Introduction
Donike and co-workers laid the foundation of comprehensive steroid profiling for
sports drug testing purposes already in the early 1980s by introducing a suitable
derivatization technique for steroids [1]. Based on
this approach, a method for the sensitive quantification of urinary testosterone (T)
and epitestosterone (E) was established, and the ratio between both urinary
concentrations (T/E) was introduced as a robust indicative marker for T
administrations [2]. Also the general concept of
urinary steroid profiles dates back to this time albeit, in the beginning, mainly
abundant steroids were considered here to evaluate the health status of individuals
encompassing determinations of androsterone (A), etiocholanolone (ETIO) and
dehydroepiandrosterone (DHEA) [3]
[4].
The applicability of the T/E ratio to detect the misuse of T was corroborated
by several excretion studies employing deuterated T, and the methodology was
straightforwardly implemented in doping control analysis by the Medical Commission
of the International Olympic Committee in 1982 [5]
[6]
[7].
Initially, a reference population-derived threshold was established at T/E
≥ 6 based on a population of 97 sport students (n = 50 males and 47
females) and a series of athletes’ samples (n = 1291), mainly
encompassing male individuals [2]. This
population-derived threshold was later confirmed by several follow-up studies [8]
[9].
Obviously, due to the broad distribution of T/E ratios, the possibility of
naturally elevated T concentrations resulting in T/E ≥ 6 was taken
into consideration right from the beginning. Additionally, physiological or
pathological factors like the diminished excretion of E or a steroid-producing
tumour had to be excluded as confounding factors prior to sanctioning an athlete.
Therefore, already at the earliest stage of steroid profiling, analytical data from
follow-up doping control samples or samples collected prior to the conspicuous
finding were included in the process of result interpretation [7]
[10]
[11]. In order to ascertain naturally elevated
T/Es, endocrinological studies were carried out in the 1990s, mainly
supported by the International Cycling Union [10]. As
these studies were laborious, the number of participants was comparably low, and
this approach was not generally applicable to doping controls.
In 1994, isotope ratio mass spectrometry (IRMS) was introduced as a possible measure
to distinguish between naturally elevated T/Es and the administration of T,
and the analytical strategy was applied to samples producing suspicious urinary
steroid profiles on a routine basis since the late 1990s [12]
[13]
[14].
Ever since, the IRMS methodology has been used as a confirmation procedure for
elevated T/Es and atypical steroid profiles. Relying on this supplementary
and unambiguous confirmation, the threshold for T/E was lowered to 4 by the
World Anti-Doping Agency (WADA) in 2004 [15]. Lowering
the T/E ratio did not significantly affect the number of adverse analytical
findings. For example, in the Cologne laboratory, only 4 additional samples
representing only 0.3% of all samples investigated in the years 2005 to 2007
showed suspicious carbon isotope ratio (CIR) values [16]. By means of CIR it is possible to differentiate if urinary steroids
have been produced inside the body or if they have been administered. Some points
related to CIR will be discussed at the end of this article.
A significant increase in sensitivity of the T/E ratios and the steroid
profiles was achieved by the implementation of the steroidal module in the Athlete
Biological Passport starting in 2014 [17]. A
prerequisite for this implementation was the harmonization of steroid profile
measurements throughout the world. This harmonization of methods was the main goal
of the Technical Document on steroid profiling issued by WADA in their function as
an accreditation body. This document explained in detail which steroids should be
quantified (T, E, A, ETIO, 5α-androstane-3α,17β-diol
(5αADIOL) and 5β-androstane-3α,17β-diol
(5βADIOL), all excreted unconjugated and glucuronidated), which internal
standards should be applied, and how the T/E ratio should be determined
[17]. Additionally, other potentially significant
ratios like A/T or 5αADIOL/E were introduced but not
considered mandatory at that time. Important confounding factors to the steroid
profile such as ethanol intake or microbial contaminations were also highlighted,
and laboratories were enrolled to monitor these in conjunction with steroid profile
analyses. In 2018, a revised Technical Document for steroid profiling came into
effect, further standardizing the analytical methods and the reporting of endogenous
urinary steroid concentrations [18]. As all WADA
accredited laboratories world-wide are obliged to follow these technical
regulations, an alignment of results can easily be obtained.
The strict harmonization of quantification enables the retrospective longitudinal
data evaluation of an athleteʼs samples over years by the Bayesian statistical
approach introduced by Sottas et al. [19]
[20]. The early longitudinal or endocrinological studies
computed a mean value for the T/E based on at least 3 different samples and
an individual threshold is calculated thereof by adding the 3-fold standard
deviation (SD) and taking into consideration the coefficient of variation (CV) [10]. This straightforward statistical approach was
found to be sufficient in many cases and, especially with a larger number of the
tested person’s samples, allowed for a clear discrimination between
naturally elevated T/Es and doping scenarios. The more sophisticated
Bayesian approach starts with the population-based threshold and adopts this
individually with each new sample added to the passport. Already after the first
collected sample the results are used to adopt the thresholds for the second one
resulting in substantially more narrow decision limits [19]. After 2 to 3 samples, the individual threshold stabilizes and yields
a highly sensitive indicator for fluctuations of the T/E [20]. This statistical approach is not only applicable
to the T/E ratio but can be applied to each meaningful ratio.
Besides, this straightforward benefit allows the comprehensive and reliable
collection of data within the steroidal module for the possible detection of doping
scenarios even beyond the intended use of this application like, for example, sample
mix up or sample substitution. Within this review the reader will be provided with
several examples for both the increased sensitivity towards the detection of steroid
applications and the enhanced observation possibilities given to the Athlete
Passport Management Unit (APMU)[21] by the ABP
regarding other doping scenarios or confounding factors. Issues that may wait for
further improvements will also be discussed.
Increased sensitivity by applying the longitudinal T/E ratio
While the first systematic population-based investigations on T/E ratios
suggested a rather homogenous distribution [2], in
subsequent determinations encompassing larger populations a clear bimodal
distribution with maxima at 0.16 and 1.0 was established [8]. A first interpretation of this finding was the
differentiation between “Oriental” or “Asian”
steroid profiles with a low T/E and other steroid profiles with
T/Es around 1 [9]. More than a decade of
research was required to identify the main factor responsible for this
significant difference, and Jakobsson et al. eventually demonstrated the
considerable influence of UGT2B17 polymorphism on the glucuronidation of T and,
thus, the urinary T/E ratio [22].
Especially the del/del genotype results in very low urinary T/E
ratios, and this polymorphism is found with a higher probability in the Asian
population resulting in the “Asian” steroid profiles reported 10
years before [22]
[23]
[24]
Even without that knowledge, Ayotte et al. suggested a different T/E
threshold for the different individuals depending on their basal T/E
ratio already in the 1990s [8]. While this was
impossible to employ for a single spot urine sample at that time, it was one of
the triggers for and direct benefits of the steroidal module. As soon as the
first sample is collected, the thresholds are individualized accordingly as
demonstrated exemplarily in [Fig. 1]. This
reduces the number of unnecessary IRMS confirmations for athleteʼs with
naturally elevated T/E ratios and increases the sensitivity of the
steroid profile dramatically for those excreting low amounts of T. The number of
adverse analytical findings (AAFs) based on IRMS increased after the
implementation of the steroidal module as demonstrated in [Fig. 2]. Samples with T/E ratios below 1
were found suspicious by applying the individual thresholds and the exogenous
origin of steroids was confirmed by IRMS. While the number of samples analysed
by IRMS remained rather stable over the years, the number of AAFs increased, and
especially since 2015 samples were confirmed by IRMS that would not have been
analysed without the longitudinal approach. This ‘local’ trend
found in the Cologne laboratory is also corroborated by IRMS findings throughout
the world as summarized in WADA´s Testing Figure Report, which
illustrates a constant number of IRMS tests with a slight increase in IRMS AAFs
since 2013 [25].
Fig. 1 Adaption of the T/E threshold (red line) according
to the individual values (blue line). Samples shown as circles represent
out-of-competition, samples shown as stars in-competition testing.
Example taken from the steroidal module of ADAMS (Anti-Doping
Administration & Management System).
Fig. 2 Increasing number of IRMS-based adverse analytical findings
(AAF) per year in the Cologne laboratory. The striped pattern columns
stand for samples forwarded to IRMS triggered by population derived
thresholds, black columns represent those samples showing atypical
passport findings (mostly T/E samples). The absolute number of
samples investigated by each year is given at the top.
Increased sensitivity by applying other ratios like
5αADIOL/E
Despite the overall increased sensitivity of the steroidal module towards
individuals with naturally low T/E ratios, especially those with the
del/del polymorphism of the UGT2B17 genotype may not show any
significant response in urinary steroid concentrations upon the administration
of T. This was reported for Japanese subjects in 2012 and corroborated later on
applying the ABP from 2014 onwards [24]
[26]
[27] While the
IRMS test, where applicable, could prove the exogenous origin of T and
T-metabolites, the urinary concentrations of T were only slightly affected in
agreement with the modest effect on the T/E ratio. A similar low
response of the T/E may be recognized if only low doses of T are applied
[28]. In such cases, the ratio of
5αADIOL and E was found to offer superior performance and resulted in
atypical classifications of the steroid profiles even after the administration
of DHEA [29]
[30]. An
example for the successful implementation of 5αADIOL/E into the
steroidal module is presented in [Fig. 3]. The
athlete produced a very low T/E of ca. 0.1, and the corresponding
individual threshold was calculated with 0.28; however, this variable was not
applicable to a variety of the athlete’s doping control samples as T and
E concentrations were below the test method’s limit of detection (LOD)
and, consequently, yielded invalid T/E determinations. Therefore, no
threshold for the T/E is computed by the software ([Fig. 3], left). Nevertheless, the limits for
5αADIOL/E were calculated, and their application to the sample
number 38 identified an atypical value, corresponding to a likewise slightly
elevated T/E. All other parameters were found to be non-suspicious.
Forwarding this sample to IRMS resulted in an AAF for both 5αADIOL and
5βADIOL while T fell below the LOD of the IRMS method. Similar results
obtained on 10 different athletes have been described in the literature recently
[31].
Fig. 3 Longitudinal profile obtained on one athlete with very low
T/E ratios (left, blue line) and the corresponding ratio of
5αADIOL/E (right, blue line) and calculated individual
thresholds (red line). Further information in the text. Samples shown as
circles represent out-of-competition, samples shown as stars
in-competition testing. Example taken from the steroidal module of
ADAMS.
While these results underline the added value of including
5αADIOL/E into the steroid profile of athletes’ doping
controls, it should be mentioned that this ratio may exhibit a larger
variability compared to other urinary steroid ratios, especially concerning
urine samples from females. This effect has also been recognized for the
T/E and can be attributed to the menstrual cycle, hormonal
contraceptives, and general analytical challenges caused by the lower
concentrations of urinary steroids in urine samples collected from females [32]
[33]
[34]
[35]
[36]. Carefully considering these confounding
factors where applicable will help to avoid unnecessary confirmation procedures
by simultaneously maintaining the high probative force of the steroidal
module.
The detection of another possible endogenous steroid application,
i. e. dihydrotestosterone (DHT), was also found to be
improved by the application of the 5αADIOL/E ratio. Already in
1992, this ratio was introduced to detect the administration of DHT and verified
shortly after [37]
[38]. In addition, the absolute concentration of urinary DHT and the
DHT/E ratio were also investigated. Both ratios appeared to perform
equally well, which was recently corroborated by an excretion study performed
with a single oral dose of DHT as shown in [Fig.
4]
[39]. Directly after application, both ratios were
found significantly elevated and returned back to initial values after 35 h. The
absolute concentration of DHT and other possible markers such as
5αADIOL/5βADIOL or A/ETIO were observed beyond
individual thresholds only for 21 h. As DHT is not included as a mandatory
component of the steroid profile, administrations will preferentially be
detected by the ratio of 5αADIOL/E. Steroid profile
confirmations triggered by this ratio should therefore encompass the
concentration determination of DHT, where applicable, and samples forwarded to
IRMS should focus on 5αADIOL, A and especially epiandrosterone found in
the fraction of sulfoconjugated steroids as this metabolite will be found
influenced for a prolonged time period [39].
Fig. 4 Urinary concentration ratios of 5αADIOL/E
(black triangles) and DHT/E (grey circles) after a single oral
application of 50 mg of DHT. The dashed lines represent the respective
individual threshold for each ratio calculated from 6 pre- and
post-administration samples.
Confounding factors and the ABP
Numerous confounding factors to the steroid profile have been described and are
well summarized in the relevant literature [40]
[41]. Within this review, we will
focus on those factors detectable by the steroidal module and which can directly
be linked to a certain confounder. Therefore, ethanol consumption, hormonal
contraceptives, pathological states, and mental stress will be carefully
considered, complemented by a short discussion on problems attributed to
(microbial) sample degradation and the administration of 5α-reductase
inhibitors and other potential confounders to the steroid profile.
Ethanol consumption
The impact of ethanol ingestion on the endogenous steroid synthesis and
metabolism was already described in the 1960s [42], and further investigations on the mode of action of the
ingestion of ethanol on the urinary T/E ratio followed approximately
20 years later [43]
[44]
[45]
[46]
[47]. Especially in samples from
females, a strong increase of the T/E ratio was found and, most
probably, triggered numerous IRMS confirmations as a result of a steroid
profile-confounding factor [48]. A decrease in
urinary concentrations of A and ETIO was also described and the T/A
ratio was suggested as the most promising marker to detect the influence of
ethanol administration [46]. In order to
enhance the probative force of the steroidal module, a reporting level for
Ethyl Glucuronide (EtG) was implemented at 5 µg/mL and if a
sample shows elevated levels of EtG, the related steroid profile may be
invalidated for the ABP in order to maintain a high probative force of the
passport. EtG is a urinary metabolite of ethanol and strongly correlated
with ethanol administration [46]. The effect
of ethanol is exemplarily depicted in [Fig.
5]. The female athlete shows a very stable T/E profile
over the time period of 5 years with the exception of two values. In both
cases significantly elevated T/E ratios were reported in the initial
testing procedure and confirmed afterwards together with the strongly
elevated concentrations of EtG. In both cases, an IRMS confirmation was
automatically triggered and conducted with negative results. In order to
maintain the high probative force of the threshold calculated by the
Bayesian model, both T/E values were removed from the longitudinal
profile.
Fig. 5 Longitudinal profile obtained on a female athlete over
a time period of 5 years. The red lines represent the threshold for
T/E as calculated by the Bayesian approach, the blue stars
the T/E measurements. The 2 samples found with elevated
T/E ratios have been added to the plot but were not
considered for threshold determinations. Samples shown as circles
represent out-of-competition, samples shown as stars in-competition
testing. Example taken from the steroidal module of ADAMS.
The procedure how to handle elevated T/E ratios found in coincidence
with elevated concentrations of EtG has been adopted since then. Today, a
general recommendation is to perform IRMS for the first incidence in order
to exclude the possible co-administration of T and ethanol. For subsequent
samples showing elevated levels, the IRMS remains optional but not
mandatory.
Hormonal contraceptives
Due to the menstrual cycle, women’s longitudinal steroid profile data
tend to show a larger variation compared to male steroid profiles [33]
[36]. This
scatter is compensated for by the Bayesian approach and results in broader
confidence intervals for females and does not challenge the principle of the
steroidal module. The use of hormonal contraceptives (HC) in contrast does,
especially if initiated or ceased during the monitored time period [34]
[35]. During
the application of HC, the urinary concentration of E is found diminished,
which results in increased ratios of T/E and
5αADIOL/E.
This is exemplarily depicted in [Fig. 6].
After the first two samples, collected within 2 months, defined the
boundaries of the steroidal module, the third sample collected 4 months
later fell beyond the threshold and triggered steroid profile confirmation
and IRMS analysis. The IRMS could unambiguously demonstrate the endogenous
origin of T and T-metabolites. Follow-up samples collected (number 4 and 5)
confirmed the increased T/E and 5αADIOL/E ratios.
The German National Anti-Doping Agency contacted the athlete who confirmed
the administration of HC starting after sample number 2. The Bayesian
approach adopted to the new elevated values with increased thresholds over a
short time period. Samples collected over the next 2 years (samples
6–18) did not show any atypical values anymore.
Fig. 6 Longitudinal profile of T/E and
5αADIOL/E for a female athlete starting to use
hormonal contraceptives after sample 2. Samples shown as circles
represent out-of-competition, samples shown as stars in-competition
testing. Example taken from the steroidal module of ADAMS.
Another example for the use of HC is shown in [Fig.
7]. Between samples number 8 and 9, the steroid profile starts to
change significantly returning back to starting values at sample number 14.
The time period in-between encompasses 14 months. While the T/E
ratio is only slightly affected, especially the 5αADIOL/E
shows a strong increase. Obviously, not only the concentration of E is
diminished in this individual, but the urinary concentration of T is also
slightly affected. Absolute mean concentrations decrease from 5 to 2
ng/mL for T and from 12 to 1.5 ng /mL for E. The
fluctuations in T concentrations are reflected by the A/T ratio. The
use of HC during the described time period was confirmed by the Testing
Authority (TA) and clearly demonstrates the impact of HC on the longitudinal
steroid profile.
Fig. 7 Longitudinal profile of T/E,
5αADIOL/E, and A/T for a female athlete
presumptively starting to use hormonal contraceptives after sample
8. Samples shown as circles represent out-of-competition, samples
shown as stars in-competition testing. Example taken from the
steroidal module of ADAMS.
In general, the use of HC should be considered as a confounding factor for
female athletes as soon as atypical fluctuation become visible in a female
athletes passport as even single dose applications of emergency
contraceptives may have an impact on the steroid profile [49].
Pathological states
Several medical conditions are known to influence steroid production and
metabolism in the human body [50]. Whenever a
disease occurs that impacts steroidogenesis, the longitudinal profile may be
affected, too. Either directly by a medical condition changing the
endogenous steroid production or indirectly due to the prescribed
medication. One example for a direct impact is the presence of a
prolactinoma, i. e. a benign tumour at the pituitary gland
that substantially increases the amount of circulating prolactin which,
downstream, strongly stimulates the steroid production in females [51]
[52]
[53] In males, the steroid producing pattern
induced by hyperprolactinaemia is different and was reported to result in a
decreased T and DHT production, accompanied by diminished
5α-reductase activity [54]
[55]
[56]. Elevated
prolactin blood concentrations can also be induced by medications such as,
for example, antidepressants and can therefore occur intermittently [57].
An example for a possible benign tumour at the pituitary gland affecting
prolactin is given in [Fig. 8]. A male
athlete presents a naturally elevated T/E ratio, and the first
doping control urine sample was confirmed to contain endogenous steroids
only by IRMS. This result was once more corroborated by IRMS determinations
obtained from sample 16. The elevated T/E ratio is due to the
urinary excretion of low amounts of E (mean value of 4.3±2.1
ng/mL), and the E concentrations close to the limit of detection are
reflected by the above-average scatter of the T/E. Between sample 7
and sample 8 a significant decrease of the T/E ratio is visible,
driven by a reduction in urinary T concentrations from more than 20
ng/mL to 5 ng/mL. This decrease is accompanied by a
simultaneous trend in both the 5αADIOL/5βADIOL and
A/ETIO ratio. There is no known doping scenario that would explain
this pattern but it would fit the above-mentioned effects caused by
hyperprolactinaemia. Unfortunately, it was not possible to contact the
athlete regarding any possible medical condition during the time period of
17 months covered by samples 7–14. In any case, the privacy and data
protection of the individual athlete prevail and in cases where medical
conditions are assumed, the APMU can only inform the relevant TA and may
request to forward the suspicion in order to encourage the athlete to seek
necessary medical assistance. But the TA confirmed a medical condition for
this athlete who had to withdraw from competitive sport for a certain time
period. The general possibility that any pathological state may have an
impact of the steroidal module should always be taken into consideration
when evaluating atypical passport findings.
Fig. 8 Longitudinal profile of T/E, A/ETIO,
and 5αADIOL/5βADIOL for a male athlete
presumptively suffering from hyperprolactinaemia between samples 7
and 14 covering a time period of 17 months. Samples shown as circles
represent out-of-competition, samples shown as stars in-competition
testing. Example taken from the steroidal module of ADAMS.
Fluctuations in the ABP caused indirectly by prescribed medications can often
be followed-up by existing therapeutic use exemptions (TUE). In cases an
athlete shows any medical condition that necessitates the administration of
a drug listed as a banned substance according to the WADA Prohibited List,
the athlete can apply for a TUE. One possible medication with a strong
impact on the steroid profile is T, which may be prescribed in cases of
hypogonadism. If the medication starts or stops during the longitudinal
monitoring of the athlete, the steroidal module will flag atypical test
results and trigger a confirmation and IRMS analysis. Such processes can be
avoided if the TUE of the athlete is recognized by the APMU. If the T
administration is continuous, the athlete will be recognized with an
elevated T/E ratio and a pronounced scattering of the urinary
steroid concentrations and calculated ratios, but not necessarily with an
atypical passport finding.
Mental and physical stress
Participating in competition is, for the majority of athletes, associated
with mental stress. The response of the body to stress is multifaceted and
encompasses the upregulation of glucocorticoid hormones [58]
[59]
[60] The stimulus of the adrenal glands alone
may have an impact on the urinary concentration of excreted steroids.
Additionally, more than 20 years ago, the effect of competition stress on
plasma T-levels was investigated and significant increases after competition
were recognized [61]. The increase was
correlated to an index of self-evaluation of stress but not reflected by
urinary T/E ratios. The impact on plasma T-levels even depend on the
type of in-competition (IC) event, i. e. differences between
home and away games were measured [62]
[63].
This impact should result in a general significant difference between samples
collected IC and out-of-competition (OOC) and may be reflected by the
steroidal module depending on the individual response to
competition-provoked stress. The first question was investigated by a
statistical evaluation of n = 10031 samples collected IC and n
= 12447 samples collected OOC in the year 2016 and analysed by the
routine doping control method as applied in the Cologne laboratory [39]. The distribution of all T/E ratios
was found to depend on the collection site of the sample (IC vs OOC) as
depicted in [Fig. 9]. The overall mean value
was found to be highly significantly elevated in IC samples (median of 0.98
OOC vs 1.11 IC, p < 0.001, Kruskal-Wallis test, Calculation
performed in R) [64]. As directly visible in
the density-plot ([Fig. 9]) this was not due
to different amounts of samples showing naturally low T/E ratios or
due to a different percentage of male and female samples IC (31 % of
female samples) and OOC (33 %). Of course, these population-derived
differences can only be attributed to stress with some uncertainty as other
potential confounding factors are unknown.
Fig. 9 Density plot of T/E ratios obtained in routine
doping control samples collected out-of-competition (blue) and
in-competition (red). The differences in mean values were found to
be highly significant. Further information in the text.
An additional investigation of steroid ratios encompassing a 5α- or a
5β-configuration corroborates the hypothesis of competition
stress-induced changes in the steroid profile. Samples collected IC tend to
show elevated ratios of A/ETIO and
5αADIOL/5βADIOL as shown in [Fig. 10]. Both difference were found to be
highly significant by applying a generalized linear model in R (p <
0.001) [64]. Again, this is an observation
based only on population data and will of course be inter-individually
highly variable. For example, the increase in both ratios can either be
triggered by an increase in the urinary concentrations of A or
5αADIOL or a decrease in ETIO and 5βADIOL, both resulting in
comparable elevations.
Fig. 10 Box plot of the ratios A/ETIO (left) and
5αADIOL/5βADIOL (right) obtained in routine
doping control samples collected out-of-competition (OOC) and
in-competition (IC). The differences in mean values were found to be
highly significant. Further information in the text.
Nevertheless, in selected individual ABPs, the influence of competition is
clearly visible. This is exemplarily depicted in [Fig. 11] showing the longitudinal profile of a male athlete with
samples collected over a time period of 21 months. Samples number 2, 4, and
6 were collected IC while the others derive from OOC testing, exhibiting
elevated 5α/5β ratios. Sample number 9 was collected
OOC and demonstrates that either stress can of course also occur during
training (maybe due to high intensities) or that other confounding factors
may have influenced these ratios. Further studies focussing on different
confounding factors may be helpful to further elucidate the impact of stress
on the steroid profile.
Fig. 11 Longitudinal profile of A/ETIO and
5αADIOL/5βADIOL for a male athlete. Samples
number 2, 4, and 6 were collected in-competition. Samples shown as
circles represent out-of-competition, samples shown as stars
in-competition testing. Example taken from the steroidal module of
ADAMS.
Sample degradation
Due to inappropriate storage conditions during sample transportation in
combination with the non-sterile sampling of urine doping control specimen,
degradation of endogenous steroids and especially steroid conjugates may
occur. For the de-conjugation of glucuronides and sulphates, even ambient
temperatures during (prolonged) sample transportation and custom clearance
may be sufficient. This effect has already been reported in 1977 and was
investigated in the field of doping controls several times since then due to
the possible impact on the validity of test results [65]
[66]
[67]
[68]
[69]
[70]
[71]. In the context of the steroidal module, microbial sample
degradation was implemented as a confounding factor, and samples exhibiting
elevated levels of 5α- and/or 5β-androstanedione are
considered as invalid. The employed decisive criteria are the ratios of
androstanediones (5α-androstanedione or 5β-androstanedione)
divided by A or ETIO, respectively, which should not exceed a value of 0.1
[18]. During the steroid profile
confirmation analysis, the ratio of unconjugated T and T-glucuronide present
in urine without enzymatic hydrolysis may also be considered and invalidate
the steroid profile if found elevated. This may result in inadequate
analytical results as recently demonstrated in a case study [31]. All urine samples showing only minor signs
of microbial degradation can be considered valid regarding their IRMS
confirmation and should therefore be forwarded to this technique in any
reasonable case [72]
[73].
Other confounding factors
In principle, all compounds administered to the human body such as food,
beverage or medication can act as a confounding factor if they encompass the
ability to act on steroid genesis or metabolism [74]
[75]
[76]
[77]. A class of compounds with
a straightforward and well investigated action on steroid metabolism are
5α-reductase inhibitors like for example finasteride [78]. This commonly prescribed medication to
treat male pattern baldness or prostate enlargement shows a significant
impact on the steroid profile as exemplarily depicted in [Table 1]. After a prolonged time period
encompassing very stable steroid profile data, a sudden change was detected
with sample 13. Further investigations showed the presence of finasteride in
this specimen and all subsequently collected samples. The strong decrease in
urinary concentrations of 5αADIOL and A is accompanied by a slight
increase in 5βADIOL and ETIO resulting in the significant decrease
of the 5αADIOL/5βADIOL and A/ETIO ratios
and, thus, in an atypical steroid profile. In the presence of finasteride,
the explanation for such atypical findings is straightforward as this
inhibitor is implemented in routine screening methods. But other compounds
with 5α-reductase inhibitory properties, which may either not be
screened for or which may even not have been recognized as
5α-reductase inhibitors, will also have a significant impact on the
steroidal module. This possibility should always be considered during the
evaluation of longitudinal steroid profile data showing diminished
5α-steroid concentrations.
Table 1 Results obtained on a male athlete over the
time period of 3 years. From sample 13 on finasteride was
present in urine samples. Example taken from the steroidal
module of ADAMS (Anti-Doping Administration & Management
System).
|
Sample number
|
Days
|
A [ng/mL]
|
ETIO [ng/mL]
|
A/ETIO
|
5aADIOL [ng/mL]
|
5bADIOL [ng/mL]
|
5aADIOL/ 5bADIOL
|
|
1
|
0
|
7400
|
3100
|
2.4
|
79
|
160
|
0.49
|
|
2
|
24
|
8400
|
3600
|
2.3
|
73
|
140
|
0.52
|
|
3
|
195
|
2600
|
1400
|
1.9
|
49
|
100
|
0.49
|
|
4
|
228
|
6500
|
3500
|
1.9
|
61
|
110
|
0.55
|
|
5
|
236
|
4200
|
1600
|
2.6
|
51
|
130
|
0.39
|
|
6
|
303
|
4200
|
3000
|
1.4
|
69
|
160
|
0.43
|
|
7
|
314
|
4300
|
2300
|
1.9
|
58
|
92
|
0.63
|
|
8
|
529
|
5400
|
3200
|
1.7
|
140
|
230
|
0.61
|
|
9
|
648
|
4000
|
2000
|
2.0
|
51
|
120
|
0.43
|
|
10
|
761
|
4500
|
2600
|
1.7
|
95
|
360
|
0.26
|
|
11
|
783
|
4100
|
2400
|
1.7
|
59
|
150
|
0.39
|
|
12
|
881
|
4000
|
2000
|
2.0
|
120
|
100
|
1.2
|
|
13
|
979
|
1500
|
6200
|
0.24
|
14
|
240
|
0.06
|
|
14
|
1004
|
1100
|
4400
|
0.25
|
13
|
160
|
0.08
|
|
15
|
1124
|
2400
|
5100
|
0.47
|
30
|
170
|
0.18
|
|
16
|
1133
|
960
|
3900
|
0.25
|
14
|
160
|
0.09
|
|
17
|
1163
|
1200
|
5300
|
0.23
|
27
|
330
|
0.08
|
All of the above mentioned confounding factors should be considered for
profile evaluation by the respective APMU, especially as many of these
factors are not covered by analytical evidence as for ethanol consumption or
some 5α-reductase inhibitors. This will improve the over-all
sensitivity of the longitudinal approach and may help to minimize the number
of samples to be confirmed.
Possible sample mix up and attempts to defraud disclosed by the ABP
As the increased sensitivity of the steroidal module of the ABP increases the
probability to detect cheating athletes, some individuals may try to circumvent
atypical steroid profiles by sample substitution. Depending on how this sample
substitution was carried out, it might be detected by the longitudinal profile.
Several different scenarios will be discussed in detail.
Sample mix up by the doping control laboratory or ADAMS
Taking into account that more than 200,000 samples per year are analysed
worldwide, it is not surprising that under rare circumstances it may happen
that either a sample is mixed up in a laboratory during analysis or that a
result is not directly linked to the appropriate individual´s
passport. This is usually easily detected as the obtained profile does not
match with all previous profiles summarized in the passport. If the mistake
was by the laboratory, the mandatory confirmation of the steroid profile
directly corrects the incongruity. If a mismatch was triggered by ADAMS, an
individually performed alignment of the data solves the issue. These safety
measures proved to be adequate to avoid any further consequences for the
athlete.
Urine substitution by the athlete
Any tampering or attempt to tamper a doping control sample is forbidden
according to WADA´s Prohibited List, and sample substitution is
explicitly listed as a prohibited method [79].
If any kind of manipulation has been detected and can be circumstantiated,
this is sanctioned like any other anti-doping rule violation. In order to
detect such manipulations, the steroidal module of the ABP can be extremely
helpful.
If artificial urine or any other possible liquid (for example apple juice or
non-alcoholic beer) was used by the athlete to substitute the original
urine, this usually results in an abnormal steroid profile as no or nearly
no endogenous steroids are detected [80]
[81]. An example is given in [Table 2] summarizing the results obtained on 2
different samples provided by the same athlete. Not only the missing
endogenous steroids demonstrate sample substitution but also other
parameters like odour or the absence of compounds found regularly in human
urine samples like metabolites of social drugs or medications can support
the anti-doping rule violation. Besides steroids, other endogenous compounds
derived from any metabolic pathway and excreted via urine have to be present
at least in trace amounts and if they are covered and detected by routine
doping control methods they can be consulted to differentiate between urine
and any other liquid.
Table 2 Summary of analytical results obtained on 2
different doping control samples provided by the same
athlete.
|
Investigated parameter
|
Sample #1
|
Sample #2
|
|
A [ng/ml]
|
n.d.
|
539
|
|
ETIO [ng/ml]
|
n.d.
|
579
|
|
E [ng/ml]
|
n.d.
|
3.3
|
|
T [ng/ml]
|
n.d.
|
1.1
|
|
5αADIOL [ng/ml]
|
n.d.
|
7.9
|
|
5βADIOL [ng/ml]
|
n.d.
|
9.3
|
|
Olfactory test
|
no odor
|
urine odor
|
|
Colour
|
colourless
|
yellowish
|
|
pH
|
7
|
6.7
|
|
Specific gravity
|
1
|
1.006
|
|
Analysis by screening method to detect unconjugated,
basic and neutral substances
|
- no caffeine and nicotine
|
- caffeine
|
|
- no biological background
|
- cholesterol
|
|
- no indol/indol derivatives
|
- indol/indol derivatives
|
|
- no piperidine derivatives
|
- piperidine derivatives
|
Several cases have been reported in the literature demonstrating sample
substitution.[82]
[83] In all of these cases, the same urine specimen was divided
into several individual urine samples resulting in identical steroid
profiles leading to the detection of anti-doping rule violations. Urine
substitutions initiated by the athlete using urine aliquots of another
individual are more challenging but may be detected by discrepancies visible
in the longitudinal profile, which cannot be attributed to any of the
above-mentioned confounding factors or any doping scenario. Ten cases have
been detected in recent years in the Cologne laboratory and one of these is
exemplarily discussed here in detail. The longitudinal profile is depicted
in [Fig. 12]. After the first 2 samples did
not show any conspicuous features, the third samples was found with an
atypically elevated T/E which triggered an IRMS confirmation
demonstrating the endogenous origin of T and T metabolites. Additionally,
the ratio of 5αADIOL/5βADIOL was decreased, which
does also not match a potential T administration. Samples 4 and 5 again
showed unsuspicious T/E ratios but especially sample 5 did not fit
into the sequence with an extraordinary A/T and
5αADIOL/5βADIOL ratio. Sample 6 showed a similar
pattern to sample 3 with elevated T/E and decreased
5αADIOL/5βADIOL. From sample 6 onwards, the
longitudinal profile stabilized. As the observed scatter in the first data
points was not explainable, an investigation on possible sample substitution
was conducted. This investigation encompassed long-term stored samples (in
this case samples 4 and 6), the confirmation of relevant steroid profiles,
the IRMS analysis of the sample showing the elevated T/E (which was
found to be negative as for sample 3), and the DNA analysis of urine samples
in order to verify if the different specimen can be attributed to the same
individual or not. In this specific case, the DNA analysis could clearly
demonstrate that sample 4 and sample 6 were not derived from the same
individual. In order to clarify which of the samples belonged to the
athlete, an additional blood sample was collected from the athlete and
subjected to DNA analysis. By comparison, sample 6 was attributed to the
athlete. Considering this finding, the conclusion may be drawn that samples
number 1, 2, 4, and 5 were not from the athlete. Interestingly, samples 1,
2, and 4 were OOC samples and here the substitution may be easier, but
sample 5 was collected IC (as sample 3 and 6 were) and here the effort to
swap samples can be considered much higher than for OOC collections. This
may also explain the very different steroid profile found in sample 5
compared to the other samples used for substitution. Nevertheless, as
mentioned above, the tampering of a single doping control specimen is
considered as an anti-doping rule violation and was prosecuted in this
case.
Fig. 12 Longitudinal profile of T/E, A/T,
A/ETIO, and 5αADIOL/5βADIOL for a
male athlete covering a time period of more than 2 years
demonstrating the impact of sample substitution. Example taken from
the steroidal module of ADAMS.
As soon as an APMU has the suspicion that sample substitution may have taken
place and after verification that no administrative error exists, it is
strongly recommended to follow the described procedure encompassing:
-
Long-term storage of available samples and initiation of additional
target tests of the athlete (if necessary)
-
Confirmation of relevant steroid profiles
-
Confirmation by IRMS
-
Application of DNA analysis in order to prove a potential anti-doping
rule violation.
This procedure reaches its limitations in cases where athletes substitute
their urine samples with their own, earlier collected, samples. In these
cases, the steroid profiles will be very similar and may only become
suspicious due to their “abnormal” similarity and DNA
analysis will not provide evidence for manipulation either. But other
analytical methods can be applied to samples suspicious for this form of
substitution. As partly mentioned in [Table
2], the pattern of other urinary ingredients or endogenous
metabolites beyond the endogenous steroids should also vary over time, and
if these patterns are highly similar for different samples this will further
raise the suspicion for a possible sample substitution. Producing
unequivocal evidence by this procedure is much more complicated and to the
best of our knowledge, no sanction has been established solely on analytical
evidence alone in such a case so far. But the combination of analytical data
with information derived by Intelligence and Investigations may offer a
solution here.
Future perspectives and current complements to the steroidal module
New biomarkers and statistical approaches
The steroidal module in its present form is a powerful tool to detect the
misuse of endogenous steroids that is based on scientific research and
method development conducted in the last four decades. Multifaceted research
was conducted in order to further improve the detectability focussing mainly
on the implementation of additional endogenous metabolites and their ratios
[84]
[85]
[86]
[87]
[88]
[89]
[90]. In many of these cases,
slight improvements in sensitivity or retrospectivity were demonstrated but
this seemed not to justify spending the effort necessary to implement these
metabolites into routine screening procedures world-wide. Especially if
additional steps in sample preparation like de-conjugation of sulphated
steroids or cysteine conjugates have to be implemented or the analytical
strategy changes from gas to liquid chromatography, expenditures will become
enormous [87]
[88]
[90]. Much easier would be the
implementation of novel metabolites that are already covered by current
sample preparations but even here measurement time on the analytical
instrument may become the limiting factor. Modifying the statistical
approach would be another option, and also here research is ongoing. Until
now, several multivariate approaches have been tested and as soon as they
will be applicable in a longitudinal manner this might become a powerful aid
to the steroidal module [89].
Steroid concentrations in blood and serum
Especially in those individuals with the above-mentioned del/del
polymorphism of the UGT2B17, the sensitivity of the markers of the urinary
steroid profile was found to be limited. As this polymorphism only affects
the phase-II-metabolism, i. e. the glucuronidation prior to urinary
excretion, serum samples might offer an alternative in those individuals and
in general. Several methods were developed in recent years based on
analytical approaches coming from clinical applications employing liquid
chromatography/mass spectrometry due to the increased sensitivity
necessary for serum steroids [91]
[92]
[93]
[94]
[95]
[96]. These methods will allow for expanding the
analytical properties of the steroidal module and, taking into account first
results, will be applicable in a longitudinal manner comparable to the
steroidal module [93]
[96]. The main drawback associated with serum samples is the more
invasive and expensive sample collection compared to urine specimen. A
possible alternative here may offer the substitution of serum with dried
blood spots (DBS). But as here only 10–20 µL of whole blood
is collected, the sensitivity necessary for a steroid profile will pose a
challenge, but both approaches can be expected to become an interesting
complement to the current ABP.
Another drawback which should be considered is the missing possibility for
confirmation procedures on samples found with an atypical steroid profile as
the IRMS approach will not straightforwardly be applicable to blood due to
the low absolute amount or steroids found in either DBS or serum. Regarding
serum, concentrations of more abundant metabolites will be sufficient for an
IRMS analysis, and recently a method has been developed addressing these
challenges [97].
A different aspect for doping controls employing blood, serum and DBS as
matrices is the detection of intact steroid esters [98]
[99]
[100]
[101]
[102]. The unequivocal benefit of this approach is the exogenous
nature of all possible steroid esters. T and prohormones of T do not occur
in human metabolism esterified, and if they are detected, they can be
directly linked to an anti-doping rule violation. Therefore, this offers the
opportunity to confirm atypical passport findings if other analytical
approaches like IRMS fail. This may happen under circumstances where the
steroid ester applied by the athlete shows an endogenous or at least close
to endogenous CIR. Several preparations have been confirmed with these
unusual enriched isotopic ratios [103]
[104]. But even for T-applications with
non-esterified steroids possible solution for enriched CIR have been
considered and investigated as summarized in the next paragraph.
Enriched carbon isotope ratios found in steroid preparations
Early investigations on CIR of steroid preparations and the first results
obtained on endogenous urinary steroids demonstrated that artificial
pharmaceutical preparations show depleted CIR compared to endogenous
steroids [12]
[14]
[105]. Further investigations
encompassing a larger athlete population covering a wide spectrum of
different geographic origins indicated a broad distribution of isotopic
ratios [16]. A comparison between the CIR at
natural abundance especially for athletes of Northern European origin showed
a distinctive overlap between the distributions of values found endogenously
and in pharmaceutical preparations [16]
[103]
[104] As the
steroidal module has been set up to detect atypical steroid profiles in
order to confirm these by investigations in CIR, this overlap constitutes a
challenge for doping controls.
One possible alternative was to investigate the other abundant element in
steroids – hydrogen [106]. Analogously
to the investigations performed on CIR, a suitable method was developed,
validated and reference population-derived thresholds were established. In
principle, the approach was found suitable for sports drug testing and
already demonstrated the ability to differentiate between endogenous and
exogenous sources of urinary steroids. The main drawback which has to be
considered is the relatively strong overlap of hydrogen isotope ratios (HIR)
found between the endogenous distribution and values found for
T-preparations [103]
[106]. The HIR in the investigated reference population fall
between –240 and –280%, and pharmaceutical
preparations were found between –170 and –260% with
the majority of samples at –230%. Taking into account the
measurement uncertainty and the safety margins applied to population based
thresholds, the probability to misuse a T-preparation that will not differ
significantly from the endogenous HIR is relatively large.
Another scientific approach to enhance the detectability of preparations with
CIR close to endogenous ones was the application of a longitudinal profile
for CIR in parallel to the ABP [107]. By this
approach it is possible to define individual thresholds that are more
sensitive towards the detection of exogenous steroids. And with individual
thresholds as close as ± 0.8% to the mean the detection of
T-preparations closely related to the endogenous CIR become much more
likely. The main drawback with this technique is the relatively large number
of samples necessary to build the “isotopic module” and the
presumable larger scatter of data between different laboratories as the
harmonization of IRMS measurements was found more complicated compared to
concentration determinations.
A recent approach was aiming at a combination of steroid concentrations and
CIR especially of those endogenous steroids easy to measure regarding IRMS,
i. e. ETIO and A [108]. An example
obtained for this approach is given in [Fig.
13]. The calculated parameter named Deviation (ETIO) is
intra-individually stable and allows for the calculation of individual
thresholds by adding the threefold SD to the mean. After the administration
of 100 mg T (CIR of –28.5 %) the equilibrium of
concentration ratios and isotopic ratios is significantly disturbed,
resulting in the abnormal values for Deviation (ETIO). As this approach is
not relying on absolute CIR but only on differences in-between different
steroids it should also be applicable for administrations of steroids
encompassing a CIR comparable to endogenous ones. Further research is
ongoing here in order to evaluate the possibilities of this novel
approach.
Fig. 13 Results obtained on a combination of urinary carbon
isotope ratios and steroid concentrations (Deviation (ETIO)) after
oral application of 100 mg of T at t = 0. The dashed lines
indicate individual thresholds for Deviation (ETIO). Further
information in the text.
