s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610

available at www.sciencedirect.com

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

Development of a highly sensitive and selective

UPLC/MS/MS method for the simultaneous determination
of testosterone and 5␣-dihydrotestosterone in human
serum to support testosterone replacement therapy
for hypogonadism

Hermes Licea-Perez ∗ , Sherry Wang, Matthew E. Szapacs, Eric Yang

Worldwide Bioanalysis, Drug Metabolism and Pharmacokinetics, GlaxoSmithkline Pharmaceuticals,
709 Swedeland Road, King of Prussia, PA 19406, USA

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

Article history: A highly sensitive and selective quantitative method to accurately determine testos-
Received 28 November 2007 terone (Te) and 5␣-dihydrotestosterone (DHT) in human serum is crucial to the success
Received in revised form of Te replacement therapy for hypogonadism. To this end we have developed and vali-
14 January 2008 dated a semi-automated and relatively high-throughput method in a 96-well plate format
Accepted 15 January 2008 using ultra-performance liquid chromatography coupled with tandem mass spectrometry
Published on line 2 February 2008 (UPLC/MS/MS) for the simultaneous determination of Te and DHT in human serum. Te and
DHT along with the internal standards [2 H3 ]-Te and [2 H3 ]-DHT were extracted from 300 ␮L of
Keywords: human serum by liquid–liquid extraction using methyl tertiary-butyl ether (MTBE), followed
Testosterone by derivatization with 2,3-pyridinedicarboxylic anhydride and solid-phase extraction for
Dihydrotestosterone sample clean up. A novel chemical derivatization approach using 2,3-pyridinedicarboxylic
Derivatization, anhydride was employed to achieve the MS sensitivity and selectivity required for DHT.
2,3-Pyridinedicarboxilic anhydride Baseline separation of Te and DHT derivatives from endogenous steroid derivatives was
UPLC, Surrogate Matrix achieved using UPLC technology on a C18 stationary-phase column with 1.7 ␮m particle size.
The validity of using double charcoal-stripped female human serum as surrogate matrix for
preparation of calibration standards was demonstrated through standard addition exper-
iments. The method was validated over the concentration ranges of 0.2–40 ng/mL for Te
and 0.01–2 ng/mL for DHT. The validation and study sample analysis results show that the
method is rugged, precise, accurate, and well suited to support pharmacokinetic studies
where approximately 300 samples can be extracted and analyzed in 1 day.
© 2008 Elsevier Inc. All rights reserved.

1. Introduction tration of Te in serum is lower than 3 ng/mL [1]. Men with

hypogonadism experience symptoms such as a decrease in
Testosterone (Te), the major androgen in males, has been libido, lean body mass, muscle strength, bone density, and sex-
widely used in androgen replacement therapy in hypogonadal ual function [1]. Although Te replacement therapy alleviates
elderly men and is often recommended when the concen- these symptoms, there are some concerns about the impact

∗
Corresponding author. Tel.: +1 610 270 5116; fax: +1 610 270 5005.
E-mail address: hermes.2.licea perez@gsk.com (H. Licea-Perez).
0039-128X/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.steroids.2008.01.018
602 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
of Te therapy on prostate health in elderly men [2]. Te is exten-
sively metabolized by 5␣-reductase enzymes (type 1 and type
2) to 5␣-dihydrotestosterone (DHT) which in turn binds to the
androgen receptor [2]. 5␣-Reductase enzyme type 1 is predom-
inantly present in the skin and liver while type 2 is expressed
in the prostate [2,3]. DHT is a more potent androgen than
Te with an anabolic/androgenic ratio of approximately 4/1,
thus the concentrations of DHT in serum must be kept within
the normal range for a patient undergoing Te replacement
therapy, since high concentrations of DHT are considered to
be the major pathogenesis of benign prostatic hyperplasia
(BPH) in adult males [4,5]. A considerable volume of data from
both pre-clinical and clinical studies show that agents such
as dutasteride, an inhibitor of both type 1 and type 2 5␣-
reductase, and finasteride, an inhibitor of type 2 5␣-reductase,
block the production of DHT preventing prostate enlargement
[6–9]. Gleave et al. demonstrated that a ≥0.5 mg/day dose of
dutasteride led to a ≥90% decrease in serum DHT levels as well
as a decrease in total prostate and tumor volumes in men with
localized prostate cancer. Therefore, Te therapy is often com-
bined with dutasteride or finasteride with the DHT/Te ratio
being closely monitored as an indicator of 5␣-reductase activ-
ity.
A sensitive and accurate bioanalytical method for the anal-
ysis of Te and DHT is required for the diagnosis and treatment
of androgen deficiency in men. The development of a method
to simultaneously quantify Te and DHT led to a number of
challenges. First, Te and DHT are both endogenous steroids,
so a surrogate matrix must be used for the preparation of
calibration standards and the method accuracy using the sur-
rogate matrix needs to be proven. Second, the levels of Te
and DHT in serum from hypogonadal patients are expected
to be lower than in normal human serum (≤2.5 ng/mL for
Te and ≤0.29 ng/mL for DHT [9]); therefore, a very sensitive
method with the lower limit of quantification (LOQ) in the
pg/mL range is required. Finally, there are a number of endoge-
nous steroids with similar chemical structure which could
interfere with Te/DHT quantification, so the method needs to
be selective to prevent the overestimation of Te/DHT levels in
human serum. Traditionally, radioimmunoassay (RIA)-based
methods have been used for analysis of Te in biological matri-
ces [10–13]. However, these methods often involve complex
and tedious sample preparation procedures and also require
handling of radioactive materials. Furthermore, RIA methods
are known to suffer from the lack of sensitivity and selectiv-
ity due to cross reactivity from polar metabolites of Te, other
endogenous steroids, and non-specific binding, particularly
at low Te concentrations [12–15]. GC/MS methods employ-
ing derivatization of steroids to mono- or tri-methylsilyl and
heptafluorobutyric ether derivatives have also been developed
[16,17]. Although these GC/MS methods are selective and sen-
sitive, sample preparation is usually very time-consuming and
analysis run times may exceed 20 min per sample, making
them less suitable for high-throughput analysis. In addition,
these methods often suffer from problems associated with the
instability of the derivatives and potential thermal decompo-
sition during analysis [17].
Recently, high-performance liquid chromatography cou-
pled with tandem mass spectrometry (HPLC/MS/MS) has
proven to be a useful tool for analysis of Te in serum
[14,15,18–23]. Due to the presence of an ␣,␤-unsaturated
carbonyl and the subsequent charge delocalization on its pro-
tonated form (conjugate acid) [18,19], it was possible to reach a
LOQ in the low pg/mL range for Te without the need for chem-
ical derivatization. DHT, on the other hand, is a neutral steroid
and ionizes poorly using conventional ionization techniques
such as electrospray ionization (ESI), atmosphere chemical
ionization (APCI), and atmosphere photon ionization (APPI)
[18,19]. In an effort to validate a high-throughput method
with a LOQ of 0.01 ng/mL, DHT must first be derivatized to
enhance MS detection sensitivity. A number of derivatization
reagents such as 2-fluoro-1-methylpyridine (FMP), hydrox-
ylamine, nitrobenzylhydroxylamine, methyl hydroxylamine,
Girard reagent P hydrazone, 2-hydrazino-1-methylpyridine,
nitrobenzoyl chloride (NBC), have been reported in the lit-
erature for derivatization of carbonyl or hydroxyl functional
groups of steroids (neutral or oxosteroids) [17–25]. A recent
paper using hydroxylamine to derivatize Te and DHT reports
a LOQ of 0.029 ng/mL using 100 ␮L of human serum; however,
this method uses a large amount of solvent for liquid–liquid
extraction (LLE) that is incompatible with a 96-well plate for-
mat which is required for a high-throughput method [25].
In this paper, we describe a relatively high-throughput
UPLC/MS/MS method for the simultaneous determination of
Te and DHT from human serum in a 96-well plate format.
A three-step process consisting of LLE, derivatization with
2,3-pyridinedicarboxylic anhydride and solid-phase extrac-
tion was employed before sample analysis. Several additional
experiments, including standard addition and stability with
incurred samples, were included in the method validation to
confirm the accuracy and the integrity of the results.
2. Experimental
2.1. Chemicals and reagents
Te, DHT, androsterone, trans-androsterone, dehydroepia-
ndrosterone, 2,3-pyridine dicarboxylic anhydride, 3,4-
pyridine dicarboxylic anhydride, pyridine, and N,N-dime-
thylformamide were purchased from Sigma–Aldrich (St.
Louis, MO, USA). Internal standards, Te-[2 H3 ] and DHT-[2 H3 ]
were obtained from CDN Isotopes (Quebec, Canada). HPLC
grade acetonitrile was purchased from Burdick and Jackson
(Muskegon, MI, USA). ␤-DHT was obtained from Steraloids
(Newport, RI, USA). Methyl tertiary-butyl ether (MTBE) and
formic acid were purchased from EMD Scientific (Gibbstown,
NJ, USA). Double charcoal-stripped female (DCSF) human
serum was obtained from Bioreclamation Inc. (East Meadow,
NY, USA).
2.2. Equipment
An Eppendorf 5810R centrifuge with a rotor capacity for four
96-well plates (Brinkmann Instrument, Westbury, NY, USA)
and a Mettler UMX2 balance (Hightown, NJ, USA) were used.
A Hamilton Mircrolab STAR liquid handler (Reno, NV, USA)
was used for serum transfer, while a TomTec Quadra 3 SPE
(Hamden, CT, USA) was used for liquid transfer. Arctic White
LLC 96-well round 2-mL plates with silicone and PTFE film
 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
seal mats (Bethlehem, PA, USA) were used to extract ana-
lytes and their internal standards from serum. One milliliter
deactivated (silanized) glass vials along with 96-well plate
covers (Blue CapMat with Pre-Cut T/S Septa) from MicroLiter
Analytical Supplies (Suwanee, GA, USA) were used for sam-
ple introduction to the UPLC. The SPE sample clean up after
derivatization was carried out on the Oasis 30-mg HLB 96-well
plates from Waters (Milford, MA, USA).
2.3. Preparation of standards and quality control (QC)
samples
Stock solutions of Te, DHT, Te-[2 H3 ] and DHT-[2 H3 ] were
individually prepared in dimethylformide at concentrations
of 1.0 mg/mL and stored at room temperature. The stock
solutions of Te and DHT were combined and diluted with
50/50 acetonitrile/water to make the working solution WS1
containing Te at 80 ␮g/mL and DHT at 4 ␮g/mL. WS1 was
diluted to make working solution WS2 at a concentration
of 800/40 ng/mL of Te/DHT in 50/50 acetonitrile/water. The
WS2 was used to make calibration standards at 40/2, 20/1,
10/0.5, 4/0.2, 2/0.1, 1/0.05, 0.4/0.02, and 0.2/0.01 ng/mL of
Te/DHT using a serial dilution procedure. Working solutions
(WQ1–WQ4) for the QC samples were prepared at concen-
trations of 80,000/4000, 4000/200, 400/20, and 40/2 ng/mL of
Te/DHT in 50/50 acetonitrile/water. The WQ1–WQ4 were used
to make QC samples at 200/10, 40/2, 32/1.6, 4/0.2, 0.8/0.04,
and 0.2/0.01 ng/mL of Te/DHT. QC samples were divided into
2.5-mL aliquots and frozen at −80 ◦ C or extracted immedi-
ately. In the first validation run, freshly prepared QC samples
were analyzed against freshly prepared calibration stan-
dards. For each subsequent validation run, frozen replicate
aliquots of the QC samples were thawed at room tem-
perature and analyzed against a freshly prepared standard
curve.
2.4. Sample preparation
MTBE (1 mL) was added to each well of the 2-mL ArcticWhite
96-well polypropylene plate. The plate was sealed with the
ArctiSeal mat and vortex mixed in an inverted position for
3 min. Subsequently, the MTBE was discarded and the plate
was allowed to dry in the chemical hood. This wash step
was used to remove any plastic residue from the plates and
plate seals. Serum samples (300 ␮L) were then transferred to
the washed 96-well plate using a Hamilton STAR liquid han-
dler. A 25-␮L aliquot of internal standard solution (20 ng/mL
for Te-[2 H3 ], and 4 ng/mL for DHT–[2 H3 ]) was added to all
tubes with the exception of the blanks, which received 25 ␮L
of 50/50 acetonitrile/water, followed by recapping and vortex
mixing for 1 min. Following addition of 1 mL of MTBE, the wells
were capped with the ArctiSeal mat; vortex mixed for 3 min,
and centrifuged at 3220 × g for 5 min. The MTBE layer was
transferred to a 2-mL polypropylene 96-well plate with 1-mL
glass-inserts using a TomTec liquid handler and the MTBE was
evaporated under a stream of nitrogen at 45 ◦ C. After evapora-
tion, 100 ␮L of 100 mg/mL 2,3-pyridine dicarboxylic anhydride
in anhydrous pyridine was added to all wells. The plate was
sealed and vortex mixed for 3 min, followed by incubation at
60 ◦ C for 60 min. After incubation, 900 ␮L of water was added
603
to all wells to hydrolyze the excess 2,3-pyridinedicarboxylic
anhydride. The samples were then loaded onto a 30-mg HLB
SPE 96-well plate, the SPE plate was washed with 2× 1000 ␮L of
40% methanol in water (v/v) and eluted with 2× 400 ␮L of 80%
methanol in water into a 2-mL 96-well plate containing 1 mL
deactivated (silanized) glass-inserts. These sample extracts
were evaporated at 45 ◦ C under a stream of nitrogen, recon-
stituted with 100 ␮L of 30/70 acetonitrile/water, capped and
vortex mixed before analysis.
2.5. Chromatographic conditions
An ACQUITYTM UPLC integrated system from Waters (Milford,
MA, USA), consisting of an autosampler combined with a sam-
ple organizer capable of holding ten 96-deep well plates, binary
solvent manager, and an ACQUITY UPLCTM BEH C18 column
(100 mm × 2.1 mm, 1.7 ␮m particle size) was used. The column
temperature was maintained at 50 ◦ C and the sample com-
partment was maintained at 10 ◦ C. Mobile phase A was water
and mobile phase B consisted of 2 mM ammonium acetate
(native pH) in acetonitrile/water 98/2. The LC system was held
at 0% B for 0.1 min followed by a non-linear gradient (convex
with steep initial gradient, curve 9) profile from 0% B to 30% B
for 1.0 min and then from 30% B at 1.0 min to 32% B at 2.0 min.
The system then was held at 32% B until 3.4 min. The column
then was washed with 90% B from 3.4 to 4.0 min and the LC was
equilibrated at 0% B from 4.01 to 4.5 min. The total run time
including the sample load was 5 min and the flow rate was
maintained constant at 0.5 mL/min throughout the run. A typ-
ical injection volume was 5 ␮L using the partial loop injection
mode. To achieve better assay performance we recommend fil-
tering the mobile phase before use and preparing it fresh every
2 days because of a shift in the retention times observed when
using unfiltered or old mobile phases. Also, the authors advise
to wash the analytical column with methanol:water (50:50) for
1 h before starting the analysis.
2.6. Mass spectrometric conditions
A triple quadrupole mass spectrometer API 5000 (Applied
Biosystems/MDS Sciex, Concord, Ontario, Canada) with an
electrospray ionization interface (ESI), operated in the nega-
tive ionization mode was used. The instrument was optimized
for the 2,3-pyridinedicarboxilic anhydride derivatives of Te,
Te-[2 H3 ], DHT, and DHT-[2 H3 ] by infusing a 10 ng/mL solu-
tion of SPE purified derivatives in acetonitrile:water (50:50,
v/v) at 500 ␮L/min through an Agilent pump 1100 series (Palo
Alto, CA, USA) directly connected to the mass spectrome-
ter. The MRM transitions of m/z 437–286, m/z 440–288, m/z
438–287, and m/z 441–288 were chosen for Te, Te-[2 H3 ], DHT,
and DHT-[2 H3 ] derivatives, respectively. The m/z 437–286 and
m/z 440–288 transitions for Te and Te-[2 H3 ], respectively, were
chosen instead of m/z 436–285 and m/z 439–287 to maintain MS
response linearity and avoid MS detection saturation at high
concentrations of Te.
The optimized mass spectrometric conditions were: ion
source temperature, 700 ◦ C; ion spray voltage, −4500 V; curtain
gas, 30 psi (nitrogen); nebulizing gas (GS1), 60 psi (zero air); TIS
gas (GS2), 80 psi (zero air); collision energy, −50 eV for Te/Te-
[2 H3 ] (detuned) and −35 eV for DHT and DHT-[2 H3 ] derivatives.
604 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
The dwell times were 150 and 100 ms for the analytes and the
internal standards, respectively. The longer dwell time for the
analyte channel was used to obtain a consistent peak shape
at the LOQ.
2.7. Data analysis
MS data were acquired and processed (integrated) using the
proprietary software application AnalystTM (Version 1.4.1,
Applied Biosystems/MDS Sciex, Canada). Calibration plots of
analyte/internal standard peak area ratio versus Te and DHT
concentrations were constructed and a weighted 1/x2 linear
regression was applied to the data. Concentrations of Te and
DHT in validation samples were determined from the appro-
priate calibration line, and used to calculate the bias and
precision of the method with an in-house LIMS (Study Man-
agement System, SMS2000, Version 1.6, GlaxoSmithKline).
3. Results and discussion
3.1. Method development
Numerous challenges were faced during method development
including: (1) the low limit of quantification of 0.01 ng/mL for
DHT requiring chemical derivatization to enhance MS sensi-
tivity, (2) chromatographic separation of Te/DHT derivatives
from other endogenous steroid derivatives, (3) only a two mass
unit difference between Te and DHT requiring baseline sepa-
ration between analytes, (4) difficulty finding a native serum
matrix with acceptable levels of Te and DHT for preparation
of QC and calibration standard samples, and (5) a three mass
unit difference between the analytes and their respective
internal standards leading to contribution from the internal
standards to the analyte signals requiring special attention
to the amount of the internal standard that was used. The
approaches used to resolve these challenges will be discussed
in detail below.
3.2. Underivatized Te and DHT
Since the expected ranges for Te were relatively high
(0.2–40 ng/mL) it was not difficult to develop a method for
Te. Moreover, the presence of an ␣,␤-unsaturated carbonyl
group in the structure of Te facilitates its ionization and sta-
bilizes these ions in the mass spectrometer. A method for
analysis of Te alone was validated in our laboratory over the
range of 0.05–10 ng/mL without derivatization using just 50 ␮L
of serum (data not shown). Briefly, Te was extracted from
serum using liquid–liquid extraction with 1 mL of MTBE. Te
was analyzed using isocratic conditions with a 50:50 mix-
ture of acetonitrile/water on an ACQUITY UPLC equipped
with a BEH C18 column (50 mm × 2 mm, 1.7 ␮m particle size)
and detected with an API-5000 in the positive ionization
mode. Unfortunately, this method could not be applied for
analysis of DHT due to insufficient sensitivity. The low-
est quantifiable level for DHT using the above method was
0.2 ng/mL, a 20-fold difference from the desired LOQ of
0.01 ng/mL.
3.3. Derivatization of Te and DHT
A number of derivatization reagents were tested to achieve
the LOQ of 0.01 ng/mL for DHT. Initially we aimed to derivatize
the carbonyl moiety of Te and DHT. A variety of reagents such
as hydroxylamine, nitrobenzylhydroxylamine, and methyl
hydroxylamine were tested but did not provide the sensitiv-
ity needed for the desired LOQ for DHT using a 96-well plate
format. Then derivatization of the hydroxyl moiety of Te and
DHT was tested. Since the hydroxyl group is a poor leaving
group it is difficult for Te and DHT to undergo SN 2 substitu-
tion reactions, thus the most effective way to derivatize the
hydroxyl moiety is to use oxygen as a nucleophile. Alcohols in
general are weak nucleophiles and often difficult to derivatize.
Normally high concentrations of the reagent are required with
long incubation times. A number of derivatization reagents
such as 2-fluoro-1-methylpyridine, nitrobenzoyl chloride,
N-methyl isotonic anhydride, 3,4-pyridine dicarboxylic anhy-
dride, 2,3-pyridine dicarboxylic anhydride, phthalic anhy-
dride, 3-nitrophthalic anhydride, 2,3-pyrazinedicarboxylic
anhydride were evaluated. The biggest challenge using high
concentrations of the reagent is the removal of the excess
reagent after the reaction is complete to avoid MS detec-
tion interference. In most cases the excess reagent can be
hydrolyzed by the addition of water to the reaction mixture.
In some cases, the reagent was ruled out because of the large
amount of precipitation observed after hydrolysis (e.g. NBC),
while other reagents were ruled out due to the lack of sensi-
tivity they provided for DHT detection (e.g. N-methylisatonic
anhydride, 3-nitrophthalic anhydride, phthalic anhydride, 2,3-
pyrazinedicarboxylic anhydride, and FMP).
The reaction of alcohols with cyclic anhydrides has
been reported previously in the literature [26]. The reac-
tion of methanol with 3,4-pyridinedicarboxylic anhydride
results in a mixture of two amphoteric derivatives where
4-methoxycarbonyl-3-carboxylic acid predominates (84%)
[27]. These amphoteric derivatives can be detected in both
positive and negative ionization modes. Derivatization of
Te and DHT with 3,4-pyridinedicarboxylic anhydride was
tested to determine if this derivatization reagent would allow
the quantitation of DHT at the desired LOQ in serum. The
excess reagent was hydrolyzed with water and the precipitate
was separated from the supernatant by centrifugation. Two
positional isomers were observed for Te and DHT derivatives.
This derivatization method provided excellent MS sensi-
tivity allowing DHT detection in DCSF at the desired LOQ
(0.01 ng/mL). However, chromatographic separation between
the DHT derivative and other endogenous steroid derivatives
such as androsterone, trans-androsterone, and ␤-DHT present
in incurred samples was not achieved, leading to an overes-
timation of DHT concentrations in the incurred samples. The
complexity of the DHT chromatogram is partially related to
the fact that two peaks are expected for each interference
compound leading to a higher chance for peak overlap when
running short chromatographic separations. A symmetrical
derivatization reagent like 2,3-pyrazinedicarboxylic anhydride
generates only one peak for each compound and would make
the LC separation simpler, but the MS sensitivity of the DHT
derivative was insufficient to reach a LOQ of 0.01 ng/mL for
DHT.
 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
Finally, derivatization of Te and DHT with 2,3-
pyridinedicarboxylic anhydride was tested. This reagent
generated two positional isomeric products (Te-1,2 and DHT-
1,2; see Fig. 1) with the amphoteric derivatives being detected
in both positive and negative ionization mode. The intensity
of peaks Te-1 and DHT-1 was the greatest with the second
peak (Te-2 and DHT-2) eluting as a hump (see Fig. 2A and B).
The analysis was performed in the negative ion mode due
to a lower base line and less interference. The derivatization
reaction efficiency of the 2,3-pyridinedicarboxylic anhydride
reaction was evaluated by monitoring the disappearance
of underivatized Te after extraction and derivatization.
This was done because pure standards for derivatized Te
and DHT were not available. The presence of underiva-
tized Te observed was <1% of the original concentration
of Te after a 1-h reaction time indicating a high-reaction
efficiency. This reagent allowed the quantitation of DHT
at the desired LOQ of 0.01 ng/mL and intensive LC opti-
mization (see below) using incurred samples allowed the
separation of Te/DHT derivatives from endogenous steroid
derivatives.
3.4. LC optimization
Our objective was to develop an UPLC method with the short-
est run time, highest sensitivity, but without compromising
the selectivity of the assay. A baseline separation between
the DHT derivative and other endogenous steroid derivatives
with the same MS/MS transition as DHT is required. UPLC
parameters such as pH, flow rate, column type, and buffer
concentration were optimized to achieve the best sensitiv-
ity, peak shape and selectivity. The presence of ammonium
acetate in the mobile phase was crucial for the separa-
tion of all components of the mixture as the lack of the
ammonium acetate resulted in much less retention of the
peaks and a poor separation. The higher the concentra-
tion of ammonium acetate, the better the peak resolution
and the longer the peak retention for the analytes. Unfor-
tunately, the higher concentration of ammonium acetate
also led to decreased sensitivity. Changing the mobile phase
from pH 3 to pH 7 increased the MS sensitivity in the
negative ion mode for the DHT derivative, but the selectiv-
ity and peak shape got worse. Two millimolar ammonium
acetate in a mixture of acetonitrile/water 98/2 was found to
offer the optimal balance between sensitivity and selectiv-
ity.
The column selection was also important because of the
complexity of the separation. A 100 mm × 2.1 mm UPLC col-
umn with 1.7 ␮m particle size showed the highest resolving
power which allowed for the use of a high-linear velocity thus,
shortening run time to 5 min without sacrificing peak resolu-
tion. In addition, we tested conventional analytical columns
with larger particle size, however, we were unable to sepa-
rate all peaks in this complex mixture without significantly
extending the run time.
3.5. Surrogate matrix
Non-stripped human serum cannot be used for preparation
of calibration standards as endogenous levels of Te and DHT
605
in the control human serum significantly exceeded the tar-
geted LOQ of the method. The testes in healthy human males
produce about 0.24 ␮mol of Te daily [1] leading to serum con-
centrations between 2.5 and 9.5 ng/mL [9]. For this reason
DCSF human serum was chosen as the surrogate matrix for
preparation of calibration standards and quality control sam-
ples. Multiple lots of DCSF human serum, were screened for
trace amounts of Te and DHT before being used for preparation
of QC and calibration standards. From 50 lots of DCSF human
serum screened only three lots were found to contain Te/DHT
levels low enough (less than 20% of the LOQ) to be acceptable
as a matrix blank.
3.6. Recovery
The recovery of Te from DCSF human serum using LLE was
assessed using the validated method for underivatized Te
described in Section 3.2. The recovery was calculated as the
difference in the peak area ratios (analyte/internal standard)
for Te spiked in DCSF human serum either prior to LLE or
in DCSF human serum extracts post-LLE at 0.2 and 4 ng/mL.
The experiment was carried out in replicates of six and the
recovery was found to be 92.1% and 87.7% at 0.2 and 4 ng/mL,
respectively.
3.7. Selectivity and linearity
The characteristic precursor [M−H]− to product ion transi-
tions, m/z 436–285, m/z 439–287, and m/z 438–287, 441–289 are
consistent with the structures of the derivatives of Te, [2 H3 ]-Te,
DHT, and [2 H3 ]-DHT, respectively. However, less sensitive pre-
cursor to product ion transitions of m/z 437–286 and 440–288
were chosen for the Te and Te-[2 H3 ] derivatives, respectively, to
maintain MS response linearity and to avoid MS detector satu-
ration at high concentrations. Since all human serum samples
contain detectable levels of Te and DHT, it is a challenge
to demonstrate that the method is selective. The selectivity
of this method was established by the analysis of incurred
samples with a variety of LC conditions (isocratic and long gra-
dients). UPLC/MS/MS chromatograms were visually examined
and compared for chromatographic integrity and potential
interferences. The retention times of Te and DHT derivatives
were the same as that of their respective internal standard
derivatives for all LC programs used. Moreover, the area ratios
between the analyte and internal standard were also con-
stant. Potential interference from endogenous steroids such as
androsterone, trans-androsterone, ␤-DHT, androstanediol, and
dehydroepiandrosterone was also investigated. This was done
by spiking DCSF human serum with the above-mentioned
reference standards and running the corresponding deriva-
tives with the validated LC method. The retention times of
the derivatives from all investigated compounds differ from
that of the derivatives of Te and DHT and their internal stan-
dards.
The linearity of the method was evaluated by analyzing
eight calibration standards in duplicate over the nominal con-
centration range of 0.2–40 ng/mL for Te and 0.01–2 ng/mL for
DHT. The correlation coefficients obtained using 1/x2 weighted
linear regressions were better than 0.9983 for both Te and DHT.
606 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610

Fig. 1 – Reaction scheme for the reaction of Te and DHT with 2,3-pyridinedicarboxilic anhydride in the presence of pyridine.
The LC elution order of Te-1/Te-2 and DHT-1/DHT-2 is unknown.

3.8. Bias and precision
At all QC concentrations examined, the bias and precision val-
ues were less than 15% over three analytical runs (see Table 1).
The maximum bias observed was −10.8% for Te and −8.2% for
DHT. The maximum within-run precision value observed was
5.5% for Te and 10.1% for DHT. The maximum between-run
precision values observed were 9.4% for Te and 5.7% for DHT.
As defined by the lower and upper validation sample concen-
trations possessing acceptable accuracy and precision, the val-
idated range of this method based on 300 ␮L of EDTA human
serum is 0.01–2.0 ng/mL for DHT and 0.2–40.0 ng/mL for Te.

Table 1 – Bias, precision and mean validation sample concentrations for Te and DHT in DCSF human serum
Concentration (ng/mL) Te DHT

0.20 0.80 4.00 32.00 40.00 0.01 0.04 0.200 1.60 2.00

Run 1 (n = 6)
Mean 0.21 0.84 4.11 31.04 38.41 0.010 0.042 0.210 1.565 2.002
%CV 4.9 2.8 2.7 2.3 1.7 10.1 5.4 3.8 3.1 3.6
Bias% 7.0 4.8 2.6 −3.0 −4.0 4.2 4.2 5.1 −2.2 0.1

Run 2 (n = 6)
Mean 0.18 0.82 4.04 30.07 38.00 0.092 0.039 0.197 1.575 1.920
%CV 5.4 5.5 4.9 3.4 4.6 3.7 4.0 3.2 3.4 4.7
Bias% −10.8 2.8 1.0 −6.0 −5.0 −8.2 −2.2 −1.3 −1.6 −4.0

Run 3 (n = 6)
Mean 0.19 0.79 3.96 30.41 37.48 0.099 0.041 0.202 1.544 1.922
%CV 3.4 1.9 3.0 2.6 2.5 3.2 4.1 3.8 3.0 1.7
Bias% −6.0 −1.4 −1.1 −5.0 −6.3 −1.0 1.7 1.2 −3.5 −3.9

Overall totals
Mean 0.19 0.82 4.03 30.50 37.96 0.098 0.041 0.203 1.561 1.948
Average bias% −3.3 2.1 0.9 −4.7 −5.1 −1.7 1.2 1.7 −2.4 −2.6

Between-run %CV (n = 18) 9.4 2.7 1.1 1.1 Negligible 5.7 2.6 2.8 Negligible 1.9
 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610 607

Fig. 2 – (A) Chromatograms for Te after derivatization with 2,3-pyridine dicarboxilic anhydride from blank DCSF (A), DCSF
spiked with 0.2 ng/mL of Te (B), 40 ng/mL of Te (C) and incurred sample determined at a concentration of 33.4 ng/mL (D). (B)
Chromatograms for DHT after derivatization with 2,3-pyridine dicarboxilic anhydride from blank DCSF (A), DCSF spiked
with 0.01 ng/mL (B), 2 ng/mL (C) and an incurred sample determined at a concentration of 0.130 ng/mL (D).

3.9. Stability of Te and DHT during freeze–thaw cycles
The stability of Te and DHT in spiked DCSF samples dur-
ing freeze–thaw cycles from −80 ◦ C to room temperature was
assessed. This was done by subjecting QC samples at 0.8/0.04
and 32/1.6 ng/mL Te/DHT to three freeze–thaw cycles from
−80 ◦ C to room temperature and then comparing the mean
concentrations (in replicates of 6) against those of the freshly
prepared spiked QC samples. This experiment gave a percent
difference of −4.4 and −2.0 for Te at 0.8 and 32 ng/mL and −1.7
and −3.8 for DHT at 0.04 and 1.6 ng/mL DHT, respectively. The
maximum CV (%) observed in this experiment was 3.4 for Te
and 5.3 for DHT. In addition, since the stability in DCSF human
serum may not reflect stability in native serum, the stability
of Te/DHT in pooled incurred human serum also was assessed
after three freeze–thaw cycles from −80 ◦ C to room temper-
ature at two concentration levels (in replicates of 6). These
pooled samples were prepared from samples collected from a
study where normal volunteers received Lupron to simulate
hypogonadism (low level of Te). This experiment gave a per-
cent difference of −3.7 and −2.2 for Te at 0.818 and 31.4 ng/mL
and −8.0 and −0.5 for DHT at 0.015 and 0.118 ng/mL, respec-
tively. The maximum CV (%) observed in this experiment was
4.2 for Te and 7.3 for DHT. The difference between samples
that underwent three freeze–thaw cycles compared to either
freshly spiked serum or pooled incurred serum samples had a
percent difference and CV (%) between replicates of less than
15% which indicates that Te and DHT are stable in human
serum after at least three freeze–thaw cycles from −80 ◦ C to
room temperature.
3.10. Matrix dilution
The ability to dilute samples containing Te and DHT at con-
centrations above the high limit of quantitation (HOQ) was
demonstrated by performing a 10-fold dilution of DCSF human
serum validation sample spiked at 200/10 (Te/DHT) ng/mL (in
replicates of 6). Since it was difficult to obtain clean DCSF
human serum, both serum and water were tested as viable
diluents. Concentrations of Te and DHT in these matrix dilu-
tion samples were determined and corrected for dilution
factor. The bias and within-run precision values using DCSF
human serum as the diluent were less than 15% (0.1% and
1.1%, respectively for Te, and 3.4% and 3.8%, respectively for
DHT) indicating that a 10-fold dilution of human serum sam-
ples containing Te and DHT above the HOQ is valid. Similarly,
the bias and within-run precision values using water as the
diluent were also less than 15% (actually −1.0% and 2.0%,
respectively for Te, and −1.5% and 2.8%, respectively for DHT)
indicating that water can be used for dilution of human serum
608 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610

samples containing Te and DHT above the HOQ. In addition,

a pooled incurred sample with a concentration above HOQ for
Te and 0.03 ng/mL for DHT was subjected to a 10-fold dilution
using either DCSF or water as the diluent (in replicates of 6).
Again the within-run precision values using DCSF or water as
the diluent were less than 15% (4.0% and 2.7%) for Te. This
indicates that a 10-fold dilution of the pooled incurred human
serum samples containing Te above the HOQ is valid. Data
for DHT was unavailable in this experiment due to its low
concentration in the pooled sample.

3.11. Stability in processed samples

The stability of Te and DHT in processed samples of human

serum was assessed by re-injecting extracted validation sam-
ples from a previous run after storage at 10 ◦ C for 48 h along
with a freshly prepared standard curve. The accuracy and
precision of these samples were found to be acceptable (less
than 15%) and no changes in the sensitivity for Te and DHT
derivatives were observed on re-injection, indicating that pro-
cessed samples are stable when stored at 10 ◦ C for at least
48 h.

3.12. Accuracy determination using standard addition

method

There are concerns about the accuracy of the results in

incurred samples since the standards and QC may not ade-
quately mimic the samples from dosed subjects. Potential
interference from metabolites or endogenous steroids, pro-
tein binding, and matrix effects, is unknown and needs to
be evaluated. In addition, a surrogate matrix is used for
preparation of calibration standards and quality control sam-
ples in our method. Therefore, it is necessary to confirm
the accuracy of the data for Te and DHT in incurred sam-
ple using another method. To test this issue a standard
addition method was used [28]. The concentrations of Te
(0.800 ng/mL, n = 4, %CV = 1.8) and DHT (0.149 ng/mL, n = 4,
%CV = 2.6) in a pooled incurred sample were determined
using the validated method (refer to as ‘actual concentra-
tion’). Then known quantities of Te/DHT were added to this
pooled sample to create a “calibration line” with Te/DHT con-
centrations of 0.4/0.2, 0.8/0.4, 1.2/0.6, 1.6/0.8, 2.0/1.0, 2.4/1.2,
2.8/1.4, and 3.2/1.6 ng/mL in addition to the actual concentra-
tions. These spiked “standards” were extracted and analyzed
using the validated method. The analyte/internal standard
peak area ratios were plotted against the added concentra-
tions of Te/DHT using GraphPad Prisma 5 software and linear
regression was performed with 1/x2 weighting (see Fig. 3).
The standard addition lines then were extrapolated to the
x-intercept, with the concentration of Te and DHT in the
pooled incurred sample being equal to the absolute value
of the x-intercept. The x-intercept for Te was −0.863 indi-
cating that the extrapolated Te concentration of this pooled
incurred sample was 0.863 ng/mL. Comparing the extrapo-
lated value with the actual concentration (0.800 ng/mL) gives
a percent difference of 7.9%. Similarly, the extrapolated DHT
concentration in the pooled incurred sample was 0.155 ng/mL
corresponding to a 4.0% difference from the actual concentra-
tion of DHT (0.149 ng/mL). The excellent agreement between
Fig. 3 – Determination of the accuracy of the concentration
of Te and DHT in a pooled incurred sample using the plots
of analyte/internal standard area ratios against the
additional amount of Te (A) and DHT (B) added.
the extrapolated and actual concentrations shows that the val-
idated method using DCSF as the surrogate matrix is accurate
for determination Te/DHT concentrations in incurred sam-
ples.
3.13. Application to pharmacokinetic study
Following validation the method was used to analyze samples
from a pharmacokinetic efficacy and safety evaluation study
of Te combined with dutasteride. It was an open-label study
with twice daily oral doses of Te (150–400 mg) co-administered
with 0.25 mg dutasteride compared with 400 mg Te alone and
0.25 mg dutasteride alone to the hypogonadal patients. The
serum concentrations of Te and DHT were determined as
described in the method. Blood samples were drawn pre-dose
and at intervals from 0.5 to 24 h post-dose. The pharmacoki-
netic profiles of two subjects are shown (Fig. 4). Subject 1
was dosed twice with 400 mg Te only while subject 2 was
dosed twice with 400 mg Te combined with 0.25 mg dutas-
teride. The maximum concentrations (Cmax ) of Te were 26.4
and 34.7 ng/mL for subjects 1 and 2, respectively, with a Tmax at
0.5 h for both patients. The maximum concentrations (Cmax ) of
DHT were 4 ng/mL at 0.5 h (Tmax ) and 0.13 ng/mL at 2 h for sub-
 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
Fig. 4 – Pharmacokinetic profiles for Te (A) and DHT (B) from
the hypogonadal patients after daily administration of
400 mg Te Alone or 400 mg Te co-administered with
0.25 mg dutasteride.
jects 1 and 2, respectively. As expected, dutasteride effectively
suppressed DHT formation. To date, approximately 1300 clin-
ical samples have been analyzed using the method described
here.
4. Conclusion
A semi-automated sample preparation method in 96-well
plate format for the determination of Te and DHT concentra-
tions in human serum was developed and validated over the
range of 0.2–40 and 0.01–2 ng/mL, respectively. The method
was proven to be highly sensitive, accurate and selective,
making it suitable for high-throughput quantitative analy-
sis of Te and DHT in clinical studies. The LOQ for analysis
of Te can be modified to 0.01 ng/mL if greater sensitivity is
needed. Also, the method can be adapted for the simultaneous
determination of Te, DHT, androsterone, trans-androsterone,
dehydroepiandrosterone, and ␤-DHT from the same aliquot.
Acknowledgements
The authors would like to acknowledge Matt Cyronak,
Jacob Dunbar, Michael Leonard, and Russ Yeager from the
Department of Drug Metabolism and Pharmacokinetics at
GlaxoSmithKline for their assistance with this project.
609
references
[1] Basaria S, Dobs AS. Hypogonadism and androgen
replacement therapy in elderly men. Am J Med
2001;110:563–72.
[2] Qoubaitary A, Swerdloff RS, Wang C. Advances in male
hormone substitution therapy. Expert Opin Pharmacother
2005;6:1493–6.
[3] Steers WD. 5Alpha-reductase activity in the prostate.
Urology 2001;58:17–24.
[4] Heracek J, Richard H, Martin H, Luboslav S, Jana S, Jitka K, et
al. Tissue and serum levels of principal androgens in benign
prostatic hyperplasia and prostate cancer. Steroids
2007;72:375–80.
[5] Bartsch G, Rittmaster RS, Klocker H. Dihydrotestosterone
and the concept of 5alpha-reductase inhibition in human
benign prostatic hyperplasia. World J Urol 2002;19:413–25.
[6] Page ST, Amory JK, Bowman FD, Anawalt BD, Matsumoto
AM, Bremner WJ, et al. Exogenous testosterone (T) alone or
with finasteride increases physical performance, grip
strength, and lean body mass in older men with low serum
T. J Clin Endocrinol Metab 2005;90:1502–10.
[7] Thompson IM. Chemoprevention of prostate cancer: agents
and study designs. J Urol 2007;178:9–13.
[8] Gleave M, Qian J, Andreou C, Pommerville P, Chin J, Casey R,
et al. The effects of the dual 5alpha-reductase inhibitor
dutasteride on localized prostate cancer—results from a
4-month pre-radical prostatectomy study. Prostate
2006;66:1674–85.
[9] Amory JK, Page ST, Bremner WJ. Oral testosterone in oil:
pharmacokinetic effects of 5alpha reduction by finasteride
or dutasteride and food intake in men. J Androl 2006;27:72–8.
[10] Shrivastav TG, Kanaujia PK. Direct radioimmunoassay for
the measurement of serum testosterone using 3H as label. J
Immunoassay Immmunochem 2007;28:127–36.
[11] Funderburgh LJ, Zipf WB, Sotos JF. Direct measurement of
testosterone in a pediatric center, with use of a
radioimmunoassay kit and unextracted serum. Clin Chem
1983;29:1796–8.
[12] Warner MH, Kane JW, Atkin SL, Kilpatrick ES.
Dehydroepiandrosterone sulphate interferes with the
Abbott Architect direct immunoassay for testosterone. Ann
Clin Biochem 2006;43:196–9.
[13] Middle JG. Dehydroepiandrostenedione sulphate interferes
in many direct immunoassays for testosterone. Ann Clin
Biochem 2007;44:173–7.
[14] Cawood ML, Field HP, Ford CG, Gillingwater S, Kicman A,
Cowan D, et al. Testosterone measurement by
isotope-dilution liquid chromatography–tandem mass
spectrometry: validation of a method for routine clinical
practice. Clin Chem 2005;51:1472–9.
[15] Turpeinen U, Linko S, Itkonen O, Hämäläinen E.
Determination of testosterone in serum by liquid
chromatography–tandem mass spectrometry. Scand J Clin
Lab Invest; in press.
[16] Shackleton CL. Mass spectrometry: application to steroid
and peptide research. Endocr Rev 1985;6:441–86.
[17] Salermo R, Moneti G, Forti G, Magini A, Natali A, Saltutti C,
et al. Simultaneous determination of testosterone,
dihydrotestosterone and 5␣-androstan-3␣,-17␤-diol by
isotopic dilution mass spectrometry in plasma and prostatic
tissue of patients affected by benign prostatic hyperplasia
effects of 3-month treatment with a GnRH analog. J Androl
1988;9:234–40.
[18] Griffiths WJ, Liu S, Alvelius G, Sjovall J. Derivatisation for the
characterisation of neutral oxosteroids by electrospray and
matrix-assisted laser desorption/ionisation tandem mass
610 s t e r o i d s 7 3 ( 2 0 0 8 ) 601–610
spectrometry: the Girard P derivative. J Rapid Commun Mass
Spectrom 2003;17:924–35.
[19] Nishio T, Higashi T, Funaishi A, Tanaka J, Shimada K.
Development and application of electrospray-active
derivatization reagents for hydroxysteroids. J Pharma
Biomed Anal 2007;44:786–95.
[20] Higashi T, Shimada K. Derivatization of neutral steroids to
enhance their detection characteristics in liquid
chromatography–mass spectrometry. Anal Bioanal Chem
2004;378:875–82.
[21] Higashi T, Yamauchi A, Shimada K, Koh E, Mizokami A,
Namiki M. Determination of prostatic androgens in 10 mg of
tissue using liquid chromatography–tandem mass
spectrometry with charged derivatization. Anal Bioanal
Chem 2005;382:1035–43.
[22] Kushnir MM, Rockwood AL, Roberts WL, Pattison EG, Bunker
AM, Fitzgerald RL, et al. Performance characteristics of a
novel tandem mass spectrometry assay for serum
testosterone. Clin Chem 2006;52:120–8.
[23] Kirk JM, Tarbin J, Keely BJ. Analysis of androgenic steroid
Girard P hydrazones using multistage tandem mass
spectrometry. Rapid Commun Mass Spectrom
2006;20:1247–52.
[24] Mao L, Sun C, Zhang H, Li Y, Wu D. Determination of
environmental estrogens in human urine by high
performance liquid chromatography after fluorescent
derivatization with p-nitrobenzoyl chloride. Anal Chim Acta
2004;522:241–6.
[25] Kalhorn TF, Page ST, Howald WN, Mostaghel EA, Nelson PS.
Analysis of testosterone and dihydrotestosterone from
biological fluids as the oxime derivatives using
high-performance liquid chromatography/tandem mass
spectrometry. Rapid Commun Mass Spectrom 2007;21:
3200–6.
[26] Quirke JME, Adams CL, Van Berkel GJ. Chemical
derivatization for electrospray ionization mass
spectrometry. 1. Alkyl halides, alcohols, phenols, thiols, and
amines. Anal Chem 1994;66:1302–
15.
[27] Ashcroft WR, Beal MG, Joule JA. Synthesis of the pyridine
analogues of phthalide. J Chem Soc Perkin Trans
1981;1:3012–
5.
[28] Robison JW. Undergraduate instrumental analysis. New
York: Marcel Dekker;
2004.