Identification of the Source of Androgen Excess in Hyperandrogenic Type 1 Diabetic Patients

We recently reported that the prevalence of hyperandrogenic disorders is markedly increased in women with type 1 diabetes (1). The polycystic ovary syndrome (PCOS) defined by endocrine criteria was found in 18.8% of the type 1 diabetic women who followed-up in our hospital (1), as compared with the 6.5% prevalence in nondiabetic women from similar ethnic and genetic backgrounds (2). The prevalence of hirsutism in type 1 diabetic women was 30.6% (1), which is markedly higher than the 7.1% prevalence of hirsutism found in nondiabetic women (2).

In the present study, we evaluated the adrenal and ovarian steroidogenic profiles of hyperandrogenic and nonhyperandrogenic type 1 diabetic women and compared them with those of nondiabetic hyperandrogenic women and healthy control subjects.

A total of 24 women with type 1 diabetes were recruited for the study (1). Fourteen diabetic patients (age [mean ± SD] 20.6 ± 4.0 years, BMI 24.8 ± 2.9 kg/m²) were considered to have hyperandrogenism. The other 10 women with type 1 diabetes (age 19.0 ± 3.0 years, BMI 23.3 ± 2.6 kg/m²) had no evidence of clinical or biochemical hyperandrogenism and had regular menstrual cycles. Both groups of type 1 diabetic patients had similar HbA_1c levels (7.4 ± 1.2 vs 7.8 ± 1.2% in nonhyperandrogenic and hyperandrogenic diabetic patients, respectively, F = 0.591, P = 0.450), and there were no differences in the mean daily insulin dose used for their treatment (0.82 ± 0.27 vs. 0.66 ± 0.28 U · kg⁻¹ body wt · day⁻¹ in nonhyperandrogenic and hyperandrogenic diabetic patients, F = 1.875, P = 0.185).

A total of 86 nondiabetic women were included as control subjects. Nondiabetic women were matched for BMI and age with the diabetic patients to avoid any influence of age and obesity on the results. Thirteen regularly menstruating women (age 23.2 ± 3.2 years, BMI 24.6 ± 5.1 kg/m²) without signs or symptoms of hyperandrogenism served as healthy control subjects; 73 untreated nondiabetic hyperandrogenic patients (age 20.6 ± 3.8 years, BMI 23.7 ± 3.2 kg/m²) were included as hyperandrogenic control subjects.

Basal and adrenocorticotropic hormone (ACTH)-stimulated samples were obtained and assayed as previously described (1,3). The study was conducted according to the principles expressed in the Declaration of Helsinki.

The group of hyperandrogenic type 1 diabetic patients comprised seven women with PCOS and seven women with hirsutism and regular menstrual cycles. The percentage of patients with PCOS was not different among the groups of diabetic and nondiabetic hyperandrogenic patients (50.0 vs 38.4%, χ² = 0.662, P = 0.553).

Both groups of hyperandrogenic patients had higher hirsutism scores compared with nonhyperandrogenic diabetic patients and healthy control subjects (Fig. 1), but the hirsutism score was higher in nondiabetic hyperandrogenic patients compared with hyperandrogenic type 1 diabetic women (Fig. 1).

Compared with healthy women, both hyperandrogenic type 1 diabetic patients and nondiabetic hyperandrogenic women had increased basal serum total and free testosterone concentrations, as well as basal Δ⁴-androstenedione concentrations (Fig. 1). Nondiabetic hyperandrogenic patients had increased free testosterone levels and decreased sex hormone–binding globulin concentrations compared with all of the other groups (Fig. 1). No differences in sex hormone–binding globulin concentrations were found between the groups of diabetic women and healthy control subjects (Fig. 1).

Nonhyperandrogenic type 1 diabetic women had intermediate values of total testosterone, free testosterone, and Δ⁴-androstenedione that were not significantly different than those of hyperandrogenic diabetic patients and healthy control subjects (Fig. 1). No differences were observed among the groups in the serum concentrations of dehydroepiandrosterone-sulfate, luteinizing hormone, follicle-stimulating hormone, and estradiol (Fig. 1).

Hyperandrogenic type 1 diabetic patients had higher ACTH-stimulated Δ⁴-androstenedione levels than healthy control subjects, whereas ACTH-stimulated Δ⁴-androstenedione and 17-hydroxyprogesterone levels were higher in nondiabetic hyperandrogenic patients than healthy control subjects and nonhyperandrogenic diabetic patients (Δ⁴-androstenedione 12.3 ± 3.0, 15.7 ± 3.1, 12.4 ± 4.3, and 16.1 ± 4.4 nmol/l, F = 4.84, P < 0.005; 17-hydroxyprogesterone 6.1 ± 1.5, 9.2 ± 5.2, 6.7 ± 1.9, and 10.1 ± 5.4 nmol/l, F = 3.30, P < 0.05, in nonhyperandrogenic diabetic patients, hyperandrogenic diabetic patients, healthy control subjects, and nondiabetic hyperandrogenic patients, respectively).

However, the net increments in Δ⁴-androstenedione and 17-hydroxyprogesterone after ACTH stimulation were not statistically different among the groups (Δ⁴-androstenedione 1.4 ± 2.2, 3.3 ± 2.3, 3.1 ± 2.5, and 3.4 ± 2.6 nmol/l, F = 1.904, P = 0.134; 17-hydroxyprogesterone 3.5 ± 2. 1, 6.3 ± 4.3, 4.4 ± 2.2, and 6.2 ± 4.6 nmol/l, F = 1.761, P = 0.159 in nonhyperandrogenic diabetic patients, hyperandrogenic diabetic patients, healthy control subjects, and nondiabetic hyperandrogenic patients, respectively).

Basal and ACTH-stimulated cortisol and 11-deoxycortisol levels were not different among the groups (data not shown), whereas basal 17-hydroxyprogesterone concentrations showed a near-significant tendency (P = 0.056) to higher levels in nondiabetic hyperandrogenic patients.

Finally, because ovulation may normalize many reproductive variables, we used analysis of covariance to rule out a significant impact of the presence or absence of oligomenorrhea on the differences in hormone analyses described above. None of these differences were influenced by oligomenorrhea (data not shown).

Our present results demonstrate that hyperandrogenic type 1 diabetic women have increased serum levels of total and free testosterone and Δ⁴-androstenedione comparable with those found in nondiabetic hyperandrogenic women. Considering that hyperandrogenic type 1 diabetic women had normal dehydroepiandrosterone-sulfate concentrations and that the increase in ACTH-stimulated Δ⁴-androstenedione levels found in these patients possibly reflects a normal adrenocortical response taking place in addition to an increased basal secretion of this steroid (the net increment of Δ⁴-androstenedione was not different compared with that of healthy control subjects), a significant contribution of the adrenal gland to the androgen excess of these patients is not supported by our present results. Nevertheless, because of the small sample size, we cannot exclude that lack of statistically significant differences in dehydroepiandrosterone-sulfate concentrations and in the net increment of Δ⁴-androstenedione after ACTH-stimulation could reflect a type II error.

Virdis et al. (4) recently reported functional ovarian hyperandrogenism (defined by an exaggerated response of 17-hydroxyprogesterone to the gonadotropin-releasing hormone analog leuprolide) in five of nine type 1 diabetic adolescents with oligomenorrhea, which is in conceptual agreement with the main ovarian source for androgen excess in type 1 diabetic patients suggested by our present results.

Surprisingly, serum sex hormone–binding globulin levels were normal in hyperandrogenic type 1 diabetic patients. The regulation of serum sex hormone–binding globulin levels depends on the inhibitory influence of insulin and androgens and the stimulatory effect of estrogens (5). The hyperinsulinism resulting from insulin resistance together with increased androgen levels explains the decrease in sex hormone–binding globulin levels found in most nondiabetic hyperandrogenic patients (5). On the contrary, as stated above, sex hormone–binding globulin levels were normal in hyperandrogenic type 1 diabetic women.

Sex hormone–binding globulin concentrations are mainly regulated by portal vein insulin concentrations (6). However, in type 1 diabetes, insulin is administered subcutaneously instead of being released directly to the portal circulation, as in insulin-resistant patients.

This difference may help to explain the normal sex hormone–binding globulin levels in hyperandrogenic type 1 diabetic patients. In addition to ameliorating the increase in free testosterone levels, a normal sex hormone–binding globulin concentration may decrease the tissue availability of circulating testosterone. This mechanism might contribute to the lower hirsutism scores found in hyperandrogenic type 1 diabetic women compared with nondiabetic hyperandrogenic patients, despite similar serum total testosterone concentrations in both groups.

In summary, our present results suggest that the ovary is the main source of androgen excess in hyperandrogenic type 1 diabetic patients. The normal serum sex hormone–binding globulin levels found in hyperandrogenic type 1 diabetic patients might partially protect these patients against androgen excess by reducing the delivery of androgens to tissues.