Association between hyperandrogenism and insulin resistance is well recognized in women with polycystic ovary syndrome (PCOS) (1). However, earlier evidence (2) suggesting an insulin-antagonizing effect of androgens has been overshadowed by more recent studies demonstrating that antiandrogen treatment with flutamide (3) or GnRH agonists (4,5) does not alter insulin resistance in PCOS. Conflicting results have been reported in non-PCOS women, with some studies (1,6–9) suggesting that testosterone may be related to insulin resistance and others (10,11) showing no correlation. Recent data suggest that some of these discrepancies may be explained by racial disparities, since only obese African-American women exhibited a positive relationship between insulin resistance and gonadal androgens (6). Inconclusive data have also been reported in men given testosterone in replacement or supraphysiologic doses, with some studies (12) suggesting a sensitizing effect of testosterone on glucose metabolism and others (13–16) showing no effect.
Nonetheless, androgens can influence body composition, which is associated with insulin sensitivity. Thus, it is conceivable that testosterone might indirectly influence insulin sensitivity via its effects on body composition. We report the results of hormonal, metabolic, and body composition studies before and 1 month and 9 months after a Leydig cell tumor removal in a postmenopausal woman.
RESEARCH DESIGN AND METHODS
A 64-year-old gravida 7, para 7, Hispanic woman was referred for evaluation of virilization starting ∼10 years earlier and progressing over the past 3 years. Menses were regular before menopause (age 50). Diabetes was diagnosed 2 months before presentation and was well controlled by 1.5 mg glyburide daily (HbA1c 4.8%). She had a 22-year history of hypertension, treated with benazepril and amlodipine. A physical examination revealed male pattern alopecia, masculine habitus, abdominal obesity, clitoromegaly, and breast atrophy but no palpable adnexal masses.
Laboratory studies revealed extreme hyperandrogenism (Table 1). A computed tomography scan and pelvic ultrasound did not detect ovarian masses. Nonetheless, she underwent total hysterectomy with bilateral salpingo-oophorectomy because of the increased risk for endometrial cancer (endometrial thickening, 8.6 mm) and possible virilizing ovarian tumor. Microscopic examination revealed a 0.9-cm Leydig cell tumor, nonhilar type, in the right ovary. Northern blot analysis (17) showed abundant expression of mRNA for P450SCC and P45017a in the tumor but no expression in stromal tissue from the contralateral ovary, indicating absence of hyperthecosis or PCOS.
Glyburide was held pre- and postoperatively, and fasting glucose remained normal (Table 1). Approximately 6 months postoperatively, fasting glucose increased to 204 mg/dl, and glyburide was reinstituted. There were no changes in blood pressure, antihypertensive medications, or self-reported diet or physical activity. Virilization decreased postoperatively.
After institutional review board approval, the patient gave written informed consent and was admitted to the General Clinical Research Center for metabolic studies 1 week before and 1 month and 9 months after surgery. Glyburide was held for at least 72 h before each admission. We measured body composition by dual-energy X-ray absorptiometry and 40K counting, hormones by radioimmuno assay, and insulin sensitivity using a 75-g oral glucose tolerance test (OGTT), an insulin tolerance test (ITT; 0.10 units/kg), and a hyperinsulinemic (prime: 5.4 mU/kg; infusion 0.9 mU · kg−1 · min−1)- euglycemic glucose clamp with measurement of steady-state glucose kinetics at baseline and during clamp ([6,6-2H2]-glucose, prime: 17.2 μmol/kg, infusion: 0.2 μmol · kg−1 · min−1) (18).
Total and free testosterone levels were markedly elevated preoperatively and declined dramatically following surgery (Table 1). Androstenedione was slightly elevated preoperatively and returned within the normal range postoperatively. Dehydroepiandrosterone-S and 17-OH-progesterone were low before and after surgery, suggesting a normal adrenal androgen production. Luteinizing hormone and follicle-stimulating hormone were low preoperatively and increased postoperatively, reaching values close to the postmenopausal normal range after 9 months, possibly due to a slow recovery of the gonadotrophs from the 10-year suppression by testosterone.
There were no changes in weight or individual compartments 1 month postoperatively. Nine months postoperatively, weight increased by 7%, with marked increases in total and abdominal fat and decreased body cell mass.
Fasting glucose concentrations were normal, and fasting insulin was moderately elevated before surgery and 1 month and 9 months postoperatively. OGTT revealed moderate insulin resistance and glucose intolerance preoperatively that remained unchanged 1 month postoperatively; 9 months postoperatively, OGTT became diagnostic for type 2 diabetes. Conversely, ITT and glucose clamp indicated deterioration of peripheral insulin sensitivity 1 month postoperatively (decreased kITT and glucose utilization), which remained unchanged 9 months postoperatively. Interestingly, the response of hepatic glucose production to insulin was incomplete preoperatively and improved 1 month postoperatively, indicating increased liver insulin sensitivity.
These data from a postmenopausal woman before and after surgical correction of extreme hypertestosteronemia suggest that testosterone may affect insulin sensitivity. The progressive worsening of insulin sensitivity following tumor removal indicates that in this patient the general effect of testosterone was sensitizing. This was the integrated result of opposite actions of testosterone on liver and peripheral insulin sensitivity, as insulin-stimulated glucose utilization and hepatic glucose production were concomitantly higher with high testosterone concentrations and decreased following testosterone withdrawal. In this patient, increased glucose utilization prevailed during hypertestosteronemia, leading to enhanced insulin sensitivity. However, testosterone concentration in this patient was among the highest reported for women with androgen-producing tumors (19,20). Since the dramatic reduction of these extreme concentrations produced relatively modest changes in glucose tolerance, the overall effect of testosterone on insulin sensitivity appears to be mild. Because of the dual action of testosterone on glucose metabolism, it is also possible that in different conditions the effect of testosterone on insulin sensitivity is neutral, which could explain the variable results of previous studies (12–16,21).
Our data also suggest that testosterone may affect insulin sensitivity both directly and indirectly. One month postoperatively, ITT and glucose clamp revealed deterioration of insulin sensitivity despite unchanged body composition, suggesting a direct effect of testosterone. This effect was subsequently overshadowed by profound changes in body composition that occurred 9 months postoperatively and led to the development of overt diabetes. Loss of lean body mass and gains in fat, particularly abdominal fat, were likely results of testosterone withdrawal, since testosterone increases lean body mass (13–15) and decreases fat mass and abdominal fat (12,13,22,23). The discrepancy between OGTT and ITT and clamp data are likely due to the higher sensitivity of the latter two techniques to detect small changes in insulin sensitivity (24).
It is important to underscore that our patient’s disease, involving autonomous production of testosterone by a Leydig cell tumor that resolved with surgical removal of the tumor, was fundamentally distinct from PCOS, in which insulin resistance is the primary abnormality stimulating ovarian androgen production (25), and treatment of hyperandrogenism does not affect insulin resistance (3–5). Finally, it is unlikely that other androgens played any role in the worsening of the patient’s insulin sensitivity following surgery, since androstenedione, whose potential effects on insulin sensitivity parallel those of testosterone, mildly decreased after surgery, and dehydroepiandrosterone-S, which has been linked to increased insulin sensitivity (9), slightly increased postoperatively to the low-normal range.
In summary, the hormonal, metabolic, and body composition changes following correction of extreme hyperandrogenism in this patient indicate that testosterone may improve insulin sensitivity both directly and through changes in body composition. Our data suggesting that testosterone is not unequivocally sensitizing, and that sex or other characteristics may influence the response of glucose metabolism to testosterone, underscore the need for further investigations in this area.
This work was supported in part by the National Institutes of Health/National Institute of Child Health and Human Development Grant no. R01 HD36092 (to R.J.U.), the National Institutes of Health/National Cancer Institute Grant no. R01 CA45181 (to M.N.), the National Institutes of Health/National Institute on Aging Grant no. R01 AG18311 (to E.V.), the National Institutes of Health/National Center for Research Resources General Clinical Research Center Program no. M01 RR00073, and the Brookdale Foundation (to E.V.).
The authors thank Drs. Andrew Coggan and Cathy Weickart for their advice and assistance with the clamp procedures, Dr. Fernando Cesani for the dual-energy X-ray absorptiometry analyses, Dr. Oliver Esch for the 40K analyses, and the staff of the General Clinical Research Center at the University of Texas Medical Branch for their assistance and dedication.
A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.