Women with PCO have a unique but poorly characterized disorder of insulin action. Obese (n = 16) and nonobese (n = 14) PCO women and age- and weight-matched normal, nondiabetic ovulatory women (obese, n = 15; nonobese, n = 17) had insulin action determined in vivo with sequential multiple insulin dose euglycemic clamps and in isolated abdominal adipocytes to clarify the mechanisms of insulin resistance. PCO resulted in significant increases in the ED50 insulin for glucose utilization in vivo (P less than 0.001) and in adipocytes (P less than 0.01), without significant changes in adipocyte insulin-binding sites. PCO also resulted in significant decreases in maximal insulin-stimulated rates of glucose utilization in vivo (P less than 0.01) and in adipocytes (P less than 0.01). Obesity resulted in smaller decreases in insulin sensitivity than PCO (ED50 insulin, P less than 0.001 in vivo and P less than 0.05 in adipocytes), but greater decreases in insulin responsiveness (Vmax, P less than 0.001 in vivo and in adipocytes). The ED50 insulin for suppression of HGP was increased only in obese PCO women (P less than 0.001), and the interactions between PCO and obesity on this parameter were statistically significant. No significant correlations between androgen or estrogen levels and adipocyte insulin binding or action were found. Because insulin binding was not changed, we conclude that the major lesion causing insulin resistance in PCO is a striking decrease in insulin sensitivity secondary to a defect in the insulin receptor and/or postreceptor signal transduction. PCO also is associated with modest but significant decreases in glucose transport. These defects in insulin action appear to represent intrinsic abnormalities that are independent of obesity, metabolic derangements, body fat topography, and sex hormone levels. Conversely, changes in hepatic insulin sensitivity appear to be acquired with obesity.

Women with the polycystic ovary syndrome (PCO) have significant insulin resistance and are at risk to develop noninsulin-dependent diabetes mellitus. It remains controversial, however, whether hyperandrogenism directly decreases insulin action. Hence, we performed 2-h euglycemic glucose (approximately 772 pmol/L steady state insulin levels) clamps in nine PCO women with insulin resistance basally and after the 12th week of therapy with a superagonist GnRH analog (40 micrograms every 8 h, sc). Diet, activity, and weight were kept constant. Despite significant decreases in plasma testosterone and androstenedione levels (both P less than 0.05), there was no significant change in insulin-mediated glucose disposal, plasma insulin levels, or hepatic glucose production. The sample size was adequate to detect a clinically significant change in insulin-stimulated glucose disposal (i.e. approximately 3.3 mumol/kg.min; P less than or equal to 0.05). We conclude that suppressing androgen levels into the normal range did not result in significant changes in insulin resistance in PCO. Thus, controlling hyperandrogenemia is not a clinically effective modality to improve insulin action and thereby decrease the risk of noninsulin-dependent diabetes in PCO.

Acanthosis nigricans (AN) is a frequent clinical finding in hyperandrogenic women. Its presence has been used to subgroup such women. We performed this study in order to determine the actual histological prevalence of AN and its relationship to sex hormone levels and insulin action. Insulin-mediated glucose disposal was determined by the euglycemic clamp technique, and neck or axillary skin biopsies were graded blind for the presence and severity of AN in lean and obese women with the polycystic ovary syndrome (PCO) and in age- and weight-matched normal ovulatory controls. AN was present on clinical examination in 11 of 13 obese PCO, 3 of 6 lean PCO, 4 of 14 obese normal, and 0 of 4 lean normal women. AN was present on histological examination in 13 of 13 obese PCO, 5 of 6 lean PCO, 13 of 14 obese normal, and 1 of 4 lean normal women. The severity of histological AN was most highly correlated with insulin-mediated glucose disposal (r = -0.61; P less than 0.001) rather than fasting (r = 0.46; P less than 0.05) or glucose-stimulated insulin levels (r = 0.48; P less than 0.01). The only sex steroid correlated with histological AN was dehydroepiandrosterone sulfate (r = 0.46; P less than 0.01). We conclude that 1) clinical skin examination was very insensitive for detecting AN; 2) the best biochemical correlate of histological AN was decreased insulin action, rather than insulin or androgen levels per se; and 3) AN is a very common epiphenomenon of insulin resistance, and its clinical presence should not be used as a criterion for stratifying hyperandrogenic women.