Serum testosterone levels and reference ranges in reproductive-age women

To determine the levels of serum testosterone and dehydroepiandrosterone sulfate (DHEAS) in women with no clinical signs of hyperandrogenism and no history of glucocorticoid or oral contraceptive use and to compare these levels with the reference ranges provided by commercial laboratories.

We undertook a cross-sectional retrospective study of 271 reproductive-age women encountered at an endocrinology clinic for complaints of potential thyroid problems. Serum testosterone and DHEAS levels were determined, and statistical analyses were performed.

The serum testosterone level in women with no acne, hirsutism, or menstrual dysfunction was 14.1 +/- 0.9 ng/dL (mean +/- standard error of the mean) (95% confidence interval [CI] = 12.4 to 15.8). This group was considered our study reference population. In women with menstrual dysfunction but no acne or hirsutism, the mean testosterone level was significantly higher (17.9 +/- 1.1 ng/dL; 95% CI = 15.7 to 20.0; P<0.002); with mild hirsutism, it further increased (38.4 +/- 5.1 ng/dL; 95% CI = 27.4 to 49.4; P<0.005); and with moderate to severe hirsutism, it was still higher (49.0 +/- 2.3 ng/dL; 95% CI = 44.4 to 53.6; P<0.003). Serum DHEAS levels showed similar patterns. The upper limit (mean + 2 standard deviations) of testosterone in our study reference population was 28 ng/dL, a level that provided a sensitivity of 84% for detecting hyperandrogenemia. The detection of hyperandrogenemia is essentially impossible when the upper limit of the reference range for testosterone from commercial laboratories (95 ng/dL) is used.

The testosterone levels reported herein and in the literature for hyperandrogenic women both are within the reference (normal) ranges provided by commercial laboratories. These observations demonstrate why diagnosis of hyperandrogenemia in hyperandrogenic women is difficult when commercial laboratories are used and why this condition is not detected in most affected women. Commercial laboratories should reevaluate the methods used for establishing their reference ranges for serum testosterone.