The contributions of oestrogen and growth factors to increased adrenal androgen secretion in polycystic ovary syndrome

Human Reproduction vol.14 no.2 pp.307 311, 1999 The contributions of oestrogen and growth factors to increased adrenal androgen secretion in polycystic ovary syndrome E.Carmina 1, F.Gonzalez 2, A.Vidali 1, F.Z.Stanczyk 3, M.Ferin 1 and R.A.Lobo 1,4 1 Department of Obstetrics & Gynecology, College of Physicians & Surgeons of Columbia University, 630 West 168th Street, New York, NY, 2 Department of Obstetrics and Gynecology, State University of New York at Buffalo School of Medicine & Biomedical Sciences, Buffalo, NY and 3 Department of Obstetrics & Gynecology, University of Southern California School of Medicine, Los Angeles, CA, USA 4 To whom correspondence should be addressed Adrenal hyperandrogenism is prevalent in many women with polycystic ovary syndrome (PCOS), although the expression of this enhanced secretion may be heterogeneous. Since no single factor acts in isolation, this study was performed to assess the influence of oestradiol (total and unbound), insulin, insulin-like growth factor (IGF)-I, IGF- II and the binding proteins IGFBP-I, and IGFBP-3, on basal and adrenocorticotrophic hormone (ACTH) stimulated adrenal androgen secretion in 25 women with PCOS and 10 matched ovulatory controls. Women with PCOS exhibited elevations of all androgens as well as unbound oestradiol, insulin and non-igfbp-1 bound IGF-I. Positive correlations were noted between oestrogen and basal and ACTH stimulated 5 adrenal androgens. Serum IGF-I was only correlated with basal dehydroepiandrosterone sulphate (DHEA-S), while insulin exhibited a strong correlation with the 4 pathway and androstenedione formation in particular. This correlation was also confirmed by dividing the PCOS group into those women with and without hyperinsulinaemia. The activity of 17,20 lyase favouring androstenedione was increased in the hyperinsulinaemic women. By multivariate analyses, body mass index did not influence these findings. Although there are inherent difficulties in making major conclusions based on correlative analyses, it is suggested that oestrogen may have a greater influence on enhancing 5 adrenal androgen secretion, and insulin a greater effect on the 4 pathway. In turn, the relative importance of these influences may contribute to the heterogeneous nature of adrenal hyperandrogenism in PCOS. Key words: adrenal androgens/igf-i/insulin/oestrogen/pcos Introduction It is well known that adrenal hyperandrogenism (AH) is a prevalent feature of polycystic ovary syndrome (PCOS) (Hoffman et al., 1984; Lobo, 1984; Gonzalez et al., 1991; Carmina et al., 1992, 1997a). Although many studies have been carried out examining possible causes of AH in PCOS (Lobo, 1984; Carmina, 1997a,b), its pathogenesis remains unclear. Among the several factors which may be involved, hyperoestrogenism and hyperinsulinaemia have been the most studied (Lobo et al., 1982; Ditkoff et al., 1995; Gonzalez et al., 1996; Moghetti et al., 1996; Rosenfield, 1996; Carmina, 1997b; Eldar-Geva et al., 1998). However, conflicting results have been reported regarding the relative importance of these influences (Ditkoff et al., 1995; Rosenfield, 1996). Eldar-Geva et al. (1998) have suggested that with ovarian stimulation, adrenal secretion of progesterone is enhanced, although no specific factors could be implicated. Insulin-like growth factors (IGF-I and IGF-II) also are known to be potent in-vitro stimulators of adrenal secretion (Pham et al., 1991; Mesiano and Jaffe, 1993) and have not been adequately studied. Inhibition of both insulin and IGF-I with somatostatin has been shown to decrease AH in PCOS (James et al., 1994). Therefore the aim of this study was to understand the relative importance of oestrogen and growth factors which may influence adrenal androgens in women with PCOS. Since it is clear that no factor acts in isolation, several of these factors were examined simultaneously in the same patient. Given the difficulties inherent in correlative analyses, in this report different types of relationships have been observed for oestrogen and growth factors: oestrogen correlated with basal and stimulated adrenal androgens of the 5 pathway while insulin correlated with the adrenal 4 pathway. Materials and methods Subjects Twenty-five women with PCOS and 10 normal controls, matched for age and body mass index (BMI) were studied. Women with PCOS were diagnosed on the basis of hyperandrogenism and chronic anovulation, after the exclusion of tumours, Cushing s syndrome and adrenal enzymatic deficiencies. Polycystic ovaries on ultrasound were present in all of the women with PCOS in this study but this was not a required inclusion criterion. Protocol and measurements During the mid-follicular phase of the cycle, between days 5 and 8, at 8 9 a.m. after an overnight fasting, a blood sample was obtained for measurements of oestradiol, unbound oestradiol, androstenedione, dehydroepiandrosterone sulphate (DHEA-S), insulin, IGF-I, IGF-II, IGFBP-3 and IGFBP-1. The following day, patients and controls underwent an adrenocorticotrophic hormone (ACTH) stimulation test (ACTH 1 17, 0.25 mg i.v. with blood samples at 0, 60, 120 and 180 min). Dexamethasone (1 mg) was administered orally the night before European Society of Human Reproduction and Embryology 307

E.Carmina et al. Table I. Basal hormone concentrations mean SE in controls and women with PCOS Insulin IGF-I IGF-II Unbound (µu/ml) (ng/ml) (ng/ml) oestradiol (pg/ml) Controls 9.1 0.8 292 22 702 64 22.2 4.1 PCOS 14.5 1.9* 279 23 822 49 37.7 5.2* the ACTH test (at 11 p.m.). After ACTH, blood samples were obtained for 17-OH pregnenolone (17-OHPe), 17-OH progesterone (17-OHP), androstenedione, DHEA and DHEA-S. After ACTH, steroid ratios were calculated as approximate estimates of various enzymatic activities. Serum hormone concentrations were quantified by well-established methods which were validated previously in our laboratory. All steroids were measured by specific radioimmunoassay (RIA) after extraction using previously described methods (Lobo and Goebelsmann, 1981; Hoffman et al., 1984). Non-sex hormone-binding globulin (SHBG)-bound oestradiol was measured by adding 500 c.p.m. of tritiated oestradiol to 0.3 ml of serum prior to precipitation with ammonium sulphate. The percentage of tritiated oestradiol in the supernatant was calculated, as previously reported (Lobo et al., 1981). Insulin and IGFBP-3 were quantified by use of direct RIA kits (Diagnostic Systems Laboratories, Inc., Webster, TX, USA). IGF- I and IGF-II were measured by RIA using material provided by Diagnostic Systems Laboratories. Both assays were performed after an acid ethanol extraction step. IGFBP-1 was measured by an immunoreactive assay using a kit provided by Diagnostic Systems Laboratories. The methods for the assay of IGF-I, IGFBP-3, IGFBP- 1 have been previously described (Carmina et al., 1995). There was no cross reactivity of the IGF-II antiserum, IGF-I, IGF binding proteins (1, 2, 3, 4, 5) insulin or growth hormone. The sensitivity of the IGF-II assay was 12 ng/ml. The interassay coefficient of variation ranged from 4.3 to 7.2%, while the interassay coefficient of variation ranged from 6.3 to 10.4%. In all other assays, intraassay and interassay coefficients of variation did not exceed 6 and 15% respectively. Statistics Analysis of the data was performed using Student s t-test for comparisons between the two groups: PCOS and controls. Correlations were analysed by Pearson product moment correlations and stepwise multivariate linear regression analysis with forward selection. P 0.05 was considered significant. Results are expressed as mean SE. Results Controls and PCOS were of similar age (24.1 1 versus 23.6 1.2 years) and BMI (27.4 0.8 versus 28.2 1.6). As depicted in Table I, PCOS patients had increased concentrations of insulin (P 0.05), while serum IGF-I and IGF-II were similar in the two groups. Total serum oestradiol was also similar but unbound oestradiol was significantly (P 0.05) higher in PCOS (Table I). Patients with PCOS had lower concentrations of serum IGFBP-1 (20.4 3 versus 124 12 ng/ml, P 0.01) and higher concentrations of basal DHEA-S (3.4 0.2 versus 1.9 0.2 µg/ml, P 0.01) and -A (3.8 0.2 versus 1.5 0.2 ng/ml, P 0.01). Serum IGFBP-3 was similar in controls and PCOS. 308 Table II depicts the adrenal androgen responses to ACTH. In PCOS, patients exhibited an exaggerated response (P 0.01) to all steroids measured. Serum insulin did not correlate with basal DHEA-S or -A but showed a significant (P 0.05) negative correlation with the increment of 17OHP ( 17-OHP) after ACTH (Table III). Serum insulin was positively correlated with the ratios of A/ 17-OHP and A/A 17-OHPe, suggesting the capacity to increase adrenal 17,20 lyase activity (Table IV). Serum IGF-I was significantly correlated (P 0.05) with the basal concentrations of DHEA-S but did not show any correlations with the adrenal androgen responses to ACTH (Tables III and IV). Both total and unbound oestradiol correlated significantly with basal DHEA-S but showed a negative correlation with the androstenedione response to ACTH (Table III). Moreover, total and unbound oestradiol correlated negatively with the ratio of A/ DHEA and positively with the ratio ( DHEA/ 17-OHP) (Table IV), suggesting a relative increase in 17,20 lyase activity and reduction in 3β-dehydroxysteroid dehydrogenase activity favouring the formation of 5 steroids. IGF-II did not show any correlation with adrenal androgen secretion. Thirteen women with PCOS were hyperinsulinaemic (mean insulin 21.3 2 µiu/ml) with values 12 µiu/ml, which was the upper 95% confidence value of the controls. Twelve women with PCOS were normoinsulinaemic (mean insulin 7.1 0.7 µiu/ml, range 3.1 10 µiu/ml). The two groups were significantly different in terms of BMI (hyperinsulinaemic PCOS: 32.6 2, normoinsulinaemic PCOS: 23.4 1, P 0.01). As shown in Table V, the hyperinsulinaemic patients with PCOS had significantly (P 0.05) higher ratios of A/ 17-OHP and A/ 17- OHPe compared to the normoinsulinaemic group. The possibility that BMI may have influenced the results was assessed directly. In PCOS, BMI correlated positively with serum insulin (r 0.57, P 0.01) and negatively with serum IGF-I (r 0.55, P 0.01) but not with unbound oestradiol (r 0.04) or with IGF-II (r 0.34). BMI did not correlate with basal or stimulated adrenal androgen concentrations or with any steroid ratios. Multivariate analysis showed that BMI did not influence the correlations of unbound oestradiol and IGF-I with adrenal androgens. In spite of a strong influence of BMI on serum insulin, BMI did not affect the negative correlation of insulin with 17-OHP, or the positive correlation with the A/ 17- OHP ratio. Correcting for BMI, the correlation of insulin with 17-OHP was r 0.43 (P 0.05) and with A/ 17-OHP was r 0.40 (P 0.05). Multivariate analyses did not show any influence of insulin on the correlation between oestradiol and unbound oestradiol with adrenal androgens. Discussion The data in this study confirm that several extra-adrenal hormones may influence adrenal androgen secretion and may participate in the pathogenesis of AH in PCOS. In this study, serum IGF-II did not correlate with circulating adrenal

Effect of oestrogen and growth factors in PCOS Table II. Adrenal androgen responses to ACTH in controls and PCOS represented as maximal increments from baseline ( ) 17-OHPe ng/ml 17-OHP ng/ml A ng/ml DHEA ng/ml DHEA-S ng/ml Normal 7.5 1 0.8 0.1 1.2 0.2 8.2 0.6 0.6 0.1 PCOS 13.5 1** 1.3 0.1** 2.4 0.2** 13.8 1** 1.1 0.1** **P 0.01. ACTH adrenocorticotrophic hormone, PCOS polycystic ovary syndrome, 17-OHPe 17- hydroxypregnenolone, 17-OHP 17-hydroxyprogesterone, A androstenedione, DHEA dihydroepiandrosterone and DHEA-S dihydroepiandrosterone sulphate. Table III. Correlations between serum concentrations and basal concentrations of DHEA-S and A; and responses to ACTH in PCOS Basal DHEA-S Basal A 17-OHPe 17-OHP A DHEA DHEA-S Insulin 0.21 0.33 0.35 0.43* 0.15 0.36 0.31 IGF I 0.46* 0.23 0.27 0.19 0.23 0.31 0.01 Oestradiol 0.47* 0.27 0.29 0.32 0.48* 0.28 0.26 Unbound oestradiol 0.50* 0.28 0.22 0.26 0.44* 0.27 0.26 For abbreviations, see Table II. Table IV. Correlations between baseline concentrations and adrenal enzymatic activity 3β-Hydroxysteroid dehydrogenase activity 17,20-Lyase activity 17-OHP/ 17OHPe A/ DHEA DHEA/ 17OHPe A/ 17-OHP Insulin 0.29 0.29 0.21 0.47* IGF-I 0.11 0.09 0.06 0.34 Oestradiol 0.18 0.46* 0.57* 0.18 Unbound oestradiol 0.16 0.44* 0.51* 0.21 BMI 0.34 0.08 0.35 0.28 For abbreviations, see Table II. Table V. 17,20-Lyase activity in normoinsulinaemic and hyperinsulinaemic PCOS DHEA/ A/ 17-OHP A/ 17-OHPe 17OHPe Normoinsulinaemic PCOS 1.0 0.11 1.62 0.3 0.14 0.05 Hyperinsulinaemic PCOS 1.09 0.1 2.51 0.3* 0.24 0.11** **P 0.01. androgen concentrations suggesting that it may have a paracrine, or direct role in adrenal androgen regulation (Mesiano and Jaffe, 1993). However, serum IGF-I, insulin and oestradiol (both total and unbound) correlated positively with adrenal androgens in the same group of patients with PCOS. These positive findings are of significance in that elevations in unbound oestradiol as well as hyperinsulinaemia appear to be consistent findings in PCOS. Nevertheless, there are subsets of patients who may have a greater or lesser elevation in insulin or unbound oestradiol. Novel implications of findings in this study are that these extra-adrenal hormones correlated differently with various adrenal androgens suggesting that these factors may induce AH by different mechanisms. Serum IGF-1 correlated with basal DHEA-S, but not with adrenal androgen responses to ACTH while insulin did not correlate with basal adrenal androgens but showed a positive correlation with the A/ 17- OHP ratio and the A/ 17-OHPe ratio after ACTH stimulation. In evaluating steroid ratios in blood after ACTH, only the effect on true enzyme activity was estimated. Nevertheless, these data suggested an important influence of insulin on adrenal 4 A production. These findings were further enhanced by our findings in the subgroup of hyperinsulinaemic patients with PCOS whose responses were different from the normoinsulinaemic group. In this study, single time point measurements of fasting insulin for correlative analyses were used. Clearly, fasting insulin alone is inadequate to characterize the existence of insulin-resistance in these women. Oestrogen (oestradiol and unbound oestradiol) on the other hand, correlated with both basal DHEA-S and adrenal androgen responses to ACTH (positively with the ratio of DHEA/ 17-309

E.Carmina et al. OHPe and negatively with the ratio of A/ DHEA). This suggests a positive influence of oestrogen on the 5 pathway, predominantly resulting in an enhancement of DHEA/DHEA-S production. In this study, BMI was similar in patients and controls, and multivariate analyses were carried out. It was shown that BMI did not influence the correlations observed. In PCOS, it has been shown that the ovarian theca hypersecretes androstenedione (Gilling-Smith et al., 1994). Also, a genetic defect in serine phosphorylation in PCOS may lead to enhanced 17,20 lyase activity in both the ovary and adrenal (Zhang et al., 1995). While hypersecretion of adrenal androgen in PCOS is heterogeneous, in ~50% of patients a more isolated increase in either the 4 or 5 pathway may be exhibited (Carmina et al., 1992; Carmina 1997a). Accordingly, it might be postulated that the more dominant the influence of one factor (oestradiol versus insulin), the more different may be the pattern of AH ( 5 versus 4 steroids). Since in this study it was found that only serum IGF-I correlated positively with DHEA-S, this influence may be artifactual or may involve other factors not studied here such as peripheral or adrenal sulphatase activity. It is interesting to observe that while the data regarding the influence of oestrogen are consistent with observations previously made using different experimental approaches (Lobo et al., 1982; Gonzalez et al., 1991; Ditkoff et al., 1995), and in-vitro data showing the influence of oestrogen of adrenal cell cultures, the data for insulin are not as consistent. It has been previously reported that insulin administration reduces 17,20 lyase activity in both normal (Nestler et al., 1992) and hyperandrogenic (Moghetti et al., 1996) women. However, it has also been suggested that insulin may in fact increase 17,20 lyase activity in PCOS (Rosenfield, 1996). It has also been reported that A responses to ACTH are increased in hyperinsulinaemic patients with PCOS (Lanzone et al., 1992). It is possible that this discrepancy may be explained by the extent of insulin resistance. In PCOS, hyperinsulinaemia may be secondary to tissue insulin, which may include the adrenals as a site of this relative resistance. Interestingly, diet and other factors (such as diazoxide) which reduce serum insulin do not change adrenal androgen concentrations in PCOS (Nestler and Jakubowicz, 1989; Holte et al., 1994) while insulin sensitizing agents and troglitazone, which decrease serum insulin by improving tissue sensitivity have been shown to reduce adrenal androgen concentrations (Dunaif et al., 1996; Nestler et al., 1997). It should also be noted that the reduction in insulin also results in an increase in SHBG. As a consequence of this decrease, serum unbound oestradiol (Lobo and Carmina, 1997) may be reduced and contribute to a reduction in AH. This study was conducted to evaluate several factors by multivariate analyses. This approach was necessary because of the heterogeneous nature of AH and of the various factors influencing it in patients with PCOS. Thus, the results have to be viewed with the problems inherent in performing correlative analyses. In addition, use of steroid ratios to estimate true enzyme activities also should be interpreted with some caution. Nevertheless, it has been shown that several extra-adrenal 310 hormones, specifically oestrogen and insulin, correlate with the presence of AH in PCOS and that these hormones may influence adrenal androgen pathways differently, thus explaining the heterogeneous presentation of AH in PCOS. References Carmina, E., Koyama, T., Chang, L. et al. (1992) Does ethnicity influence the prevalence of adrenal hyperandrogenism and insulin resistance in polycystic ovary syndrome? Am. J. Obstet. Gynecol., 167, 1807 1812. Carmina, E (1997a) Prevalence of adrenal androgen excess in PCOS. In Azziz, R., Nestler, J.E. and Dewailly, D. (eds), Androgen Excess Disorders in Women. Lippincott-Raven, Philadelphia, pp. 385 393. Carmina, E. (1997b) The role of extra-adrenal factors in adrenal androgen excess: in vivo studies in androgen excess disorders in women. In Azziz, R., Nestler, J.E. and Dewailly, D. (eds) Androgen Excess Disorders in Women. Lippincott-Raven, Philadelphia, 425 434. Carmina, E., Stanczyk, F.Z., Morris, R.S. et al. (1995) Altered regulation of insulin-like growth factor binding protein 1 in patients with polycystic ovary syndrome. J. Soc. Gynecol. Invest., 2, 743 747. Ditkoff, E.C., Fruzzetti, F., Chang, L. et al. (1995) The impact of estrogen on adrenal androgen sensitivity and secretion in polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 80, 603 607. Dunaif, A., Scott, D., Finegood, D. et al. (1996) The insulin sensitizing agent troglitazone improves metabolic and reproductive abnormalities in the polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 81, 3299 3306. Eldar-Geva, T., Margalioth, E.J., Brooks, B. et al. (1998) The origin of serum progesterone during the follicular phase of menotropin-stimulated cycles. Hum. Reprod., 13, 9 14. Gonzalez, F., Hatola, D.A. and Speroff, L. (1991). Adrenal and ovarian steroid hormone responses to gonadotropin-releasing hormone agonist treatment in polycystic ovary syndrome. Am. J. Obstet. Gynecol., 165, 535 545. Gonzalez, F., Chang, L., Horab, T. and Lobo, R.A. (1996) Evidence for heterogeneous etrologies of adrenal dysfunction in polycystic syndrome. Fertil. Steril. 66, 354 361. Gilling-Smith, C., Willis, D.S., Beard, R.W. et al. (1994) Hypersecretion of androstenedione by isolated theca cells from polycystic ovaries. J. Clin. Endocrinol. Metab., 79, 1158 65. Hoffman, D.L., Klove, K., and Lobo, R.A. (1984) The prevalence and significance of elevated dehydroepiandrosterone sulfate in anovulatory women. Fertil. Steril., 42, 76 81. Holte, J., Bergh, T., Gennarelli, G. et al. (1994) The independent effect of polycystic ovary syndrome and obesity on serum concentrations of gonadotropins and sex steroid hormones in premenopausal women. Clin. Endocrinol., 41, 473 481. James, W.M., Morris, R.S., Gentzschein, E. et al. (1994) The effects of octreotide on adrenal response to corticotropin releasing factor (CRF) in polycystic ovary syndrome (PCO). 50th Ann. Meet. Am. Fertil. Soc., San Antonio, Texas. Abstract P154. Lanzone, A., Fulghesu, A.M., Guido, M. et al. (1992) Differential androgen response to adrenocorticotropic hormone stimulation in polycystic ovarian syndrome relationship with insulin secretion. Fertil. Steril., 58, 296 301. Lobo, R.A. (1984) The role of the adrenal in polycystic ovary syndrome. Semin. Reprod. Endocrinol., 2, 251 62. Lobo, R.A., and Carmina, E. (1997) Polycystic ovary syndrome. In Lobo, R.A., Paulson, D.R. and Shoupe, D. (eds), Mishell s Textbook of Infertility, Contraception and Reproductive Endocrinology, Fourth Edition., Blackwell Scientific, Malden, Massachusetts, pp. 363 383. Lobo, R.A., and Goebelsmann, U. (1981) Evidence for reduced 3β-olhydroxysteroid dehydrogenase activity in some hirsute women thought to have polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 53, 394 400. Lobo, R.A., Granger, L., Goebelsmann, U. et al. (1981) Elevations in unbound serum estradiol as a possible mechanism for inappropriate gonadotropin secretion in women with PCO. J. Clin. Endocrinol. Metab., 52, 156 158. Lobo, R.A., Goebelsmann, U., Brenner, P.F. et al. (1982) The effects of estrogen on adrenal androgens in oophorectomized women. Am. J. Obstet. Gynecol., 142, 471 478. Mesiano, S., and Jaffe, R.B. (1993) Interaction of insulin like growth factor II and estradiol directs steroidogenesis in the human fetal adrenal toward dehydroepiandrosterone sulfate production. J. Clin. Endocrinol. Metab., 77, 754 758.

Effect of oestrogen and growth factors in PCOS Moghettii, P., Castello, R., Negri, C. et al. (1996) Insulin infusion amplifies 17α-hydroxycorticosteroid intermediates response to adrenocorticotropin in hyperandrogenic women: apparent relative impairment of 17,20 lyase activity. J. Clin. Endocrinol. Metab., 81, 881 886. Nestler, J.E., and Jakubowicz, D.J. (1997) Lean women with polycystic ovary syndrome respond to insulin reduction with decreases in ovarian P450C17α activity and serum androgens. J. Clin. Endocrinol. Metab., 82, 4075 4079. Nestler, J.E., Barlascini, C.O., Matt, D.W. et al. (1989) Suppression of serum insulin by diazoxide reduces serum testosterone levels in obese women with polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 68, 1027 1032. Nestler, J.E., McClanahan, M.A., Clore, J.N. et al. (1992) Insulin inhibits adrenal 17,20 lyase activity in man. J. Clin. Endocrinol. Metab., 74, 362 367. Pham, H.U.U., Trung, M.T., Villette, J.M. et al. (1991) Effects of insulin like growth factor I (IGF-I) on enzymatic activity in human adrenocortical cells. Interactions with ACTH. J. Steroid Biochem. Mol. Biol., 39, 903 909. Rosenfield, R.L. (1996) Evidence that idiopathic functional adrenal hyperandrogenism is caused by dysregulation of adrenal steroidogenesis and that hyperinsulinemia may be involved. J. Clin. Endocrinol. Metab., 81, 878 880. Zhang, L.H., Rodriguez, H., Ohno, S. et al. (1995) Serine phosphorylation of human P450c17 increases 17,20-lyase activity: implications for adrenarche and the polycystic ovary syndrome. Proc. Natl. Acad. Sci. USA, 92, 10619 10629. Received on July 13, 1998; accepted on October 2, 1998 311