Serum leptin levels in women with polycystic ovary syndrome: the role of insulin resistance/hyperinsulinemia.

Polycystic ovary syndrome (PCOS) is associated with chronic anovulation, hyperandrogenemia, insulin resistance (IR)/hyperinsulinemia, and a high incidence of obesity. Thus, PCOS serves as a useful model to assess the role of IR and chronic endogenous insulin excess on leptin levels. Thirty-three PCOS and 32 normally cycling (NC) women of similar body mass index (BMI) were studied. Insulin sensitivity (S(I)) was assessed by rapid ivGTT in a subset of 28 PCOS and 29 NC subjects; percent body fat was determined by dual-energy x-ray absorptiometry (DEXA) in 14 PCOS and 17 NC. Fasting (0800 h) and 24-h mean hourly insulin levels were 2-fold higher (P < 0.0001), and S(I) was 50% lower (P = 0.005) in PCOS than in NC, while serum androstenedione (A), testosterone (T), 17-alpha hydroxyprogesterone (17OHP), and estrone (E1) levels were elevated (P < 0.0001), and sex hormone-binding globulin (SHBG) levels were decreased (P < 0.01). Twenty-four hour LH pulse frequency, mean pulse amplitude, and mean LH levels were elevated in PCOS (P < 0.001) as compared with NC. Serum leptin levels for PCOS (24.1 +/- 2.6 ng/mL) did not differ from NC (21.5 +/- 3.5 ng/mL) and were positively correlated with BMI (r = 0.81) and percent body fat (r = 0.91) for the two groups (both P < 0.0001). Leptin levels for PCOS and NC correlated positively with fasting and 24-h mean insulin levels (r = 0.81, P < 0.0001 for both PCOS and NC) and negatively with S(I) and SHBG levels. Leptin concentrations for PCOS, but not NC, correlated positively with 24-h mean glucose levels and inversely with 24-h mean LH levels and 24-h mean LH pulse amplitude. Leptin levels were not correlated with estrogen or androgen levels for either PCOS or NC, although leptin levels were positively related to the ratios of E1/SHBG and E2/SHBG for both PCOS and NC and to the ratio of T/SHBG for PCOS only. In stepwise multivariate regression with forward selection, only 24-h mean insulin levels contributed significantly (P < 0.01) to leptin levels independent of BMI and percent body fat for both PCOS and NC. Given this relationship and the presence of 2-fold higher 24-h mean insulin levels in PCOS, the expected elevation of leptin levels in PCOS was not found. This paradox may be explained by the presence of adipocyte IR specific to PCOS, which may negate the stimulatory impact of hyperinsulinemia on leptin secretion, a proposition requiring further study.