0021-972X/91/7201-0083$02.00/0 Journal of Clinical Endocrinology and Metabolism Copyright © 1991 by The Endocrine Society
Vol. 72, No. 1 Printed in U.S.A.
A Direct Effect of Hyperinsulinemia on Serum Sex Hormone-Binding Globulin Levels in Obese Women with the Polycystic Ovary Syndrome* JOHN E. NESTLER, LINDA P. POWERS, DENNIS W. MATT, KENNETH A. STEINGOLD, STEPHEN R. PLYMATE, ROGER S. RITTMASTER, JOHN N. CLORE, AND WILLIAM G. BLACKARD Departments of Medicine (J.E.N., L.P.P., J.N.C., W.G.B.) and Obstetrics and Gynecology (D.W.M., K.A.S.), Medical College of Virginia/Virginia Commonwealth University, Richmond, Virginia 23298; the Department of Clinical Investigation (S.R.P.), Madigan Army Medical Center, Tacoma, Washington 98431; and the Division of Endocrinology, Halifax Infirmary (R.S.R.), Halifax, Nova Scotia, Canada
ABSTRACT. To determine whether hyperinsulinemia can directly reduce serum sex hormone-binding globulin (SHBG) levels in obese women with the polycystic ovary syndrome, six obese women with this disorder were studied. Before study, ovarian steroid production was suppressed in each woman by the administration of 7.5 mg of a long-acting GnRH agonist, leuprolide depot, im, on days —56, —28, and 0. This resulted in substantial reductions in serum concentrations of testosterone (from 1.72 ± 0.29 nmol/L on day -56 to 0.32 ± 0.09 nmol/L on day 0), non-SHBG-bound testosterone (from 104 ± 16 pmol/L on day —56 to 19 ± 5 pmol/L on day 0), androstenedione (from 7.25 ± 1.65 nmol/L on day -56 to 2.78 ± 0.94 nmol/L on day 0), estrone (from 371 ± 71 pmol/L on day -56 to 156 ± 29 pmol/ L on day 0), estradiol (from 235 ± 26 pmol/L on day —56 to 90 ± 24 pmol/L on day 0), and progesterone (from 0.28 ± 0.12 nmol/L on day -56 to 0.08 ± 0.02 nmol/L on day 0). Serum SHBG levels, however, did not change (18.8 ± 2.8 nmol/L on day -56 vs. 17.8 ± 2.6 nmol/L on day 0).
P
While continuing leuprolide treatment, the women were administered oral diazoxide (300 mg/day) for 10 days to suppress serum insulin levels. Diazoxide treatment resulted in suppressed insulin release during a 100-g oral glucose tolerance test (insulin area under the curve, 262 ± 55 nmol/min • L on day 0 vs. 102 ± 33 nmol/min • L on day 10; P < 0.05) and deterioration of glucose tolerance. Serum testosterone, androstenedione, estrone, estradiol, and progesterone levels did not change during combined diazoxide and leuprolide treatment. In contrast, serum SHBG levels rose by 32% from 17.8 ± 2.6 nmol/L on day 0 to 23.5 ± 2.0 nmol/L on day 10 (P < 0.003). Due primarily to the rise in serum SHBG levels, serum non-SHBG-bound testosterone levels fell by 43% from 19 ± 5 pmol/L on day 0 to 11 ± 4 pmol/L on day 10 (P = 0.05). These observations suggest that hyperinsulinemia directly reduces serum SHBG levels in obese women with the polycystic ovary syndrome independently of any effect on serum sex steroids. {J Clin Endocrinol Metab 72: 83-89, 1991)
the hyperandrogenism of PCO (9). Both in vitro (10) and in vivo evidence (11-15) suggest that hyperinsulinemia contributes to hyperandrogenism by augmenting ovarian androgen production. An additional mechanism by which hyperinsulinemia might lead to hyperandrogenism is the reduction of serum sex hormone-binding globulin (SHBG) levels. Since testosterone is highly bound to SHBG, a decrease in serum SHBG concentration would increase the tissue availability of circulating testosterone. Several lines of evidence suggest that hyperinsulinemia indeed reduces serum SHBG levels. In vitro studies indicate that insulin suppresses SHBG production by cultured hepatoma cells (16). Furthermore, epidemiological studies demonstrate an inverse correlation between serum levels of insulin and SHBG in obese MexicanAmerican women (17) and obese healthy women (18). These inverse correlations are independent of serum
OLYCYSTIC ovary syndrome (PCO) is a poorly understood disorder, characterized clinically in its purest form by menstrual dysfunction, hirsutism, infertility, obesity, and polycystic ovaries. Notably, while not all women with PCO manifest all of these features, the frequency of obesity is high (1), and insulin resistance with hyperinsulinemia has been repeatedly demonstrated in both nonobese and obese women with this disorder (2-8). Hyperinsulinemia may play a key role in producing Received June 13, 1990. Address all correspondence and requests for reprints to: John E. Nestler, M.D., Division of Endocrinology and Metabolism, Medical College of Virginia, MCV Station, Box 111, Richmond, Virginia 232980111. * This work was supported in part by NIH Grants RR-00065 and DK-18904, a Research Award from the Virginia Affiliate of the American Diabetes Association, and the Thomas F. and Kate Miller Jeffress Memorial Trust. 83
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
NESTLER ET AL.
84
androgens or degree of adiposity. To date, however, no prospective study exists demonstrating a direct action of hyperinsulinemia on serum SHBG levels in vivo. This study was designed to test the hypothesis that hyperinsulinemia directly reduces serum SHBG levels in obese women with PCO. To accomplish this, diazoxide was employed to inhibit insulin release, and serum SHBG levels were measured before and after 10 days of diazoxide administration. To eliminate the influence of sex steroids on serum SHBG levels, ovarian steroid hormone production was suppressed by the administration of leuprolide, a GnRH agonist, for 56 days before and concurrently during the study. Finally, recent evidence suggests that hyperinsulinemia, by its action as a growth factor, might further heighten hyperandrogenism at the tissue level by stimulating tissue 5a-reductase activity, thereby increasing the conversion of testosterone to its biologically more active metabolite dihydrotestosterone (19). To examine this possibility as well, we monitored serum levels of 3a:androstanediol glucuronide, a dihydrotestosterone metabolite, as an index of 5o:-reductase activity (20). Materials and Methods
JCE & M • 1991 Vol 72 • No 1
remaining women were not taking any medications. All had normal serum glucose responses to a standard 3-h 100-g oral glucose tolerance test [National Diabetes Data Group criteria (23)]. The study was approved by the Committee on the Conduct of Human Research of the Medical College of Virginia, and signed informed consent was obtained from each subject. To suppress ovarian steroid hormone production, each woman received 7.5 mg leuprolide acetate depot (Lupron Depot, TAP Pharmaceuticals, North Chicago, IL), a long-acting GnRH agonist, im, on days —56, —28, and 0. The study began on day 0, and each woman took 100 mg diazoxide (Proglycem, Medical Market Specialties, Boonton, NJ), orally, three times daily (300 mg/day) for 10 days from days 0-10 to inhibit insulin release. On days —56, —28, 0, and 10 the women were admitted to the General Clinical Research Center of the Medical College of Virginia after an overnight fast. Weight and height were recorded, and a heparin lock was inserted into a hand vein for blood withdrawal. Blood samples were drawn at 0800, 0815, and 0830 h, and the sera were pooled for determination of serum SHBG, sex steroid, gonadotropin, insulin, and glucose concentrations. On days —56, 0, and 10 the women drank 100 g glucose (Glucola, Whitaker General Medical, Richmond, VA) at 0830 h, and blood samples were drawn every 15 min from 0830-1130 h for determination of serum glucose and insulin levels.
Study design
Assays
Six obese women with PCO were entered into and completed the study. Each woman had oligo- or amenorrhea, hirsutism, and elevated serum free testosterone levels. The characteristics of the women are given in Tables 1 and 2. Although subjects 2 and 3 had serum LH/FSH ratios greater than 2 on initial screening (Table 1), this was not evident in basal blood collected on day —56 (Table 2). Nonclassical congenital adrenal hyperplasia was excluded in four women by determination of serum 17-hydroxyprogesterone and 17-hydroxypregnenolone levels either in the basal state (21) (subjects 4-6) or 1 h after stimulation with cosyntropin (Cortrosyn, Organon, West Orange, NJ; subject 2) (22). Subject 1 was taking nadolol, hydrochlorothiazide, and triamterene for hypertension, and subject 3 was taking propranolol for hypertension. These subjects continued their medications throughout the study period. The
Serum glucose concentrations were determined by the glucose oxidase method (Beckman Glucose Analyzer 2, Fullerton, CA). Serum insulin levels were determined by RIA (24). The intra- and interassay coefficients of variation for the insulin assay were 5.0% and 8.0%, respectively. The integrated insulin response during the oral glucose tolerance test was analyzed by calculating the area under the curve for insulin (AUCinsuHn) during the 3 h after glucose ingestion by the trapezoidal rule using absolute values. Serum testosterone, androstenedione, estrone, 17/3-estradiol, and progesterone concentrations were measured by RIA, as previously described (15, 25). Serum (1.0 mL) was extracted with diethyl ether, and steroids were isolated by Celite column chromatography. Highly specific antisera were used to measure testosterone, androstenedione, estrone, 170-estradiol, and progesterone. 3H-Labeled steroids were added to serum before extraction to correct for procedural losses. The antiserum to testosterone was kindly provided by Dr. Gordon Niswender, Department of Physiology, Colorado State University (Fort Collins, CO), and the antisera to androstenedione, 17/3-estradiol, and progesterone by Dr. John Resko, Department of Physiology, University of Oregon School of Medicine (Portland, OR). Antiserum to estrone was purchased from Steranti Research Ltd. (London, England). Serum dehydroepiandrosterone sulfate (DHEA-S) levels were determined using a commercial antibody-coated tube kit (Diagnostic Products, Los Angeles, CA). The intraassay coefficients of variation for all of the above steroid assays were less than 8.0%. To eliminate interassay variation, which ranges between 10-15%, all samples for each steroid were determined in a single assay. Serum 3a-
TABLE
1. Characteristics of the six obese women with PCO who were
studied o u- j. Subject no. 1 2 3 4 5 6
A Body mass Age . , . . index (yr) ., . ,, (kg/nr) 26 23 32 27 25 24
34.3 35.6 40.5 31.2 38.9 27.5
TT. Serum TLH/ . FSH >2
o
TOTT
No
Polycystic . ,, . . , Acanthosis ovaries by . . ., , nigncans ultrasound Yes
Yes
No
No
No
Yes Yes
Yes Yes
No No No
Yes
Yes
Yes
Body mass index values were determined on day —56. Pelvic ultrasound examinations were not performed in subjects 2 and 5. The presence of acanthosis nigricans was assessed visually.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
INSULIN AND SHBG
85
TABLE 2. Basal biochemical parameters on day —56 of the six obese women with PCO who were studied Subject no.
Testosterone (nmol/L)
Normal range (mean ± SE)°
Non-SHBG-bound SHBG Testosterone Androstenedione (nmol/L) (pmol/L) (nmol/L)
2.58 1.82 0.84 2.51 1.32 1.23
19.7 18.4 28.1 24.0 12.0 10.5
144 105 30 114 114 120
9.38 5.02 1.99 5.30 13.58 8.25
0.71 ± 0.07
75.0 ± 9.9
13 ± 4
13.53 ± 0.25
Estrone (pmol/ L) 650 374 187 494 290 232
Estradiol DHEA-S (pmol/ L) L) 354 220 232 227 164 212
376 ± 40 279 ± 30
4.87 3.13 0.88 1.95 11.24 2.80
LH (IU/L)
FSH (IU/L)
Insulin (pmol/ L)
11.5 117 17.7 9.7 20.5 25.5
7.1 7.8 12.0 10.2 9.1 12.5
149 165 101 106 151 176
11.6 ± 0.9 10.0 ± 0.5 115 ± 19
" Adapted from data obtained in five nonobese normal women on two occasions separated by 28 days during the follicular phase of the menstrual cycle (n = 10 for all values except insulin, for which n = 5) (30).
androstanediol glucuronide concentrations were determined by RIA of androstanediol after enzymatic hydrolysis of the etherextracted serum, followed by Celite chromatography of the resulting unconjugated steroids (26). The intra- and interassay coefficients of variation for the 3a-androstanediol glucuronide assay were 3.2% and 9.0%, respectively. Serum SHBG concentrations were determined by an immunoradiometric assay with materials obtained from Farmos Diagnostica (Oulunsalo, Finland). This assay does not crossreact with any other identified circulating serum proteins, including albumin, transcortin, T4-binding globulin, and transferrin. The immunoradiometric assay correlates with the dihydrotestosterone binding assay for a functional SHBG with an r value of 0.96 (16). The intraassay coefficient of variation was 3.5%, and all samples were measured in a single assay to eliminate the 12% interassay variation. Non-SHBG-bound testosterone was calculated from the total molar concentrations of testosterone and SHBG, as previously described (27). Serum LH and FSH concentrations were determined by double antibody RIA (Diagnostic Products). The intraassay coefficients of variation for LH and FSH were 12.1% and 6.5%, respectively. The interassay coefficients of variation for LH and FSH were 7.2% and 5.2%, respectively. All hormone determinations from each woman were determined in a single assay to avoid interassay variation. Statistical analysis All results are reported as the mean ± SE. Results before diazoxide treatment (day 0) were compared to those after diazoxide treatment (day 10) by first testing for normality using the Wilk-Shapiro test and then employing Student's two-tailed paired t test. Analysis of variance (Dunnett's multiple comparison test) was used to compare AUCinsuiin values and gonadotropin levels at baseline (day -56) with those on days 0 and 10. Univariate analysis was performed using CLINFO, a statistical package made available by the NIH. P < 0.05 was considered significant.
Results Leuprolide treatment significantly reduced the mean serum LH level from 16.1 ± 2.5 IU/L at baseline (day
-56) to 5.4 ± 0.7 IU/L on day 0 (P < 0.05) and 4.2 ± 0.7 IU/L on day 10 (P < 0.05). Similarly, the mean serum FSH level was reduced from 9.8 ± 0.9 IU/L on day -56 to 5.6 ± 0.2 IU/L on day 0 (P < 0.05) and 6.5 ± 0.5 IU/ L on day 10 (P < 0.05). Before leuprolide treatment (day —56), glucose tolerance, as determined by the oral glucose tolerance test, was normal in all six women (Fig. 1). However, as expected in these patients, the stimulated insulin response was exaggerated (Fig. 1). Treatment with leuprolide for 56 days altered neither the glycemic nor the insulin profile during the oral glucose tolerance test (Fig. 1), and AUCinsuiin on days -56 and 0 were similar (290 ± 63 us. 262 ± 55 nmol/min-L; P = NS). In contrast, diazoxide administration uniformly inhibited insulin release during the oral glucose tolerance test and caused deterioration of glucose tolerance (Fig. 1). The AUCinsuiin after diazoxide administration (day 10) was significantly less than that on day 0 (102 ± 33 us. 262 ± 55 nmol/min-L; P < 0.05). Neither fasting serum glucose nor insulin concentrations differed significantly on days 0 and 10. All six women gained weight during diazoxide treatment (mean weight gain, 2.2 ± 0.4 kg), and four of the six women complained of diffuse bloating, palpitations, and worsened hot flashes. The women's weights returned to baseline and symptoms promptly abated after cessation of diazoxide treatment. Serum progesterone levels on day —56 were low (0.28 ± 0.12 nmol/L), consistent with the anovulatory status of the women. Leuprolide treatment resulted in an 82% decrease in serum concentrations of both total testosterone (from 1.72 ± 0.29 nmol/L on day -56 to 0.32 ± 0.09 nmol/L on day 0; Fig. 2) and non-SHBG-bound testosterone (from 104 ± 16 pmol/L on day -56 to 19 ± 5 pmol/L on day 0). Substantial reductions also occurred in serum androstenedione, estrone, estradiol, and progesterone levels, whereas serum levels of the adrenal androgen DHEA-S did not change (Table 3). In spite of
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
NESTLER ET AL.
86
14 -i
i o
JCE & M • 1991 Vol 72 • No 1
- BASELINE(DAY-56) •LEUPROLIDE (OAYO) -LEUPROLIDE a DIAZOXIDE (DAY 10)
,-rTTTl
12-
E 10-
8-
I 6 UJ CO
4-
2500o 2000 A a. ?
1500-
_i CO
-
IOOO H
£j co
500-
-56 DAY OF
30
60
90
120
150
180
MINUTES
FlG. 1. Serum insulin and glucose levels in six obese women with PCO at baseline (day —56; • ) , after 56 days of leuprolide alone (day 0; D), and after 10 days of combined treatment with diazoxide and leuprolide (day 10; • ) . Values are the mean ± SE.
declines in both ovarian androgens and estrogens, serum SHBG levels did not change and were 18.8 ± 2.8 nmol/ L on day -56 and 17.8 ± 2.6 nmol/L on day 0 (P = NS, by Student's t test; Fig. 2). The administration of diazoxide for 10 days while continuing leuprolide treatment did not alter serum testosterone levels (0.32 ± 0.09 nmol/L on day 0 vs. 0.26 ± 0.09 nmol/L on day 10; P = NS; Fig. 2) or serum androstenedione, estrone, estradiol, progesterone, or DHEA-S levels (Table 3). In contrast, serum SHBG levels rose significantly by 32% from 17.8 ± 2.6 nmol/L on day 0 to 23.5 ± 2.0 nmol/L on day 10 (P < 0.003; Figs. 2 and 3). Due primarily to the rise in serum SHBG levels, serum non-SHBG-bound testosterone levels fell by 43% from 19 ± 5 pmol/L on day 0 to 11 ± 4 pmol/L on day 10 (P = 0.05). Percent incremental changes in serum SHBG levels from days 0-10 correlated inversely with basal serum SHBG levels on day 0 (r = -0.86; P < 0.03; n = 6), indicating that those women in whom serum SHBG levels were most suppressed experienced the greatest rise. A strong positive correlation was also noted between incremental changes in serum SHBG levels from days
STUDY
FIG. 2. Serum testosterone and SHBG levels in six obese women with the polycystic ovary syndrome at baseline (day —56), after 28 and 56 days of treatment with leuprolide alone (days —28 and 0, respectively), and after 10 days of combined treatment with diazoxide and leuprolide (day 10). Values are the mean ± SE.
0-10 and insulin values on day 0 (r = 0.70; P = 0.12; n = 6), suggesting that the more insulin-resistant women experienced the greatest rise in serum SHBG levels during inhibition of insulin release with diazoxide. Leuprolide treatment resulted in a significant fall in serum 3a-androstanediol glucuronide levels from 20.3 ± 4.0 nmol/L on day -56 to 16.4 ± 3.5 nmol/L on day 0 (P < 0.02, by Student's t test). In contrast, serum 3aandrostanediol glucuronide levels before and after diazoxide administration were not statistically different (16.4 ± 3.5 nmol/L on day 0 us. 22.6 ± 7.4 nmol/L on day 10; P = NS).
Discussion Hyperinsulinemia may play a pathogenic role in the hyperandrogenism of some women with PCO (9). In support of this hypothesis, insulin resistance and elevated serum insulin levels have been well documented in women with this disorder (2-8), and basal serum insulin levels correlate directly with basal serum testosterone and androstenedione levels (2, 6). The insulin resistance of women with PCO does not appear to be due solely to the obesity often associated with this syndrome, since nonobese women with PCO also manifest hyperinsulinemic insulin resistance (7, 8). Moreover, it has recently been shown by the insulin-glucose clamp technique that
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
INSULIN AND SHBG
87
TABLE 3. Serum steroid concentrations in six obese women with PCO at baseline (day —56), after 28 and 56 days of treatment with leuprolide alone (days —28 and 0, respectively), and after 10 days of combined treatment with diazoxide and leuprolide (day 10)
Baseline (day -56) Leuprolide alone Day -28 DayO Diazoxide + leuprolide (day 10)"
Androstenedione (nmol/L)
Estrone (pmol/L)
Estradiol (pmol/L)
Progesterone (nmol/L)
DHEA-S (/umol/L)
7.25 ± 1.65
371 ± 71
235 ± 26
0.28 ± 0.12
4.14 ± 1.52
3.20 ± 1.25 2.78 ± 0.94 2.43 ± 0.76
147 ± 25 156 ± 29 158 ± 52
83 ± 13 90 ±24 85 ± 18
0.07 ± 0.02 0.08 ± 0.02 0.05 ± 0.02
3.84 ± 1.08 3.79 ± 1.12 4.28 ± 1.74
Values are the mean ± SE. ° P = NS for all steroid values on day 10 compared to those on day 0.
SERU VI SEX HORMONE (nmol/
NDING GLOB
35 -i
P
Vol. 72, No. 1 Printed in U.S.A.
A Direct Effect of Hyperinsulinemia on Serum Sex Hormone-Binding Globulin Levels in Obese Women with the Polycystic Ovary Syndrome* JOHN E. NESTLER, LINDA P. POWERS, DENNIS W. MATT, KENNETH A. STEINGOLD, STEPHEN R. PLYMATE, ROGER S. RITTMASTER, JOHN N. CLORE, AND WILLIAM G. BLACKARD Departments of Medicine (J.E.N., L.P.P., J.N.C., W.G.B.) and Obstetrics and Gynecology (D.W.M., K.A.S.), Medical College of Virginia/Virginia Commonwealth University, Richmond, Virginia 23298; the Department of Clinical Investigation (S.R.P.), Madigan Army Medical Center, Tacoma, Washington 98431; and the Division of Endocrinology, Halifax Infirmary (R.S.R.), Halifax, Nova Scotia, Canada
ABSTRACT. To determine whether hyperinsulinemia can directly reduce serum sex hormone-binding globulin (SHBG) levels in obese women with the polycystic ovary syndrome, six obese women with this disorder were studied. Before study, ovarian steroid production was suppressed in each woman by the administration of 7.5 mg of a long-acting GnRH agonist, leuprolide depot, im, on days —56, —28, and 0. This resulted in substantial reductions in serum concentrations of testosterone (from 1.72 ± 0.29 nmol/L on day -56 to 0.32 ± 0.09 nmol/L on day 0), non-SHBG-bound testosterone (from 104 ± 16 pmol/L on day —56 to 19 ± 5 pmol/L on day 0), androstenedione (from 7.25 ± 1.65 nmol/L on day -56 to 2.78 ± 0.94 nmol/L on day 0), estrone (from 371 ± 71 pmol/L on day -56 to 156 ± 29 pmol/ L on day 0), estradiol (from 235 ± 26 pmol/L on day —56 to 90 ± 24 pmol/L on day 0), and progesterone (from 0.28 ± 0.12 nmol/L on day -56 to 0.08 ± 0.02 nmol/L on day 0). Serum SHBG levels, however, did not change (18.8 ± 2.8 nmol/L on day -56 vs. 17.8 ± 2.6 nmol/L on day 0).
P
While continuing leuprolide treatment, the women were administered oral diazoxide (300 mg/day) for 10 days to suppress serum insulin levels. Diazoxide treatment resulted in suppressed insulin release during a 100-g oral glucose tolerance test (insulin area under the curve, 262 ± 55 nmol/min • L on day 0 vs. 102 ± 33 nmol/min • L on day 10; P < 0.05) and deterioration of glucose tolerance. Serum testosterone, androstenedione, estrone, estradiol, and progesterone levels did not change during combined diazoxide and leuprolide treatment. In contrast, serum SHBG levels rose by 32% from 17.8 ± 2.6 nmol/L on day 0 to 23.5 ± 2.0 nmol/L on day 10 (P < 0.003). Due primarily to the rise in serum SHBG levels, serum non-SHBG-bound testosterone levels fell by 43% from 19 ± 5 pmol/L on day 0 to 11 ± 4 pmol/L on day 10 (P = 0.05). These observations suggest that hyperinsulinemia directly reduces serum SHBG levels in obese women with the polycystic ovary syndrome independently of any effect on serum sex steroids. {J Clin Endocrinol Metab 72: 83-89, 1991)
the hyperandrogenism of PCO (9). Both in vitro (10) and in vivo evidence (11-15) suggest that hyperinsulinemia contributes to hyperandrogenism by augmenting ovarian androgen production. An additional mechanism by which hyperinsulinemia might lead to hyperandrogenism is the reduction of serum sex hormone-binding globulin (SHBG) levels. Since testosterone is highly bound to SHBG, a decrease in serum SHBG concentration would increase the tissue availability of circulating testosterone. Several lines of evidence suggest that hyperinsulinemia indeed reduces serum SHBG levels. In vitro studies indicate that insulin suppresses SHBG production by cultured hepatoma cells (16). Furthermore, epidemiological studies demonstrate an inverse correlation between serum levels of insulin and SHBG in obese MexicanAmerican women (17) and obese healthy women (18). These inverse correlations are independent of serum
OLYCYSTIC ovary syndrome (PCO) is a poorly understood disorder, characterized clinically in its purest form by menstrual dysfunction, hirsutism, infertility, obesity, and polycystic ovaries. Notably, while not all women with PCO manifest all of these features, the frequency of obesity is high (1), and insulin resistance with hyperinsulinemia has been repeatedly demonstrated in both nonobese and obese women with this disorder (2-8). Hyperinsulinemia may play a key role in producing Received June 13, 1990. Address all correspondence and requests for reprints to: John E. Nestler, M.D., Division of Endocrinology and Metabolism, Medical College of Virginia, MCV Station, Box 111, Richmond, Virginia 232980111. * This work was supported in part by NIH Grants RR-00065 and DK-18904, a Research Award from the Virginia Affiliate of the American Diabetes Association, and the Thomas F. and Kate Miller Jeffress Memorial Trust. 83
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
NESTLER ET AL.
84
androgens or degree of adiposity. To date, however, no prospective study exists demonstrating a direct action of hyperinsulinemia on serum SHBG levels in vivo. This study was designed to test the hypothesis that hyperinsulinemia directly reduces serum SHBG levels in obese women with PCO. To accomplish this, diazoxide was employed to inhibit insulin release, and serum SHBG levels were measured before and after 10 days of diazoxide administration. To eliminate the influence of sex steroids on serum SHBG levels, ovarian steroid hormone production was suppressed by the administration of leuprolide, a GnRH agonist, for 56 days before and concurrently during the study. Finally, recent evidence suggests that hyperinsulinemia, by its action as a growth factor, might further heighten hyperandrogenism at the tissue level by stimulating tissue 5a-reductase activity, thereby increasing the conversion of testosterone to its biologically more active metabolite dihydrotestosterone (19). To examine this possibility as well, we monitored serum levels of 3a:androstanediol glucuronide, a dihydrotestosterone metabolite, as an index of 5o:-reductase activity (20). Materials and Methods
JCE & M • 1991 Vol 72 • No 1
remaining women were not taking any medications. All had normal serum glucose responses to a standard 3-h 100-g oral glucose tolerance test [National Diabetes Data Group criteria (23)]. The study was approved by the Committee on the Conduct of Human Research of the Medical College of Virginia, and signed informed consent was obtained from each subject. To suppress ovarian steroid hormone production, each woman received 7.5 mg leuprolide acetate depot (Lupron Depot, TAP Pharmaceuticals, North Chicago, IL), a long-acting GnRH agonist, im, on days —56, —28, and 0. The study began on day 0, and each woman took 100 mg diazoxide (Proglycem, Medical Market Specialties, Boonton, NJ), orally, three times daily (300 mg/day) for 10 days from days 0-10 to inhibit insulin release. On days —56, —28, 0, and 10 the women were admitted to the General Clinical Research Center of the Medical College of Virginia after an overnight fast. Weight and height were recorded, and a heparin lock was inserted into a hand vein for blood withdrawal. Blood samples were drawn at 0800, 0815, and 0830 h, and the sera were pooled for determination of serum SHBG, sex steroid, gonadotropin, insulin, and glucose concentrations. On days —56, 0, and 10 the women drank 100 g glucose (Glucola, Whitaker General Medical, Richmond, VA) at 0830 h, and blood samples were drawn every 15 min from 0830-1130 h for determination of serum glucose and insulin levels.
Study design
Assays
Six obese women with PCO were entered into and completed the study. Each woman had oligo- or amenorrhea, hirsutism, and elevated serum free testosterone levels. The characteristics of the women are given in Tables 1 and 2. Although subjects 2 and 3 had serum LH/FSH ratios greater than 2 on initial screening (Table 1), this was not evident in basal blood collected on day —56 (Table 2). Nonclassical congenital adrenal hyperplasia was excluded in four women by determination of serum 17-hydroxyprogesterone and 17-hydroxypregnenolone levels either in the basal state (21) (subjects 4-6) or 1 h after stimulation with cosyntropin (Cortrosyn, Organon, West Orange, NJ; subject 2) (22). Subject 1 was taking nadolol, hydrochlorothiazide, and triamterene for hypertension, and subject 3 was taking propranolol for hypertension. These subjects continued their medications throughout the study period. The
Serum glucose concentrations were determined by the glucose oxidase method (Beckman Glucose Analyzer 2, Fullerton, CA). Serum insulin levels were determined by RIA (24). The intra- and interassay coefficients of variation for the insulin assay were 5.0% and 8.0%, respectively. The integrated insulin response during the oral glucose tolerance test was analyzed by calculating the area under the curve for insulin (AUCinsuHn) during the 3 h after glucose ingestion by the trapezoidal rule using absolute values. Serum testosterone, androstenedione, estrone, 17/3-estradiol, and progesterone concentrations were measured by RIA, as previously described (15, 25). Serum (1.0 mL) was extracted with diethyl ether, and steroids were isolated by Celite column chromatography. Highly specific antisera were used to measure testosterone, androstenedione, estrone, 170-estradiol, and progesterone. 3H-Labeled steroids were added to serum before extraction to correct for procedural losses. The antiserum to testosterone was kindly provided by Dr. Gordon Niswender, Department of Physiology, Colorado State University (Fort Collins, CO), and the antisera to androstenedione, 17/3-estradiol, and progesterone by Dr. John Resko, Department of Physiology, University of Oregon School of Medicine (Portland, OR). Antiserum to estrone was purchased from Steranti Research Ltd. (London, England). Serum dehydroepiandrosterone sulfate (DHEA-S) levels were determined using a commercial antibody-coated tube kit (Diagnostic Products, Los Angeles, CA). The intraassay coefficients of variation for all of the above steroid assays were less than 8.0%. To eliminate interassay variation, which ranges between 10-15%, all samples for each steroid were determined in a single assay. Serum 3a-
TABLE
1. Characteristics of the six obese women with PCO who were
studied o u- j. Subject no. 1 2 3 4 5 6
A Body mass Age . , . . index (yr) ., . ,, (kg/nr) 26 23 32 27 25 24
34.3 35.6 40.5 31.2 38.9 27.5
TT. Serum TLH/ . FSH >2
o
TOTT
No
Polycystic . ,, . . , Acanthosis ovaries by . . ., , nigncans ultrasound Yes
Yes
No
No
No
Yes Yes
Yes Yes
No No No
Yes
Yes
Yes
Body mass index values were determined on day —56. Pelvic ultrasound examinations were not performed in subjects 2 and 5. The presence of acanthosis nigricans was assessed visually.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
INSULIN AND SHBG
85
TABLE 2. Basal biochemical parameters on day —56 of the six obese women with PCO who were studied Subject no.
Testosterone (nmol/L)
Normal range (mean ± SE)°
Non-SHBG-bound SHBG Testosterone Androstenedione (nmol/L) (pmol/L) (nmol/L)
2.58 1.82 0.84 2.51 1.32 1.23
19.7 18.4 28.1 24.0 12.0 10.5
144 105 30 114 114 120
9.38 5.02 1.99 5.30 13.58 8.25
0.71 ± 0.07
75.0 ± 9.9
13 ± 4
13.53 ± 0.25
Estrone (pmol/ L) 650 374 187 494 290 232
Estradiol DHEA-S (pmol/ L) L) 354 220 232 227 164 212
376 ± 40 279 ± 30
4.87 3.13 0.88 1.95 11.24 2.80
LH (IU/L)
FSH (IU/L)
Insulin (pmol/ L)
11.5 117 17.7 9.7 20.5 25.5
7.1 7.8 12.0 10.2 9.1 12.5
149 165 101 106 151 176
11.6 ± 0.9 10.0 ± 0.5 115 ± 19
" Adapted from data obtained in five nonobese normal women on two occasions separated by 28 days during the follicular phase of the menstrual cycle (n = 10 for all values except insulin, for which n = 5) (30).
androstanediol glucuronide concentrations were determined by RIA of androstanediol after enzymatic hydrolysis of the etherextracted serum, followed by Celite chromatography of the resulting unconjugated steroids (26). The intra- and interassay coefficients of variation for the 3a-androstanediol glucuronide assay were 3.2% and 9.0%, respectively. Serum SHBG concentrations were determined by an immunoradiometric assay with materials obtained from Farmos Diagnostica (Oulunsalo, Finland). This assay does not crossreact with any other identified circulating serum proteins, including albumin, transcortin, T4-binding globulin, and transferrin. The immunoradiometric assay correlates with the dihydrotestosterone binding assay for a functional SHBG with an r value of 0.96 (16). The intraassay coefficient of variation was 3.5%, and all samples were measured in a single assay to eliminate the 12% interassay variation. Non-SHBG-bound testosterone was calculated from the total molar concentrations of testosterone and SHBG, as previously described (27). Serum LH and FSH concentrations were determined by double antibody RIA (Diagnostic Products). The intraassay coefficients of variation for LH and FSH were 12.1% and 6.5%, respectively. The interassay coefficients of variation for LH and FSH were 7.2% and 5.2%, respectively. All hormone determinations from each woman were determined in a single assay to avoid interassay variation. Statistical analysis All results are reported as the mean ± SE. Results before diazoxide treatment (day 0) were compared to those after diazoxide treatment (day 10) by first testing for normality using the Wilk-Shapiro test and then employing Student's two-tailed paired t test. Analysis of variance (Dunnett's multiple comparison test) was used to compare AUCinsuiin values and gonadotropin levels at baseline (day -56) with those on days 0 and 10. Univariate analysis was performed using CLINFO, a statistical package made available by the NIH. P < 0.05 was considered significant.
Results Leuprolide treatment significantly reduced the mean serum LH level from 16.1 ± 2.5 IU/L at baseline (day
-56) to 5.4 ± 0.7 IU/L on day 0 (P < 0.05) and 4.2 ± 0.7 IU/L on day 10 (P < 0.05). Similarly, the mean serum FSH level was reduced from 9.8 ± 0.9 IU/L on day -56 to 5.6 ± 0.2 IU/L on day 0 (P < 0.05) and 6.5 ± 0.5 IU/ L on day 10 (P < 0.05). Before leuprolide treatment (day —56), glucose tolerance, as determined by the oral glucose tolerance test, was normal in all six women (Fig. 1). However, as expected in these patients, the stimulated insulin response was exaggerated (Fig. 1). Treatment with leuprolide for 56 days altered neither the glycemic nor the insulin profile during the oral glucose tolerance test (Fig. 1), and AUCinsuiin on days -56 and 0 were similar (290 ± 63 us. 262 ± 55 nmol/min-L; P = NS). In contrast, diazoxide administration uniformly inhibited insulin release during the oral glucose tolerance test and caused deterioration of glucose tolerance (Fig. 1). The AUCinsuiin after diazoxide administration (day 10) was significantly less than that on day 0 (102 ± 33 us. 262 ± 55 nmol/min-L; P < 0.05). Neither fasting serum glucose nor insulin concentrations differed significantly on days 0 and 10. All six women gained weight during diazoxide treatment (mean weight gain, 2.2 ± 0.4 kg), and four of the six women complained of diffuse bloating, palpitations, and worsened hot flashes. The women's weights returned to baseline and symptoms promptly abated after cessation of diazoxide treatment. Serum progesterone levels on day —56 were low (0.28 ± 0.12 nmol/L), consistent with the anovulatory status of the women. Leuprolide treatment resulted in an 82% decrease in serum concentrations of both total testosterone (from 1.72 ± 0.29 nmol/L on day -56 to 0.32 ± 0.09 nmol/L on day 0; Fig. 2) and non-SHBG-bound testosterone (from 104 ± 16 pmol/L on day -56 to 19 ± 5 pmol/L on day 0). Substantial reductions also occurred in serum androstenedione, estrone, estradiol, and progesterone levels, whereas serum levels of the adrenal androgen DHEA-S did not change (Table 3). In spite of
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
NESTLER ET AL.
86
14 -i
i o
JCE & M • 1991 Vol 72 • No 1
- BASELINE(DAY-56) •LEUPROLIDE (OAYO) -LEUPROLIDE a DIAZOXIDE (DAY 10)
,-rTTTl
12-
E 10-
8-
I 6 UJ CO
4-
2500o 2000 A a. ?
1500-
_i CO
-
IOOO H
£j co
500-
-56 DAY OF
30
60
90
120
150
180
MINUTES
FlG. 1. Serum insulin and glucose levels in six obese women with PCO at baseline (day —56; • ) , after 56 days of leuprolide alone (day 0; D), and after 10 days of combined treatment with diazoxide and leuprolide (day 10; • ) . Values are the mean ± SE.
declines in both ovarian androgens and estrogens, serum SHBG levels did not change and were 18.8 ± 2.8 nmol/ L on day -56 and 17.8 ± 2.6 nmol/L on day 0 (P = NS, by Student's t test; Fig. 2). The administration of diazoxide for 10 days while continuing leuprolide treatment did not alter serum testosterone levels (0.32 ± 0.09 nmol/L on day 0 vs. 0.26 ± 0.09 nmol/L on day 10; P = NS; Fig. 2) or serum androstenedione, estrone, estradiol, progesterone, or DHEA-S levels (Table 3). In contrast, serum SHBG levels rose significantly by 32% from 17.8 ± 2.6 nmol/L on day 0 to 23.5 ± 2.0 nmol/L on day 10 (P < 0.003; Figs. 2 and 3). Due primarily to the rise in serum SHBG levels, serum non-SHBG-bound testosterone levels fell by 43% from 19 ± 5 pmol/L on day 0 to 11 ± 4 pmol/L on day 10 (P = 0.05). Percent incremental changes in serum SHBG levels from days 0-10 correlated inversely with basal serum SHBG levels on day 0 (r = -0.86; P < 0.03; n = 6), indicating that those women in whom serum SHBG levels were most suppressed experienced the greatest rise. A strong positive correlation was also noted between incremental changes in serum SHBG levels from days
STUDY
FIG. 2. Serum testosterone and SHBG levels in six obese women with the polycystic ovary syndrome at baseline (day —56), after 28 and 56 days of treatment with leuprolide alone (days —28 and 0, respectively), and after 10 days of combined treatment with diazoxide and leuprolide (day 10). Values are the mean ± SE.
0-10 and insulin values on day 0 (r = 0.70; P = 0.12; n = 6), suggesting that the more insulin-resistant women experienced the greatest rise in serum SHBG levels during inhibition of insulin release with diazoxide. Leuprolide treatment resulted in a significant fall in serum 3a-androstanediol glucuronide levels from 20.3 ± 4.0 nmol/L on day -56 to 16.4 ± 3.5 nmol/L on day 0 (P < 0.02, by Student's t test). In contrast, serum 3aandrostanediol glucuronide levels before and after diazoxide administration were not statistically different (16.4 ± 3.5 nmol/L on day 0 us. 22.6 ± 7.4 nmol/L on day 10; P = NS).
Discussion Hyperinsulinemia may play a pathogenic role in the hyperandrogenism of some women with PCO (9). In support of this hypothesis, insulin resistance and elevated serum insulin levels have been well documented in women with this disorder (2-8), and basal serum insulin levels correlate directly with basal serum testosterone and androstenedione levels (2, 6). The insulin resistance of women with PCO does not appear to be due solely to the obesity often associated with this syndrome, since nonobese women with PCO also manifest hyperinsulinemic insulin resistance (7, 8). Moreover, it has recently been shown by the insulin-glucose clamp technique that
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 May 2015. at 16:21 For personal use only. No other uses without permission. . All rights reserved.
INSULIN AND SHBG
87
TABLE 3. Serum steroid concentrations in six obese women with PCO at baseline (day —56), after 28 and 56 days of treatment with leuprolide alone (days —28 and 0, respectively), and after 10 days of combined treatment with diazoxide and leuprolide (day 10)
Baseline (day -56) Leuprolide alone Day -28 DayO Diazoxide + leuprolide (day 10)"
Androstenedione (nmol/L)
Estrone (pmol/L)
Estradiol (pmol/L)
Progesterone (nmol/L)
DHEA-S (/umol/L)
7.25 ± 1.65
371 ± 71
235 ± 26
0.28 ± 0.12
4.14 ± 1.52
3.20 ± 1.25 2.78 ± 0.94 2.43 ± 0.76
147 ± 25 156 ± 29 158 ± 52
83 ± 13 90 ±24 85 ± 18
0.07 ± 0.02 0.08 ± 0.02 0.05 ± 0.02
3.84 ± 1.08 3.79 ± 1.12 4.28 ± 1.74
Values are the mean ± SE. ° P = NS for all steroid values on day 10 compared to those on day 0.
SERU VI SEX HORMONE (nmol/
NDING GLOB
35 -i
P
