Content uploaded by Martin Hill
Author content
All content in this area was uploaded by Martin Hill
Content may be subject to copyright.
Sex- and Age-Related Changes in Epitestosterone in
Relation to Pregnenolone Sulfate and Testosterone in
Normal Subjects
HELENA HAVLI
´
KOVA
´
, MARTIN HILL, RICHARD HAMPL, AND LUBOS
ˇ
STA
´
RKA
Institute of Endocrinology, CZ 116 94 Prague, Czech Republic
Epitestosterone has been demonstrated to act at various lev-
els as a weak antiandrogen. So far, its serum levels have been
followed up only in males. Epitestosterone and its major cir-
culating precursor pregnenolone sulfate and T were mea-
sured in serum from 211 healthy women and 386 men to find
out whether serum concentrations of epitestosterone are suf-
ficient to exert its antiandrogenic actions. In women, epites-
tosterone exhibited a maximum around 20 yr of age, followed
by a continuous decline up to menopause and by a further
increase in the postmenopause. In men, maximum epitestos-
terone levels were detected at around 35 yr of age, followed by
a continuous decrease. Pregnenolone sulfate levels in women
reached their maximum at about age 32 yr and then declined
continuously, and in males the maximum was reached about
5 yr earlier and then remained nearly constant. Epitestoster-
one correlated with pregnenolone sulfate only in males. In
both sexes a sharp decrease of the epitestosterone/T ratio
around puberty occurred. In conclusion, concentrations of
epitestosterone and pregnenolone sulfate are age dependent
and, at least in prepubertal boys and girls, epitestosterone
reaches or even exceeds the concentrations of T, thus sup-
porting its role as an endogenous antiandrogen. The dissim-
ilarities in the course of epitestosterone levels through the
lifespan of men and women and its relation to pregnenolone
sulfate concentrations raise the question of the contribution
of the adrenals and gonads to the production of both steroids
and even to the uniformity of the mechanism of epitestoster-
one formation. (J Clin Endocrinol Metab 87: 2225–2231, 2002)
T
HE 17
␣
-EPIMER OF T (epitestosterone, EpiTe) has pre-
viously been considered to be a by-product of the ⌬
5
-
steroid pathway, without biological significance (1). In 1987,
Nuck and Lucky (2) reported the effect of EpiTe on the flank
organ of a golden hamster, in which it inhibited the T effect
on the pilosebaceous unit; they hypothesized that EpiTe
could act as an antiandrogen. This finding encouraged the
authors of this study, among others, to investigate EpiTe
antiandrogenic effects in vitro as well as in vivo, using various
rodent and human models.
Besides the ability of EpiTe to inhibit 5
␣
-reductase in rats,
reported already previously by others (3), the authors have
demonstrated that EpiTe can reduce the weight of androgen-
dependent organs in rats and/or mice and compete with
synthetic androgen methyltrienolone for ARs in rat prostate
cytosol with K
i
about half of that of dihydrotestosterone (4).
Furthermore, EpiTe appears to act as a competitive inhibitor
of the testicular P450C17
␣
enzyme in rats and humans (5, 6)
and, at least in rats, influence the secretion and production
of LH and FSH (7, 8). Recently, significantly lower EpiTe
concentrations were found in hair samples from balding men
(9); it was concluded that EpiTe may act as a weak antian-
drogen, the efficiency of which could be potentiated by the
complexity of its actions at various levels.
However, it is unclear whether circulating EpiTe levels are
sufficient to exert antiandrogenic actions. In previous studies
the authors have focused on the role of EpiTe and androgens
in males with respect to their plausible involvement in the
pathogenesis of typical androgen-dependent diseases of
older men, such as benign prostate hyperplasia (BPH). It has
been shown that in childhood the EpiTe/T ratio is close to
or even higher than one but that it decreases sharply during
the prepubertal period and puberty, remaining nearly con-
stant in adulthood (10). Similar results have been obtained
by others (11). This points to some importance, at least, of
EpiTe before puberty in males, when it may attenuate the
effects of T.
In contrast to the very low serum EpiTe levels in older
men, relatively high concentrations of this steroid have
been found in human prostatic tissue from men operated
on for BPH. EpiTe concentrations in this tissue were about
half those of dihydrotestosterone but twice as high as those
of T (12).
In humans, the biosynthesis of EpiTe arises from preg-
nenolone, a certain portion of which is converted to 5-
androstene-3

,17
␣
-diol, which in turn serves as a substrate
for 3

-hydroxysteroid dehydrogenase/⌬
4,5
-isomerase giv
-
ing rise to EpiTe, thus avoiding the DHEA and androstenedi-
one involved in the T pathway (13). As has been demon-
strated by Dehennin (14), the main sources of EpiTe are
testicular Leydig cells, but a certain proportion of the circu-
lating steroid may come from the adrenals.
The majority of circulating pregnenolone is sulfated;
though mainly of adrenal origin, it may serve as a supply for
EpiTe precursors. Changes of pregnenolone sulfate levels
during life in both sexes were reported as early as 1983 by de
Peretti and Mappus (15). More recently, Morley et al. (16)
followed up pregnenolone sulfate and six other hormonal
variables, along with cognitive and physical tests, in a cohort
of exceptionally healthy men in search of suitable parameters
of predicative value for the development of age-related dys-
Abbreviations: BPH, Benign prostate hyperplasia; EpiTe, epitestos-
terone; LSD, least significant differences.
0013-7227/02/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 87(5):2225–2231
Printed in U.S.A. Copyright © 2002 by The Endocrine Society
2225
regulations. Taking into consideration that pregnenolone
and its more abundant sulfate are indeed the major precur-
sors in EpiTe biosynthesis, it may be of interest whether some
relationship between circulating EpiTe and pregnenolone
sulfate does exist and, if so, whether it is confined to males
and what changes there are in the ratio of the two circulating
steroids through life.
The present study provides data on serum EpiTe, preg-
nenolone sulfate, and T in a large group of male and female
subjects of normal population during their lives. The fol-
lowing questions were addressed: 1) What are the levels of
circulating EpiTe during life; 2) do they differ according to
sex; 3) are serum concentrations of EpiTe sufficient to exert
its antiandrogenic actions with respect to actual T levels; and
4) is there any relationship between circulating EpiTe and
pregnenolone sulfate?
Materials and Methods
Subjects
The study included 211 females (10 –77 yr) and 386 males (1–91 yr),
randomly selected in the framework of iodine deficiency screening in the
Czech Republic. The subjects with major apparent medical problems
such as endocrinopathies, oligo- or amenorrhea in women of reproduc-
tive age, or those receiving medications known to affect endocrine status
with particular respect to steroid metabolism were excluded. All par-
ticipants signed informed consent to the use of their blood samples for
research purposes.
Steroid determination
Blood was withdrawn from the cubital vein between 0800 h and
1000 h. Not more than 2 h later, the serum was separated and stored in
a freezer at ⫺20 C until processed. The serum EpiTe was determined by
the method of Bı´lek et al. (17). Specific antiserum raised against 17
␣
-
hydroxy-4-androstene-3-one-3-(O-carboxymethyl)-oxime-BSA and radio-
iodinated epitestosterone-3-(O-carboxymethyl)-oxime tyrosine methyl
ester derivative as a tracer were used as reagents. The sensitivity of the
analysis in serum was 0.033 nmol/liter, and inter- and intraassay co-
efficients of variation were 10.9% and 7.7%, respectively. The sample
volume was 350
l. The identity of measured EpiTe was validated by
analyzing 15 samples of pooled sera from males and females as follows:
serum (1 ml), spiked with 10,000 cpm of [1,2,6,7(n)
3
H]testosterone (Ra
-
diochemical Center, Amersham Pharmacia Biotech, Uppsala, Sweden)
was extracted twice with diethyl ether (3 ml); the extract after evapo-
ration was partitioned between 80% methanol with water (1 ml) and 1
ml light petroleum (boiling point 60 – 80 C). The water-methanol phase
after evaporation was subjected HPLC. A binary gradient was used at
a constant flow rate of 1 ml/min. Mobile phase A was 15% acetonitrile
in water containing 100 mg/liter ammonium bicarbonate. Mobile phase
B was methanol. The gradient was as follows: 1-min delay at 40% B,
switch to 65% B, and delay up to the 8th minute followed by switch to
100% B and delay up to the 12th minute when the mobile phase was
adjusted to 40% B; rinsing proceeded up to the 18th minute. The tem-
perature was kept at 40 C. The column was an ET 250/4, NUCLEOSIL
100⫺5 C18 from Macherey-Na¨gel (Du¨ ren, Germany) The fractions con-
taining EpiTe and T (reverse transcription ⫽ 10.4 and 11.6 min, respec-
tively) were collected and following evaporation of the solvent, the
fraction containing EpiTe was analyzed by gas chromatography-mass
spectroscopy (GC-MS) after derivatization with O-(2,3,4,5,6-pentaflu-
orobenzyl)-hydroxylamine hydrochloride at 60 C for 60 min and sub-
sequent derivatization with a mixture of bis-(trimethysilyl)trifluoroac-
FIG. 1. Polynomial regression of the age dependencies of EpiTe in the
serum of 211 women aged 10 –77 yr (A) and 386 men aged 1–91 yr (B).
Because of the skewed data distribution on the y-axis, the original
data were transformed to the minimum skewness of the studentized
residuals. The curves of the mean prediction (the solid line with the
95% confidence interval, the dashed lines closer to mean prediction)
and the 95% confidence intervals of prediction (the dashed lines fur-
ther from the mean prediction) were obtained by the retransformation
of the results to the original scale. All of the parameters of the poly-
nomial were significant (t tests). r, correlation coefficient of the mul-
tiple regression; p, level of statistical significance of the model, m,
degree of polynomial, n, number of subjects (the value in parentheses
represents the number of outliers and high leverage points excluded
from computation);
, power of the transformation (
⫽ 0 denotes
logarithmic transformation).
FIG. 2. Age and sex differences of EpiTe in the serum of 211 women
aged 10 –77 yr (empty circles), and in that of 386 men aged 1–91 yr
(solid circles). The circles with error bars represent group mean values
with 95% confidence intervals, calculated using LSD multiple com-
parison. Overlapping of the confidence intervals denotes statistical
insignificance between individual groups and vice versa. As confirmed
using two-way ANOVA with sex and age group as the first and the
second factors, respectively, both sex and age differences were highly
significant (P ⬍ 0.0001), as were the differences in the shapes of the
age dependencies (age/sex interactions).
2226 J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 Havlı´kova´ et al. • Age, Sex, and Epitestosterone
etamide and trimethychlorosilane (99:1) at 60 C for 45 min. GC
separation was carried out with a ZEBRON ZB-50 (15m ⫻ 0.25 mm)
middle polar capillary column with 0.15-
m film thickness, catalog no.
7e.g.-G004-05, (PHENOMENEX, St. Torrance, CA). The temperature of
the injection port was 300 C.
The following protocol was used: temperature gradient: plateau at
120 C (1 min), linear gradient 40 C/min from 120 C to 240 C (3 min),
linear gradient 10 C/min from 240 C to 300 C (6 min), and plateau at 300
C (1 min); pressure gradient: high-pressure (pulsed splitless) injection at
60 kPa (1 min), linear gradient 10.5 kPa/min from 30 to 62 kPa (3.05 min),
linear gradient 2.6 kPa/min from 62 to 77.5 kPa (5.95 min), and plateau
at 300 C (1 min). The duration of the analysis was 11 min.
Responses were recorded in selected ion monitoring mode monitor-
ing the molecular ion (m/z ⫽ 555.5). In addition, three different GC
gradients were used to check the correct identification of the substances.
The detector voltage was set at 2 kV and the sampling rate was 0.25 sec.
The temperatures of the interface and the ion source were 310 C and 240
C, respectively.
The results were corrected to losses during extraction and chroma-
tography using recovered [
3
H]testosterone. In parallel, the samples were
analyzed by RIA. The correlation of RIA (y) with GC-MS (x) was ex-
pressed by a linear equation y ⫽ 0.160 ⫹ 0.983 • x with a correlation
coefficient of R ⫽ 0.924.
T was determined using standard RIA (18), the only modification
being that radioiodinated testosterone-3-O(carboxymethyl)oxime ty-
rosine methyl ester derivative was used as a tracer. Pregnenolone sulfate
was determined as has been described elsewhere (19).
Statistical analyses
The dependencies of EpiTe, pregnenolone sulfate, and the steroid
ratios on age were evaluated using stepwise polynomial regression.
Because of the non-Gaussian distribution in the concentrations of all
three measured steroids, the original data were subjected to power
transformation (20, 21) to attain the minimum skewness of the studen-
tized residuals. The retransformed 95% confidence intervals of predic-
tion were considered to be the age-dependent limits of the reference
range of the steroid. The degree of polynomial was determined using the
correlation coefficient of multiple regression adjusted to degrees of
freedom, se of estimation (the square root of the mean squared error),
mean absolute error (the average of the absolute values of the residuals),
and Akaike information criterion (20, 21). The statistical significance of
the model was determined using Fisher’s test. Regression diagnostic
plots were used for the detection of outliers and high leverage points (20,
21). However, all outliers and high leverage points excluded from the
computation of predictions and confidence intervals were retained in the
figures for completeness.
Two-way ANOVA with sex as the first and age as the second factors
was used for evaluation of age and sex relationships in steroids and their
ratios. The original data were transformed to minimum skewness of
residuals to stabilize the group variances and to approximate a Gaussian
distribution of the data. For the detection of outliers, an analysis of
residuals was used.
The individual differences between the two groups were tested using
the method of least significant differences (LSD). The group means with
95% confidence intervals obtained from ANOVA treatment of the trans-
formed data were retransformed to the original scale. All computations
described above were performed using Statgraphics Plus version 3.3
software (Manugistics Inc., Rockville, MA).
Pearson’s method was used for the evaluation of mutual correlations
between the measured steroids. To avoid nonconstant variance, a non-
Gaussian distribution of the data, and to straighten the simple mono-
tonic curvilinear relationships between the variables, power transfor-
mation to minimum skewness in each of the two dimensions was
FIG. 3. Polynomial regression of the age dependencies of the EpiTe/T
ratio in the serum of 174 women aged 10–70 yr (A) and in that of 201
men aged 10 – 69 yr (B). Because of the skewed data distribution on
the y-axis, the original data were transformed to minimum skewness
of the studentized residuals. The curves of the mean prediction (the
solid line), the 95% confidence interval (the dashed lines closer to
the mean prediction), and the 95% confidence intervals of prediction
(the dashed lines further from the mean prediction) were obtained by
retransformation of the results to the original scale. All of the pa-
rameters of the polynomial were significant (t tests). r, correlation
coefficient of the multiple regression; p, level of statistical significance
of the model; m, degree of polynomial; n, number of subjects (the value
in parentheses representing the number of outliers and high leverage
points excluded from computation);
, power of the transformation
(
⫽ 0 denotes logarithmic transformation).
FIG. 4. Age and sex differences of the EpiTe/T ratio in the serum of
174 women aged 10 –70 yr (empty circles) and 201 men aged 10 – 69
yr (solid circles). The circles with error bars represent group mean
values with 95% confidence intervals that were calculated using LSD
multiple comparison. Overlapping of the confidence intervals denotes
statistical insignificance between individual groups and vice versa. As
confirmed using two-way ANOVA with sex and age group as the first
and the second factors, respectively, both the sex and the age differ-
ences were highly significant (P ⬍ 0.0001), as were the differences in
the shapes of the age dependencies (age/sex interactions).
Havlı´kova´ et al. • Age, Sex, and Epitestosterone J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 2227
applied (20). The principal axis and 95% confidence ellipsoids were
computed in Excel 97 (Microsoft Corp., Redmont, WA) using a method
described elsewhere (21). The results obtained were retransformed to the
original scale.
Results
Serum EpiTe levels in women and men are shown in Fig.
1. The peak of EpiTe levels in women was found around the
20th year of age (Fig. 1A), and in men it was shifted to around
the 35th year (Fig. 1B). In contrast to men, an increasing and
accelerating trend starting in menopause and proceeding in
the postmenopause is apparent in women. The differences
among the mean levels of EpiTe in various age groups are
shown in Fig. 2. The confidence intervals of the group means,
which are not overlapped, imply statistically significant dif-
ferences in intergroup mean values.
The course of the age dependence of the EpiTe/T ratio in
men and women is shown in Fig. 3. In both sexes a sharp
decrease with age occurred, remaining nearly constant in
adulthood. As shown in Fig. 4, there was no overlap between
confidence intervals in males and females, which means that
the differences in each age group were significant.
The polynomial curves reflecting the dependence of preg-
nenolone sulfate on age in both sexes are shown in Fig. 5. In
females (A) only one distinct maximum was found, at around
30 yr, and in males an indistinct maximum was found at
around 25 yr; subsequently, it did not change much until
approximately 52 yr, after which it decreased continuously
again. The differences in individual age groups according to
sex are shown in Fig. 6.
The age dependence of the EpiTe/pregnenolone sulfate
ratio in both sexes as expressed by straight-lined (B) or poly-
nomial regression (A) is shown in Fig. 7. Although in men (B)
the ratio displayed a slight but significant (P ⬍ 0.01) increas-
ing trend, in women (A) a U-shaped age dependence was
found, with a decreasing trend up to the third decade, from
which time a continuous increase followed. The differences
were highly significant (ANOVA, P ⬍ 0.0001). The differ-
ences in individual age groups according to sex are shown
in Fig. 8. In men, the increasing overall trend of the EpiTe/
pregnenolone sulfate ratio in senescence was still significant
but was much less pronounced than in women.
As shown in Fig. 9, EpiTe significantly correlated with
pregnenolone sulfate in men, but no correlation was found
in women. No significant correlations were found between
the levels of EpiTe and T in the two sexes.
Discussion
Hitherto, EpiTe levels in serum have been systematically
investigated only in men, with respect to the possible role of
this metabolite as an endogenous antiandrogen (9), partic-
ularly in relation to its potential role in the pathogenesis of
BPH (10, 12). Here, circulating EpiTe and pregnenolone sul-
FIG. 5. Polynomial regression of the age dependencies of preg-
nenolone sulfate in the serum of 230 women aged 10 –70 yr (A) and
179 men aged 4 – 69 yr (B). Because of the skewed data distribution
on the y-axis, the original data were transformed to minimum skew-
ness of the studentized residuals. The curves of the mean prediction
(the solid line), the 95% confidence interval (the dashed lines closer to
the mean prediction), and the 95% confidence intervals of prediction
(the dashed lines further from the mean prediction) were obtained by
retransformation of the results to the original scale. All of the pa-
rameters of the polynomial were significant (t tests). r, correlation
coefficient of the multiple regression; p, level of statistical significance
of the model; m, degree of polynomial; n, number of subjects (the value
in parentheses representing the number of outliers and high leverage
points excluded from computation);
, power of the transformation.
FIG. 6. Age and sex differences of pregnenolone sulfate in the serum
of 230 women aged 10–70 yr (empty circles) and 179 men aged 4 –79
yr (solid circles). The circles with error bars represent group mean
values with 95% confidence intervals that were calculated using LSD
multiple comparison. Overlapping of the confidence intervals denotes
statistical insignificance between individual groups and vice versa. As
confirmed using two-way ANOVA with sex and age group as the first
and second factors, respectively, both the sex and age differences were
highly significant (P ⬍ 0.0001), as were the differences in the shapes
of the age dependencies (age/sex interactions).
2228 J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 Havlı´kova´ et al. • Age, Sex, and Epitestosterone
fate were measured in a population sample of both men and
women from all age groups.
The age dependence curves of the two steroids differed in
men and in women. For EpiTe in men, a similar age curve
(Fig. 1) as that reported previously (10) was obtained, with
a maximum at around 35 yr. In women the maximum was
found about 15 yr earlier, and an increasing trend appeared
after the 60th year. Stepwise age-related changes (Fig. 2)
revealed significantly higher levels in males with the excep-
tion of the prepubertal period and senescence (after the 60th
year). With the exception of the prepubertal levels, circulat-
ing T levels in adult males are 5 to 10 times higher than those
of EpiTe, but in females, from the end of puberty until se-
nescence, the levels of both steroids are close to each other,
and in prepubertal girls and postmenopausal women EpiTe
is even prevalent (Fig. 3). In both sexes a sharp decline of the
EpiTe/T ratio before puberty was found. The sex differences
between corresponding age groups were significant (Fig. 4).
Concerning pregnenolone sulfate, the prepubertal and
adult levels of this steroid in women were close to those
reported by de Peretti and Mappus (15), with a maximum
near the 30th year, but in males only an indistinct maximum
was found about the 30th year, followed by a slight decline
until 55 yr, after which the decline becomes distinct, as re-
ported by others (16).
This study aimed to contribute to the question of whether
EpiTe might play the role of an endogenous antiandrogen.
The relationship of circulating EpiTe to pregnenolone sulfate
may add some information as to whether in men and women
the contribution of the adrenals and gonads to the production
of EpiTe is similar or whether these sources undergo some
changes during the lifespan. Taking into account the fact that
EpiTe is bound to AR with one-tenth to one-third affinity as
T (5, 22), it may be concluded that in women in reproductive
or postmenopausal age, EpiTe could only marginally coun-
teract the effect of circulating androgens. However, in pre-
pubertal children of both sexes, it might function as a factor
of hormonal homeostasis. Questions remain as to the role of
EpiTe formation and accumulation in target tissues (e.g. in
the prostate in which it may be involved in the regulation of
intracellular hormone levels and may also provide a contri-
bution to the cellular response).
Another question is whether the mechanism of EpiTe for-
mation proposed by Weusten et al. (13) for EpiTe production
in testicular tissue is the only one prevailing in humans or
whether other biosynthetic mechanisms are operating in
other tissues. Weusten et al. (13) have suggested a mechanism
of EpiTe biosynthesis not involving 17
␣
-hydroxylase/
FIG. 7. Polynomial regression of the age dependencies of the EpiTe/
pregnenolone sulfate ratio in the serum of 183 women aged 10–70 yr
(A) and 175 men aged 4–69 yr (B). Because of the skewed data
distribution on the y-axis, the original data were transformed to
minimum skewness of the studentized residuals. The curves of the
mean prediction (the solid line), the 95% confidence interval (the
dashed lines closer to the mean prediction), and the 95% confidence
intervals of prediction (the dashed lines further from the mean pre-
diction) were obtained by retransformation of the results to the orig-
inal scale. All of the parameters of the polynomial were significant (t
tests). r, correlation coefficient of the multiple regression; p, level of
statistical significance of the model; m, degree of polynomial; n, num-
ber of subjects (the value in parentheses representing the number
of outliers and high leverage points excluded from computation);
, power of the transformation (
⫽ 0 denotes logarithmic trans-
formation).
FIG. 8. Age and sex differences in the EpiTe/pregnenolone sulfate
ratio in the serum of 183 women aged 10–70 yr (empty circles) and 175
men aged 4 – 69 yr (solid circles). The circles with error bars represent
group mean values with 95% confidence intervals that were calcu-
lated using LSD multiple comparison. Overlapping of the confidence
intervals denotes statistical insignificance between individual groups
and vice versa. As confirmed using two-way ANOVA with sex and age
group as the first and the second factors, respectively, the age dif-
ferences were highly significant (P ⬍ 0.0001). The sex differences
were nonsignificant, but the difference in the shape of age dependence
(age/sex interaction) was highly significant (P ⬍ 0.0001).
Havlı´kova´ et al. • Age, Sex, and Epitestosterone J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 2229
C
17–20
-lyase. According to these authors, pregnenolone is
directly metabolized to 5-androstene-3

,17
␣
-diol, which in
turn serves as a substrate for 3

-hydroxysteroid dehydro-
genase/⌬
4,5
-isomerase, thus yielding EpiTe. This hypothesis
has been indirectly supported by the discovery of a tight
correlation between EpiTe and 5-androstene-3

,17
␣
-diol lev-
els in both peripheral and spermatic blood in men (13).
A different situation is seen when the concentrations of
EpiTe and its precursor pregnenolone sulfate are correlated.
When the EpiTe/pregnenolone sulfate ratio was plotted as
a function of age (Fig. 7) or stepwise age-related changes
were calculated (Fig. 8), quite different patterns were ob-
tained for men and women: In men, the ratio displayed a
slight but significant increasing tendency, but in women a
U-shape dependence was found.
Circulating pregnenolone sulfate is mostly if not entirely
of adrenal origin (23). The major proportion of EpiTe in men
is believed to be formed in the testis (13), even though a
recent report has demonstrated that in hypogonadal (but not
in eugonadal) men, it responded to ACTH, indicating adre-
nal participation (11). These facts were a further reason for
considering pregnenolone sulfate as well as EpiTe and T to
find out whether some relationship between the levels of
pregnenolone sulfate and EpiTe does exist.
Although in men a positive correlation between EpiTe and
pregnenolone sulfate was found, no such a correlation could
FIG. 9. Correlation between EpiTe and
pregnenolone sulfate in the serum of 180
women (A and B) and 174 men (C and D)
aged 4 –70 yr. Because of the skewed data
distribution on both axes violating the as-
sumption of a Gaussian distribution in the
data to be correlated, the data were trans-
formed by logarithmic transformation be-
fore correlation. The principal axis (the solid
line) and the 95% confidence ellipsoid (the
dashed line) obtained (A and C) were re-
transformed to the original scale (B and D).
r, Pearson’s correlation coefficient; p, level of
statistical significance of the model; n, num-
ber of subjects.
2230 J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 Havlı´kova´ et al. • Age, Sex, and Epitestosterone
be demonstrated in women (Fig. 9). No definite conclusion
may be drawn from these results. However, it may be spec-
ulated that EpiTe in women is produced by a different mech-
anism than it is in men or that it is derived by the same
mechanism but is metabolized differently. The absence of a
correlation between pregnenolone sulfate and EpiTe levels in
women excludes a higher contribution by the adrenals. The
different mechanism of EpiTe formation means that EpiTe is,
at least partially, a by-product of the classical route of an-
drogen biosynthesis, including P450C17
␣
action. The inter-
conversion of T to EpiTe via androstenedione and conse-
quent 17
␣
-reduction, which can be seen in some species,
must also be taken into account. However, Thijssen et al. (24)
have demonstrated that in men the peripheral conversion of
labeled androstenedione and labeled T to EpiTe can account
for less than 5% of the total urinary excretion of EpiTe.
In a study of the regulation of the P450C17
␣
enzyme in the
direction toward 17
␣
-hydroxylation (typical of adrenals) or
C
17–20
-cleavage operating predominantly in the testis, Miller
et al. (25) pointed to the importance of the availability of
reducing equivalents, which depend on electron-donating
partners formed by other cooperating enzymes occurring in
the respective tissue. Our previous finding that EpiTe acts as
a competitive inhibitor of both activities in the human testis
(6) indicates that EpiTe is, at least partially, a by-product of
T biosynthesis in this tissue.
In conclusion, the authors’ measurements have demon-
strated that the concentrations of EpiTe and pregnenolone
sulfate are age dependent and that at least in prepubertal
boys and girls, EpiTe concentration approaches, or even sur-
passes, those of T, thus leaving it free to seek its role in
hormonal homeostasis. In both sexes, the dissimilarities in
the course of EpiTe levels during life, and its relation to
pregnenolone sulfate concentrations, raise the question of the
contribution of the adrenals and gonads to the production of
both steroids and even of the uniformity of the mechanism
of EpiTe formation.
Acknowledgments
We express our sincere thanks to Karel Paca´k, M.D., D.Sc. (NIH,
Bethesda, MD), for his valuable help in preparation of the manuscript.
Received January 29, 2001. Accepted January 24, 2002.
Address all correspondence and requests for reprints to: Martin Hill,
Ph.D., Institute of Endocrinology, Na´rodnı´ 8, 116 94 Praha 1, Czech
Republic. E-mail: mhill@endo.cz.
This work was supported by Grant 5398-3 from the Internal Grant
Agency of the Czech Ministry of Health.
References
1. Clark LC, Kochakian CD 1947 The in vitro metabolism of testosterone to
4-androstenedione-3,17, cis-testosterone and other steroids by rabbit liver
slices. J Biol Chem 170:23–33
2. Nuck BA, Lucky AW 1987 Epitestosterone: a potential new antiandrogen.
J Invest Dermatol 89:209 –211
3. Monsalve A, Blaquier JA 1977 Partial characterization of epididymal 5 alpha
reductase in the rat. Steroids 30:41–51
4. Sta´rka L, Bicˇı´kova´ M, Hampl R 1989 Epitestosterone—an endogenous anti-
androgen? J Steroid Biochem 33:1019 –1021
5. Bicˇı´kova´ M, Hampl R, Sta´rka L 1992 Epitestosterone—a potent competitive
inhibitor of C21-steroid side chain cleavage in the testis. J Steroid Biochem Mol
Biol 43:721–724
6. Bicˇı´kova´ M, Hampl R, Hill M, Sta´rka L 1993 Inhibition of steroid 17 alpha-
hydroxylase and C17,20-lyase in the human testis by epitestosterone. J Steroid
Biochem Mol Biol 46:515–518
7. Lapcˇı´k O, Perheentupa A, Bie`ı´kova´ M, Huhtaniemi I, Hampl R, Sta´rka L 1994
The effect of epitestosterone on gonadotrophin synthesis and secretion. J
Endocrinol 143:353–358
8. Bicˇı´kova´ M, Kanceva R, Lapeˇı´k O, Hill M, Sta´rka L 1993 The effect of
epitestosterone on the plasma levels of LH and FSH in ovariectomized im-
mature rats. J Steroid Biochem Mol Biol 45:321–324
9. Choi MH, Yoo YS, Chung BC 2001 Biochemical roles of testosterone and
epitestosterone to 5alpha-reductase as indicators of male-pattern baldness.
J Invest Dermatol 116:57– 61
10. Lapcˇı´k O, Hampl R, Hill M, Sta´rka L 1995 Plasma levels of epitestosterone
from prepuberty to adult life. J Steroid Biochem Mol Biol 55:405–408
11. Kicman AT, Coutts SB, Cowan DA, Handelsman DJ, Howe CJ, Burring S,
et al. 1999 Adrenal and gonadal contributions to urinary excretion and plasma
concentration of epitestosterone in men—effect of adrenal stimulation and
implications for detection of testosterone abuse. Clin Endocrinol (Oxf) 51:
661–668
12. Hill M, Hampl R, Petrˇı´k R, Sta´rka L 1996 Concentration of the endogenous
antiandrogen epitestosterone and androgenic C19-steroids in hyperplastic
prostatic tissue. Prostate 28:347–351
13. Weusten JJ, Legemaat G, van der Wouw MP, Smals AG, Kloppenborg PW,
Benraad T 1989 The mechanism of the synthesis of 16-androstenes in human
testicular homogenates. J Steroid Biochem 32:689– 694
14. Dehennin L 1993 Secretion by the human testis of epitestosterone, with its
sulfoconjugate and precursor androgen 5-androstene-3beta,17alpha-diol. J Ste-
roid Biochem Mol Biol 44:171–177
15. de Peretti E, Mappus E 1983 Pattern of plasma pregnenolone sulfate levels in
humans from birth to adulthood. J Clin Endocrinol Metab 57:550 –556
16. Morley JE, Kaiser F, Raum WJ, Perry III HM, Flood JF, Jensen J, Silver AJ,
Roberts E 1997 Potentially predictive and manipulatable blood serum corre-
lates of aging in the healthy human male: progressive decreases in bioavailable
testosterone, dehydroepiandrosterone sulfate, and the ration of insulin-like
growth 1 to growth hormone. Proc Natl Acad Sci USA 94:7537–7542
17. Bı´lek R, Hampl R, Putz Z, Sta´rka L 1987 Radioimmunoassay of epitestos-
terone: methodology, thermodynamic aspects and applications. J Steroid Bio-
chem 28:723–729
18. Hampl R, Dvorˇa´k P, Lukesˇova´ S, Koza´k I, Chrpova´ M, Sta´rka L 1978 The use
of iodinated steroid as radioligand for testosterone radioimmunoassay. J Ste-
roid Biochem 9:771–773
19. Bieˇı´kova´ M, Sza´mel I, Hill M, Tallova´ J, Sta´rka L 2001 Allopregnanolone,
pregnenolone sulfate, and epitestosterone in breast cyst fluid. Steroids 66:55–57
20. Meloun M, Militky´ J, Forina M 1994 Chemometrics for analytical chemistry.
vol. 2. PC-aided regression and related methods. Chichester: Horwood
21. Cook RD, Weisberg S 1999 Applied regression including computing and
graphics. London: John Wiley & Sons
22. Verhoeven G, Heyns W, De Moor P 1975 Testosterone receptors in the pros-
tate and other tissues. Vitam Horm 33:265–242
23. McKenna TJ, Jennings AS, Liddle GW, Burr IM 1976 Pregnenolone, 17-OH-
pregnenolone, and testosterone in plasma of patients with congenital adrenal
hyperplasia. J Clin Endocrinol Metab 42:918–925
24. Thijssen JHH, de Boer D, van Heeswijk R, Donker GH 1996 Peripheral
production of epitestosterone by metabolism of androstenedione and testos-
terone. Abstracts PP3– 488 of the 10th International Congress of Endocrinol-
ogy, June 14 –15, 1996, Program and Abstracts Vol II, p 876
25. Miller WL, Auchus RJ, Geller DH 1997 The regulation of 17,20 lyase activity.
Steroids 62:133–142
Havlı´kova´ et al. • Age, Sex, and Epitestosterone J Clin Endocrinol Metab, May 2002, 87(5):2225–2231 2231
