Insulin, androgens, and obesity in women with and without polycystic ovary syndrome: a heterogeneous group of disorders

FERTILITY AND STERILITY VOL. 72, NO. 1, JULY 1999 Copyright 1999 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A. Insulin, androgens, and obesity in women with and without polycystic ovary syndrome: a heterogeneous group of disorders Pedro Acién, M.D.,* Francisco Quereda, M.D.,* Pilar Matallín, M.D., Encarnación Villarroya, M.D.,* Jose A. López-Fernández, M.D., Maribel Acién, M.D., Monserrat Mauri, M.D., and Rocío Alfayate, M.D. Department of Obstetrics and Gynecology, School of Medicine, Miguel Hernández University, Campus de San Juan, Alicante, Spain Received June 24, 1998; revised and accepted February 17, 1999. Reprint requests: Pedro Acién, M.D., Departamento de Ginecologia, Facultad de Medicina, Campus de San Juan, Apartado de Correos 18, 03550 San Juan, Alicante, Spain (FAX: 34-965919550; E-mail: acien@umh.es). * Service of Obstetrics and Gynecology, San Juan University Hospital. Service of Obstetrics and Gynecology, Alicante University General Hospital. Institute of Gynecology P.A.A. Hormone Laboratory, Alicante University General Hospital. 0015-0282/99/$20.00 PII S0015-0282(99)00184-3 Objective: To analyze the correlations among insulin, androgens, body mass index (BMI), and other related metabolic anomalies in women with and without polycystic ovary syndrome (PCOS). Design: Retrospective study of normal and obese women with and without PCOS. Setting: Gynecologic endocrinology units of Elche, San Juan, and Alicante Hospitals and Hormone Laboratory at Alicante University Hospital ( Miguel Hernández University). Patient(s): A total of 212 women were studied: 137 with PCOS and 75 without PCOS. Intervention(s): BMI, gonadotropins, insulin, androgens (T, androstenedione, DHEAS), 17 -hydroxyprogesterone, sex hormone-binding globulin, and triglycerides were studied. Glycemia and insulin response to the tolerance test (GTT) with a 100-g oral glucose load were also assessed in 103 women. Result(s): A good correlation between insulin and BMI was found in normal and obese women without hormonal dysfunction and in patients with or without PCOS. Good correlations, although lower, between insulin and T, and BMI, insulin, and T with triglycerides were also found in patients with PCOS. These patients fell into clearly distinct categories: with or without insulin resistance and with or without obesity, but slim women with PCOS had insulin and metabolic variables similar to those without PCOS, and most obese women with PCOS were insulin-resistant and more hyperandrogenic and hypertriglyceridemic. Conclusion(s): Insulin, androgens, and BMI are related in women both with PCOS and without PCOS, especially in obese ones. Insulin and metabolic indices are similar in lean women with PCOS and those without PCOS, but obese women with PCOS are more insulin-resistant, hyperandrogenic, and hypertriglyceridemic. Three types of disorders can be distinguished: simple nonhyperandrogenic obesity, typical nonhyperinsulinemic PCOS, and insulin-resistant PCOS. (Fertil Steril 1999;72:32 40. 1999 by American Society for Reproductive Medicine.) Key Words: Insulin, insulin resistance, hyperandrogenism, obesity, polycystic ovary syndrome Polycystic ovary syndrome (PCOS) is a complex clinical entity that probably includes several different pathologies. The condition affects a great number of women, especially infertile women. It is clinically indicated by menstrual disorders (oligomenorrhea or amenorrhea), hirsutism, and/or other signs of hyperandrogenism. Patients frequently are obese and, as mentioned, infertile. The ovaries appear thickened both ultrasonographically and laparoscopically. On histopathologic examination, sometimes the polycystic aspect predominates (multiple half-grown, 6 8-mm peripheral follicles), and in some other cases, hyperthecosis prevails (with fewer small follicles but with a large thecal hyperplasia), the latter being the real cause of hyperandrogenism. In terms of hormonal analysis, hyperandrogenism usually prevails in patients with PCOS ( 4-androstenedione [A] and/or T increased), but an alteration of the FSH/LH ratio (with predominance of the latter gonadotropin) can also be seen, together with a decrease in sex hormone-binding globulin (SHBG), an increase in 17 -hydroxy progesterone (17-OHP) and 32

DHEAS, an increase in insulin, and sometimes hyperinsulinemiaandinsulinresistance. Thesehyperinsulinemicinsulinresistant patients with PCOS are the most obese, hyperandrogenic, hypercholesterolemic, and hypertriglyceridemic. These women may develop hypertension, diabetes, or cardiovascular disease and suffer other severe vital and reproductive risks in the future. Therefore, insulin, androgens, and obesity are implicated in patients with PCOS, especially in the most severe cases. The definitions of these symptoms, the correlations between them, their mutual influences, their origins or causes, the characterization of the syndrome, and the consideration of PCOS as a heterogeneous group of disorders with different pathologies are controversial issues. In some studies it has been reported that hyperinsulinemia promotes hyperandrogenism in women with PCOS but not in normal ones and that hyperinsulinemia represents a universal feature of PCOS (1). Other reports show that obesity is related to hyperandrogenism and hyperinsulinemia (2, 3). However, other studies (4, 5) have not reported any relation between hyperinsulinemia and hyperandrogenism, and it is evident that not all obese women are hyperandrogenic or hyperinsulinemic. Furthermore, it does not appear that hyperandrogenism causes hyperinsulinemia or obesity. After ovary castration or the administration of GnRH analogues (6), insulin resistance continues in those women who suffer from it, and men with clear hyperandrogenism do not show any insulin resistance (unless pathologically). To help clarify these pathogenic aspects of PCOS; to evaluate the mutual influences among insulin, hyperandrogenism, and obesity; and to classify the different disorders included in PCOS, we studied and sought correlations among the following: [1] basal levels of insulin and insulin resistance (hyperresponse to 75 or 100 g oral glucose load), [2] androgens ( A, T, 17-OHP, DHEAS) and levels of hyperandrogenism (of T), and [3] body mass index (BMI) and grades of obesity (based on BMI), as well as some related metabolic anomalies (triglycerides), in normal women, patients with PCOS, and those with other non- PCOS endocrinopathies. MATERIALS AND METHODS Subjects and Study Design A total of 212 women were studied. One hundred thirteen patients with PCOS (88 noninsulin-resistant and 25 insulinresistant) and 61 women without PCOS were studied at the University Hospitals of Elche and San Juan. Women without PCOS included 26 normal women, 18 with obesity (BMI 30 kg/m 2 ), 12 with hyperprolactinemia, 10 with hypothalamic dysfunction or amenorrhea, and 13 with slight ovarian dysfunction or essential hirsutism. All women were examined clinically (e.g., general and gynecologic examinations, weight, height, blood pressure, Ferriman-Gallwey index). Hormonal and biochemical analyses, including gonadotropins, insulin, androgens ( A, T, DHEAS), 17-OHP, SHBG, triglycerides, and others, were performed during the first days of the cycle, or in any case, during the follicular phase. The women were analyzed retrospectively, although many had been studied previously in a prospective manner as part of other studies. SHBG had not been studied in all cases. Fifty-one patients in the PCOS group and 14 without PCOS had undergone a test of tolerance to a 100-g oral glucose load (or 75 g in 10 cases) to determine insulin and glycemia at baseline and at 30, 60, 120, and 180 minutes. Other determinations or studies (such as cortisol or ACTH test, or those performed in the luteal phase) were not considered here. To study the insulin response (or insulin resistance) to the glucose load in a wider number of patients, we included another 14 normal women and 24 women with PCOS who had been studied prospectively at the University Hospital of Alicante with a different analysis objective by two of the authors (P.M. and J.A.L.-F.). All of these additional subjects had also been studied at the beginning of the follicular phase, and apart from gonadotropin, insulin, androgen, 17-OHP, and SHBG levels, all of them had undergone an insulin and glycemia response test to a 100-g oral glucose load, and the blood samples were analyzed in the same Hormone Laboratory. Therefore, we analyzed 75 women without PCOS, 28 of them with a response test or curve of glycemia and insulin (with insulin resistance in 2), and 137 patients with PCOS, 75 of them with a curve of insulin and glycemia (with insulin resistance in 27). To catalogue and achieve a suitable distribution of the cases studied, we followed several criteria. [1] PCOS was diagnosed according to the characteristic clinical findings (various cycle alterations, obesity, hirsutism, and acne), endocrine data (high LH levels, LH/FSH ratio of 2, high androgen levels), and the appearance of polycystic ovaries on transvaginal ultrasonography according to the criteria of Adams et al. (1986) and Takahashi et al. (1994) (7 9). Specifically, PCOS was defined when the patient exhibited at least three of the following five signs or symptoms: LH/FSH ratio 2, A 4.5 ng/ml and/or T 0.8 ng/ml, ovaries with PCOS criteria in the transvaginal ultrasound examination (presence of 10 small [2 8 mm] peripheral follicles bilaterally, without dominance, around a dense core of stroma), amenorrhea or oligomenorrhea, and hirsutism or acne. [2] The diagnosis of PCOS was also based on the exclusion of other PCOS-like syndromes, including adrenal dysfunction, Cushing s syndrome, congenital adrenal hyperplasia, androgen-producing tumors, and thyroid dysfunction. However, within the PCOS group, we included patients with PCOS who had a slight partial deficiency of 21-OH (if the basal 17-OHP was 3 ng/ml and 6 ng/ml, or if at 1 hour FERTILITY & STERILITY 33

in the New s test [ACTH 250 U, IV] it was 6 ng/ml and 10 ng/ml) and patients with PCOS with an adrenal component if the DHEAS was 4 ng/ml and/or cortisol was 25 g/dl. In the group without PCOS, we included normal women and obese women with normal hormonal analysis (some of them were considered as a control group in other studies; others were obese women studied for this disorder). This group also included [1] women with hyperprolactinemia (if PRL was 25 ng/ml without other clinical or analytic data of PCOS); [2] those with hypothalamic dysfunction, including cases of delayed puberty, chronic psychic anovulation, and those with hypogonadism-hypogonadotropism; and [3] women with slight ovarian dysfunction and essential hirsutism, to refer to cases with menstrual delay, slight increase in androgens, ultrasonographic appearance of polycystic ovaries, or hirsutism, without the three criteria required to be included as PCOS. [3] Hyperinsulinemia was considered when basal insulin was 25 U/mL. Insulin resistance was recorded when the basal insulin was 30 U/mL and/or some point of the curve in the response test (at 1 hour) was 175 U/mL (10, 11). The 95th percentile and means 2 SDs of insulin levels in our studied normal women without hormonal dysfunction were 23 U/mL (baseline level), 157 U/mL (at 30 minutes), 175 U/mL (at 60 minutes), 93 U/mL (at 120 minutes), and 44 U/mL (at 180 minutes). [4] The BMI was calculated as weight in kg/m 2 and was considered normal at BMI 25 kg/m 2. Patients were considered clearly obese at BMI 30 kg/m 2. Setting and Assays Most of the patients were studied clinically and using transvaginal ultrasound by one of the authors (P.A.) at the University Hospital of Elche (until 1992) and San Juan (after 1992). Other patients were studied by other authors (P.M. and J.L.L.-F.) at the University Hospital of Alicante. However, all hormonal analyses were performed at the Hormone Laboratory of the General University Hospital of Alicante (M.M. and R.A.). FSH, LH, and PRL concentrations were measured by an immunoenzymatic assay (Abbott Laboratories, Madrid, Spain); E 2 and T were measured by a double-antibody RIA (ICN Biomedical Inc., Costa Mesa, CA). The intraassay and interassay coefficients of variation (CVs) were 10.6% and 11.9% for E 2 and 7.5% and 9% for T, respectively. Delta A was measured by a solid-phase RIA (Coat-a-Count; Diagnostic Products Corporation, Los Angeles, CA) with intraassay and interassay CVs of 9.4% and 11.6%, respectively. Insulin was determined by an immunoradiometric method with reagents provided by Medgenix Diagnostics (Biosource, Fleurus, Belgium). The intraassay and interassay CVs were 4.5% and 12.2%, respectively. DHEAS and 17- OHP were determined by a solid-phase RIA (Diagnostic Products Corporation). The intraassay and interassay CVs were 5% and 8% for 17-OHP and 6.5% and 8.2% for DHEAS. Finally, SHBG was measured by an immunoradiometric method (Spectria; Orion Diagnostica, Finland) with intraassay and interassay CVs of 4.9% and 5.5%, respectively. Statistical Analysis The data from the 212 subjects were introduced into a computer with the statistical package R-Sigma (Horus Hardware, 1990). First, a Kolmogorov-Smirnov test was applied to all groups and variables to evaluate the adjustment to a normal or symmetrical distribution. In cases of adjustment to a normal distribution, means and SDs were obtained, and then an unpaired t-test for independent variables was applied. If the Kolmogorov-Smirnov test was significant (P.05), showing an asymmetrical distribution, the median and 25th and 75th percentiles were obtained, and the Mann- Whitney U test was applied for the differences between groups. Correlation coefficients between different variables were obtained using Pearson s and Spearman s methods. Differences and correlations were considered statistically significant at P.05. RESULTS Table 1 shows the mean values of BMI and of serum concentrations in the follicular phase of gonadotropins, A, T, DHEAS, 17-OHP, insulin, SHBG, and triglycerides in normal and obese women and in PCOS and non-pcos endocrinopathies. Women without hormonal dysfunction were separated into nonobese (normal women) and obese women (BMI 30 kg/m 2 ), and the latter showed higher values of insulin and lower values of SHBG. There was also a statistically significant increase in T and a nonsignificant increase in triglycerides. There were no significant differences between normal or obese women and other non-pcos endocrinopathies, except for triglycerides (which were lower). In women without hormonal dysfunction (normal and obese women), correlations among BMI, insulin, T, SHBG, and triglycerides were highly significant between BMI and insulin, and BMI and T. Correlations were also statistically significant, but negative, between BMI and SHBG, and T and SHBG. In women with non-pcos endocrinopathies, the correlation between BMI and insulin was also statistically significant, but not the other correlations. Table 1 also shows the values of the same variables and analysis of correlations for patients with PCOS, with or without insulin resistance. These patients were compared with normal and obese women without hormonal dysfunction and between the groups. We found the following results. [1] Women with PCOS without insulin resistance had significantly higher LH, A, T, 17-OHP, and DHEAS, and decreased SHBG, in comparison with normal and obese women. [2] Women with PCOS and insulin resistance had normal LH but significantly higher T (even higher than 34 Acién et al. Insulin, androgens, and obesity Vol. 72, No. 1, July 1999

TABLE 1 Body mass index and serum levels in the follicular phase of gonadotropins, androgens, insulin, SHBG, and triglycerides, as well as correlations among them, in normal and obese women and in PCOS and non-pcos endocrinopathies. Studied variables (normal values) Normal women (n 22) Obese women without hormonal dysfunction (n 18) Women with non-pcos endocrinopathies (n 35) Women with PCOS without insulin resistance (n 110) Women with PCOS with insulin resistance (n 27) All women with PCOS (n 137) BMI ( 25 kg/m 2 ) 22.1 3.2 37.3 5* 24.3 4.7 25.4 4.7 33.3 4.3 27 5.6 FSH level (3 14 miu/ml) 4.9 3 4.4 1.1 5.7 2 5.1 1.6 4.8 1.4 5.1 1.6 LH level (3 20 miu/ml) 8.6 7.9 8.6 8.8 6 2.4 9.1 (6.3 14.5) 8.6 4.7 8.7 (6.2 14) A level (0.4 4.5 ng/ml) 2.7 0.7 2.4 0.8 3 1.1 4.2 1.5 4.6 1.7 4.3 1.6 T level (0.2 0.8 ng/ml) 0.3 0.2 0.5 0.2 0.5 0.2 0.8 0.4 1.2 0.6# 0.9 0.4 Insulin level (8 25 U/mL) 7.8 3.7 18.8 4.3* 10.3 4.9 11.1 5.3 35.7 19.8 16 13.9 SHBG level (11 124 nmol/l) 70.8 42.7 31.7 14.4 82.9 52.8 42.3 21.7 15.9 8.2 37.5 22.4 17-OHP level (0.3 3 ng/ml) 1.5 1.5 1 0.6 0.9 0.4 1.4 (1 1.8) 1.3 0.7 1.3 (1 1.8) DHEAS level (0.7 3.9 ng/ml) 2.1 1 2.3 1.3 2.2 1.1 3.1 1.4 2.9 1.7 3.1 1.4 Triglycerides (40 128 mg/dl) 95.6 67 134.6 79.3 67.9 32.1 96.3 50.5 162 61 112.5 60.1 Obese women (BMI 30 kg/m 2 ) (%) 100 11.4 14.5 77.7 27 Insulin resistance (%) 0 11 0 0 100 20 Significant correlations (r): Normal obese women BMI/insulin 0.803 0.552 0.349 0.428 0.578 BMI/T 0.505 0.247 Insulin/T 0.209 0.373 BMI/SHBG 0.413 T/SHBG 0.506 BMI/triglycerides 0.397 0.434 Insulin/triglycerides 0.469 0.523 T/triglycerides 0.486 0.440 Insulin/SHBG 3.343 0.368 Note: Values are expressed as means SD or median (25th and 75th percentiles). SHBG was studied in 19, 7, 10, 36, 8, and 44 patients of each group. r 0.413 correlation coefficient. * P.001 versus normal women. P.01 versus normal obese women. P.05 versus normal obese women. P.001 versus patients with PCOS without insulin resistance. P.001 versus normal obese women. P.01 versus normal women. # P.01 versus patients with PCOS without insulin resistance. PCOS without insulin resistance) and of course BMI, insulin, and triglycerides, whereas SHBG was decreased. [3] In the total group of PCOS versus normal and obese women, we observed that A, T, and DHEAS, and insulin, were significantly higher, and that SHBG was lower. Twenty percent of these women had insulin resistance versus 2.7% in non-pcos (only in obese women). Correlations were again positive and statistically significant between BMI and insulin in all women with PCOS, between insulin and T in those patients without insulin resistance, and between BMI, T, and triglycerides in the total group of patients with PCOS. The correlation between insulin and SHBG was still negative and statistically significant in women with PCOS without insulin resistance and in the total group of patients with PCOS. Table 2 shows the mean values and SDs (or the median and 25th and 75th percentiles) of the same indices as in Table 1 and correlations among them, comparing obese versus nonobese women in all patients with and without PCOS. Compared with nonobese women, obese women without PCOS had significantly higher insulin and triglycerides and lower SHBG. Among patients with PCOS, the obese ones had higher T, insulin, and triglycerides compared with the nonobese ones, and lower LH and SHBG. Both groups of PCOS, compared with the corresponding non- PCOS group (obese and nonobese), had higher A, T, and insulin; and in nonobese women with PCOS (vs. non- PCOS), besides, higher LH, 17-OHP, DHEAS, triglycerides, and BMI, and less SHBG. The percentage of insulin resistance in nonobese patients with PCOS was 6%, versus 0% in those without PCOS, and it was 57% in obese women with PCOS versus 8.7% in obese women without PCOS. Again, the correlation between BMI and insulin was significantly high in both women with FERTILITY & STERILITY 35

TABLE 2 Body mass index and serum levels in the follicular phase of gonadotropins, androgens, insulin, SHBG, and triglycerides, as well as correlations among them, in the total cases studied with or without PCOS, and in obese and nonobese women. Women without PCOS (n 75) Women with PCOS (n 137) Studied variables (normal values) Not obese (BMI 30 kg/m 2 ) (n 52) Obese (BMI 30 kg/m 2 )(n 23) Not obese (BMI 30 kg/m 2 ) (n 100) Obese (BMI 30 kg/m 2 ) (n 37) BMI ( 25 kg/m 2 ) 22.5 3 36.6 4.9* 24.2 3.1 34.5 3.4 FSH level (3 14 miu/ml) 5.4 2.6 4.8 1.4 5.2 1.5 4.8 1.5 LH level (3 20 miu/ml) 6.5 (4.5 8.2) 7.6 7.6 11.8 8 7.4 4.6 A level (0.4 4.5 ng/ml) 2.8 1 2.5 1 4.2 1.5 4.4 1.9 T level (0.2 0.8 ng/ml) 0.4 0.2 0.5 0.2 0.8 (0.6 1.1) 1.04 0.5 Insulin level (8 25 U/mL) 8.8 4.2 18.5 4.1* 9.4 (7.7 14.1) 29.3 20 SHBG level (11 124 nmol/l) 75.3 46.4 31.7 13.3* 41.5 22.2 19.4 12.8# 17-OHP level (0.3 3 ng/ml) 1 (0.7 1.3) 0.9 0.5 1.4 (1 1.8) 1.3 0.7 DHEAS level (0.7 3.9 ng/ml) 2.1 0.9 2.2 1.2 3.1 1.4 2.9 1.6 Triglycerides (40 128 mg/dl) 71.1 42.1 130.4 74.4** 96.4 54 145.5 58.5 Obese women (BMI 30 kg/m 2 ) (%) 29.3 27 Insulin resistance (%) (0) 2.7 (2/23) (6/100) 20 (21/37) Significant correlations (r) BMI/insulin 0.746 0.582 BMI/T 0.285 0.254 Insulin/T 0.373 BMI/SHBG 0.40 T/SHBG 0.34 BMI/triglycerides 0.524 0.434 Insulin/triglycerides 0.393 0.524 T/triglycerides 0.440 Insulin/SHBG 0.370 Note: Values are expressed as means SD or median (25th and 75th percentiles). SHBG was studied in only 27, 9, 36, and 8 patients of each group. r 0.40 correlation coefficient. * P.001 versus nonobese women. P.01 versus the corresponding group, obese and nonobese women who do not have PCOS. P.001 versus nonobese women with PCOS. P.001 versus the corresponding group, obese and nonobese women who do not have PCOS. P.05 versus nonobese women with PCOS. P.05 versus the corresponding group, obese and nonobese women who do not have PCOS. # P.01 versus nonobese women with PCOS. ** P.01 versus nonobese women. PCOS and without PCOS, but less between BMI and T; and only significant between insulin and T in patients with PCOS. SHGB showed a negative correlation with BMI in non-pcos, and with insulin in patients with PCOS. Triglycerides also showed a statistically significant positive correlation with BMI and insulin in women with PCOS and without PCOS, and with T only in patients with PCOS. Although nonobese patients with PCOS had slightly more insulin and insulin resistance than the corresponding group without PCOS, this was probably due to the higher BMI in the former group. If only women with BMI 23 kg/m 2 were compared, there were no significant differences in BMI, insulin, and triglycerides, but there were differences in androgens and SHBG (not shown in any Table). On the other hand, a good correlation between T and insulin was found in those patients with PCOS who had a slight partial deficiency of 21-OH (n 15) or adrenal involvement (n 26) (r 0.71 and r 0.46, respectively), and also with the insulin response in those insulin-resistant patients (not shown in any Table). Figures 1 and 2 analyze the values of insulin, T, SHBG, and triglycerides, and the insulin response to the overload of glucose, as a function of the different values of BMI. In women without PCOS, obesity (higher BMI, especially 30 kg/m 2 ) was moderately related to higher insulin and T, lower SHBG, and higher triglycerides. With or without the inclusion of the two cases of insulin resistance, obesity in these women without PCOS also increased the insulin response to the glucose overload (Fig. 2A). In patients with PCOS, when cases of insulin resistance were excluded, obesity had less effect than in women without PCOS. Insulin scarcely increased; T showed no change, although there was an increase in triglycerides; and the 36 Acién et al. Insulin, androgens, and obesity Vol. 72, No. 1, July 1999

FIGURE 1 Variations in mean values of insulin (A), T(B), SHBG (C), and triglycerides (D) as a function of different values of BMI in women without PCOS (o), patients with PCOS without insulin resistance (Œ), and all women with PCOS ( ). FIGURE 2 (A), Insulin response to 100 g of glucose load in women without PCOS considering BMI 25 kg/m 2 ( ) versus BMI 35 kg/m 2 (}). (B), Insulin response to 100 g of glucose load in patients with PCOS according to the four groups of BMI: 25 kg/m 2 ( ), 25 30 kg/m 2 (Œ ), 30 35 kg/m 2 (FE), and 35 kg/m 2 (}{). Dotted lines PCOS without insulin resistance; solid lines all patients with PCOS. insulin response was hardly influenced. Nevertheless, if all patients with PCOS were considered (because of the insulinresistant patients), obesity or increased BMI clearly led to higher insulin and T, lower SHBG, and higher triglycerides (Fig. 1). The insulin response increased in a correlated and proportional way with the increase in BMI, making clear the mutual influence between obesity and insulin resistance in patients with PCOS (Fig. 2B). If the analysis was performed as a function of the different values of basal insulin, we observed that in women without PCOS, an increase in insulin (even within the normal limits in all cases) increased BMI and triglycerides, had little effect on T, and lowered SHBG. Perhaps this occurred because there were many obese women in the non-pcos group, and perhaps it could also be the other way around: Obese women had higher insulin and triglycerides and lower SHBG. FERTILITY & STERILITY 37

FIGURE 3 (A), Variations in insulin response to 100 g of glucose load in patients with PCOS according to four groups of basal insulin: 15 U/mL ( ), 15 25 U/mL (Œ ), 25 35 U/mL (E), and 35 U/mL ({). Dotted lines PCOS without insulin resistance; solid lines all patients with PCOS. (B), Variations in insulin response to 100 g of glucose load in patients with PCOS according to four groups of T levels: 0.5 ng/ml ( ), 0.5 1 ng/ml (Œ ), 1 1.5 ng/ml (FE), and 1.5 ng/ml (}{). Dotted lines PCOS without insulin resistance; solid lines all patients with PCOS. increase in T. The insulin response also increased slightly with T, but this could be because there were more obese women included. In fact, women with PCOS without insulin resistance showed exactly the same insulin response to the overload of glucose independent of the T values (Fig. 3B), thus showing that T had no influence on the insulin values or on the response to the overload of glucose in patients with PCOS. Nor did BMI, insulin, or SHBG change in relation to T in noninsulin-resistant patients with PCOS. However, in the total group of patients with PCOS (including the cases with insulin resistance) and with hyperandrogenism (T 1 ng/ml), there was a slight but statistically significant increase in BMI, insulin, and triglycerides, as well as in the insulin response (Fig. 3B). This could be due, however, to the hyperinsulinemia, the increased BMI, and the associated hyperandrogenism. DISCUSSION Thus, in women without PCOS (noninsulin resistant), when insulin increased (from a BMI increase), the insulin response to an overload of glucose increased as well. This did not occur in patients with PCOS without insulin resistance. However, in the total group of patients with PCOS (insulin-resistant included), when insulin increased, we observed a very strong hyperresponse of this (corresponding to insulin-resistant patients) (Fig. 3A). BMI, T, and triglycerides also increased when basal insulin did. Finally, if the analysis was performed as a function of the different values of T, then when there was a slight increase in T, in women without PCOS, there was a slight increase in BMI and a decrease in SHBG, without any modification of insulin and triglycerides. These findings again suggest that few changes were a consequence of obesity and not of an Prelevic (12) has recently pointed out that hyperinsulinemia was observed in 30% of slim women and in 75% of obese women with PCOS. Despite certain conceptual problems about PCOS, hyperinsulinemia, and insulin resistance, it seems evident that insulin and obesity are strongly correlated and that both influence each other regardless of whether women are normal or have PCOS, are normoandrogenic or hyperandrogenic. In some studies (4, 6, 13), it has been reported that both obese and nonobese women with PCOS were more insulin-resistant and hyperinsulinemic than age- and weight-matched normal women. However, other studies have failed to find hyperinsulinemia in slim women with PCOS, and with a thorough methodologic approach, insulin sensitivity was found to be normal in muscle, liver, and adipose tissue in slim women with PCOS (14). In our study, all hyperinsulinemic and insulin-resistant patients had BMI 25 kg/m 2 (and in 76% of cases, it was 30 kg/m 2 ), and although most of them had PCOS, a positive and significant correlation between BMI and insulin was present in all cases with or without PCOS. Slim patients with PCOS had insulin values similar to normal women of the same weight, even though the former were hyperandrogenic and had low SHBG. However, it is also true that insulin and insulin resistance are higher with a higher BMI, and the other way around, especially in women with PCOS (Figs. 1 and 2). Among normal women, and women without PCOS in general, there was a high correlation between BMI and insulin (r 0.75), so that obese women without PCOS had significantly higher insulin, and, in some cases, insulin resistance (with BMI of 38 kg/m 2 ). These normal obese women also had slightly higher T, lower SHBG, and higher triglycerides, suggesting that all this was due to obesity, with 38 Acién et al. Insulin, androgens, and obesity Vol. 72, No. 1, July 1999

a consequent decrease in sensitivity to insulin (increase in resistance). Among women with PCOS, there are both obese and slim women, and there was a correlation between BMI and insulin (and SHBG) that was lower than that in women without PCOS (r 0.58). In these women, there was also a smaller correlation between insulin and T, and BMI, insulin, and T with triglycerides. However, among these women with PCOS, two clearly differentiated groups can be distinguished according to the presence or absence of insulin resistance. [1] Women with PCOS without insulin resistance are simply more hyperandrogenic than normal women or those without PCOS: LH, A, T, 17-OHP, and DHEAS were higher and SHBG was lower; but there were no clear differences in BMI, insulin, insulin response, or even in triglycerides. [2] Women with PCOS with insulin resistance had normal values of LH but were more hyperandrogenic, with high BMI, hyperinsulinemia, and higher triglycerides. All of them were obese, and they naturally showed insulin hyperresponse, which increased with higher BMI, basal insulin, and (to a lesser extent) T. As mentioned earlier, slim women with PCOS showed insulin and metabolic indices similar to those of women without PCOS, although they had significantly higher LH and androgens and lower SHBG. Insulin seemed to have no influence. However, the obesity in most obese women with PCOS seemed to be due to or related to insulin resistance, and although some patients are simply obese (as among normal ones), the obese insulin-resistant women with PCOS are more hyperandrogenic with higher metabolic risks. Therefore, two different types of obesity seem to exist. Primary obesity is due perhaps to psychological and nutritional factors, to an alteration of leptin, or to an alteration in -endorphins of the endogenous opioid system. This type is probably a global obesity (type 1) or a gynecoid one (type 4) of Lefebvre et al. (3). Secondary obesity may be due to insulin resistance and related hormonal (androgens) and metabolic alterations (lipid metabolism); this would more likely be an android obesity (types 2 and 3 of Lefebvre et al.) (3). Primary obesity in severe cases also leads to insulin resistance, alterations of lipid metabolism, hyperleptinemia (increase in triglycerides, decrease in insulin receptors, and hyperinsulinemia), higher -endorphin levels, and lower SHBG (2, 15 17). Therefore, it also seems likely that two kinds of insulin resistance exist. The primary type is probably due to a genetic anomaly (genes involved in the secretion and action of insulin) and is related to hyperandrogenism. The secondary type may be a consequence of a pronounced global or gynecoid obesity. Based on our results, at least three entities or clearly defined disorders must be differentiated in patients with obesity and/or hyperandrogenism and/or hyperinsulinemia and insulin resistance. [1] Simple obesity is not hyperandrogenic but involves increased levels of insulin (sometimes with insulin resistance) and triglycerides, and decreased SHBG levels. [2] Typical PCOS is not hyperinsulinemic and is probably due to anomalies in the genetic coding of steroidogenic enzymes, therefore affecting the regulation of the secretion and action of androgens (18, 19). The increase in androgens will lead to various events, such as a decrease in SHBG, an increase in estrogens, an alteration in the pulsatile release of LH, an increase in serum LH, arrest of follicular development, and hyperthecosis. These events contribute to the vicious cycle that causes polycystic ovaries, apart from the possible intervention of genes regulating folliculogenesis, leading to higher hyperandrogenism (more follicles, more inhibin, more estrogens, less FSH, and more LH) and anovulation. Hyperinsulinemia is not evident in these patients, although it increases with BMI, but clinical signs are evident, including hyperandrogenism (acne, hirsutism), anovulation-amenorrhea, and the usual ultrasound appearance of the polycystic ovaries. Perhaps the most suitable treatment is antiandrogens (e.g., cyproterone acetate, flutamide, finasteride), the combination of ethinyl estradiol and cyproterone acetate as a contraceptive pill (7), or ovulation inducers. [3] As mentioned, insulin-resistant PCOS could be due to anomalies in the genes involved in the secretion and action of insulin. Franks et al. (18) concluded that the VNTR (variable number tandem repeats) of the insulin gene is a major susceptibility locus for certain patients with PCOS and may contribute to the mechanism of hyperinsulinemia and to the high risk of diabetes type 2 in women with PCOS. These patients may have hyperinsulinemia and insulin resistance, and therefore secondary android obesity, hyperleptinemia, alterations of lipid metabolism, anovulation, hyperthecosis, and secondary hyperandrogenism because of the decrease in SHBG and insulin-like growth factor binding protein-1 and the increase in -endorphins (1, 2, 18). Naturally, these women have high metabolic, atherogenic, hypertensive, renal, and cardiovascular risks. The most appropriate treatment is probably the combination of diet and oral antidiabetic drugs (metformin; or with less indication, troglitazone or diazoxide) (20, 21). Antagonist opioids (naloxone and naltrexone) also could be useful in these cases; the combination of GnRH analogues and a contraceptive pill containing ethinyl estradiol and cyproterone acetate has also been useful (7). Barbieri (22), Dale et al. (23), and other investigators have also described two groups of patients with PCOS, similarly differentiated. However, there can also be some mixed cases, and in PCOS groups with or without insulin resistance, some kind of adrenal involvement may occur. Moghetti et al. (24) reported that hyperinsulinemia may stimulate cytochrome P450c 17 activity in the adrenal glands of women with PCOS. We have observed the best correlation between insulin and T in women with PCOS who FERTILITY & STERILITY 39

had a slight partial deficit of 21-OH or adrenal involvement. But the three entities mentioned could have a defined genetic origin, show clinical differences, and have different consequences. Likewise, the therapeutic possibilities will be different. In short, insulin, androgens, and BMI are related in women with PCOS and without PCOS, especially in obese ones; and three types of disorders can be differentiated: simple nonhyperandrogenic obesity, typical nonhyperinsulinemic PCOS, and insulin-resistant PCOS. The latter occurs in the most hyperandrogenic obese women and causes the highest metabolic, renal, and cardiovascular risks. References 1. Nestler JE. Insulin regulation of human ovarian androgens. Hum Reprod 1997;12(Suppl 1):53 62. 2. Pasquali R, Casimirri F, Vicennati V. Weight control and its beneficial effect on fertility in women with obesity and polycystic ovary syndrome. Hum Reprod 1997;12(Suppl 1):82 7. 3. Lefebvre P, Bringer J, Renard E, Bonlet F, Clonet S, Jaffiol C. Influences of weight, body fat patterning and nutrition on the management of PCOS. Hum Reprod 1997;12(Suppl 1):72 81. 4. Dunaif A, Graf M. Insulin administration alters gonadal steroid metabolism independent of changes in gonadotropin secretion in insulinresistant women with the polycystic ovary syndrome. J Clin Invest 1989;83:23 9. 5. Rajkhowa M, Bicknell J, Jones M, Clayton R. Insulin sensitivity in women with polycystic ovary syndrome: relationship to hyperandrogenemia. Fertil Steril 1994;61:605 12. 6. Dunaif A, Green G, Futterweit W, Dobrjansky A. Suppression of hyperandrogenism does not improve peripheral or hepatic insulin resistance in the polycystic ovary syndrome. J Clin Endocrinol Metab 1990;70:699 704. 7. Acién P, Mauri M, Gutierrez M. Clinical and hormonal effects of the combination gonadotropin-releasing hormone agonist plus oral contraceptive pills containing ethinyl-oestradiol (EE) and cyproterone acetate (CPA) versus the EE-CPA pill alone on polycystic ovarian diseaserelated hyperandrogenism. Hum Reprod 1997;12:423 9. 8. Falsetti L, Eleftherion G. Hyperinsulinemia in the polycystic ovary syndrome: a clinical, endocrine and echographic study in 240 patients. Gynecol Endocrinol 1996;10:319 26. 9. Ciampelli M, Fulghesu AM, Murgia F, Guido M, Cucinelli F, Apa R, et al. Acute insulin response to intravenous glucagon in polycystic ovary syndrome. Hum Reprod 1998;13:847 51. 10. Barbieri R, Ryan K. Hyperandrogenism, insulin resistance, and acanthosis nigricans syndrome: a common endocrinopathy with distinct pathophysiologic features. Am J Obstet Gynecol 1983;147:90 101. 11. Millet-Serrano A, Hernández A, Millet A, Dolz M, Rodríguez M. Ovario poliquístico: resistencia a la insulina. Obstet Ginecol Esp 1998; 7:71 7. 12. Prelevic G. Insulin resistance in polycystic ovary syndrome. Curr Opin Obstet Gynecol 1997;9:193 201. 13. Ehrmann DA, Sturis J, Byrne MM, Karrison T, Rosenfield RL, Polonsky KS. Insulin secretory defects in polycystic ovary syndrome. Relationship to insulin sensitivity and family history of non-insulin-dependent diabetes mellitus. J Clin Invest 1995;96:520 7. 14. Ovensen P, Moller J, Ingerslev HJ, Jorgensen JO, Mengel A, Schmitz O, et al. Normal basal and insulin-stimulated fuel metabolism in lean women with the polycystic ovary syndrome. J Clin Endocrinol Metab 1993;77:1636 40. 15. Gennarelly G, Holte J, Wide L, Berne C, Lithell H. Is there a role for leptin in the endocrine and metabolic aberrations of polycystic ovary syndrome? Hum Reprod 1998;13:535 41. 16. Geisthövel F, Meysing A, Brabant G. C-peptide and insulin, but not C19-steroids, support the predictive value of body mass index on leptin in serum of premenopausal women. Hum Reprod 1998;13:547 53. 17. Segal K, Landt M, Klein S. Relationship between insulin sensitivity and plasma leptin concentration in lean and obese men. Diabetes 1996;45: 988 91. 18. Franks S, Gharani N, Waterworth D, Batty S, White D, Williamson R, et al. The genetic bases of polycystic ovary syndrome. Hum Reprod 1997;12:2641 8. 19. Gharani N, Waterworth DM, Batty S, White D, Gilling-Smith C, Conway GS, et al. Association of the steroid synthesis gene CYP11a with polycystic ovary syndrome and hyperandrogenism. Hum Mol Genet 1997;6:397 402. 20. Van Montfrans JM, van Hoof MHA, Hompes PGA, Lambalk CB. Treatment of hyperinsulinaemia in polycystic ovary syndrome? Hum Reprod 1998;13:5 6. 21. Morin-Papunen LC, Koivunen RM, Ruokonen A, Martikainen HK. Metformin therapy improves the menstrual pattern with minimal endocrine and metabolic effects in women with polycystic ovary syndrome. Fertil Steril 1998;69:691 6. 22. Barbieri RL. Hyperandrogenic disorders. Clin Obstet Gynecol 1990; 33:640 54. 23. Dale PO, Tanbo T, Vaaler S, Abyholm T. Body weight, hyperinsulinemia, and gonadotropin levels in the polycystic ovarian syndrome: evidence of two distinct populations. Fertil Steril 1992;58:487 91. 24. Moghetti P, Castello R, Negri C, Tosi F, Spiazzi G, Brun E, et al. Insulin infusion amplifies 17-alpha-hydroxy-corticosteroid intermediate response to adrenocorticotropin in hyperandrogenic women: apparent relative impairment of 17,20-lyase activity. J Clin Endocrinol Metab 1996;81:881 6. 40 Acién et al. Insulin, androgens, and obesity Vol. 72, No. 1, July 1999