Measuring serum testosterone is essential to identify hyperandrogenaemia as one of the criteria for diagnosing PCOS. Testosterone is produced either directly by the ovary or adrenal gland, or indirectly through metabolism of secreted androstenedione or DHEA (or its sulphate DHEAS) in peripheral tissues, such as fat and skin. Approximately 50% of testosterone arises from peripheral conversion of androstenedione, whereas 25% is secreted by the ovaries and an additional 25% by the adrenal. LH is the stimulus for ovarian testosterone production.
High serum testosterone values in the adult- male range may indicate the presence of an androgenic tumour, while only a moderate increase of this androgen is a poor predictor of androgenic tumours. Androgen- producing ovarian tumours are uncommon, occurring only in about 1 in 500 hirsute women. Ovarian tumours that produce androgens directly include Sertoli– Leydig cell tumours (androblastoma, arrhenoblastoma), granulosa- theca cell (stromal cell) tumours, and hilus cell tumours. Nevertheless, any large ovarian tumour (i.e. cystic teratomas, Brenner tumours, serous cystadenomas, and Krukenberg tumours) can produce androgens indirectly by causing hyperplasia of the surrounding normal stroma.
Several limitations exist, however, in measuring testosterone, which include lack of a standardized assay; poor sensitivity and accuracy in the female range; and variability based upon menstrual cycle phase, time of day, and feeding. Avoiding interference by steroids of similar configuration requires extraction and separation of other circulating steroids chromatographically before subjecting the sample to immunoassay or mass spectrometry.
Serum total testosterone levels are above the upper normal limit in about 50% of women with PCOS by 1990 NIH criteria. There is an overlap in total testosterone values between PCOS patients and normal women. When measured by liquid chromatography– mass spectrometry (LC- MS/ MS), serum total testosterone levels that best distinguish PCOS from normal women are approximately 50 ng/ dl in the general population (10% estimated PCOS prevalence) and 35 ng/ dL in the clinical or referred, setting (70% estimated PCOS prevalence).
Measuring free testosterone is recommended by the 2004 Rotterdam consensus to detect hyperandrogenism in PCOS women. Free testosterone is more sensitive than total testosterone in assessing hyperandrogenaemia. As a surrogate of free testosterone by equilibrium dialysis, the free androgen index (FAI) is calculated by measuring sex hormone binding globulin (SHBG) and total testosterone. The FAI is elevated in about 54– 89% of women with PCOS by 1990 NIH criteria. The FAI also is elevated in 30% of women with PCOS by Rotterdam criteria who have normal total testosterone levels, thereby identifying a subset of these PCOS women who should be considered as hyperandrogenic.
Serum levels of total testosterone in excess of 200 ng/ dL are suspicious for an ovarian androgen- producing tumour. However, the best predictor of an androgen secreting neoplasm is virilization defined as male temporal balding, deepening of the voice, and enlargement of the clitoris, which occurs in 98% of such tumours, regardless of circulating testosterone levels.
The value of routinely measuring circulating androstenedione or DHEA as precursors of testosterone in hyperandrogenic women is more controversial. Serum androstenedione and total testosterone levels are elevated in 88% and 65% of PCOS women by Rotterdam criteria, respectively. The combined use of total testosterone, androstenedione and FAI identifies hyperandrogenaemia in about 90% of PCOS patients.
Elevated serum DHEAS levels occur in 40– 70% of hyperandrogenic women with PCOS. Although increased in women with symptoms of androgen excess, only 16% of hyperandrogenic women have increased serum DHEAS in the presence of normal total and free testosterone levels. Conversely, low serum DHEAS levels can occur in adolescents and young women with functional hypothalamic amenorrhea from anorexia nervosa, despite hypercortisolism and adequate adrenocorticotropic hormone (ACTH). Measuring serum DHEAS, however, should be considered to rule out adrenal aetiologies. In general, measurements of DHEAS are of lower yield unless indicated by clinical findings of severe hirsutism with high concentrations far exceeding the normal range (>6000 ng/ dl), suggestive of an adrenal tumour.
