-reported
Anabolic androgenic steroids (AAS) are compounds that mimic the motives is the desire to become more aggressive or brave, often
structure and biological effects of the naturally occurring male sex in a criminal context. A combination of AAS and other drugs of
hormone testosterone (T). They are commonly used among cer- abuse is often seen, as drug addicts use AAS in the belief that
tain groups, especially young males striving for a greater muscle they counteract the negative effects of their drug abuse, and
mass. Up to 4 million people are currently using AAS (1), even abusers motivated by the anabolic effects apply stimulants to
though they are associated with well-known adverse effects (2). train harder (22).
Testing for AAS in sports was introduced at the Olympic The method presented here is designed for a forensic laboratory
Games in 1976 in Montreal using a radioimmunoassay analysis. primarily relying on LC–MS equipment. Exogenous compounds
This technique was subsequently replaced by gas chromatogra- were selected on the basis of the findings in the other Nordic
phy (tandem) mass spectrometry (GC– MS-MS) and later by liq- countries (20) and from police seizures presented to our laborato-
uid chromatography – tandem mass spectrometry (LC – MS-MS). ry. Because high concentrations of AAS are expected, a fast and ef-
The common approach among the World Anti-Doping Agency ficient automated screening method with high selectivity was
(WADA) laboratories is an enzymatic hydrolysis, followed by prioritized, rather than a more laborious method aiming at fulfill-
liquid–liquid extraction, and finally GC–MS(/MS) or LC–MS-MS ing the sensitivity demands required for doping in sports analysis.
analysis (3). Trimethylsilyl derivatization has proven to greatly The method presented here provides a fully automated sample
improve the sensitivity of GC – MS assays (4). A major topic in preparation followed by LC–MS-MS analysis for common exoge-
the analysis of AAS is the detection of externally administered nous AAS and their metabolites and also a determination of the
T. In the 1980s, an upper limit of 6.0 was set for the ratio between T/E ratio. The method presented by Pozo et al. (15) proved
T and epitestosterone (E) (5), and WADA later lowered the that LC – MS using ESI and adduct formation could be used as a
threshold to the current 4.0 limit. However, due to large varia- qualitative screening tool for exogenous steroids. Automated
tions in naturally occurring levels of T, this ratio can exceed SPE sample preparation was incorporated into this method,
6.0 in some cases (6). Therefore, a subsequent analysis by GC/ which was further modified to accommodate the analytes of in-
C/IRMS is needed to confirm the presence of pharmaceutically terest. A full qualitative validation for the screening of 18 exoge-
produced T (7). Newer studies include more endogenous com- nous AAS including a determination of the lower limit of
pounds for a complete steroid profile (8 –10). identification (LLOI) was performed. For determination of the

# The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]
T/E ratio, we compared the concentration ratio determined by methanol–water), samples were transferred and sealed in a new
using stable isotope-labeled internal standards (SIL-IS) of T and 96-well plate ready for injection on the analytical instrument.
E with a simpler approach using only the peak areas of T and
E. The method was used in the analysis of 580 urine samples re-
ceived from the jurisdiction of Eastern Denmark. Samples from LC– MS-MS analysis
both males and females were collected in a continuous time pe- Samples were analyzed by HPLC (1100 series; Agilent,
riod to establish the occurrence as well as concentration levels of Wilmington, DE, USA) with an upgraded autosampler (1260 se-
AAS in a forensic routine material. ries; Agilent, Wilmington, Germany) coupled to the triple quad-
rupole mass spectrometer (Micromass; Waters, Milford, MA,
USA). Separation was performed on a C18 Kinetex column
Materials and methods (50  2.10 mm, particle size 2.6 mm; Phenomenex) with aque-
ous ammonium acetate (1 mM) and methanolic ammonium ace-
Chemicals
tate (1 mM) as mobile phase. The flow rate of the mobile phase
Nandrolone (ND) was purchased from Organon (Oss, the was 0.50 mL/min and a gradient varying the percentage of organ-
Netherlands). 19-Norandrosterone (19NAN), 19-noretiocholanolone ic solvent linearly: 0 min, 35%; 7 min, 40%; 10 min, 50%; 18.5 min,
(19NET) and oxandrolone (OXA) were obtained from Steraloids 60%; 18.5 min, 100%; 20.5 min, 100%; 20.5 min, 35% and 23 min,
(Newport, RI, USA). Stanozolol (STAN), methandrostenolone 35%. The injection volume was 30 mL. The source temperature of
(MD), methyltestosterone (MT) and epitestosterone (E) were the mass spectrometer was held at 1208C and the desolvation
from Sigma-Aldrich (St Louis, MO, USA). 30 -Hydroxystanozolol temperature at 3008C. Compound-specific settings used for the
(3STAN) and dihydroboldenone (BOLDm) were from LGC ionization and fragmentation of analytes are listed in Table I,
Standards (Teddington, UK). 16b-Hydroxystanozolol (16STAN) along with the selected precursor and product ions.
and methenolone (MET) were from Cerilliant (Austin, TX, USA).
Boldenone (BOLD), 17-epimethyltestosterone (17MT), drost-
anolone (DROS), trenbolone (TREN) and 17-epitrenbolone Table I
(17TREN) were from Toronto Research Chemicals (Toronto, Mass Spectrometer Settings and Retention Times for the Selected Analytes and Internal Standards
Canada). Mesterolone (MEST) and testosterone (T) were pur-
Analyte Precursor ion Cone voltage Product ion Collision energy Retention time
chased from Schering Plough (Levallois Perret Cedex, France). (m/z) (V) (m/z) (eV) (min)
3a-Hydroxy-1a-methyl-5a-androstane-17-one (MESTm), testos-
ND 275 32 257 16 7.1
terone-d3 (T-d3) and epitestosterone-d3 (E-d3) were from NMI 239 18
(Sydney, Australia). Methanol, 2-propanol and water (LC – MS 19NAN 277 22 241 14 12.5
grade) were obtained from Fisher Scientific (Leicestershire, 201 16
19NET 277 22 241 14 12.0
UK). Ammonium acetate was from Merck (Darmstadt, Germany). 201 16
b-Glucuronidase/arylsulfatase from Helix pomatia was obtained STAN 329 60 81 44 15.6
from Roche Diagnostics GmbH (Mannheim, Germany). 121 34
3STAN 345 66 97 42 10.4
107 42
16STAN 345 65 81 44 11.7
95 40
Sample preparation MD 301 18 149 15 8.2
The sample preparation procedure was based on automated SPE 283 11
BOLD 287 20 135 14 6.6
using a previously described setup using a Freedom Evo 200 liq- 173 16
uid handling robot (Tecan, Männerdorf, Switzerland) equipped BOLDm 321 28 187 24 11.0
121 24
with several add-ons (23). All samples and solvents for the glucu- MET 303 30 83 21 11.3
ronidase treatment were placed in the robot, and 400 mL of urine 187 21
was transferred from a primary tube to a 96-well plate. The iden- MT 303 32 285 16 10.9
267 16
tities of the samples were confirmed using the barcode reader. 17MT 303 31 109 28 14.0
Diluted glucuronidase enzyme (20 mL, 1:4 enzyme – water, 189 20
b-glucuronidase activity: 1.65 U, arylsulfatase activity: 5.13 U) DROS 305 29 215 17 14.6
161 21
was added to each sample followed by ammonium acetate buffer MEST 305 34 269 16 12.9
(200 mL, 0.1 M, pH 5.3). Samples were shaken and incubated 229 20
MESTm 322 16 161 12 15.1
overnight on the orbital shaker. 305 16 229 18
After incubation at 408C, the remaining solvents were placed TREN 271 35 178 50 5.7
in the robot, and ammonium acetate buffer (400 mL, 1 M, pH 199 23
17TREN 271 35 178 50 6.6
4.0) was added to each sample tube and mixed thoroughly. An 199 23
aliquot (900 mL) of each sample was transferred to a 96-well OXA 307 24 289 10 7.1
plate, and 20 mL of internal standard (176 ng/mL T-d3 and 135 12
T 289 30 109 25 9.3
176 ng/mL E-d3) was added to each sample. The samples were 97 21
centrifuged and were extracted by reverse-phase SPE using a E 289 30 109 25 11.5
97 21
Strata-X plate (30 mg/well; Phenomenex, Torrance, CA, USA) T-d3 292 33 109 24 9.3
using water (900 mL) and 40:60 water – methanol (900 mL) as 97 21
washing solvents and isopropanol – methanol (1:1, 2  300 mL) E-d3 292 33 109 24 11.5
97 21
for elution. After evaporation and reconstitution (60 mL 35:65

638 Andersen and Linnet

Method validation T and E were validated for quantitative analysis using T-d3 and
Validation was performed following the guidelines of Eurachem E-d3 as internal standards. Linearity was validated by spiking
for qualitative analysis (24). Twenty samples were used to deter- water samples at seven different levels with six replicates over
mine the LLOI of the 18 exogenous steroids and to confirm the 2 days. Validation of precision and accuracy was performed by
selectivity of the applied method. The 20 samples were fortified spiking urine samples at six levels with eight replicates, thus de-
at various concentrations ranging from 2 to 80 ng/mL and includ- termining the lower limit of quantification. Baseline concentra-
ed samples from both males and females as well as samples from tions of T and E were determined for each urine sample, and
living and postmortem cases. Identification criteria when using the spiked levels were adjusted accordingly. ME, ER and PE
LC –MS-MS were adapted from Rivier (25), which specifies that were tested by the procedure described for the qualitative vali-
the S/N ratio should be above 3; the retention time should differ dation of the exogenous AAS.
by ,2% from a reference standard and the relative ion intensity The determination of the T/E ratio by two different approach-
should be within +20% to +50% that of the reference standard es was compared for 45 randomly selected samples from the
depending on the relative abundance of the measured ion to the forensic material, including postmortem samples:
base peak. The method validation hereby also complied with the † A simple method only using the area of the chromatographic
Scientific Working Group for Forensic Toxicology (SWGTOX) peak of the analyte: T/E ratio (area) ¼ area (T)/area (ET).
recommendations in which experiments for interference (mini- † A method that used the response of the analyte as corrected
mum 10 sources of sample matrix), testing of carryover and by the area of the internal standard multiplied by the concen-
LLOD are required for a qualitative method (26). tration (C) of the internal standard: response (T) ¼ area (T)/
Matrix effects (ME), extraction recovery (ER) and process effi- area (T-d3)  C (T-d3). This method thereby calculated the
ciency (PE) were tested by spiking both urine samples and water T/E ratio (SIL-IS) ¼ response (T)/response (ET).
samples before extraction, after extraction and after evaporation
following the procedure suggested by Matuszewski et al. (27).
ME was calculated as ME (%) ¼ B/A  100, ER (%) ¼ C/B  Results
100 and PE (%) ¼ C/A  100, where A corresponds to the area Automated sample preparation
of the peak obtained from a neat standard, B is the area of the All operations during the SPE sample preparation were automat-
peak from a sample spiked after extraction and C corresponds ed. Sample preparation was performed in ,3 h, not including the
to the area of the peak from a sample spiked before extraction. enzyme treatment and incubation of the samples that was per-
Sets of eight replicates were used to assess ME, ER and PE. formed overnight. Manual interactions were limited to the initial
Furthermore, carryover and both short- and long-term stabilities loading of samples, solvents, plates and standards and transferal of
were tested to complete the validation. Ion suppression was fur- the 96-well plate to the analytical instrument. The initial loading
ther investigated by post-column infusion. When screening the process of samples, etc., was performed in ,30 min, and addi-
forensic cases for exogenous AAS, a semi-quantitative assessment tional time spent on manual labor was minimal.
was applied using calibration points in two replicates at three lev- Standard solutions for calibrators, standard addition experi-
els at 5, 20 and 80 ng/mL. The method was not validated as a ments and SIL-IS were dispensed by the automated liquid handler
quantitative method, but this semi-quantitative approach was with an imprecision and bias of ,3%. The pipetting procedure
used to display approximate concentrations in the forensic mate- and performance of the automated system have been previously
rial. Quality control samples at three levels containing all analytes published (23). The ER ranged from 77 to 95% for all of the com-
were included in each batch to validate long-term stability and pounds (Table II). The ME was between 48 and 78% yielding an
precision. overall PE of 40– 54%.

Table II
Extraction Recovery, Matrix Effect, Positive Findings Score After Analysis and Lower Limit of Identification for the Exogenous AAS

Analyte ER (%) ME (%) Positive findings LLOI (ng/mL)

2 ng/mL 5 ng/mL 10 ng/mL 20 ng/mL 40 ng/mL 80 ng/mL
ND 90 53 20/20 20/20 20/20 20/20 20/20 20/20 2
19NAN 86 58 8/20 20/20 20/20 20/20 20/20 20/20 5
19NET 95 63 11/20 20/20 20/20 20/20 20/20 20/20 5
STAN 84 67 20/20 20/20 20/20 20/20 20/20 20/20 2
3STAN 77 78 12/20 15/20 20/20 20/20 20/20 20/20 10
16STAN 87 66 20/20 20/20 20/20 20/20 20/20 20/20 2
MD 90 48 17/20 17/20 20/20 20/20 20/20 20/20 10
BOLD 88 53 20/20 20/20 20/20 20/20 20/20 20/20 2
BOLDm 89 57 3/20 8/20 14/20 20/20 20/20 20/20 20
MET 88 60 20/20 20/20 20/20 20/20 20/20 20/20 2
MT 95 50 20/20 20/20 20/20 20/20 20/20 20/20 2
17MT 87 64 20/20 20/20 20/20 20/20 20/20 20/20 2
DROS 84 68 5/20 20/20 20/20 20/20 20/20 20/20 5
MEST 90 67 20/20 20/20 20/20 20/20 20/20 20/20 2
MESTm 83 66 2/20 5/20 10/20 13/20 20/20 20/20 40
TREN 88 51 4/20 11/20 11/20 17/20 20/20 20/20 40
17TREN 86 55 15/20 17/20 20/20 20/20 20/20 20/20 10
OXN 88 50 10/20 17/20 20/20 20/20 20/20 20/20 10

Screening for Anabolic Steroids in Urine of Forensic Cases 639

Screening for exogenous AAS ratios were determined for all cases and are presented in
Liquid chromatography with full separation of analytes was per- Table III. The limit for an elevated T/E ratio was set to 6.0 as a
formed in 23 min including equilibration of the system. Baseline more conservative approach compared the current limit of 4.0
separation of isobaric compounds such as 19NAN and 19NET set by WADA. A limit at 4.0 would strongly increase the number
was achieved. For selectivity reasons, the most abundant product of false-positive samples, which would need to be further ana-
ions were discarded for BOLDm (m/z: 253) and OXA (m/z: 271) lyzed. Eleven samples (all female) were discarded as the concen-
due to the presence of interferences from matrix components. tration of E was below the limit of quantification, and therefore,
Two compounds were ionized by adduct formation. The pres- the T/E could not be determined. The observed values of T/E
ence of a 1-ene-3-keto function in BOLDm led to a [M þ H þ ranged from 0.1 to 260 (median value: 1.3; mean value: 2.8).
MeOH]þ adduct formation and the unconjugated keto group in The median values of the T/E ratios are in accordance with
MESTm yielded a [M þ NH4]þ adduct in accordance with the other studies of Caucasian populations (8), whereas the mean
results reported by Pozo et al. (14). values are slightly higher due to the presence of very high T/E
Retention times were consistent, and the S/N ratio exceeded ratios in samples from abusers of T. Seventeen samples were
three in all samples for all compounds. For a few samples, the ion found positive for AAS abuse (Table IV). Exogenous AAS and/or
ratio criteria could not be met at low levels for some compounds their metabolites were found in 15 samples, whereas 12 samples
due to severe ME. These levels were therefore excluded from the had an elevated T/E ratio ranging from 14.5 to 260 (median value:
identification range. This confirmed that at least 20 different sam- 30.9; mean value: 59.7). Eight additional samples had a slightly el-
ples should be used for the validation as suggested by Dadgar and evated T/E ratio above 6 (T/E: 6.4 – 8.1; all male samples), but
Burnett (28) and Peters et al. (29). The LLOI was established by they were not registered as positive samples. These samples
fulfillment of all identification criteria for all of the 20 fortified needed additional investigations by GC–IR-MS to confirm wheth-
samples (Table II). The LLOI ranged from 2 to 40 ng/mL with er the elevated ratio originated from abuse of T.
8 of the 18 analytes having a LLOI at 2 ng/mL. Quality control All positive samples were from male subjects between the age of
samples in each batch verified long-term stability at levels 18 and 40 years. Of 13 (46%) antemortem samples, six were directly
LLOI. This proved valid for all analytes including analytes linked to an act of violence, three (23%) were from suspected sex
where adduct precursor ions were used. offenders and the remaining four samples were from other types of
cases. Drugs of abuse were found in 11 of the 17 positive samples
(65%). The compounds included tetrahydrocannabinol, amphet-
amine, morphine, 3,4-methylenedioxy-N-methylamphetamine
Quantitative validation of T and E and comparison
of methods for determination of T/E ratio
A linear calibration model with R 2 . 0.99 was applied for both T
and E with a linear range from 2 to 160 ng/mL for both com-
pounds. MEs were 60 and 64% for T and E, respectively, and
ERs were 88 and 84%. The PE was 52% for T and 54% for E,
which was considered acceptable and within the same range as
for the exogenous AAS. Precision and accuracy were determined
by spiking a human urine sample at six levels over the linear
range and analyzing each level in replicates of six over 2 days.
Accuracy ranged from 93 to 111%, and the overall coefficient
of variation (CV) ranged from 2 to 9%.
The two approaches for determining the T/E ratio listed in
‘Method validation’ were compared by a paired Student’s t-test.
Logarithmic data transformation was applied due to the larger
differences between methods at higher levels of the T/E ratio.
The comparison indicated that the method using SIL-IS yielded
Figure 1. Data points plotted with the identity line for comparison of methods for
a T/E ratio that was 16 – 22% higher than the measurement determining the T/E ratio by the SIL-IS method versus the simple method.
based on only the peak areas (95% confidence limits, n ¼ 45,
P , 0.0001) (Figure 1).
Table III
Summary of T/E Ratio Findings in the 580 Forensic Cases
Findings in the forensic material T/E ratio Postmortem cases (n ¼ 300) Living subjects (n ¼ 280) Total
A total of 580 consecutive urine samples from forensic cases (n ¼ 580)
Male Female Total Male Female Total
were analyzed including 300 postmortem samples (247 male, (n ¼ 247) (n ¼ 53) (n ¼ 166) (n ¼ 114)
53 female) and 280 samples from living subjects (166 male,
Mean 2.1 1.8 2.0 4.8 1.8 3.5 2.8
114 female). The ethnicity of the subjects was primarily Median 1.2 1.5 1.3 1.6 1.2 1.3 1.3
Caucasian. Postmortem cases included autopsies of drug addicts Minimum 0.1 0.1 0.1 0.1 0.1 0.1 0.1
Maximum 90 4.8 90 260 6.6 260 260
with supposed poisoning as cause of death, suspicious deaths 10th 0.2 0.4 0.2 0.2 0.2 0.2 0.2
and fire-related deaths. Samples from antemortem cases included percentile
cases of violence, driving under the influence of drugs and cases 90th 3.6 3.4 3.6 4.7 4.5 4.5 4.0
percentile
of sexual assault (both samples from suspects and victims). T/E

640 Andersen and Linnet

Table IV
Findings and Case Information in 17 Forensic Cases Positive of AAS Abuse

Case Sex (F/M) Age (years) Urinary T/E (ratio) and exogenous Blood ethanol concentration (g/kg) Additional case information
AAS detected (ng/mL) and other compounds (ng/mL)
Postmortem subjects
1 M 34 T/E: 47.8 Amphetamine 551 Drug addict in methadone treatment. Found in bed by a domestic partner.
TREN: 14 Chlorprothixen 64 Cause of death was determined to be poisoning by methadone and
17TREN: 200 Methadone 445 amphetamine
Tramadol 265
Duloxetine 212
Ritalinic acid 12
Ethanol 0.5
2 M 35 T/E: 14.5 Methadone 329 Thickening of the heart muscle and moderate calcified changes in the
ND: 3 Benzoylecgonine 92 coronary arteries and aorta. Enlarged lymph nodes around the liver and
19NAN: 89 anthracosis around the lungs. Cause of death was established to be
19NET: 34 poisoning by methadone
BOLD: 20
BOLDm: 28
3 M 23 T/E: 2.1 Amphetamine 233 Diseased were affiliated with organized crime and was found in car with a
ND: 18 MDMA 11 hose taped from the exhaust to the cabin. Drugs found in blood samples
19NAN: 400 Benzoylecgonine 170 were at nontoxic levels, and the cause of death was carbon monoxide
19NET: 99 Carbon monoxide saturation 71% poisoning
4 M 21 T/E: 90,3 Tetrahydrocannabinol 16 Diseased were attacked with a knife and was brought to the hospital. A
ND: 450 knife wound in the abdomen region had caused severe damage to liver and
19NAN: .3,000 spleen, and death was caused by bleeding
19NET: .1,000
Living subjects
5 M 22 T/E ¼ 15.1 None detected Suspected driving under the influence of drugs (DUID)
ND: 43
19NAN: 800
19NET: 200
6 M 28 T/E: 5.9 Clonazepam 3 Suspected sex offender
BOLD: 4 Ethanol (urine) 0.13
BOLDm: 17 (LLOI 20)
7 M 36 T/E: 1.0 Alprazolam 6 DUID
ND: 2 Tetrahydrocannabinol 0.6
19NAN: 52 Ketobemidone 16
19NET: 9 Clonazepam 28
8 M 23 T/E: 110 None detected Convicted of criminal activities
ND: 12
19NAN: 200
19NET: 45
BOLD: .1,000
BOLDm: 500
TREN: 700
17TREN: .3,000
9 M 38 T/E: 49.3 None detected Convicted of criminal activities
ND: 300
19NAN: .4,000
19NET: 1,600
10 M 31 T/E: 260 None detected Suspected of knife stabbing
STAN: 200
3STAN: .3,000
16STAN: 120
TREN: 600
17TREN: .4,000
DROS: .2,000
11 M 22 T/E: 30.5 Tetrahydrocannabinol 3 Suspected sex offender
ND: 500 Benzoylecgonine 100
19NAN: .3,000 Ethanol (urine) 0.44
19NET: 600
TREN: 100
17TREN: 53
12 M 27 T/E: 19.6 Paracetamol 7,400 Attacked by person involved in organized crime
19NAN: 20 Tramadol 127
19NET: 8
DROS: 500
13 M 23 T/E: 22.6 MDMA 30 Suspected sex offender
Ethanol 0.35
Ethanol (urine) 0.75
14 M 30 T/E: 25.0 Amphetamine 106 Suspected of violence
(continued)

Screening for Anabolic Steroids in Urine of Forensic Cases 641

Table IV Continued

Case Sex (F/M) Age (years) Urinary T/E (ratio) and exogenous Blood ethanol concentration (g/kg) Additional case information
AAS detected (ng/mL) and other compounds (ng/mL)

15 M 40 T/E: 31.2 Morphine 7 Examined at hospital after being stabbed with a knife and beaten with a
STAN:150 Metoclopramide 92 baseball bat
3STAN: .2,000 Tramadol 180
16STAN: 100 Pethidine 61
16 M 34 T/E: 2.9 Amphetamine 70 Examined at hospital after being stabbed with knife
19NAN: 125 Oxazepam 10
19NET: 51 Oxycodone 16
MET: .1,500 Ethanol (urine) 0.22
TREN: 82
17TREN: .1,000
DROS: 61
17 M 18 T/E: 5.9 None detected Suspected of severe violence
DROS: 100

(MDMA), benzodiazepines and ethanol (Table IV). Two cases are arylsulfatase from H. pomatia for enzymatic preparation of
more closely described below as examples. urine samples for steroid analysis (32). Enzymatic treatment
One case involved a 40-year-old male who involved in an with H. pomatia can produce artifacts by converting endoge-
act of violence. The T/E ratio was 31.2, and STAN was detected nous androst-5-ene-3b,17b-diol into T, thus leading to a slightly
in the urine sample together with the two metabolites distorted T/E ratio (33, 34). However, the use of sodium ascor-
[16-hydroxystanozolol (16STAN) and 3 0 -hydroxystanozolol bate has been shown to diminish the issues associated with
(3STAN)]. The semi-quantitative method estimated the concen- H. pomatia (35).
trations to be 150, 100 and .2,000 ng/mL for STAN, 16STAN
and 3STAN, respectively. The male subject had been beaten by
a baseball bat and stabbed with a knife and had therefore been Detection of exogenous compounds
hospitalized and had received analgesics, which were seen in ad- According to recent developments, common anabolic steroids
ditional analyses. can be measured by LC–MS-MS (4, 12, 15). In case of poor ioni-
A second case involved a young male aged 22 years suspected zation for some compounds, adduct formation may make detec-
of being a sex offender. ND and two metabolites 19NAN and tion possible (14). The use of adduct ions was necessary for the
19NET were detected with estimated concentrations of 200, boldenone metabolite (BOLD DH) and the mesterolone metabo-
.3,000 and 600 ng/mL, respectively. A T/E ratio of 30.5 was lite (MESTm) to obtain the desired selectivity and sensitivity.
also found together with TREN and 17TREN at concentrations Even though all anabolic steroids may not be detected by LC –
of 100 and 53 ng/mL, respectively. In additional analyses per- MS-MS with sufficient sensitivity to meet WADA requirements,
formed in blood, tetrahydrocannabinol was found in a concentra- commonly occurring anabolic steroids can be detected in con-
tion of 3 ng/mL and benzoylecgonine at 100 ng/mL. Ethanol was centrations usually observed in forensic cases (Table IV) (20,
present in the urine sample in a concentration of 0.44 g/kg. 36). When analyzing the samples spiked at multiple levels with
exogenous AAS, the identification criteria established by Rivier
(25) were followed and included retention time, S/N ratio and
ion ratio. The main reason that compounds did not fulfill the
Discussion identification criteria at low concentrations was the lack of com-
Determination of T/E ratio pliance of the ion ratios with the established limits. This discrep-
The simple method for determining the T/E ratio using peak ancy was mainly caused by high levels of background noise in
areas was the least comprehensive of the methods applied, and some samples. Concerning specificity, one should be careful
in contrast to the SIL-IS method, it does not account for any sup- when selecting m/z values. Thorough examination of back-
pression of the signal or variation during sample preparation. ground signals in blank urines is necessary in at least 20 samples.
Student’s t-test revealed that the area ratio approach gave lower We observed interference for several m/z values presented in
values of the T/E ratio compared with the SIL-IS method. the literature, making use of alternative values necessary. For
However, comparison of the two procedures gave a relatively OXA, a product ion of m/z: 229 was initially tested but endoge-
narrow confidence interval and the simple approach based on nous interference was observed (37), and product ions of m/z:
peak areas proved to be valid as a screening method, taking 289 and 135 were used instead (38). Thus, the present method
this inaccuracy into account. Samples with a T/E ratio above a may be less sensitive for some compounds, but, on the other
set threshold in a screening method using only the raw peak hand, a high degree of selectivity has been attained with a very
areas should be further investigated using the method including low chance of false positives.
SIL-IS. This outcome suggests that a fairly low threshold should
be set in the screening method to avoid false negatives. As an al-
ternative, more endogenous compounds could be included to Determination of anabolic steroids in forensic
provide a full steroid profile (30, 31). cases by LC –MS-MS
As of January 2014, WADA requires purified b-glucuronidase The fully automated extraction system presented here was able
from Escherichia coli and does not accept b-glucuronidase/ to prepare samples that were ready for injection into 96-well

642 Andersen and Linnet

plates, even including enzyme treatment, incubation, SPE, evapo- incubation, centrifugation and SPE. The qualitative method for
ration and reconstitution. The offline automated system is an the exogenous compounds had a LLOI that ranged from 2 to
ideal platform for sample preparations prior to LC – MS. 40 ng/mL. Selectivity was thoroughly tested and emphasized
However, this methodology requires expensive instrumentation the need for at least 20 samples during validation when using
and highly trained experts (39). LC–MS-MS for analyzing AAS. A total of 580 samples from foren-
Avoiding a derivatization step in the sample preparation is one sic cases were analyzed, and 17 (2.6%) samples were found pos-
advantage of choosing LC–MS over GC– MS (40), but it may also itive of AAS. Only six different exogenous AAS were observed,
be practical to use this technique in a laboratory like ours, where and in 80% of the positive cases the concentration of one or
the predominating instrumentation is LC –MS-MS. Recent devel- more AAS exceeded 100 ng/mL.
opments in UHPLC systems have furthermore increased the sen-
sitivity and decreased run times significantly, thereby supporting
the LC approach (12). Commercially available columns with References
sub-2 mm particles have enabled better and faster chromato- 1. McArdle, W.D., Katch, F.I., Katch, V.L. Exercise Physiology. 6th edi-
graphic separation, with UHPLC systems being able to handle tion. Lippincott Williams & Wilkins: Baltimore, MD, 2007; pp.
the high pressures. This development has been particularly inter- 559–565.
esting when analyzing isobaric compounds and has led to a 5- to 2. Amsterdam, J.V., Opperhuizen, A., Hartgens, F. (2010) Adverse health
10-fold increase in sensitivity (41, 42). Furthermore, TOF MS and effects of anabolic—androgenic steroids. Regulatory Toxicology and
Orbitrap MS techniques have expanded the applications of LC– Pharmacology, 57, 117– 123.
3. Thevis, M., Thomas, A., Schänzer, W. (2011) Current role of LC–MS(/MS)
MS including analysis of alternative biological matrices found in in doping control. Analytical and Bioanalytical Chemistry, 401,
forensic cases (43). This includes hair analysis (44) and direct 405–420.
analysis of phase II metabolites of AAS using high-resolution 4. Parr, M., Schänzer, W. (2010) Detection of the misuse of steroids in
mass spectrometry (45). doping control. Journal of Steroid Biochemistry and Molecular
In the forensic material, 17 of 580 (2.9%) samples were posi- Biology, 121, 528– 537.
tive of AAS abuse. The samples included samples from both 5. Donike, M.L., Bärwald, K.R., Klosterman, K., Schanzer, W.,
Zimmermann, J. Nachweis von exogenem Testosteron. In: Heck, H.,
males and females and both postmortem and antemortem sam-
Hollmann, W., Leisen, H., Rost, R. (eds). Sport: Leistung und
ples. A study of 500 both competitive and noncompetitive body- Gesundheit. Deutsche Artzte-Verlag: Koln, Germany, 1982; pp.
builders conducted by Parkinson et al. (46) suggests that 80% of 293–298.
the AAS users are young males aged between 18 and 35 years. 6. Oftebro, H. (1992) Evaluating an abnormal urinary steroid profile.
When looking at samples from this demographic subgroup, 14 Lancet, 339, 941– 942.
of 227 (5.7%) samples were positive. Thus, screening of this se- 7. Becchi, M., Aquilera, R., Farizon, Y., Flament, M., Casabianca, H., James,
lected group could be preferred. However, the forensic material P. (1994) Gas chromatography/combustion/isotope-ratio mass spec-
trometry analysis of urinary steroids to detect misuse of testosterone
also involves samples from drug addicts using AAS, which tend to in sport. Rapid Communications in Mass Spectrometry, 8, 304–308.
widen the age interval of the target group. 8. Renterghem, P.V., Eenoo, P.V., Geyer, H., Schänzer, W., Delbeke, F.T.
In the forensic material, only six different exogenous AAS were (2010) Reference ranges for urinary concentrations and ratios of en-
found. ND was the most predominant AAS, as it and/or its metab- dogenous steroids, which can be used as markers for steroid misuse,
olites were found in 10 cases. The exogenous AAS were in high in a Caucasian population of athletes. Steroids, 75, 154–163.
concentration, as 12 of 15 (80%) positive samples had a concen- 9. Sottas, P., Saugy, M., Saudan, C. (2010) Endogenous steroid profiling in
the athlete biological passport. Endocrinology Metabolism Clinics
tration above 100 ng/mL for one or more compounds. Six of 13
of North America, 39, 59 –73.
(46%) positive antemortem samples were directly linked to an 10. Mazzarino, M., Abate, M.G., Alocci, R., Rossi, F., Stinchelli, R., Molaioni,
act of violence not including sexual assault. The use of AAS is sug- F. et al. (2011) Urine stability and steroid profile: towards a screening
gested to be related to heightened levels of violent behavior (47), index of urine sample degradation for anti-doping purpose.
but it is uncertain whether AAS act as an immediate trigger for a Analytica Chimica Acta, 683, 221– 226.
violent act (36). The observed T/E ratio for samples positive of T 11. Leinonen, A., Kuuranne, T., Kostiainen, R. (2002) Liquid chromatog-
raphy/mass spectrometry in anabolic steroid analysis—optimization
abuse ranged from 14.5 to 260 (mean: 59.7) also suggesting that
and comparison of three ionization techniques: electrospray ioniza-
the abusers in the forensic material do not try to conceal the tion, atmospheric pressure chemical ionization and atmospheric
abuse as opposed to a professional athlete subjected to doping pressure photoionization. Journal of Mass Spectrometry, 37,
control. These findings are in agreement with similar investiga- 693– 698.
tions (20, 48), confirming that only a limited selection of com- 12. Gosetti, F., Mazzucco, E., Gennaro, M.C., Marengo, E. (2013) Ultra
pounds are being abused and that the concentrations of AAS in high performance liquid chromatography tandem mass spectrome-
try determination and profiling of prohibited steroids in human bio-
forensic cases generally are much higher than in cases concern-
logical matrices. A review. Journal of Chromatography B, 927,
ing doping in sports. 22– 36.
13. Leinonen, A., Kuurannea, T., Kotiaho, T., Kostiainen, R. (2004)
Screening of free 17-alkyl-substituted anabolic steroids in human
urine by liquid chromatography – electrospray ionization tandem
Conclusion mass spectrometry. Steroids, 69, 101–109.
A screening method for detection of AAS in urine using a fully 14. Pozo, O.J., Eenoo, P.V., Deventer, K., Delbeke, F.T. (2007) Ionization of
anabolic steroids by adduct formation in liquid chromatography elec-
automated sample preparation and LC–MS-MS has been present- trospray mass spectrometry. Journal of Mass Spectrometry, 42,
ed and validated. The method analyzed for 18 exogenous AAS and 497– 516.
also determined the ratio of the endogenous T and E. All actions 15. Pozo, O.J., Eenoo, P.V., Deventer, K., Delbeke, F.T. (2007) Development
during sample preparation were fully automated including and validation of a qualitative screening method for the detection

Screening for Anabolic Steroids in Urine of Forensic Cases 643

 of exogenous anabolic steroids in urine by liquid chromatography– 33. Hauser, B., Schulz, D., Boesch, C., Deschner, T. (2008) Measuring uri-
tandem mass spectrometry. Analytical and Bioanalytical Chemistry, nary testosterone levels of the great apes—problems with enzymatic
389, 1209–1224. hydrolysis using Helix pomatia juice. General and Comparative
16. Ho, E.N.M., Leung, D.K.K., Wan, T.S.M., Yu, N.H. (2006) Endocrinology, 158, 77 –86.
Comprehensive screening of anabolic steroids, corticosteroids, and 34. Gomes, R.L., Meredith, W., Snape, C.E., Sephton, M.A. (2009) Analysis
acidic drugs in horse urine by solid-phase extraction and liquid chro- of conjugated steroid androgens: deconjugation, derivatisation and
matography – mass spectrometry. Journal of Chromatography A, associated issues. Journal of Pharmaceutical and Biomedical
1120, 38– 53. Analysis, 49, 1133– 1140.
17. Kolmonen, M., Leinonen, A., Kuuranne, T., Pelander, A., Ojanperä, I. 35. Christakoudi, S., Cowan, D.A., Taylor, N.F. (2008) Sodium ascorbate
(2009) Generic sample preparation and dual polarity liquid chroma- improves yield of urinary steroids during hydrolysis with Helix
tography time-of-flight mass spectrometry for high-throughput pomatia juice. Steroids, 73, 309–319.
screening in doping analysis. Drug Testing and Analysis, 1, 250–266. 36. Lundholm, L., Käll, K., Wallin, S., Thiblin, I. (2010) Use of anabolic an-
18. Kaklamanos, G., Theodoridis, G.A., Dabalis, T., Papadoyannis, I. (2011) drogenic steroids in substance abusers arrested for crime. Drug and
Determination of anabolic steroids in bovine serum by liquid chroma- Alcohol Dependence, 111, 222– 226.
tography–tandem mass spectrometry. Journal of Chromatography 37. Jeon, B.W., Yoo, H.H., Jeong, E.S., Kim, H.J., Jin, C., Kim, D.H. et al.
B, 879, 225–229. (2011) LC– ESI/MS/MS method for rapid screening and confirmation
19. Saito, K., Yagi, K., Ishizaki, A., Kataoka, H. (2010) Determination of an- of 44 exogenous anabolic steroids in human urine. Analytical and
abolic steroids in human urine by automated in-tube solid-phase Bioanalytical Chemistry, 401, 1353–1363.
microextraction coupled with liquid chromatography-mass spec- 38. Pozo, O., Deventer, K., Van Eenoo, P., Delbeke, F.T. (2007) Presence of
trometry. Journal of Pharmaceutical and Biomedical Analysis, endogenous interferences in the urinary detection of selected ana-
52, 727–733. bolic steroids by liquid chromatography/electrospray tandem mass
20. Lood, Y., Eklund, A., Garle, M., Ahlner, J. (2012) Anabolic androgenic spectrometry. Rapid Communications in Mass Spectrometry, 21,
steroids in police cases in Sweden 1999 – 2009. Forensic Science 2785–2796.
International, 219, 199–204. 39. Vogeser, M., Kirchhoff, F. (2011) Progress in automation of LC–MS in
21. Petersson, A., Bengtson, J., Voltaire-Carlsson, A., Thiblin, I. (2010) laboratory medicine. Clinical Biochemistry, 44, 4 –13.
Substance abusers’ motives for using anabolic androgenic steroids. 40. Vogeser, M., Parhofer, K. (2007) Liquid chromatography tandem-
Drug and Alcohol Dependence, 111, 170–172. mass spectrometry (LC– MS/MS)—technique and applications in en-
22. Klötz, F., Petersson, A., Isacson, D., Thiblin, I. (2007) Violent crime docrinology. Experimental and Clinical Endocrinology & Diabetes,
and substance abuse: a medico-legal comparison between deceased 115, 559– 570.
users of anabolic androgenic steroids and abusers of illicit drugs. 41. Ventura, R., Roig, M., Montfort, N., Sáez, P., Bergéz, R., Segura, J. (2008)
Forensic Science International, 173, 57 –63. High-throughput and sensitive screening by ultra-performance liquid
23. Andersen, D., Rasmussen, B., Linnet, K. (2012) Validation of a fully au- chromatography tandem mass spectrometry of diuretics and other
tomated robotic setup for preparation of whole blood samples for doping agents. European Journal of Mass Spectrometry, 14,
LC – MS toxicology analysis. Journal of Analytical Toxicology, 36, 191– 200.
280–287. 42. Badoud, F., Grata, E., Perrenoud, L., Avois, L., Saugy, M., Rudaz, S. et al.
24. Eurachem Working Group. A Laboratory Guide to Method (2009) Fast analysis of doping agents in urine by ultra-high-pressure
Validation and Related Topics. Eurachem: Teddington, Middlesex, liquid chromatography-quadrupole time-of-flight mass spectrometry
UK, 1998. I. Screening analysis. Journal of Chromatography A, 1216,
25. Rivier, L. (2003) Criteria for the identification of compounds by liquid 4423–4433.
chromatography – mass spectrometry and liquid chromatography – 43. Strano-Rossi, S., Fiore, C., Chiarotti, M., Centini, F. (2011) Analytical
multiple mass spectrometry in forensic toxicology and doping techniques in androgen anabolic steroids (AASs) analysis for antidop-
analysis. Analytica Chimica Acta, 492, 69 –82. ing and forensic purposes. Mini-Reviews in Medicinal Chemistry,
26. Scientific Working Group for Forensic Toxicology. (2013) Scientific 11, 451– 458.
Working Group for Forensic Toxicology (SWGTOX) standard prac- 44. Strano-Rossi, S., Castrignanò, E., Anzillotti, L., Odoardi, S., De-Giorgio,
tices for method validation in forensic toxicology. Journal of F., Bermejo, A. et al. (2013) Screening for exogenous androgen ana-
Analytical Toxicology, 37, 452– 474. bolic steroids in human hair by liquid chromatography/orbitrap-high
27. Matuszewski, B.K., Constanzer, M.L., Chavez-Eng, C.M. (2003) resolution mass spectrometry. Analytica Chimica Acta, 793, 61 –71.
Strategies for the assessment of matrix effect in quantitative bioana- 45. Badoud, F., Grata, E., Boccard, J., Guillarme, D., Veuthey, J.L., Rudaz, S.
lytical methods based on HPLC – MS/MS. Analytical Chemistry, 75, et al. (2011) Quantification of glucuronidated and sulfated steroids
3019– 3030. in human urine by ultra-high pressure liquid chromatography
28. Dadgar, D., Burnett, P.E. (1995) Issues in evaluation of bioanalytical quadrupole time-of-flight mass spectrometry. Analytical and
method selectivity and drug stability. Journal of Pharmaceutical Bioanalytical Chemistry, 400, 503– 516.
and Biomedical Analysis, 14, 23–31. 46. Parkinson, A., Evans, N. (2006) Anabolic androgenic steroids: a survey
29. Peters, F.T., Drummer, O.H., Musshoff, F. (2007) Validation of new of 500 users. Medicine and Science in Sports and Exercise, 38,
methods. Forensic Science International, 165, 216–224. 644– 651.
30. Saugy, M., Cardis, C., Robinson, N., Schweizer, C. (2008) Test meth- 47. Beaver, K., Vaughn, M., Delisi, M., Wright, J.P. (2008) Anabolic-
ods: anabolics. Best Practice & Research: Clinical Endocrinology & androgenic steroid use and involvement in violent behavior in a
Metabolism, 14, 111– 133. nationally representative sample of young adult males in the United
31. Mareck, U., Geyer, H., Opfermann, G., Thevis, M., Schanzer, W. (2008) States. American Journal of Public Health, 98, 2185– 2187.
Factors influencing the steroid profile in doping control analysis. 48. Petersson, A., Garle, M., Holmgren, P., Druid, H., Krantz, P., Thiblin, I.
Journal of Mass Spectrometry, 43, 877– 891. (2006) Toxicological findings and manner of death in autopsied users
32. WADA Laboratory Expert Group. (2013) WADA Technical of anabolic androgenic steroids. Drug and Alcohol Dependence, 81,
Document—TD2014EAAS 1.0 3. 241– 249.

644 Andersen and Linnet