Changes in luteinizing hormone and insulin secretion in polycystic ovarian syndrome

A.M. Fulghesu, F. Cucinelli, V. Pavone, F. Murgia, M. Guido, A. Caruso, S. Mancuso and A. Lanzone1

Department of Obstetrics and Gynecology, Catholic University of Sacred Heart, L.go A. Gemelli 8, 00168 Rome, Italy


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Uncertainties regarding the pathogenetic changes underlying the polycystic ovarian syndrome (PCOS) have been reported. The aim of this study was to investigate the endocrine and metabolic features of PCOS patients in relation to luteinizing hormone (LH) secretion. Androgen assays, oral glucose tolerance tests, hyperinsulinaemic euglycaemic clamps and gonadotrophin releasing hormone (GnRH) tests were performed in 100 patients. Sixty-six patients scheduled as hyperinsulinaemic and 34 as normoinsulinaemic showed similar concentrations of LH, follicle stimulating hormone (FSH), LH/FSH ratio, and LH response to GnRH testing. Hyperinsulinaemic subjects showed higher body mass index (BMI), insulin resistance, testosterone and free androgen index levels compared with those of normoinsulinaemic subjects; when clustered in relation to their LH basal concentrations, the two groups obtained differed only in androstenedione concentrations. Considering both insulin and LH plasma concentrations, four groups were obtained. Hyperinsulinaemia and hyper-LH secretion were not related in 54% and coexisted in the same subjects in 26% of cases. Hyperinsulinaemia as well as hyper-LH secretion affected the expression of the syndrome; the insulinaemia was directly correlated with testosterone concentrations and all metabolic parameters that affected the free androgen index. The LH concentrations were related to androgen production and were independent of BMI and insulin concentrations. It is concluded that the degree of hormonal alteration is the final sum of such pathogenetic factors.

Key words: androgen/insulin/LH/obesity/PCOS


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Since Stein and Leventhal (1935) first described the hyperandrogenic syndrome that later bore their name, the polycystic ovarian syndrome (PCOS) has been diagnosed variably on the basis of multiple biochemical features associated with ovarian morphology, clinical symptoms and hormonal findings, including elevated gonadotrophins (Keetel et al., 1957Go; Rebar et al., 1976Go), serum androgens (Bergman et al., 1985Go), and insulin concentrations (Burghen et al., 1980Go; Dunaif et al., 1989Go; Fulghesu et al., 1993Go). Altered gonadotrophin secretion, which in the past was considered to be the causal factor in the pathogenesis of the syndrome (Yen and Jaffer, 1986Go), is no longer considered a universal finding in PCOS and should not be evaluated in the diagnosis (Hall and Crowley, 1996Go). A number of studies of families with several cases of PCOS have produced results suggesting an autosomal dominant trait, even if a more complex aetiology seems more likely (Franks et al., 1997Go). On the other hand the development of PCOS is not associated with obesity and insulin. In fact, ~ 50% of PCOS women are not obese (Yen, 1980Go; Franks, 1989Go), and 40% of PCOS subjects show increased insulin concentrations independent of obesity (Dunaif et al., 1988Go).

Recent studies have reported hyperandrogenaemia and hyperinsulinaemia in obese PCOS subjects with relatively normal luteinizing hormone (LH) concentrations, whereas higher LH concentrations with relatively normal insulin concentrations have been verified in normal-weight PCOS subjects (Dale et al., 1992Go; Antilla et al., 1993). This isolated observation led to the proposal for the subdivision of PCOS subjects on the basis of either endocrine or metabolic inputs leading to the development of ovarian hyperandrogenism. However, several authors failed to divide their PCOS patients into these two distinct subgroups (Grulet et al., 1993Go; Tropeano et al., 1994Go).

Thus, the assessment of the relative contributions of obesity, insulin resistance and changes in gonadotrophins (or neuroendocrine changes) and their impact on hyperandrogenism and chronic anovulation are pivotal in the understanding of this complex syndrome. Moreover, the alteration of insulin sensitivity and secretion constitute a specific risk factor in PCOS patients for developing abnormalities of glucose metabolism in pregnancy (Paradisi et al., 1998Go). The aim of this study is to investigate simultaneously the endocrine and metabolic features in relation to LH secretion and its effects on ovarian steroid concentrations in a large PCOS population classified according to their insulinaemic status.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Subjects
A total of 100 recruitable consecutive patients ranging from 19–36 years old who were affected by PCOS participated in the study. PCOS was diagnosed by the findings of bilaterally normal or enlarged ovaries (maximal ovarian volume >12 mm3x103) with the presence of at least 7–10 microcysts per ovary (<5mm diameter) at the time of ultrasonography and laparoscopy, and the presence of the following clinical and endocrine factors: amenorrhoea or oligomenorrhoea, and elevated plasma androgen concentrations. The presence of impaired glucose tolerance or diabetes was considered an exclusion criterion.

None of the patients had hyperprolactinaemia (normal range of prolactin, 3.5–26.5 µg/l; conversion factor, 1.0) or clinical evidence of hypercorticism or thyroid dysfunction. Late-onset congenital hyperplasia was excluded by an adrenocorticotrophic hormone stimulation test. All had spontaneous onset of puberty and sexual development. No patient had taken medication known to affect carbohydrate metabolism or plasma steroid concentrations for at least 3 months before the study. Informed consent was obtained from each of the patients.

Clinical protocol
On a random day (at least 7 days after menstrual bleeding) PCOS women were hospitalized. Recent ovulation was excluded by retrospective measurement of serum progesterone concentrations on the day of the study. After 3 days of a standard 300-g carbohydrate diet, patients were subjected to a determination of hormonal and sex hormone binding globulin (SHBG) baseline concentrations and, after a 12-h overnight fasting, to a 75-g oral glucose tolerance test (OGTT).

Insulin (I) sensitivity was assessed by the use of the hyperinsulinaemic euglycaemic clamp technique (Bergman et al., 1985Go). The test was performed the day after the OGTT, after a 12-h overnight fasting. A retrograde i.v. catheter was inserted into a forearm vein for blood sampling and kept in a warming device at 60°C to arterialize the venous blood samples. Another indwelling catheter was inserted in the contralateral forearm vein (Cavafix; B.Braun, Melsungen, Germany) for the glucose and insulin infusions. A two-step primed constant infusion of human insulin (Actrapid HM; Novo Nordisk, Denmark) was administered at a rate of 40 mIUxm2xmin–1. Having reached insulin circulatory concentrations of about 717 pmol/l within 10 min, the steady-state velocity of insulin infusion was fixed in order to maintain these levels during the clamp. Blood samples were taken every 5 min from the arterialized line, and blood glucose concentrations were immediately measured on a glucose analyser. The exogenous glucose infusion was adjusted according to a standard algorithm to maintain the blood glucose concentration between 4.4 and 4.9 mmol/l (Bergman et al., 1985Go). On the following day, a gonadotrophin releasing hormone (GnRH) test was performed by i.v. infusion of 100 µg of GnRH (Relisorm; Serono, Rome, Italy). Blood samples were obtained at 0, 30, 60, 90 and 120 min after the bolus.

The study protocol was approved by the department ethical committee.

Data analysis
The body mass index (BMI) was calculated according the formula: body weight (kg)/height (m2). The presence of obesity was considered for BMI values >25. Plasma baseline concentrations of gonadotrophins, oestradiol, sex hormone binding globulin, androstenedione, testosterone, 17-hydroxyprogesterone (17-OHP) and dihydroepiandrosterone sulphate (DHEAS) were measured. The basal gonadotrophin concentrations were obtained by pooling two blood samples obtained on two different days. The free androgen index (FAI) was calculated as follows: [T]x(6.11–2.38x(log10 [SHBG]) (Rajkhowa et al., 1994Go).

Insulin and glucose serum concentrations were analysed in all samples under oral glucose stimulus. A normal glycaemic response to OGTT was defined according to the criteria of the National Diabetes Data Group (1979) (normal glucose tolerance: plasma glucose at 0 min less than 115, peak value less than 200, 120 min less than 140 mg/dl; impaired glucose tolerance: 0 min 115–140, peak value >200, 120 min >200 mg/dl; diabetes mellitus: 0 min >140, peak value >200, 120 min >200 mg/dl, conversion factor to SI unit, 0.05551). Insulinaemic and glycaemic responses to glucose load were calculated as area under the curve (AUC) calculated by the trapezoidal rule. The insulinaemic response to the oral glucose tolerance test was considered normal when the insulin concentrations were under 717 pmol/l (conversion factor 7.175) at 30 and 60 min and the area under the curve lower than 107 550 pmol/lx240 min which corresponds to the mean + 2SD of a normal lean control population aged from 20 to 40 years (as established by the standard procedures of our laboratory) (Fulghesu et al., 1997Go).

During the glucose clamp, total body glucose utilization (M) was determined between 60 and 240 min, and expressed as mgxkg BW–1 x min–1. We have preferred this index for the measure of the insulin sensitivity because the M/I ratio fails to narrow the range of the individual sensitivity values (Bergman et al., 1985Go). Serum LH and follicle stimulating hormone (FSH) responses to GnRH were determined and expressed as AUC calculated by the trapezoidal rule.

Data are presented as mean ± SD. Data were stored and analysed using SPSS (Statistical Package for Social Sciences; release 5.0) on a IBM-compatible computer. The Kolmogorow–Smirnow test was performed to assess differences in the general shapes of distribution. The Wilcoxon and Mann–Whitney paired and unpaired test were used for non-parametric comparisons. A P-value < 0.05 was considered significant.

The K-means cluster analysis was performed to identify relatively homogeneous groups of cases based on significantly different LH concentrations. Simple and multiple correlation analysis was performed by Spearman and stepwise methods.

Assays
Glucose concentrations were determined by the glucose oxidase technique. Plasma samples for hormone determinations were maintained at –20°C until assayed. All hormones were measured by radioimmunoassay methods using commercial kits. For LH and FSH assays, a monoclonal double-antibody technique was used. Steroids were assayed by the dextran–charcoal technique. Intra-assay and inter-assay coefficients of variation were below 8 and 15% respectively for all hormones (Radim, Pomezia, Italy).


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Based on the insulinaemic response to glucose load, 66 patients were hyperinsulinaemic (42 obese and 24 lean subjects) and 34 were normoinsulinaemic (14 obese and 20 lean subjects). The BMI values were higher in hyperinsulinaemic patients than in the normoinsulinaemic group (27.6 ± 5.2 and 24.5 ± 5.7 kg/m2 respectively; P < 0.001). Both fasting and after-load glucose concentrations were similar in the two groups (fasting glucose: hyperinsulinaemic 79.6 ± 7.2 mg/dl, normoinsulinaemic 75.6 ± 9 mg/dl; glucose AUC: hyperinsulinaemic 27 529 ± 5870, normoinsulinaemic 24 832 ± 3243; conversion factor, 0.05551).

Table IGo shows the metabolic and hormonal characteristics of the hyper- and normoinsulinaemic groups. The insulin fasting levels as well as the insulin resistance index were higher in the hyperinsulinaemic group than in the normal group. The two populations showed similar concentrations of LH and FSH, LH/FSH ratio and LH-AUC under GnRH. The circulating androstenedione, 17-OHP, and DHEAS concentrations were similar in both groups. Hyperinsulinaemic patients had higher testosterone and decreased SHBG concentrations, so that the FAI index was also elevated in this group.


View this table:
[in this window]
[in a new window]
 
Table I. Metabolic and endocrine features in normo- and hyperinsulinaemic polycystic ovarian syndrome subjects
 
Table IIGo shows the metabolic and endocrine features of the patients grouped in relation to their LH baseline level. In the absence of a cut-off, two groups with significantly different concentrations were obtained by cluster analysis: in the first, made up of 63 subjects and characterized by the lower LH concentrations, the mean concentration was 6.7 ± 2.3 IU/l; and in the second, made up of 37 subjects, the mean LH concentration was 14.9 ± 4.8 IU/l (P < 0.01). The LH-AUC and the LH/FSH ratio were significantly different between the two groups, whereas the BMI and all metabolic parameters were superimposable. The androstenedione concentrations were higher in the high-LH group. Moreover, the patients were grouped both in relation to LH basal concentration and insulin-AUC. The four groups were characterized by: low concentrations of both insulin-AUC and LH (NN 20 pts); normal concentrations of insulin-AUC and high LH (NH 14 pts); high concentrations of insulin-AUC and normal LH (HN 40 pts); and high concentrations of both insulin-AUC and LH (HH 26 pts).


View this table:
[in this window]
[in a new window]
 
Table II. Metabolic and endocrine features in polycystic ovarian syndrome subjects clustered by luteinizing hormone (LH) concentrations
 
Figure 1Go shows the BMI, the fasting insulin, the M value, and the LH/FSH ratio as well as the LH-AUC and the other hormonal parameters in the four groups. The androstenedione concentrations were significantly elevated in the HH group compared with those of the other groups, while no difference was observed in testosterone, DHEAS and 17-OHP concentrations. The SHBG concentration was reduced in the presence of elevated insulin concentrations (HN versus NN and NH, P < 0.01) so that the FAI indexes were higher in HN and HH groups than in the NN.



View larger version (44K):
[in this window]
[in a new window]
 
Figure 1. Left panel: from top to bottom, body mass index (BMI), fasting insulin, insulin area under curve after glucose load (I-AUC), androstenedione (A) and free androgen index (FAI) are shown in normoinsulinaemic polycystic ovarian syndrome (PCOS) subjects with low basal luteinizing hormone (LH) (NN) {square} and high basal LH plasma concentrations (NH) {blacksquare}, in hyperinsulinaemic PCOS subjects with low basal LH (HN) {square}, and high basal LH (HH) {square}. Right panel: from top to bottom, LH/follicle stimulating hormone (FSH) ratio (LH/FSH), LH area under curve after gonadotrophin releasing hormone i.v. bolus (100 µg) (LH-AUC), sex hormone binding globulin (SHBG), dihydroepiandrosterone sulphate (DHEAS), testosterone (T) and 17-{alpha} hydroxyprogesterone (17-OHP) are shown in normoinsulinaemic PCOS subjects with low basal LH (NN) {square} and high basal LH plasma levels (NH) {blacksquare}, in hyperinsulinaemic PCOS subjects with low basal LH (HN) {square} and high basal LH (HH) {square}. Significance: *NN versus NH, P <= 0.01; {dagger}NN versus HN, P <= 0.01; {ddagger}NN versus HH, P <= 0.01; $NH versus HN, P <= 0.01; #NH versus HH, P <= 0.01; §HN versus HH, P <= 0.01.

 
In order to evaluate the relevance of obesity, insulin concentrations and LH or LH-AUC in the pathogenesis of the syndrome, both linear and multiple correlations were studied. Table IIIGo shows the linear correlations between obesity, insulin secretion, LH secretion and the androgen plasma concentrations in our PCOS population. The BMI was positively related with both fasting and glucose-stimulated insulin secretion. Moreover, a positive relation with testosterone and a negative relation with SHBG concentrations determined a positive significant correlation with the FAI.


View this table:
[in this window]
[in a new window]
 
Table III. Linear correlation between steroid and metabolic status of all patients
 
The I-AUC was significantly related with the fasting insulin, negatively with SHBG concentrations, and positively with testosterone and FAI. There was no correlation between insulin and the androstenedione, 17-OHP and DHEAS concentrations. The baseline LH concentration was significantly related with LH-AUC, androstenedione, testosterone and FAI, whereas the LH-AUC value was only related with baseline LH concentrations.

Table IVGo shows the impact of obesity, I-AUC and basal LH concentration on the PCOS hormonal characteristics. It is well known that the frequent coexistence of these factors in the same subject make it very difficult to understand the pathogenetic role of the single factors. To overcome this problem, we applied a partial correlation analysis to the three variables significantly related to the testosterone and FAI values. This approach makes it possible to assess the importance of a single parameter by removing the other two confounding factors. The table indicates that BMI itself was related only to the FAI index, probably through the decrease of the SHBG concentrations. Circulating insulin influenced both the total and free testosterone values, whereas the LH was related only with testosterone ovarian production.


View this table:
[in this window]
[in a new window]
 
Table IV. Direct and partial coefficients of correlation between androgens, and gonadotrophic and metabolic factors
 
Owing to the absence of a linear correlation with androgen concentrations, it was not possible to analyse the impact of LH-AUC.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
PCOS has been one of the most explored and controversial areas in reproductive medicine. The spectrum of PCOS clinical signs may appear very heterogeneous, from minimal symptoms of hyperandrogenism in lean women with regular menses (Franks, 1987Go), to severe disease in obese and amenorrhoeic patients. Tonic hypersecretion of LH, associated with hyperpulsatility of GnRH over 24 h and with an exaggerated LH pituitary response to GnRH (Morales et al., 1996Go) were for a long time considered pathognomonical factors for PCOS.

Over the last 10 years there has been increasing interest in the role of insulin and growth factors in patients with PCOS. Obesity associated with insulin resistance and hyperinsulinaemia are well-recognized features of this syndrome; however, changes in insulin concentrations have also been reported in lean subjects by several authors (Dunaif et al., 1988Go; Fulghesu et al., 1995Go). Our PCOS population reflects the syndrome heterogeneity: we found that hyperinsulinaemia affected 66% of the patients, while obesity was present in 56%. In 14 subjects, obesity did not coexist with alteration of insulin secretion, whereas 24 lean women were hyperinsulinaemic.

Contrasting data exist about the importance of obesity in determining the insulin resistance in PCOS. Dunaif et al. (1989) demonstrated that insulin resistance in PCOS is associated with a unique cellular glucose transport defect independent of, but amplified by, obesity. Several papers also reported reduced insulin sensitivity in lean PCOS subjects (Ovesen et al., 1993Go), whereas other authors reported that insulin resistance was entirely related to the adiposity in obese PCOS women and that hyperinsulinaemia was secondary to other factors in lean patients (Mahabeer et al., 1990Go; Siegel et al., 1990Go; Ciampelli et al., 1997Go). It was demonstrated that the reduced response of glucose transport to a given concentration of insulin was greater in obese than in non-obese PCOS patients. In agreement with these studies, which demonstrate a greater degree of insulin resistance in obese PCOS patients (Dunaif et al., 1989Go, 1992Go; Mahabeer et al., 1990Go; Fulghesu et al., 1998Go), we found lower M values in the hyperinsulinaemic group and a negative significant correlation between obesity and the insulin sensitivity index. The bulk of these observations indicates that hyperinsulinaemia in obese PCOS subjects is due to two factors: one characteristic of PCOS, and the other obesity-specific. Considering these observations, we prefer to analyse the influence of hyperinsulinaemia itself, whatever cause may trigger it, and not necessarily the association with obesity, which could represent a confounding factor.

The presence of insulin receptor messenger ribonucleic acid and protein in all cellular compartments of both normal and PCOS ovaries (El-Roeiy et al., 1993Go, 1994Go) supports the proposition that insulin excess may exert an endocrine impact in the ovary. In vitro, insulin has been shown to stimulate androgen secretion directly and to enhance LH-mediated responses in isolated thecal tissue to a greater degree than in normal ovaries (Bergh et al., 1993Go; Nahum et al., 1995Go). In vivo, the frequent coexistence of elevated LH and insulin concentrations leads to the more severe expression of the syndrome.

Recently, several authors proposed a subclassification of the PCOS population in two subgroups: those with obesity, hyperinsulinaemia, insulin resistance, and minimally elevated LH concentrations; and those with normal BMI, elevated LH concentrations, normoinsulinaemia, and no insulin resistance (Buyalos et al., 1992Go; Dale et al., 1992Go; Insler et al., 1993Go; Meirow et al., 1995Go). In our large study group, we failed to find a clear-cut division between the two groups described above. The presence of an altered LH axis is not easily predictable in such patients. In fact, in our patient group with elevated baseline LH concentrations, a great many of the subjects (64%) showed accompanying elevated insulin concentrations. Moreover, we found that 14% of patients presented normal concentrations of insulin and elevated LH concentrations, and 38% elevated insulinaemia and normal LH secretion, whereas the remaining 48% presented neither or both hormonal characteristics.

In our study group, altered LH basal concentrations were present in 28% of obese patients, and no relation with BMI was found. Our results are consistent with those of several authors (Srzedncka et al., 1991Go; Takai et al., 1991Go; Toscano et al., 1992Go; Buyalos et al., 1993Go; Grulet et al., 1993Go; Takahaschi et al., 1994; Norman et al., 1995Go). In all these papers, obese and non-obese PCOS patients demonstrated a similar LH and/or LH/FSH ratio. Moreover, in a paper on the importance of obesity in the determinism of PCOS, Dunaif et al. (1988) demonstrated that both obese and lean PCOS women had similar plasma LH concentrations, LH pulse amplitude and integrated LH responses to GnRH. In recent studies on the spectrum of gonadotrophin defects in a large PCOS women group, BMI correlated strongly and inversely with pool and LH pulse amplitude but not with LH pulse frequency (Morales et al., 1996Go; Arroyo et al., 1997Go). All these data raise the possibility that an obesity-associated factor may suppress the GnRH pulse amplitude or pituitary LH responsivity. Moreover, the negative correlation between LH and BMI, which has been found by some authors, suggests that the hypothetical factor may be related to adipose tissue volume. The difference in the degree of obesity reported by the authors could account for the differences in results between reports. A second possibility is represented by the fact that the fewer number of cases reported from some authors reduced the statistical impact of the smaller subgroups, influencing the study conclusions.

On the other hand, concerning the GnRH test, the LH hyper-responsivity to GnRH is not related to BMI or any androgen or metabolic parameter in our population. These data partially agree with recent results (Dale et al., 1992Go; Morales et al., 1996Go; Arroyo et al., 1997Go), but we failed to find any reduction of the LH secretory response in obese PCOS women, as described by some authors, following the injection of a low dose of GnRH (10 µg i.v.). It is important to note that in our study we used a maximal GnRH stimulus. This difference, not important in normal-weight patients, could be relevant in very obese subjects.

In conclusion, hyperinsulinaemia and hypersecretion of LH seem to be present in an independent way in our study group. Moreover, data from our study indicate that both elevated insulin and elevated LH circulating concentrations affected the clinical expression of hyperandrogenism. The insulinaemia correlated with testosterone and with all metabolic parameters, influencing markedly the free androgen index, and the LH concentrations were correlated to androgen production and were independent from BMI and insulin concentrations, so that the degree of clinical symptoms could be the final sum of such pathogenetic factors.


    Notes
 
1 To whom correspondence should be addressed Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Anttila, R., Koskinen, P., Jaatinen, T.A. et al. (1993) Insulin hypersecretion together with high luteinizing hormone concentration augments androgen secretion in oral glucose tolerance test in women with polycycstic ovarian disease. Hum. Reprod., 8, 1179–1183.[Abstract]

Arroyo, A., Laughlin, G.A., Morales, A.J. and Yen, S.S.C. (1997) Inappropriate gonadotropin secretion in polycystic ovary syndrome, influence of adiposity. J. Clin. Endocrinol. Metab., 82, 3728–3733.[Abstract/Free Full Text]

Bergh, C., Carlsson, B., Olsson, J.H., et al. (1993) Regulation of androgen production in cultured human thecal cells by insulin-like growth factor I and insulin. Fertil. Steril., 59, 323–331.[ISI][Medline]

Bergman, R.M., Finegood, D.T. and Ader, M. (1985) Assessment of insulin sensitivity in vivo. Endocr. Rev., 6, 45–75.[ISI][Medline]

Burghen, J.A., Givens, J.R. and Kitabchi, A.E. (1980) Correlation of hyperandrogenism with hyperinsulinism in polycystic ovarian disease. J. Clin. Endocrinol. Metab., 50, 113–116.[Abstract]

Buyalos, R.P., Geffner, M.E., Bersch, N. et al. (1992) Insulin and insulin like-growth factor-I responsiveness in polycycstic ovary syndrome. Fertil. Steril., 57, 796–803.[ISI][Medline]

Buyalos, R.P., Geffener, M.E., Watanabe, R.M. et al. (1993) The influence of luteinizing hormone and insulin on sex steroids and sex hormone-binding globulin in the polycystic ovarian syndrome. Fertil. Steril., 60, 626–633.[ISI][Medline]

Ciampelli, M., Fulghesu, A.M., Cucinelli, F. et al. (1997) Heterogeneity in ß cell activity, hepatic insulin clearance and peripheral insulin sensitivity in women with polycystic ovary syndrome. Hum. Reprod., 12, 1897–1901.[Abstract]

Dale, P.O., Tanbo, T., Vaaler, S. and Abyolm, H. (1992). Body weight, hyperinsulinemia and gonadotropin levels in the polycystic ovarian syndrome, evidence of two distinct population. Fertil. Steril., 3, 487–491.

Dunaif, A., Mandeli, J., Fluhr, H. and Dobrjansky, A. (1988) The impact of obesity and chronic hyperinsulinemia on gonadotropin release and gonadal steroid secretion in the polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 66, 131–139.[Abstract]

Dunaif, A., Segal K.R., Futterweit, W. and Dobrjansky, A. (1989) Profound peripheral insulin resistance, independent of obesity, in polycystic ovary syndrome. Diabetes, 38, 1165–1168.[Abstract]

Dunaif, A., Segal, K.R., Schelley, D.R. et al. (1992) Evidence for distinctive and intrinsic defects in insulin action in polycystic ovary syndrome. Diabetes, 41, 1257–1266.[Abstract]

El-Roeiy A., Chen, X., Roberts, V.J. et al. (1993) Expression of the insulin-like growth factor (IGF)-I and IGF-II and the IGF-I, IGF-II and insulin receptor genes and localization of the gene products in the human ovary. J. Clin. Endocrinol. Metab., 77, 1411–1418.[Abstract]

El-Roeiy, A., Roberts, V.J. Shimasaki, S. et al. (1994) Expression of the gene encoding the insulin-like growth factors (IGF-I and IGF-II), the IGF-I and insulin receptors and IGF binding proteins (IGFBPs 1–6), and the localization of their gene products in normal and polycystic ovary syndrome ovaries. J. Clin. Endocrinol. Metab., 78, 1488–1496.[Abstract]

Franks, S. (1987) Polycystic ovary syndrome. Ann. NY Acad. Sci., 499, 201–205.

Franks, S. (1989) Polycystic ovary syndrome, a changing perspective. Clin. Endocrinol., 31, 87–120.[ISI][Medline]

Franks, S., Gharani, N., Waterworth, D. et al. (1997) The genetic basis of polycystic ovary syndrome. Hum. Reprod., 12, 2641–2648.[Abstract]

Fulghesu, A.M., Lanzone, A., Cucinelli, F. et al. (1993) Long term naltrexone treatment reduces the exaggerated insulin secretion in patients with polycystic ovarian disease. Obstet. Gynecol., 82, 191–197.[Abstract]

Fulghesu, A.M., Ciampelli, M., Fortini, A. et al. (1995) Effect of opioid blockade on insulin metabolism in polycystic ovarian disease. Hum. Reprod., 10, 2253–2257.[Abstract]

Fulghesu, A.M., Villa, P., Pavone, V. et al. (1997) The impact of insulin secretion on the ovarian response to exogenous gonadotropins in polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 82, 644–648.[Abstract/Free Full Text]

Fulghesu, A.M., Ciampelli, M., Guido, M. et al. (1998) Role of opioid tone in the pathophysiology of hyperinsulinemia and insulin resistance in polycystic ovarian disease. Metabolism, 47, 1–5.

Grulet, H., Hecart, A.C., Delemer, B. et al. (1993) Roles of LH and insulin resistance in lean and obese polycystic ovary syndrome. Clin. Endocrinol., 38, 621–626.[ISI][Medline]

Hall, J.E. and Crowley, W.F. (1996) Elucidation of hypothalamic-pituitary-gonadal interactions in polycystic ovarian syndrome. In Filicori, M. and Flamigni, C. (eds), The Ovary, Regulation Disfunction and Treatment. Elsevier Science B.V., Marco Island, FL, pp. 287–293.

Insler, V., Shoham, Z., Barash, A. et al. (1993) Polycystic ovaries in non-obese and obese patients, possible pathophysiological mechanism based on new interpretation of facts and findings. Hum. Reprod., 8, 379–384.[Abstract]

Keetel, W.C., Bradbury, J.T. and Stoddard, F.J. (1957) Observations on the polycystic ovary syndrome. Am. J. Obstet. Gynecol., 73, 954–957.[ISI]

Mahabeer, S., Naidoo, C., Joubert, S.M. (1990) Glucose, insulin and c-peptide secretion in obese and non obese women with polycystic ovarian disease. Diabetes Res., 14, 79–82.[ISI][Medline]

Meirow, D., Yossepowitch, O., Rosler, A. et al. (1995) Insulin resistant and non-resistant polycycstic ovary syndrome represent two clinical and endocrinological subgroups. Hum. Reprod., 10, 1951–1956.[Abstract]

Morales, A.J., Laughlin, J.A., Butzow, T. et al. (1996). Insulin, somatotropin, and luteinizing hormone axis in lean and obese women with polycystic ovary syndrome, common and distinct features. J. Clin. Endocrinol. Metab., 81, 2854–2863.[Abstract]

Nahum, R., Thong, K.J. and Hillier, S.G. (1995) Metabolic regulation of androgen production by human thecal cells in vitro. Hum. Reprod., 10, 75–81.[Abstract]

National Diabetes Data Group (1979) Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes, 28, 1039–1057.[ISI][Medline]

Norman, R.J., Masters, S.C., Hague, W. et al. (1995) Metabolic approaches to the subclassification of polycystic ovary syndrome. Fertil. Steril., 63, 629–635.

Ovesen, P., Moller, J., Ingerslev, H.J. et al. (1993) Normal basal and insulin-stimulated fuel metabolism in lean women with the polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 77, 1636–1640.[Abstract]

Paradisi, G., Fulghesu, A.M., Ferrazzani, S. et al. (1998) Endocrino-metabolic features in women with polycycstic ovary syndrome during pregnancy. Hum. Reprod., 13, 542–546.[Abstract]

Rajkhowa, M., Bicknell, J., Jones, M. and Clayton, R.N. (1994) Insulin sensitivity in women with polycystic ovary syndrome, relationship to hyperandrogenemia. Fertil. Steril., 61, 605–612.[ISI][Medline]

Rebar, R., Judd, H.L., Yen, S.S.C. et al. (1976) Characterization of the inappropriate gonadotropin secretion in polycystic ovary syndrome. J. Clin. Invest., 57, 1320–1329.[ISI][Medline]

Siegel, S.F., Finegold, D.N., Lanes, R. and Lee, P.A. (1990) ACTH stimulation test and plasma dihydroepiandrosterone-sulfate levels in women with hirsutism. New Engl. J. Med., 23, 849–854.

Srzedncka, J.S., Zgliczynski, S., Wierzbicki, M. et al. (1991) The role of hyperinsulinemia in the development of lipid disturbances in nonobese and obese women with the polycystic ovary syndrome. J. Endocrinol. Invest., 14, 569–575.[ISI][Medline]

Stein, I. and Leventhal, M. (1935)Amenorrhea associated with bilateral polycystic ovaries. Am. J. Obstet. Gynecol., 9, 181–191.

Takahashi, K., Eda, Y., Abu-Musa, A. et al. (1994) Transvaginal ultrasound imaging, histopathology and endocrinopathy in patients with polycystic ovarian syndrome. Hum. Reprod., 7, 1231–1236.

Takai, I., Taii, S., Takakura, K. and Mori, T. (1991) Three types of polycystic ovarian syndrome in relation to androgenic function. Fertil. Steril., 56, 856–862.[ISI][Medline]

Toscano, V., Bianchi, P. Balducci, R. et al. (1992) Lack of linear relationship between hyperinsulinaemia and hyperandrogenism. Clin. Endocrinol., 36, 197–202.[ISI][Medline]

Tropeano, G., Lucisano, A. and Liberale, I. (1994) Insulin, c-peptide, androgen and ß-endorphin response to oral glucose in patients with polycystic ovary syndrome. J. Clin. Endocrinol. Metab., 78, 305–309.[Abstract]

Yen, S.S.C. (1980) The polycystic ovary syndrome. Clin. Endocrinol., 12, 177–207.[ISI][Medline]

Yen, S.S.C. and Jaffer, R.B. (1986) Reproductive Physiology, Pathophysiology and Clinical Management. Saunders, Philadelphia.

Submitted on May 15, 1998; accepted on November 20, 1998.