Clinical Chemistry 43: 731-735, 1997;
(Clinical Chemistry. 1997;43:731-735.)
© 1997 American Association for Clinical Chemistry, Inc.
Detection of anabolic steroid administration: ratio of urinary testosterone to epitestosterone vs the ratio of urinary testosterone to luteinizing hormone
Paul J. Perry1,3,a,
John H. MacIndoe2,
William R. Yates1,
Shane D. Scott3 and
Timothy L. Holman1
1
Departments of Psychiatry and
2
Internal Medicine, College of Medicine, and
3
Department of Clinical Pharmacy, College of Pharmacy, University of Iowa, Iowa City, IA 52246
a Address correspondence to this author, at: S415 Pharmacy, University of Iowa, Iowa City, IA 52246. Fax 319-353-5646; e-mail paul-perry{at}uiowa.edu
 |
Abstract
|
|---|
Our goal in this study was to determine whether the urinary ratio of
testosterone to luteinizing hormone (T/LH) as an indicator of exogenous
anabolic steroid (AS) use is superior to the urinary ratio of
testosterone to epitestosterone (T/E). After 2 weekly placebo
injections, 19 subjects were given testosterone cypionate (TC)
injections of 250 or 500 mg/week for 14 weeks followed by 14 weekly
placebo injections. Patients were considered to have ceased taking TC
if they tested negative 9 weeks after their last injection. For
detection of illicit or supraphysiological TC (AS) use, the urinary T/E
ratio of 
6 yielded a false-negative rate of 46% and a false-positive
rate of 4%. However, a urinary T/LH ratio of 30 produced a
false-negative rate of only 24% and a false-positive rate of 13%. We
conclude that the urinary T/LH ratio of 30 is a more sensitive marker
of AS use than the urinary T/E ratio of 6 and remains sensitive for
twice as long as urinary T/E.
Key Words: indexing terms: abused drugs • sports medicine • GC-MS • androgens • anabolic steroids
|
Introduction
|
|---|
The primary method for detecting illicit anabolic steroid (AS) use
has been the analysis of urinary
steroids.1
This methodology has been successful for the majority of
steroids, especially the synthetic variety that have specific
structures that are easily identified by GC-MS. However, the detection
and monitoring of anabolic compounds is not fail-safe. Detection of the
illicit use of testosterone (T), a naturally occurring AS, has become a
difficult clinical problem. Methods for detecting administration of
exogenous T depend on distortions of the normal hormone profile in the
user's urine (1). Attempts to identify optimal markers of
exogenous T administration from untimed urine samples in male athletes
have uncovered several compounds as possible indicators of T abuse. In
1982, the ratio of androgen glucuronides to epitestosterone (E;
17
-hydroxy-4-androsten-3-one) was adopted by the Medical Commission
of the International Olympic Committee (IOC) in Los Angeles, with a
cutoff point 6 being the sole test for illicit T self-administration
(2)(3); the expected urinary ratio of T/E
among healthy subjects not using AS is ~1 (1). However,
analyses from all IOC-accredited laboratories in 1991 suggested that
the majority of athletes who were using AS had switched from synthetic
compounds to T pharmaceuticals to evade detection
(4). Consequently, covert AS use has
become more difficult to detect.
The overall incidence of urinary T/E 6 in the general population of
healthy males not abusing steroids is <0.8%, as evaluated by Catlin
and Hatton (5) and confirmed by Dehennin (4).
In general, the increase of the T/E ratio after high-dose T
administration results from increased T excretion and a decrease of E
output (6). However, some athletes have produced
false-positives, i.e., T/E ratios 6 with subsequent verification that
no exogenous T had been administered (7). Dehennin and
Matsumoto (6) indicated that this problem could be reduced
by taking into account the sulfate excretions of E (ES) in the ratio
T/(ES + E), the relevant threshold being 2.85. Accordingly, Dehennin
(4) suggested that using a T/(ES + E) ratio of 3.0 would
be a more sensitive marker of covert T use.
Dehennin (7) also noted that the joint misuse of T and E
could also lead to false-negative test results, and the IOC in 1992
recommended that urinary E concentrations >150 µg/L should be noted
as abnormally high and therefore suspicious. False-negative results can
also be produced by stimulation of testicular steroidogenesis by
administering human chorionic gonadotropin, which would result in a
concomitant increase in the urinary excretion rate of T and E but with
no significant change in the T/E ratio (1). Dehennin and
Matsumoto (6) confirmed earlier reports of false negatives
by demonstrating that, despite their determination that an average dose
of 47 mg of exogenous testosterone per week would equal or exceed the
IOC cutoff, 2 of 9 subjects receiving 72 mg of testosterone (100 mg of
testosterone enanthate) per week for 6 months did not produce a T/E
ratio 6.
Dehennin (4) suggested that when a T/E ratio of 6 to 12 is
found for the first time in subjects for whom no documentation of a
previously normal ratio exists, some complimentary criteria should be
examined. He found that the ratio of urinary T and E glucuronides to
5-androstene-3ß,17-diol glucuronide was increased in the use of
exogenous T and E use despite the T/E ratio being <6. These findings
indicate a need for further study of additional markers for detecting
the administration of T.
Because the secretion of T is under the control of luteinizing hormone
(LH), Brooks et al. (8) suggested that the urinary T/LH
ratio might be a potentially useful marker for detecting administration
of exogenous T. Kicman et al. (9) observed that high-dose
T administration resulted in dose-dependent suppression of both serum
and urinary LH. This was confirmed by Matsumoto (2), who
found that the urinary LH excretion was reduced to a lesser extent than
was the decrease in both E and T conjugates, such that the T/LH values
were lower than those reported by Kicman et al.—also suggesting a need
for more study.
Palonek et al. (3) reported significantly increased T/LH
ratios in 11 healthy sedentary men participating in a WHO
investigational program for male contraception. Each subject received
144 mg of T per week (200 mg of testosterone enanthate) for 9 months.
The T/LH ratio increased from a mean of 0.052 (range 0.002–0.108) at
baseline to 45.16 (1.28–252) at 3 months, 85.7 (8.3–238) at 6 months,
and 71.7 (5.3–344) at 9 months. The authors indicated that, among all
the different ratios or proposed markers they investigated, the urinary
T/LH ratio showed the most dramatic increase (~1000-fold). Of the
other markers, the increase in the serum T/LH ratio was of similar
magnitude as that of the urinary T/LH ratio, whereas the urinary T/E
ratio had only a 50-fold increase. The investigators also reported that
1 of the 11 subjects produced a T/E ratio below the IOC cutoff at 3 and
9 months of administration and just over the threshold at 6 months. The
T/LH ratio for the same subject was above the upper reference limit at
3, 6, and 9 months. Palonek et al. concluded that increased serum and
urinary T/LH ratios in the presence of a normal T/E ratio may indicate
self-administration of both T and E.
Unanswered is whether the T/LH ratio might be more sensitive than the
T/E ratio for identifying illicit use of AS. Thus the goal of the
present study was to determine which laboratory test is most sensitive
and specific for detecting the administration of exogenous T.
|
Materials and Methods
|
|---|
subjects and study design
Healthy male volunteers between ages 18 and 40 years were
recruited and, after an explanation of the study, gave their signed
informed consent. The study protocol was reviewed and approved by the
Human Subjects Institutional Review Board and the Clinic Research
Center of the University of Iowa. A standard drug history, developed by
the National Institute on Drug Abuse, was administered before
entry. Any subject who indicated he was currently using central nervous
system stimulants other than modest amounts of caffeine (two cups of
coffee per day) was excluded from the study.
Each subject received two weekly placebo doses of cottonseed oil, the
vehicle for testosterone cypionate (TC). At the end of the 2-week
placebo lead-in period, subjects were randomized to one of three doses
of TC (100, 250, or 500 mg/week) given for 14 consecutive weeks. In our
experience with AS users (10), the subjects' shortest
average cycle was 7 weeks, the longest 14 weeks. Thus, we decided that
subjects should be administered TC for a typical 14-week AS cycle to
mimic the maximum cycling interval.
Subjects functioned as their own controls. They received weekly
intramuscular injections of either TC or placebo (vehicle only) for 28
consecutive weeks. For the purpose of evaluating the effectiveness of
the urine T/E ratio as an indicator of recent AS use, we considered
only the subjects receiving supraphysiological TC doses (250 and 500
mg/week). A 100 mg/week dose is generally regarded as a physiological
replacement dose in the majority of patients.
To monitor the subjects medically, we assessed their liver-function
tests, fasting lipid profiles, thyroxine-binding globulin,
sex-hormone-binding globulin, 24-h urinary free cortisols, serum free
and total T, estradiol, LH, follicle-stimulating hormone,
thyroid-stimulating hormone, and free thyroxine—obtained at baseline,
at entry into the study, after the 2-week placebo injection period, and
then biweekly for the remainder of the study. All endocrine samples
were collected between 0700 and 0900 to minimize the chronotropic
secretion effects of these hormones. Depo®-testosterone
(TC), 200 g/L (200 mg/mL), was the proprietary product utilized for the
study. The diluent (0.2 mL of benzyl benzoate, 9.45 mg of benzyl
alcohol, and 560 mg of cottonseed oil per milliliter) was prepared by
the Pharmaceutical Services Division of the University of Iowa College
of Pharmacy (an FDA-approved manufacturing group). At the end of the 14
weeks of TC administration, the subjects were switched to diluent-only
injections.
assays
The urine drug screens were performed by Smith Kline Beecham
Clinical Laboratories Sports Testing Center in Tucker, GA, a laboratory
certified by the US Department of Health and Human Services. The
initial drug screen and all subsequent screens were negative for AS
(other than T), amphetamines, barbiturates, benzodiazepines, cocaine
metabolites, methadone, methaqualone, opiates, phencyclidine, and
propoxyphene.
Urine concentrations of T, E, LH, and creatinine were also determined
in the samples (assayed by Smith Kline Beecham). The urine samples were
refrigerated at 8 °C and were analyzed within 5–10 days after
collection. If the T/E ratio was <6, the sample was discarded within
30 days. Urine drug screens were routinely obtained at weeks 0, 1, 4,
8, 12, 16 or 17, 20, 24, and 28; in some follow-up cases, they were
obtained at weeks 40 and 92. AS screens and confirmations were
performed by GC-MS on separate aliquots. Samples were initially
screened for the substance abuse panel by Emit (Behring, Palo Alto,
CA); all positive results were confirmed by GC-MS (11). LH
in urine was performed by Microparticle Enzyme Immunoassay with the
Abbott Diagnostics (Chicago, IL) IMx system.
The T/E ratio was determined by GC-MS. Both free and conjugated T and E
were extracted with C18 solid-phase extraction columns
(Bond Elute LRC; Varian, Harbor City, CA), hydrolyzed with
ß-glucuronidase (Boehringer Mannheim, Mannheim, Germany), and
detected by monitoring characteristic ions with the mass spectrometer.
Quantification and identification of T and E required selected-ion mode
analysis in which the presumptive positive specimens were matched with
the retention times and ion ratios of known compounds. The T
calibration curve was linear between 2 and 400 µg/L; that for E was
linear between 2 and 500 µg/L. The CV for the T/E ratio was 13.3%.
The specificity of this method is extremely high: At the time of the
performance of the assays, no compounds were known to interfere with
either T or E.
The urine concentration of LH was determined with the IMx system
kit for serum LH as described in the 1991 IMx LH package insert. To
determine that there was no matrix effect for the assay, we added known
amounts LH to urine and serum samples and found that the resulting
calibration curves could be superimposed on each other and were linear
between 2 and 600 IU/L. The lower limit of detection for this assay is
0.5 IU/L. The CV for the serum LH assay is 8.7% at 5.37 IU/L, 6.4% at
43.2 IU/L, and 6.2% at 82.5 IU/L.
All serum samples for determining free and total T were stored at
-20 °C until assay. The T concentrations were quantified with
Coat-A-Count® kits (Diagnostic Products Corp., Los
Angeles, CA) as described in the manufacturer's package insert (1995).
The lower limits of detection were 40 ng/L for total T and 0.15 ng/L
for free T. The inter- and intraassay CVs for the free T assay were
11.2% and 5.5%, respectively; those for total T were 10.4% and
8.8%, respectively.
|
Results
|
|---|
In all, 93 urine drug screen samples were obtained from the 19
subjects participating in the study who received supraphysiological
doses of TC. Seven received 250 mg/week and 12 took 500 mg/week. None
of the subjects was positive for exogenous AS use other than for the TC
injections administered during weeks 2–15 of the study. Concentrations
of free T in serum were analyzed 5–7 times between days 3 and 21 after
the last TC injection for 17 of the 19 subjects. From these data, we
calculated for each patient the terminal elimination rate
(ke) and the elimination half-life
(t1/2) for free T in serum. To determine
ke, we fit the T concentrations
f(t) and time points (t) to the
following single exponential decay equation, where a is the
concentration of T at time 0:
 | (1) |
We determined t1/2 as follows:
 | (2) |
A 21-day T sampling period was appropriate to determine the
t1/2 of exogenous T because
gonadotropin-releasing-hormone stimulation tests indicated that the
hypothalamic–pituitary–testicular axes of the subjects did not regain
sufficient sensitivity to stimulate release of T until 4–6 weeks after
discontinuation of the TC injections. The individual elimination
half-life data are presented in Table 1
. There was no difference in half-life values between the weekly
TC doses of 250 and 500 mg (Mann–Whitney U = 25.0,
P = 0.37). The overall mean ± SD elimination
half-life for free T in serum after administration of TC was 6.6
± 2.3 days. Based on these data, an 11-day t1/2
would be 2 SD from the mean. Given that 97% of the exogenous T was
excreted in 5 half-lives (i.e., 11-day half-life x 5
half-lives = 55 days, or 8 weeks) and that pituitary sensitivity
to gonadotropin-releasing hormone returned within 4–6 weeks of the
last TC injection, subjects were assumed to have ceased taking ("be
off") exogenous T by the time of the urine drug screen performed 9
weeks after the last TC injection.
The AS urine drug screen findings indicated that the urinary T/E ratio
cutoff of 6, the traditional laboratory marker to determine the use
of exogenous T and used as such by the National Collegiate Athletic
Association and the IOC, although quite specific for determining nonuse
of T, is not a sensitive indicator for detecting illicit T usage. Table 2
illustrates this. Although the T/E ratio of >6 had 96%
specificity in identifying our subjects as being off steroids by 9
weeks after their last dose, it was correct only 54% of the time for
identifying our subjects as being on steroids during the 14 weeks of TC
injections and in the 9 subsequent weeks when they received sham
injections. As a practical matter, these data suggest that one of every
two subjects using injectable TC will, both during injection periods
and for 9 weeks afterwards, give a false-negative urine drug screen.
Receiver operating characteristic (ROC) analysis of these data
(12) identified a urinary T/E ratio of 1.2 as the cutoff
value that provided optimum sensitivity and specificity for indicating
use or nonuse of T. Resorting the data in Table 2
illustrates that use
of a T/E ratio of 1.2 for a T-positive urine improves the sensitivity
to 83% and the specificity decreases only somewhat, to 77%.
View this table:
[in this window]
[in a new window]
|
Table 2. Contingency table for various urinary T/E ratios used as
the threshold ratio for anabolic steroid use (TC 250 or 500
mg/week).
|
|
The potential usefulness of the urine T/LH ratio as an indicator of T
use and nonuse is illustrated in Table 3
. These data suggest that to maintain 100% specificity requires
a threshold T/LH ratio of 74, although the sensitivity at this cutoff
is only 52%. However, by ROC analysis (data not shown), the urinary
T/LH ratio cutpoint that optimizes sensitivity and specificity is 30.
As Table 3
shows, use of a urinary T/LH ratio 30 increases
sensitivity to 76% but decreases specificity to 87%.
View this table:
[in this window]
[in a new window]
|
Table 3. Contingency table for urinary T/LH ratios used as the
threshold ratio for anabolic steroid use (TC 250 or 500
mg/week).
|
|
|
Discussion
|
|---|
From a medical-legal standpoint the most worrisome finding of
these data is the false-positive tests. Table 4
characterizes the false positives—i.e., a test result that is
not negative at 9 weeks after the last TC injection—for the various
testing schemes. Nine weeks is equivalent to the amount of time
required to clear 97% of the exogenous TC. For the urinary T/E ratio
of 6, only two subjects did not meet this criteria, whereas for the
urine T/LH ratio of 30, four subjects did not meet this criteria. All
six patients who tested "positive" were actually tested 9–25 weeks
after their last TC injection. When contrasted with the half-life data,
this suggests that the normalization of LH concentrations may lag
behind the rate at which the exogenous T clears from the body.
Moreover, in reality there are no truely false-positive test results.
However, in no case did a subject's urine screen test positive before
the start of the TC injections.
View this table:
[in this window]
[in a new window]
|
Table 4. False-positive rates for use of supraphysiological doses
of anabolic steroids in different urine testing
schemes.
|
|
The mean ± SD urinary T/LH and T/E ratios before the start of the
TC injections were 3.8 ± 2.4 and 0.8 ± 1.3, respectively.
Other than when the subjects were receiving T injections, the only time
there was a significant difference between pre-TC injection urinary T/E
and T/LH ratios and the ratios after the start of the TC injections was
2 weeks after the last injection. For the urinary T/LH ratio, the mean
difference between the baseline value and the ratio 2 weeks after the
last injection was 29.8 (t = 2.829, P
<0.02, df = 16); for the urine T/E ratio, the mean
difference was 14.9 (t = 2.703, P <0.02,
df = 16).
As suggested in Table 4
, the urinary T/LH ratio of 30 is the screen
most likely to detect AS use the longest, i.e., as long as 25 weeks
after the last injection. Using the urinary T/LH ratio 30 as a marker
showed that 4 of 19 (21%) subjects tested positive 9–25 weeks after
their last injection of T. In contrast, use of the urinary T/E ratio
6 found only 2 of 19 (11%) patients positive for steroid usage at 9
weeks after their last T injection. Fig. 1
chronologically contrasts the mean urine T/LH and T/E ratios.
Inspection of Fig. 1
suggests that the urinary T/LH ratio returns to
baseline at a slower rate than the urinary T/E ratio does, thereby
explaining the greater number of false-positive results for T/LH in
this group. To prove this point, we regressed the mean T/E and T/LH
ratios against their timepoints at weeks 17, 20, 24, 28, and 40 and fit
this as a monoexponential decay curve. The regression line intersects
the critical T/LH ratio of 30 at 7.9 weeks after the last TC
injection (T/LH ratio = 90.2
e-0.14(week), r2 =
0.86). However, the T/E ratio fitted to the exponential equation (T/E
ratio = 12.3
e-0.1916(week),
r2 = 0.80) intersects the ratio of 6 at 3.7
weeks. Both models, therefore, demonstrate why more subjects test
positive for a longer time when assessed with the T/LH ratio.
It is not uncommon for nonpower athletes (e.g., distance runners,
swimmers, tennis players, soccer players) to utilize physiological
doses of T (i.e., TC 100 mg/week) to counter the catabolic effects of
stress and exercise on muscle. We measured urine T/E and urine T/LH in
seven subjects who were administered TC at 100 mg/week. Monoexponential
regression equations for the T/E and T/LH ratios to return to baseline
values were based on the mean ratios measured in these subjects at
weeks 17, 20, and 24 after cessation of TC injections. The urinary T/E
ratio, when fitted as an exponential decay equation (T/E ratio =
8.3 e-0.2072(week),
r2 = 0.82), intersects the ratio of 6 at
1.6 weeks, whereas the urinary T/LH ratio, fitted to the equation T/LH
ratio = 48.9 e-0.2148(week)
(r2 = 0.99), intersects the ratio of 30 at
2.3 weeks. These data suggest that is debatable whether TC at 100
mg/week is actually a physiological replacement dose: Some athletes may
test positive even at this small a dose of T.
In conclusion, we find that the urinary T/LH ratio is a more
sensitive and specific test for a longer time for investigating AS use
than is the urinary T/E ratio. Supporting this finding is the fact
that, unlike the case for E, there are no commercially available
FDA-approved LH products. This advantage alone makes the urinary T/LH
ratio a considerably more practical screening test than the urinary T/E
ratio.
|
Acknowledgments
|
|---|
This project was supported by a grant from the National
Institute on Drug Abuse at the National Institutes of Health
(RO1DA08347).
|
Footnotes
|
|---|
1 Nonstandard abbreviations: AS, anabolic steroid(s); T, testosterone; E, epitestosterone; IOC, International Olympics Committee; ES, epitestosterone sulfate; LH, luteinizing hormone; and TC, testosterone cypionate. 
|
References
|
|---|
-
Kicman A, Oftebro H, Walker C, Norman N, Cowan DA. Potential use of ketoconazole in a dynamic endocrine test to differentiate between biological outliers and testosterone use by athletes. Clin Chem 1993;39:1798-1803.
[Abstract]
-
Matsumoto AM. Effects of chronic testosterone administration in normal men: safety and efficacy of high-dosage testosterone and parallel dose-dependent suppression of luteinizing hormone, follicle-stimulating hormone, and sperm production. J Clin Endocrinol Metab 1990;70:282-287.
[Abstract/Free Full Text]
-
Palonek E, Gottlieb C, Garle M, Björkhem I, Carlstrom K. Serum and urinary markers of exogenous testosterone administration. J Steroid Biochem Mol Biol 1995;55:121-127.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Dehennin L. Detection of simultaneous self-administration of testosterone and epitestosterone in healthy men. Clin Chem 1994;40:106-109.
[Abstract/Free Full Text]
-
Catlin DH, Hatton CK. Use and abuse of anabolic and other drugs for athletic enhancement. Adv Intern Med 1991;36:399-424.
[Medline]
[Order article via Infotrieve]
-
Dehennin L, Matsumoto A. Long-term administration of testosterone enanthate to normal men: alterations of the urinary profile of androgen metabolites potentially useful for detection of testosterone misuse in sport. J Steroid Biochem Mol Biol 1993;44:179-189.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Dehennin L. On the origin of physiologically high ratios of urinary testosterone to epitestosterone: consequences for reliable detection of testosterone administration by male athletes. J Endocrinol 1994;142:353-360.
[Abstract/Free Full Text]
-
Brooks R, Jeremiah G, Webb W, Wheeler M. Detection of anabolic steroid administration to athletes. J Steroid Biochem 1979;11:913-917.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Kicman A, Brooks R, Collyer S, Cowan DA, Nanjee MN, Southan GJ, Wheeler MJ. Criteria to indicate testosterone administration. Br J Sports Med 1990;24:253-264.
[Abstract/Free Full Text]
-
Perry PJ, Andersen KH, Yates WR. Illicit anabolic steroid use in athletes: a case series analysis. Am J Sports Med 1990;18:422-428.
[Abstract/Free Full Text]
-
Sample RHB, Baenziger JC. Gas chromatography/mass spectrometric detection of anabolic steroids. Deutsch J eds. Analytical aspects of toxicology 1989:247-272 JR Wiley and Sons New York. .
-
Somoza E, Mossman D. Neuropsychiatric decision making: designing nonbinary diagnostic tests. J Neuropsychiatry Clin Neurosci 1991;3:197-200.
[Abstract/Free Full Text]
Top
Abstract
Introduction
