| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Department of Physiology, Faculty of Medicine, Dentistry, and Health Sciences, University of Melbourne (J.P.), Victoria, Australia; Lipid Research Center (J.-P.D.) and the Diabetes Research Unit (A.N.), CHUL Research Center and Laval University, and the Physical Activity Sciences Laboratory, Laval University (J.G., A.T., C.B.), Quebec, Canada G1K 7P4; and the Department of Medicine, University of Vermont (A.T.), Burlington, Vermont 05405
Address all correspondence and requests for reprints to: Claude Bouchard, Ph.D., Physical Activity Sciences Laboratory, Division of Kinesiology, Department of Social and Preventive Medicine, Faculty of Medicine, Pavillon de l’Education Physique et des Sports, Laval University, Ste-Foy, Quebec, Canada G1K 7P4. E-mail: claude.bouchard{at}kin.msp.ulaval.ca
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
,17ß-diol glucuronide (3-DIOL-G), and cortisol were
measured in 12 pairs of young, sedentary, male monozygotic twins before
and after 100 days of overfeeding. The dietary energy excess of 4.2
MJ/day (1000 Cal), 6 days a week, resulted in a total positive energy
balance of 353 MJ (84,000 Cal). Overfeeding induced significant changes
(P < 0.0001) in body weight and other measures of
body composition. Within-twin pair resemblance was observed at baseline
in all steroids, except cortisol [intraclass correlation range:
DHEA-S, 0.50 (P < 0.05); DHT, 0.77
(P < 0.001)] and was lost with overfeeding,
except for DHT and SHBG (P < 0.05). SHBG levels
fell and 3-DIOL-G rose with the gain in body fatness. The change in
testosterone was a significant correlate of the change in upper body
fat (r = -0.48; P < 0.05). The change in
3-DIOL-G correlated positively with increases in all measures of
central adiposity (r = 0.52; P < 0.01). A
decrease in DHEA-S occurred with a higher, but not with a lower, gain
in abdominal visceral fat (P < 0.05). Thus,
analysis of adrenal and gonadal steroids and of conjugated metabolites
before and after overfeeding in monozygous twins supports the idea that
there is a genotype effect on steroid circulating steroid levels and
that these blood levels are correlated with the pattern of body fat
distribution. Moreover, the baseline within-twin pairs similarity in
steroid levels was attenuated by prolonged positive energy balance and
body fat gain. |
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Dehydroepiandrosterone (DHEA), which has been found to be inversely
associated with body fatness, has also been described as an antiobesity
agent in rodents (9, 10). Androstenediol (
5-DIOL) was recently
reported as the best single steroid correlate of body fatness and
abdominal fat deposition in men (11). Sex hormone-binding globulin
(SHBG) and testosterone (TESTO) have been shown to be reduced (8) in
men and elevated (12) in women with increasing central adiposity. In
men, plasma estrone, estradiol, and 3-DIOL and its glucuronide
metabolite (3-DIOL-G) were positively associated (13, 14), and DHEA
and its sulfated metabolite (DHEA-S) (15, 16) were negatively
associated with total and regional body fatness. The high levels of
cortisol found in obese women (4) and men (17) have been postulated to
induce an insulin-related lipid-accumulating effect (18).
In an overfeeding study of Vermont state prisoners (17), cortisol was the only adrenal steroid measured. The observed increase was not significant when adjusted for the gain in body weight (19). No other overfeeding studies have reported the effect of altered body composition on gonadal or adrenal steroid and conjugated metabolite levels.
Interest in the influence of genes on blood steroid levels has arisen from reports based on twin and family studies. From a cross-sectional study of adrenal steroid levels in twins, Meikle and colleagues quantified the genetic influence on plasma steroid concentrations and tissue production (20). Familial factors were found to account for more than 50% of the variation in plasma hormone levels in identical twins, with between 1–76% accounted for by genetic effects on specific hormones. Whether these relationships are maintained when body composition phenotypes are controlled for has not been reported.
Previous reports of the Quebec long term overfeeding study of 12 pairs of monozygous twins have described the changes in body weight, fat distribution, lean body mass, adipose tissue lipolysis, energy expenditure, and plasma levels of thyroid hormones, glucose, insulin, and glucagon (21, 22, 23, 24, 25, 26, 27). Here, further analysis of the data collected in this intervention study was undertaken to describe the changes in adrenal and gonadal steroids and their associations with changes in body mass and composition. An important goal was to test the hypothesis that individual differences in steroid levels in response to chronic overfeeding are compatible with a genetic model.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Twelve pairs of young, sedentary, male twins (aged 21 ± 2 yr) gave written consent to participate in an overfeeding study approved by the Laval University medical ethics committee and the Office for the Protection from Research Risks of the NIH (Bethesda, MD). The specific aims, study design, and methodology have been previously described in detail (21, 22).
The twins were established as monozygous and had no recent illness or history of obesity, diabetes, hyperlipidemia, hypertension, or endocrinopathy. They were fed an energy surplus for 6 days/week for 100 days while maintaining a sedentary, controlled level of energy expenditure. Subjects were housed at the Laval University campus for 120 days under 24-h supervision, such that diet and physical activity were closely monitored to ensure compliance. The first 14 days constituted a baseline period involving 3 days of testing and the establishment of the energy cost of weight maintenance while subjects ate freely from prepared diets. During the intervention, the subjects consumed a total excess energy of 353 MJ (84000 Cal) above baseline maintenance energy needs on 84 of the 100 days [4.2 MJ (1000 Cal)/day]. The nutrient contribution to the energy content of the diet was 50% carbohydrate, 35% lipid, and 15% protein. The subjects’ schedule of sedentary activities included playing video games, reading, playing cards, watching television, and walking 30 min/day. There was a postoverfeeding testing period of 6 days.
Testing before and after the overfeeding period
Dietary analysis. All of the foods selected and the plate waste remaining during the baseline period were recorded and weighed for each subject, and the nutrient and energy composition were derived from a computerized analysis using the Canadian food composition tables (28) to establish energy required for weight maintenance. During the overfeeding period, daily energy intakes were monitored, and top-up portions were provided to ensure that individual energy intakes provided the excess 4.2 MJ (1000 Cal)/day required.
Body composition. Observations before and after overfeeding as well as daily measurements during the intervention involved body weight measured at the same time daily with subjects wearing light exercise shorts, body density determined before and after overfeeding by underwater weighing (29) using the helium dilution technique to measure pulmonary residual volume (30) and the Siri equation to estimate percent body fat (31), fat mass and fat-free mass obtained from percent body fat and body weight (kilograms), and skinfold thickness measurements (millimeters) at 10 sites [5 trunk (subscapular, suprailiac, abdominal, midaxillary, and chest) and 5 extremities (biceps, triceps, front midthigh, suprapatellar, and medial calf)] according to standardized procedures (32). Computed tomography (CT) was performed before and after overfeeding with a Siemens Somatom DRH scanner (Erlangen, Germany) according to the method described by Sjöström et al. (33) to determine abdominal visceral fat, abdominal subcutaneous (sc) fat, total abdominal fat, and total femoral fat areas (square centimeters).
Biochemical analyses. Blood samples were obtained 24 h after termination of the overfeeding treatment after an overnight fast between 0730–0800 h in the morning for the determination of plasma steroid levels and for fasting glucose, insulin, and glucagon levels. Plasma insulin levels were determined by RIA, as described by Oppert et al. (27). The insulin assay reliability was analyzed in a batch at completion of the study, and the coefficient of variation was 10.8%. The steroids were measured by RIA after separation of conjugated and unconjugated steroids by C18 column chromatography, as described previously by Bélanger et al. (34). Sulfate derivatives were submitted to hydrolysis. Glucuronide conjugates were also submitted to hydrolysis with ß-glucuronidase. Steroids from each fraction were further separated by elution on LH-20 columns. Levels of the steroids were measured by RIA as previously described (35). Plasma SHBG levels were measured by the direct immunoradiometric method using a commercial kit from Farmos Diagnostic (Turku, Finland).
Among all the steroids measured, six have been identified as having
high assay reproducibility. Their analytical errors ranged between
6.6–11.9% (except for 3-DIOL-G and ADT-G with 19.4% and 21.5%,
respectively) and an intraclass correlation coefficient for repeated
assays greater than 0.96 (except for 3-DIOL-G and ADT-G with
coefficients of 0.80% and 0.81%, respectively; Gagnon, J., et al.,
personal communication). The day to day variability for three of the
six steroids ranged from 8.8–13.5% (3-DIOL-G and ADT-G had
coefficients of variation of 22.3% and 23.4%, respectively), with
intraclass correlation coefficients above 0.94 (except for 3-DIOL-G
and ADT-G with coefficients of 0.73 and 0.77). Cortisol had a high day
to day variation (26.0%), but its analytical error was low (6.6%)
with an intraclass correlation coefficient for repeated assays of
0.98.
The baseline plasma levels of these steroids were within the normal range for adult men (36). The postoverfeeding level of TESTO of one subject was 5 SD above the mean for all subjects. It was removed from the statistical analysis to avoid an excessive contribution to the results from this particular individual. Its removal did not alter the significance of changes in TESTO or the direction of correlations. One testosterone postoverfeeding level was also missing.
Statistical analysis. The effects of overfeeding on the body fat phenotypes and the interaction between genotype- and intervention-induced changes were assessed with a two-way ANOVA for repeated measures on one factor (time) (21). The twins were considered nested within the pair, whereas the treatment effect was considered a fixed variable. The intraclass correlation coefficient was computed from the between-pairs and within-pairs means of squares and was used to quantify the similarity within pairs for plasma adrenal and gonadal steroids. Correlation analysis was undertaken to estimate the association between overfeeding-induced changes in body fat with 1) the preoverfeeding (baseline) plasma adrenal and gonadal steroid levels, and 2) the overfeeding-induced changes in plasma adrenal and gonadal steroid levels, with 24 subjects (22 for TESTO) considered as independent individuals. Analyses were conducted with and without adjustment for changes in total fat mass whenever appropriate. Statistical analyses were performed with the SAS statistical package (version 6.12 for Windows, SAS Institute, Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Changes in body weight and measures of body composition with
overfeeding in the Quebec Overfeeding Study have been reported
previously (21). Overfeeding induced significant changes
(P < 0.0001) in body weight, measures of body
composition, and fat distribution (Table 1
).
|
Effects of overfeeding. Plasma levels of the sex hormone
transport protein SHBG fell significantly (15.9 ± 1.6%;
P < 0.0001), whereas levels of 3-DIOL-G rose
(41.3 ± 9.7%; P < 0.01). There was no change in
DHEA-S, TESTO, DHT, ADT-G, or cortisol (Table 2). After adjustment for change in fat
mass, the changes in SHBG and 3-DIOL-G were no longer significant,
and the change in ADT-G became statistically significant
(P < 0.05). The percent changes in the steroids after
overfeeding are illustrated in Fig. 1.
|
|
shows
the within-twin pair resemblance for the absolute plasma levels of the
steroids before and after overfeeding. The similarity within twin pairs
at baseline was highest for SHBG and DHT levels (intraclass
coefficient, 0.76 and 0.77, respectively; P < 0.001).
There was more than 7 times the variance between pairs as within pairs
for these steroids. Concentrations of DHEA-S, TESTO, ADT-G, and
3-DIOL-G also showed significant within-pair resemblance. For these
steroids, the variance between pairs was 2.7- to 3.8-fold higher than
that within pairs. Cortisol was the only steroid that showed no
significant within-pair resemblance at baseline.
|
. The within-twin pair resemblance in the response of plasma
TESTO, as an example of a steroid not exhibiting twin-pair similarity,
and of plasma DHT (the only steroid characterized by an intrapair
resemblance and response) levels to the overfeeding treatment is
illustrated in Fig. 2.
|
Body composition. The associations between body composition
phenotypes and plasma steroid levels were investigated by calculating
the correlation coefficients with all men considered as independent
subjects before and after overfeeding, with and without adjustment for
the changes in fat mass after the intervention. Positive correlations
were found between baseline TESTO and baseline body weight (r =
0.40) and between baseline 3-DIOL-G and body weight and fat mass
(r = 0.52 for both), whereas correlations were negative between
baseline DHEA-S and body weight (r = -0.59) and fat mass (r
= -0.46) and between baseline ADT-G and body weight (r = -0.41;
Table 4a). Baseline cortisol showed a
negative correlation (r = -0.57) with change in body weight with
overfeeding, whereas the baseline level of 3-DIOL-G was positively
correlated with the postoverfeeding changes in body weight and fat mass
(r = 0.52 for both; Table 4b). Changes in levels of cortisol and
ADT-G were negatively correlated with the change in body weight with
overfeeding (r = -0.48 and r = -0.46, respectively).
Conversely, the change in 3-DIOL-G was positively correlated with
the change in fat mass with overfeeding (r = 0.52; Table 4c).
|
-DIOL-G, which correlated positively with changes in total
abdominal fat and trunk skinfolds, retained significance after
adjustment for fat mass (Table 4b; TESTO with trunk to extremity ratio
of skinfolds, r = -0.52; TESTO with CT assessed visceral fat,
r = -0.44; cortisol with trunk skinfolds, r = -0.43).
Correlations between changes in plasma steroid levels with changes in
body fat distribution phenotypes after overfeeding were significant and
negative for TESTO with visceral fat (r = -0.46), for cortisol
with abdominal sc and visceral fat (r = -0.49 and r =
-0.52, respectively), for DHEA-S with trunk to extremity skinfold
ratio (r = -0.48), and for ADT-G with abdominal sc fat (r =
-0.48; Table 4c). Conversely, correlations were positive for
3-DIOL-G with abdominal sc fat and total abdominal fat (r =
0.51 and r = 0.44, respectively). The change in femoral fat was
negatively correlated with the change in ADT-G (r = -0.41) and
positively with the change in 3-DIOL-G (r = 0.59). It is
noteworthy that only the correlations of TESTO and 3-DIOL-G retained
significance after adjustment for total fat gain.
Postoverfeeding levels of TESTO showed significant negative correlation
with the postoverfeeding CT measure of abdominal visceral fat (r =
-0.46) retained after adjustment for fat mass (Fig. 3a). Similarly, plasma levels of DHEA-S
after overfeeding were inversely correlated with abdominal sc fat
(r = -0.42, with adjustment for fat mass; Fig. 3b), whereas
postoverfeeding levels of 3-DIOL-G showed a positive correlation
with the postoverfeeding skinfold trunk to extremity ratio (r =
0.52, with adjustment for fat mass; Fig. 3c). Plasma SHBG was not
correlated with body fat phenotypes at baseline or after overfeeding or
with postoverfeeding values.
|
-DIOL-G was in positive correlation with total and regional fat
measures before and after overfeeding.
Individual subjects were subdivided into high gainers and low
gainers of body fat phenotypes (n = 6 in each group, without two
members of the same pair) in an attempt to assess differences with
respect to baseline steroid levels or changes with overfeeding. High
fat gainers had lower postoverfeeding change in TESTO levels than low
fat gainers (P < 0.01). ADT-G showed a similar pattern
(P < 0.05), whereas 3-DIOL-G differences almost
reached the significance level (P < 0.06), but DHEA-S
levels did not. This relationship is shown in Fig. 4a. Subjects were also subdivided on the
basis of high and low abdominal visceral fat gainers with overfeeding.
The postoverfeeding changes in DHEA-S in the high visceral fat gainers
were significantly different from those in the low gainers
(P < 0.01). There was no difference between these two
subgroups for the overfeeding-induced changes in TESTO, ADT-G, and
3-DIOL-G (Fig. 4b).
|
-DIOL-G, but was negatively
correlated with the changes in TESTO with overfeeding (r = -0.48,
adjusted for change in fat mass). A two-way ANOVA for repeated measures
on time in fasting insulin and TESTO levels indicated a significant
interaction effect (F ratio = 22.7; P < 0.0001
after adjustment for change in fat mass). Those subjects with higher
baseline insulin showed a greater overfeeding-induced decrease in
TESTO (r = -0.54; Fig. 5). However,
the correlation between the changes in insulin and the changes in TESTO
with overfeeding was not significant (r = 0.27, after adjustment
for change in fat mass).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
-DIOL-G were all
characterized by intraclass coefficients of about 0.50 or better.
With overfeeding, the twin resemblance remained significant for SHBG
and DHT. These results suggest that genetic differences play a role in
the overfeeding-associated changes in plasma steroid levels. A high
genetic effect on the plasma levels of 3-DIOL-G and a moderate
influence on TESTO and DHT have been reported in a previous study of MZ
and DZ twins (20). Earlier work with nontwin families had identified
significantly less variability within brother siblings than among
nonbrothers for SHBG and TESTO, but not DHT (37), although fathers and
sons in the same sample of families were significantly correlated for
DHT levels (38). A genetic effect has also been described for blood
levels of DHEA-S in a study of steroid concentrations in 26 families
(39). The analysis of the baseline steroid levels in the Quebec
Overfeeding Study confirms these earlier reports. Overfeeding substantially decreased the twin resemblance, particularly for those steroids with high baseline intrapair resemblance. Increases in measures of abdominal fat were correlated with decreasing TESTO levels even after adjustment for fat mass. Although there was high within-pair resemblance in abdominal fat after overfeeding (21), with the exception of DHT, the within-twin pair resemblance for the steroids, which were also highly correlated with abdominal fat measures, was lost after overfeeding. It is as if the putative genotype effect on steroid levels observed at baseline (from the intrapair resemblance) was disrupted or overwhelmed by the overfeeding or obscured by the body mass, body fat, and abdominal fat gains.
The present results also confirm previously reported relationships
between SHBG, DHEA-S, TESTO, and 3-DIOL-G steroids and body fatness
(7, 8, 11, 13, 15). In a cross-sectional study of middle-aged men,
Seidell and colleagues (8) found that overweight men with high levels
of body fat and visceral fat had lower concentrations of SHBG and
TESTO. Accordingly, in the Quebec Overfeeding Study, SHBG levels fell
in proportion to the gain in fat mass. Overall TESTO and DHEA-S levels
did not change with overfeeding, but they were well correlated with the
postoverfeeding body fat levels. In a cross-sectional analysis of 1241
randomly sampled middle-aged US men, Field et al. (15)
showed decreased plasma adrenal steroid and sex hormone levels with
increasing relative weight. Pasquali et al. (7) confirmed
this relationship in a comparison of obese men with normal weight
controls. In the Quebec Overfeeding Study, those who gained the least
amount of body fat did not decrease their TESTO or ADT-G levels.
Similarly, those who gained most abdominal fat had greater increases in
plasma 3-DIOL-G and greater decreases in plasma DHEA-S.
In a multivariate analysis of TESTO and the adrenal
C19 steroid precursor 5-DIOL in a cross-sectional study
of a sample of 80 middle-aged men, a greater association between
steroid hormones and total fatness than between steroids and visceral
fat accumulation was reported (11). The intervention design of the
Quebec Overfeeding Study provided new insights into the relationships
of the sex steroids to body fat distribution. Changes in TESTO were a
significant correlate of changes in abdominal fat. A fall in DHEA-S was
accompanied by a gain in abdominal visceral fat. The increase in
3-DIOL-G levels with overfeeding was highly correlated with all
measures of total and abdominal fat. Hence, the present study confirmed
the positive associations recently described in a cross-sectional study
of obese men (13). It also identified the baseline level of this
steroid as a strong correlate to the changes in body mass, fat mass,
and trunk skinfold thickness. However, the proposal of an association
of elevated cortisol with an increased lean body mass accompanying the
gain in body weight (1, 4) was not supported by the present study.
In men, secretion of TESTO and DHT from the testis contributes
approximately 40–50% of active DHT, whereas a significant
contribution to DHT formation comes from synthesis within the adrenal
gland and peripheral conversion of sex hormone precursors such as
DHEA-S (40). On the other hand, glucuronidation has been postulated as
the major pathway for steroid elimination (34). Major glucuronide
steroid conjugates in men (ADT-G and 3-DIOL-G) have no intrinsic
activity but, as they are more soluble, return to plasma for excretion
through the urine. It has been suggested that glucuronide levels are
markers of androgen metabolism (34, 41). The decrease in the sulfated
DHEA precursor of active steroids with increasing adiposity suggests
either a decline in the activity of the adrenal steroid synthetic
pathway or a higher metabolic clearance of this water-soluble steroid
with higher body fatness. The negative correlation of TESTO and the
positive correlation of 3-DIOL-G with increasing adipose tissue with
overfeeding is in accordance with the proposal that the enzymatic
control of active metabolism of TESTO may occur in adipose tissue (42).
This correlation is also concordant with the recent finding of the
expression of UDP-glucuronosyltransferase enzymes responsible for
metabolism and elimination of androgens in omental and sc abdominal
adipose tissue (43).
Hyperinsulinemia has been described as an endocrine correlate of high levels of body fat (44) and is consistently correlated with specific adrenal steroid levels (10). Variation in insulin levels in response to overfeeding in the present set of twins is characterized by a significant intrapair resemblance (27). It has also been postulated that hyperinsulinemia may be involved in the regulation of steroid metabolism (7). TESTO, which is known to stimulate lipid mobilization and to have an anabolic effect on lean body mass, has been reported to be negatively associated with the hyperinsulinemia and the insulin resistance state accompanying abdominal obesity in men (8, 14). The present study extends these findings to the demonstration of a relationship between the baseline insulin level and both the changes in TESTO and the changes in abdominal fat under the influence of overfeeding.
It should be pointed out that the subjects in the Quebec Overfeeding Study were young lean adults. Thus, even after a mean weight gain of about 8 kg, these young men were still normal weight at the end of the overfeeding period, with a mean body mass index of 22.4 kg/m2 and a low level of CT-measured visceral fat (average area, 58 cm2). Thus, the correlations reported herein are derived from men at the low end of the adiposity spectrum. It cannot be excluded that an overfeeding study conducted in overweight patients may have yielded different correlation patterns with the changes in body fat or abdominal fat. This study cannot rule out the hypothesis that some of the observed plasma steroid changes are actually associated in part with the overconsumption of calories rather than with the changes in body mass, body composition, or fat topography. It would be useful to design a study to specifically address the latter issue.
In conclusion, the long term Quebec Overfeeding Study in identical
twins has identified a significant within-twin pair resemblance for
C19 steroids and conjugated metabolites before the
overfeeding intervention. The twin resemblance was, however, attenuated
with overfeeding for all steroids except DHT. DHT retained its
within-twin pair resemblance after overfeeding independently of the
gain in fat mass. Overfeeding-induced changes in fat mass and fat
distribution were related inversely to cortisol, DHEA-S, and TESTO, but
positively to 3-DIOL-G. The change in plasma TESTO was a significant
correlate of abdominal fat gain with overfeeding. The study has also
confirmed an inverse relationship between hyperinsulinemia and TESTO
with body fat gain. The pattern of plasma adrenal and gonadal steroid
levels before and after overfeeding in these monozygous twins suggests
that the genotype could be an important determinant of the sex hormones
that influence the pattern of fat distribution in healthy young men.
However, this apparent genetic determinism can be overcome by prolonged
positive energy balance and the ensuing gain in body fat.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received April 2, 1998.
Revised May 27, 1998.
Accepted June 9, 1998.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
,17ß-diol glucuronide as a steroid correlate of
visceral obesity in men. J Clin Endocrinol Metab. 82:1528–1534.-reduced steroid glucuronides
in hyperandrogenic and idiopathic hirsute women. Fertil Steril. 48:948–953.[Medline]
5-4 isomerase (3ß-HSD)
and 17ß-hydroxysteroid dehydrogenase (17ß-HSD) in adipose tissue. Int J Obesity. 15:91–99.
This article has been cited by other articles:
 |
|
 |
|
  I. Ouellet-Morin, M. Boivin, G. Dionne, S. J. Lupien, L. Arsenault, R. G. Barr, D. Perusse, and R. E. Tremblay Variations in Heritability of Cortisol Reactivity to Stress as a Function of Early Familial Adversity Among 19-Month-Old Twins Arch Gen Psychiatry, February 1, 2008; 65(2): 211 - 218. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
