Ambient Circulating LH, but Not Insulin, Predicts Rise in Testosterone Levels After Recombinant hCG in Girls With Obesity

Abstract Polycystic ovary syndrome (PCOS) is associated with obesity and insulin resistance. Adolescent hyperandrogenemia (HA) may precede adult PCOS. Androgen production in females occurs in both the adrenals and the ovaries, but the relative contribution of each to adolescent HA is unknown. Both luteinizing hormone (LH) and insulin contribute to HA in adult PCOS, and both correlate with HA in obese girls, but detailed assessments of LH and insulin in combination with ovarian and adrenal androgen responses to stimulation (in the same individual) have not been described. To assess the relative roles of stimulatory factors (LH and insulin) and end organ (adrenal, ovarian) responsiveness to stimulation, we have studied 16 girls with obesity: age 13.4 (10.5–15.9) y (median [range]); Tanner 5 (2 girls 2-3; 14 girls 4-5); BMI Z 2.2 (1.7–2.7); free testosterone (T) 17.7 (6.6–88.3) pmol/L. Subjects underwent a detailed study including (a) frequent blood sampling for LH (6p–9a), to estimate mean 24-h LH; (b) sampling for insulin from 1 h before to 2 h after a standardized mixed meal (7p) and while fasting (7a–9a), to estimate mean 24-h insulin; (c) an adrenal stimulation protocol (dexamethasone [DEX] given at 10p, with 17-OHProgesterone [17OHP], T, and androstenedione (∆4A) drawn before plus 30 and 60 min after synthetic ACTH [250 mcg iv] given at 7a); and (d) an ovarian stimulation protocol (after the 8a sample above, recombinant hCG [r-hCG, 25 mcg iv] given, DEX given at 10p, with 17OHP, T, and ∆4A drawn the next morning at 8a). Responses to ACTH and r-hCG stimulation were defined as the mean value 30 and 60 min post-ACTH and the value 24 h post-hCG, respectively, minus the post-DEX morning value. Relationships between such responses and estimated mean 24-h LH and 24-h insulin were assessed using Spearman partial correlation (correcting for differences in 24-h insulin and 24-h LH, respectively). Estimated 24-h LH was 3.7 (1.8–21.5) mIU/mL in the group, while estimated 24-h insulin was 61.4 (23.2–175) uIU/mL. After correcting for differences in 24-h insulin, estimated 24-h LH predicted hCG-stimulated changes in T (r = 0.61, p = 0.02), but did not predict ACTH-stimulated changes in T. When corrected for 24-h LH, there were no significant relationships between estimated 24-h insulin and T responses to either r-hCG or ACTH. Estimated 24-h LH and 24-h insulin were not correlated with ACTH- or hCG-stimulated changes in either 17OHP or ∆4A. These data suggest that, in pubertal girls with obesity, either that ovarian T responses to stimulation are influenced by ambient LH concentrations, but not by insulin, or that ovarian hyperresponsiveness leads to increased LH. Similar relationships with 17OHP or ∆4A were not evident, for either ambient LH or insulin. Simultaneous detailed assessments of LH, insulin, and end organ (adrenal, ovarian) responsiveness to stimulation may help discriminate the determinants of HA in girls with obesity.