Special Feature: Measuring Testosterone

By Leon Speroff, MD

There are 2 important potential uses for the measurement of testosterone levels: 1) to monitor the dosage of testosterone supplementation; and 2) to diagnose hyperandrogenism, whatever the cause. Unfortunately, the measurement of testosterone is particularly difficult in women because of the low circulating levels. The situation is further complicated by how testosterone circulates in the blood and by having more than 1 assay available to measure testosterone.

In the circulation, testosterone is bound to sex hormone-binding globulin (SHBG) and to albumin. Free testosterone is that which is unbound and available for target tissue activity. In healthy women, approximately 50-60% of testosterone is bound to SHBG and 30-40% to albumin, leaving only 0.5% to 3.0% as unbound and active. Estrogen and thyroid hormone increase SHBG levels, and androgens, glucocorticoids, growth hormone, and insulin decrease levels. Bioavailable testosterone refers to the free and unbound testosterone and that which is bound to albumin (because testosterone binding to albumin is weak and thus some, but not all, of albumin-bound testosterone is available for tissue activity).1

The total testosterone assay measures by direct radioimmunoassay with commercial kits free testosterone, albumin-bound testosterone, and the testosterone that is bound to SHBG. The other available assays include the following:

Free testosterone, measured by equilibrium dialysis and ultracentrifugation-a is a laborious, time-consuming method that requires strict temperature controls and is expensive. However, this is the gold standard against which other methods estimating the amount of active testosterone must be compared. Free testosterone is also measured by direct immunoassay with commercial kits. The normal range is from 0.3-0.8 to 3-6 pg/mL in most laboratories.

Bioavailable testosterone, measured by ammonium sulfate precipitation of SHBG, followed by radioimmunoassay with commercial kits is also time-consuming and expensive.

Free androgen index, also called the free testosterone index, is calculated by dividing the total testosterone by the SHBG concentration and multiplying by 100 (T/SHBG ´ 100).

The direct immunoassay of free testosterone is very attractive because of its ease, rapidity, and relative cost. This method, however, is subject to considerable inaccuracy and variability. The results with this assay measure only 20-60% of the levels measured by the more difficult and expensive method that uses dialysis.2 In addition, this assay is affected by the changes in the levels of SHBG. The free testosterone index and measurement of the bioavailable testosterone correlate well with the gold standard dialysis method. However, these methods are also affected by the circulating amounts of SHBG and testosterone.3 With a lot of SHBG and lower levels of testosterone, there are abundant binding sites on the SHBG giving falsely elevated values for these methods, and vice-versa.

The clinical problem is the fact that there are discrepancies among the values for these various methods reported in the literature. For appropriate clinical use, each method requires the establishment and validation of normal ranges and changes with pathologic conditions. This has not been done. The salivary concentration of sex steroids represents only a very small fraction of the amount in the ciruclation.4 Salivary measurements also have not been validated; specifically, there is a lack of studies establishing the correlation between salivary levels and serum levels, and between salivary levels and clinical presentations and/or responses.

Testosterone production decreases by approximately 25% after menopause, but the postmenopausal ovary in most women secretes more testosterone than the premenopausal ovary. With the disappearance of follicles and estrogen, the elevated gonadotropins drive the remaining stromal tissue in the ovary to a level of increased testosterone secretion. The total amount of testosterone produced after menopause, however, is decreased because the amount of the primary source, peripheral conversion of androstenedione, is reduced. The early postmenopausal circulating level of androstenedione decreases approximately 62% from young adult life.5 The menopausal decline in the circulating levels of testosterone is not great, from no change in many women to as much as 15% in others.5-8 Nevertheless, compared with young women, the overall androgen exposure of postmenopausal women to androgens is less.9 After age 60, testosterone levels are about half those measured in young women. The key question is whether the free and active amount of testosterone decreases or increases.

Circulating Hormone Levels
Premenopause Postmenopause





40-400 pg/mL

30-200 pg/mL

20-80 ng/dL

60-300 ng/dL

10-20 pg/mL

30-70 pg/mL

15-70 ng/dL

30-150 ng/dL

The conclusions in the preceding paragraph are derived from studies that have been hampered by small sample sizes and by their cross-sectional nature. A prospective longitudinal study in Australia followed 172 women for 7 years as they passed through menopause.8 The circulating levels of total testosterone did not change. SHBG levels decreased about 43%, and the free androgen index increased (by 80%!) because of the decrease in SHBG. But does this have clinical meaning, and in view of the variabilities with these methods, is this observation accurate? Why wouldn’t this substantial increase in a measure indicating an increase in free testosterone produce behavioral changes? By preventing the decrease in SHBG, postmenopausal estrogen therapy would prevent this increase in free testosterone. Is this important clinically? These are questions currently without answers.

One study concluded that the free androgen index displayed a good correlation with hyperinsulinemia.12 However, the range of values was very great with considerable overlap comparing normoinsulinemic women with hyperinsulinemic women. The free testosterone level has been reported to decrease after 2 years of treatment of postmenopausal women with oral estrogen in contrast to no change with transdermal treatment.13 But again there was impressive overlap; in fact, the decrease within the group on oral therapy did not reach statistical significance. Some have argued for measuring free testosterone as a screening method for polycystic ovary syndrome. The variability found both within individuals and among the assays is a strong argument against this practice.

Testosterone supplementation might favorably affect muscle strength, sexuality, and psychological state. There is little doubt that the administration of pharmacologic amounts of testosterone can produce these favorable effects, but it remains unknown whether maintaining testosterone levels within the normal physiologic range (10-20 to 60-90 ng/dL, depending on the laboratory) can have a beneficial impact on health. Furthermore, the long-term consequences of pharmacologic amounts of testosterone are totally unknown.

I am reluctant to support the pharmacologic use of testosterone given the difficulties in monitoring dosage and the lack of knowledge regarding the long-term effects on health. 17a-Methyltestosterone is administered orally in combination with estrogen. The available doses are definitely pharmacologic. The problem is that this androgen is not demethylated in the body and cannot be measured by testosterone assays; therefore, it is impossible to monitor dosage. If a clinician and a patient choose to use supplemental androgens, my advice is to select a treatment that can be monitored with measurements of total testosterone in serum. The choices include the testosterone transdermal patch (not yet on the market), a testosterone skin gel (on the market for use in men), and testosterone compounded for individual use by a pharmacist.

Dr. Speroff is Professor of Obstetrics and Gynecology at Oregon Health Sciences University in Portland.


1. Manni A, et al. J Clin Endocrinol Metab. 1985;61: 705-710.

2. Morley JE, Perry HM. Metabolism. 2002;51:554-559.

3. Vermeulen A, et al. J Clin Endocrinol Metab. 1999;84: 3666-3672.

4. Shirtcliff EA, et al. Hormones Behavior. 2002;42:62-69.

5. Labrie F, et al. J Clin Endocrinol Metab. 1997;82: 2396-2402.

6. Rannevik G, et al. Maturitas. 1995;21:103-113.

7. Jiroutek MR, et al. Menopause. 1998;5:90-94.

8. Burger HG, et al. J Clin Endocrinol Metab. 2000;85: 2832-2838.

9. Zumoff B, et al. J Clin Endocrinol Metab. 1995;80: 1429-1430.

10. Meldrum DR, et al. Obstet Gynecol. 1981;57:624.

11. Judd HL, et al. J Clin Endocrinol Metab. 1974;39: 1020.

12. Maturana MA, et al. Metabolism. 2002;51:238-243.

13. Serin IS, et al. Eur J Obstet Gynecol Reprod Biol. 2001;99:222-225.