Clinical Endocrinology (2005)
62
, 498–503 doi: 10.1111/j.1365-2265.2005.02252.x
498
© 2005 Blackwell Publishing Ltd
ORIGINAL ARTICLE
Blackwell Publishing, Ltd.
Serum levels of sex hormone-binding globulin (SHBG) are not
associated with lower levels of non-SHBG-bound testosterone
in male newborns and healthy adult men
Willem de Ronde*
,
†, Yvonne T. van der Schouw‡, Frank H. Pierik§
,
††, Huibert A. P. Pols*
,
¶, Majon Muller‡,
Diederick E. Grobbee‡, Louis J. G. Gooren†, Rob F. A. Weber†† and Frank H. de Jong*
Departments of
*
Internal Medicine,
§
Public Health,
¶
Epidemiology and Biostatistics and
††
Andrology, Erasmus Medical Center,
Rotterdam,
†
Department of Endocrinology, Vrije Universiteit Medical Center, Amsterdam and
‡
Julius Center for Health Sciences
and Primary Care, University Medical Center, Utrecht, the Netherlands
Summary
Objective
It is generally accepted that SHBG decreases the bio-
availability and activity of testosterone (T). In
in vitro
experiments
increased levels of SHBG will be associated with decreased levels of
non-SHBG bound testosterone (non-SHBG-T). However,
in vivo
SHBG can alter both production and clearance rates and thus plasma
levels of T.
Design and patients
In order to study the effect of SHBG on
the levels of non-SHBG-T
in vivo
in the presence of an active
hypothalamo–pituitary–gonadal (HPG) axis we conducted a cross
sectional study in 400 healthy adult men with an age range of 40–
80 years and in 106 newborn boys.
Measurements
In both groups, regression coefficients (
β
) and par-
tial correlation coefficients (r) were calculated for the relationship
between SHBG and T or non-SHBG-T. Adult men were divided into
age groups per decade (40–50 years, 51–60 years, 61–70 years and
71–80 years) to study possible differences in the impact of SHBG on
the level of non-SHBG-T throughout ageing.
Results
Higher levels of SHBG were associated with higher levels
of total testosterone in neonates (
β
= 0·02
±
0·004,
r
= 0·44,
P
< 0·001) but not with non-SHBG-T (
β
=
−
0·001
±
0·001,
r
= 0·05,
P
= 0·52). In adult men there was a significant age related increase
in levels of SHBG and an age-related decrease of both total and
non-SHBG-T. Higher SHBG was strongly associated with higher total
testosterone in all age groups (
β
= 0·26, 0·26, 0·26 and 0·23 for 40–
50 years, 51–60 years, 61–70 years and 71–80 years, respectively,
P
< 0·001 for all age groups). Higher SHBG was not or only slightly
associated with higher non-SHBG-T
β
= 0·02 (
P =
0·32),
β
= 0·04
(
P =
0·03),
β
= 0·04 (
P =
0·02) and
β
= 0·02 (
P =
0·16) for 40–
50 years, 51–60 years, 61–70 years and 71–80 years, respectively.
Conclusions
In contrast to general belief, SHBG levels barely influ-
ence levels of non-SHBG-bound testosterone both in male newborns
and healthy adult men: the influence, if any, is positive. Consequently
the age related increase of SHBG does not account for the age related
decline in non-SHBG-T in healthy adult men.
(Received 3 August 2004; returned for revision 20 October 2004;
finally revised 12 January 2005; accepted 7 February 2005)
Introduction
Sex-hormone binding globulin (SHBG), corticosteroid binding
globulin (CBG) and albumin are important steroid hormone bind-
ing proteins in human plasma. In normal adult men approximately
44% of the circulating testosterone is specifically bound to SHBG,
50% is nonspecifically bound to albumin and 3·5% is bound to
CBG implicating that only 2–3% is unbound or free.
1
By binding
testosterone, SHBG decreases the metabolic clearance rate of
testosterone
2,3
and decreases peripheral conversion of testosterone to
androstenedione.
2
To exert its genomic action, testosterone must
bind to an intracellular androgen receptor. SHBG withholds bound
hormone to diffuse from the bloodstream as free hormone to this
receptor. There is some controversy whether or not albumin-bound
testosterone can dissociate freely from this carrier protein and
enter tissues;
4,5
however, non-SHBG bound testosterone (non-
SHBG-T = free + albumin bound testosterone) and non-SHBG-
nonalbumin bound testosterone (= free testosterone) are extremely
well-correlated
6
and interchangeable in most cases.
The levels of SHBG vary widely between healthy men and depend
on factors such as diet,
7
body mass index (BMI), insulin levels and
age.
8
On the basis of simple mathematics a higher level of SHBG is
expected to result in a lower level of non-SHBG-T. However, when
the non-SHBG-T is considered bioactive, variations in SHBG
concentrations will cause a variable secretion of LHRH by the
hypothalamus and LH secretion by the pituitary. Low levels of
non-SHBG-T will then lead to increased LH and testicular testoster-
one production via a decreased negative feedback signal on the
hypothalamus and pituitary. Estimation of the effect of SHBG on the
levels of non-SHBG-T
in vivo
by simply applying the laws of mass
Correspondence: Frank H. de Jong Ph.D., Endocrine Laboratory, room Ee
516, Department of Internal Medicine, Erasmus MC, PO Box 1738, 3000 DR
Rotterdam, the Netherlands. E-mail: f.h.dejong@erasmusmc.nl
SHBG and non-SHBG-bound testosterone in newborns and adult men
499
© 2005 Blackwell Publishing Ltd,
Clinical Endocrinology
,
62
, 498–503
action might thus be inappropriate in men with a functional
hypothalamo–pituitary–gonadal (HPG) axis.
In healthy males, two phases of life are characterized by high
gonadal axis activity; the neonatal period and the period during and
after puberty.
9
In the first 6 months of life, levels of both LH and T
are high, reaching levels in the low adult range.
10
At the age of
6 months, both LH and T levels have decreased to very low levels
only to increase at the start of puberty.
11
In adult men, T levels remain
high throughout life although ageing is associated with a slight
decrease of the levels of total serum testosterone and an increase of
the levels of SHBG.
12
To study the effect of SHBG on the levels of non-SHBG-T
in vivo
we conducted a cross-sectional study in 400 healthy adult men with
an age range of 40–80 years and in 106 newborn boys.
Subjects and methods
Neonates
Serum samples were collected from 112 boys with ages between 1
and 6 months who served as controls in a case–control study on
cryptorchism and hypospadias. Controls had been randomly
selected from the population of boys that visited Child Healthcare
Centers (CHCs). In the Netherlands, CHCs are notified of live births
within 2 days after registration in the municipal birth register. CHCs
invite all parents to participate free of charge in the national preven-
tive child health care programme, including growth monitoring and
vaccination, in which 95% of all parents participate with their child.
This study was approved by the ethical committee of the Erasmus
Medical Center, Rotterdam. Six subjects were excluded for analysis
because of missing data on SHBG (five patients) or age (one patient).
Eventually 106 subjects were available for analysis.
Laboratory measurements in neonates
Testosterone levels were estimated using a nonextraction coated tube
radioimmunoassay (Coat-a-Count, Diagnostic Products Corpora-
tion, Los Angeles, CA, USA). The lower limit of detection of the assay
was 0·1 nmol/l. Intra- and interassay coefficients of variation were
below 9% for the concentration range measured in these samples.
SHBG concentrations were measured using a chemoluminescence-
based immunometric method (Immulite 2000) from the same sup-
plier. The lower limit of detection was 5 nmol/l. Variations for this
method were below 7%. Non-SHBG bound T was calculated using
the method described by Sodergard
et al
.
13
using a fixed albumin
level of 40 g/l. The formulas for these calculations have been
described earlier.
14
Adult subjects
The study is a cross-sectional, single-centre study in 400 independ-
ently living men aged 40–80 years. The study was originally designed
to study the relationships between endogenous sex hormones and
risk factors or manifestations of chronic diseases. The subjects were
recruited in two ways. First, by asking female participants of other
studies conducted by the department whether they knew a possibly
interested male volunteer. Invitation letters were sent to 770 female
participants. Eventually, 240 men volunteered for participation. Sec-
ond, names and addresses of a randomly selected male population
aged 40–80 years were drawn from the municipal register of Utrecht,
a large-sized town in the central part of the Netherlands. 1230 invi-
tation letters were send. From this group, 390 men volunteered for
participation.
The subjects who did not live independently or were not physically
or mentally able to visit the study centre independently (
n
= 16) were
excluded. No additional health related eligibility criteria were used.
Of the remaining 614 men, 400 men were randomly selected to par-
ticipate. One subject was excluded from analysis because of clear
hypogonadism (total testosterone = 0·24 nmol/l). To obtain equal
numbers in each age-decade we sampled 100 men in each decade of
age. Data collection took place between March 2001 and April 2002.
All participants gave written informed consent before enrolment
and the institutional review board of the Utrecht University Medical
Centre approved the study. Height and weight were measured in
standing position without shoes. BMI was calculated as the weight
in kilograms divided by the square of height in metres. Details on
lifestyle and health of the subjects have been published earlier.
15
Laboratory measurements in adult subjects
Fasting blood samples were obtained by venepuncture. Cell free
serum was immediately stored at
−
20
°
C. T was measured after dieth-
ylether extraction using an in house radioimmunoassay employing
a polyclonal antiT-antiserum (AZG 3290 Dr JJ Pratt, Groningen,
the Netherlands). The lower limit of detection of the assay was
0·24 nmol/l and interassay variation was 6·0, 5·4, and 8·6% at 2·1,
5·6 and 23 nmol/l, respectively. Results using this assay were com-
parable to those obtained with the assay used for estimation of T in
neonates (
r
= 0·91,
n
= 37, slope of regression line 1·03, intercept on
y
-axis 0·37 nmol/l). SHBG was measured as described for the
measurement in neonates. Non-SHBG bound T was calculated as
described above.
Data analysis
All variables, except for age in neonates, were normally distributed.
In neonates, age was logarithmically transformed to obtain nor-
mality. Associations between SHBG, total and non-SHBG-bound
testosterone were assessed and tested for significance using linear
regression analysis with SHBG as the independent variable and
expressed as regression coefficient (
β
) or (partial) correlation co-
efficient (r). The linear regression coefficient
β
indicates the change
of the dependent variable for every 1 nmol/l change in SHBG. With
every regression analysis, residuals were checked for normality. In
both groups, analyses were repeated with adjustments for age (days)
in neonates, and adjustments for age (years) and BMI in adult men.
Adjustments for age and BMI were made by adding these parameters
as independent variables to the regression and correlation model.
Adjustments were made because both age and BMI were shown to
be associated with levels of SHBG and T in men.
10,15
Adult men were
studied as a group and after division into age groups per decade
(40–50 years, 51–60 years, 61–70 years and 71–80 years). Effect
500
W. de Ronde
et al.
© 2005 Blackwell Publishing Ltd,
Clinical Endocrinology
,
62
, 498–503
modification by age was tested using a linear regression model in
which BMI, age group, SHBG and an interaction term (SHBG*age
group) were used as independent variables and total testosterone
or non-SHBG-T as dependent variables. Analyses were done using
SPSS 11·0 software (SPSS, Chicago, IL, USA).
Results
The characteristics of newborns and adult men are presented in
Ta ble 1. In newborns levels of SHBG were much higher and levels
of both total and unbound testosterone were much lower compared
to adults. Higher levels of SHBG were significantly associated with
higher levels of total testosterone in neonates but levels of SHBG
were not significantly associated with levels of non-SHBG-T (Un-
adjusted, for T:
β
= 0·03
±
0·005,
r
= 0·49,
P
< 0·001; for non-SHBG-T:
β
= 0·001
±
0·001,
r
= 0·09,
P
= 0·35. After adjustment for age (days),
for T:
β
= 0·02
±
0·004,
r
= 0·45,
P
< 0·001; for non-SHBG-T:
β
= 0·001
±
0·001,
r
= 0·06,
P
= 0·52). The relations between SHBG
and total or unbound testosterone in neonates are depicted in Fig. 1.
A similar pattern was seen in adult men (Fig. 1; Unadjusted, for T:
β
= 0·22
±
0·02,
r
= 0·59,
P
< 0·001; for non-SHBG-T:
β
= 0·01
±
0·01,
r
= 0·05,
P
= 0·35). After adjustment for age and BMI the
slightly positive association between SHBG and non-SHBG-T was
statistically significant (for T:
β
= 0·25
±
0·01,
r
= 0·68,
P
< 0·001; for
non-SHBG-T:
β
= 0·03
±
0·01,
r
= 0·15,
P
= 0·01).
The characteristics of adult men after division in age groups are
presented in Table 2. There was a significant age-related increase in
levels of SHBG and a significant age-related decrease of both total
and non-SHBG-T (
P <
0·001 for all variables). Relationships
between SHBG and total or unbound testosterone after adjustments
for age and BMI are presented in Table 3 and Fig. 2. Higher levels of
SHBG were strongly associated with higher levels of total testoster-
one in all age groups. Levels of SHBG were not or marginally
positively associated with levels of non-SHBG-T. There was no
modifying effect of age on the associations between SHBG and total
testosterone (
P
for interaction = 0·36) or non-SHBG-T (
P
for
interaction = 0·94).
Discussion
This study in a large number of neonates and men across a wide age
range confirms that levels of SHBG can vary widely among individ-
uals.
8
Our results show that there is a strong and positive relation
between SHBG and total testosterone levels in both newborn boys
and adult men. Moreover there is no, or only a weak positive asso-
ciation between SHBG levels and levels of circulating non-SHBG-T.
These associations are completely different compared to the rela-
tionships anticipated on the basis of mathematical models.
1
In these
models, SHBG does not have an effect on the level of total testoster-
one. However,
in vivo
SHBG can have an effect on clearance and
Ta b le 1. Levels of (non-SHBG-bound) testosterone and SHBG in newborns
and adult men (mean ± SD)
Newborns Adult men
n 106 399
Age (days/years) 79·1 ± 27·3 60·2 ± 11·3
SHBG (nmol/ l) 118 ± 43 40·6 ± 14·5
Testosterone (nmol/l) 3·73 ± 2·30 18·6 ± 5·26
Non-SHBG-T (nmol/l) 0·97 ± 0·53 10·4 ± 2·62
Fig. 1 SHBG vs. total testosterone (open symbols)
and non-SHBG-T (closed symbols) in newborn
boys and adult men.
Age group (years) 40–50 51–60 61–70 71–80
n 100 100 100 99
Age (years) 45·0 ± 3·12 56·2 ± 2·89 65·4 ± 2·77 74·4 ± 2·69
SHBG (nmol/ l)*** 34·7 ± 13·7 38·0 ± 13·3 43·6 ± 15·0 46·1 ± 13·2
Testosterone (nmol/l)*** 20·2 ± 5·59 18·5 ± 5·22 17·9 ± 5·53 17·8 ± 4·32
Non-SHBG-T (nmol/l)*** 12·2 ± 2·75 10·6 ± 2·37 9·57 ± 2·42 9·29 ± 1·79
***P < 0·001 for differences between age groups.
Ta b le 2. Levels of (non-SHBG-bound)
testosterone and SHBG in adult men by age
decade (mean ± SD)
SHBG and non-SHBG-bound testosterone in newborns and adult men 501
© 2005 Blackwell Publishing Ltd, Clinical Endocrinology, 62, 498–503
might also affect production rates of testosterone. It can be hypoth-
esized that the SHBG-induced decreased bioavailability of testoster-
one will result in increased production of LH and subsequently
testosterone until a new equilibrium is reached. SHBG decreases the
metabolic clearance rate of testosterone in men and women.
2,3
Higher levels of SHBG will thus be associated with higher levels of
testosterone if the testosterone production rate does not change.
However, in pre- and postmenopausal women, in whom testosterone
is not directly involved in the regulation of the HPG axis, plasma
testosterone levels were shown to be negatively correlated with SHBG
levels,
16
which argues against an important effect of SHBG on tes-
tosterone levels by effects on testosterone clearance alone. These
observations support the conclusion that non-SHBG-T and not total
testosterone is driving HPG axis activity in men.
The fact that in male neonates SHBG is not negatively associated
with non-SHBG-T suggests that during the first months of life the
HPG axis is functional and sensitive to androgen feedback. This is
supported by the observation that castration of neonatal monkeys
results in elevated LH levels comparable to those found in castrated
adults.
17
In fact, androgen sensitivity appears to be a prerequisite for
neonatal HPG axis activity as the postnatal hormone surge is absent
in infants with androgen insensitivity.
18
The absent relation between
SHBG and non-SHBG-T in our subjects implies that the observed
testosterone levels are not just the result of an unleashed HPG axis
but are restrained by the feedback inhibition of non-SHBG-T. This
is supported by the observation that LH and/or testosterone secretion
in neonates can be further increased when stimulated with GnRH
or human chorionic gonadotropin.
19
Although the HPG axis is operative
in neonates, circulating (non-SHBG) T levels are much lower than
in adults. It appears that the higher SHBG levels found in neonates are
not responsible for the relatively low levels of non-SHBG-T. A dif-
ferent setpoint of the HPG axis in neonates might thus be postulated.
Ta b le 3. Multiple linear regression coefficients (β) and partial correlation coefficients (r) for the relation between SHBG, testosterone and non-SHBG-bound
testosterone in adult men for the total population (n = 399) and by age decade
Total population Age group (years)
β ± SE r
40–50
β ± SE r
51–60
β ± SE r
61–70
β ± SE r
71–80
β ± SE r
T (nmol/ l) 0·25 ± 0·02 0·68*** 0·26 ± 0·03 0·64*** 0·26 ± 0·03 0·66*** 0·26 ± 0·03 0·69*** 0·23 ± 0·03 0·66***
Non-SHBG-T (nmol/l) 0·03 ± 0·01 0·15* 0·02 ± 0·02 0·10NS 0·04 ± 0·02 0·20* 0·04 ± 0·02 0·24* 0·02 ± 0·01 0·15NS
***P < 0·001, *P < 0·05, NS, not significant. All associations have been adjusted for age and BMI.
Fig. 2 SHBG vs. total testosterone (open symbols)
and non-SHBG-T (closed symbols) in four age
groups of adult men.
502 W. de Ronde et al.
© 2005 Blackwell Publishing Ltd, Clinical Endocrinology, 62, 498–503
Our study confirms the well-known increase of SHBG and slight
decrease of total testosterone with increasing age in adult men.
12
This
age related increase of SHBG is generally believed to be an important
factor in the pronounced decline of non-SHBG-T in ageing men.
However, in our cross-sectional analysis of different age groups we
found that even in the oldest age group higher levels of SHBG are
not associated with lower levels of non-SHBG-T. Therefore, the age-
related increase of SHBG does not appear to be responsible for the
age-related decline of non-SHBG-T.
Several studies indicate that with age the sensitivity to the negative
feedback action of sex hormones increases.
20–22
Also, in our adult
men, there seems to be a change of the setpoint of the HPG axis to
testosterone feedback with ageing. The reason for this increased feed-
back sensitivity remains unknown. There are no arguments to
believe that the expression of androgen action is stronger with age;
the extent of SHBG suppression by androgens is similar in young
and aged men.
21
Ageing appears to be associated with combined hypothalamic and
gonadal defects. However, the pituitary of elderly men responds ade-
quately to stimulation by exogenous LHRH,
23
but the Leydig cell
responsiveness to LH is decreased.
24
With ageing, alterations in
hypothalamic GnRH secretion are noted; decreased LH burst ampli-
tude, increased LH burst frequency and a disordered LH burst pat-
tern. Androgen deprivation reveals impaired GnRH secretory reserve
in older men compared to young men.
25
Our study shows that these
defects of ageing do not cause an inadequate adaptation of testoster-
one concentrations to variations in SHBG levels.
In neonates, it appears that the HPG axis is even more sensitive
to androgen feedback than the HPG axis of elderly men. The mech-
anism behind this is unknown. The response of the HPG axis to var-
ying SHBG levels shows that the level of non-SHBG-T in newborns
and adults is not solely a result of randomly combined SHBG and
testosterone levels but the product of the HPG axis responding to
negative and positive feedback signals. The precise nature and
impact of these factors remain to be determined.
Our study has certain limitations. The adult volunteers for this
study were all living independently. Especially in the highest age
groups this might have led to a health selection bias.
The levels of testosterone in the neonates were low in comparison
to adult men. Because commercial testosterone assays are designed
to measure reliably in the adult range it is well known that they loose
accuracy in the lower ranges.
26
For several reasons we believe that
our conclusions drawn from the neonatal data are justified. The
method used in our study for the determination of testosterone
showed the closest correlation with values determined by liquid
chromatography-tandem mass spectrometry.
26
The intra- and inter-
assay coefficients of variation were below 9% for the concentration
range measured in the samples and the mean testosterone level in
our study was similar to levels reported by others.
10
The studied popu-
lation was large. Moreover, this study was not designed to make
statements about individual hormone levels but to analyse relations
between parameters in a group of individuals. The finding of a highly
significant association between levels of total testosterone and SHBG
in both neonates and adult men underlines the validity of our esti-
mates; if the estimates for the testosterone concentration in neonates
were highly inaccurate such a relation would not have been found.
The approach used in this study allows us to make statements
about the setpoint of the HPG axis of a group of men but cannot
be translated to individuals. For instance, it is well known that all
kinds of diseases are associated with an impaired function of the
HPG axis.
27
When the HPG axis is compromised not only by age but
also by other factors the capability to adapt to testosterone binding
by SHBG might be lost. In those cases, SHBG may become a con-
tributor to low non-SHBG-T levels in men.
The cross-sectional nature of this study inherently does not allow
us to make definitive causative statements about the observed cor-
relations between SHBG and sex hormones.
In conclusion, the results of our study show that in male neonates
and in healthy adult men levels of SHBG, if at all, barely contribute
to variations in non-SHBG-T.
References
1Dunn, J.F., Nisula, B.C. & Rodbard, D. (1981) Transport of steroid
hormones: binding of 21 endogenous steroids to both testosterone-
binding globulin and corticosteroid-binding globulin in human
plasma. Journal of Clinical Endocrinology and Metabolism, 53, 58–68.
2Vermeulen, A. & Ando, S. (1979) Metabolic clearance rate and inter-
conversion of androgens and the influence of the free androgen frac-
tion. Journal of Clinical Endocrinology and Metabolism, 48, 320–326.
3 Saez, J.M., Forest, M.G., Morera, A.M. & Bertrand, J. (1972) Meta-
bolic clearance rate and blood production rate of testosterone and
dihydrotestosterone in normal subjects, during pregnancy, and in
hyperthyroidism. Journal of Clinical Investigation, 51, 1226–1234.
4Mendel, C.M. (1989) The free hormone hypothesis: a physiologically
based mathematical model. Endocrine Reviews, 10, 232–274.
5Manni, A., Pardridge, W.M., Cefalu, W., Nisula, B.C., Bardin, C.W.,
Santner, S.J. & Santen, R.J. (1985) Bioavailability of albumin-bound
testosterone. Journal of Clinical Endocrinology and Metabolism, 61,
705–710.
6Vermeulen, A., Verdonck, L. & Kaufman, J.M. (1999) A critical eval-
uation of simple methods for the estimation of free testosterone in
serum. Journal of Clinical Endocrinology and Metabolism, 84, 3666–
3672.
7Longcope, C., Feldman, H.A., McKinlay, J.B. & Araujo, A.B. (2000)
Diet and sex hormone-binding globulin. Journal of Clinical Endo-
crinology and Metabolism, 85, 293–296.
8Vermeulen, A., Kaufman, J.M. & Giagulli, V.A. (1996) Influence of
some biological indexes on sex hormone-binding globulin and
androgen levels in aging or obese males. Journal of Clinical Endo-
crinology and Metabolism, 81, 1821–1826.
9Mann, D.R. & Fraser, H.M. (1996) The neonatal period: a critical
interval in male primate development. Journal of Endocrinology, 149,
191–197.
10 Andersson, A.M., Toppari, J., Haavisto, A.M., Petersen, J.H., Simell,
T., Simell, O. & Skakkebaek, N.E. (1998) Longitudinal reproductive
hormone profiles in infants: peak of inhibin B levels in infant boys
exceeds levels in adult men. Journal of Clinical Endocrinology and
Metabolism, 83, 675–681.
11 Belgorosky, A. & Rivarola, M.A. (1986) Progressive decrease in
serum sex hormone-binding globulin from infancy to late prepu-
berty in boys. Journal of Clinical Endocrinology and Metabolism, 63,
510–512.
12 Harman, S.M., Metter, E.J., Tobin, J.D., Pearson, J. & Blackman, M.R.
(2001) Longitudinal effects of aging on serum total and free testosterone
