Androgens During the Reproductive Years: What Is Normal for Women?

Marina A. Skiba; Robin J. Bell; Rakibul M. Islam; David J. Handelsman; Reena Desai; Susan R. Davis


J Clin Endocrinol Metab. 2019;104(11):5382-5392. 

In This Article


The results from the present study have reaffirmed that the serum testosterone and androstenedione levels increase after ovulation, with higher levels in the midcycle and luteal phases compared with the early follicular phase.[11,12] In addition, we found that in healthy women, 11KA and 11KT make major quantitative contributions to the circulating androgen pool and that the levels of these steroids are stable across the menstrual cycle. Serum levels of testosterone, androstenedione, DHEA, 11KA, and 11KT declined substantially with increasing reproductive age, highlighting the need for age-specific reference ranges to guide the clinical interpretation of androgen levels. Now that their circulating levels have been quantified, the biological effect of the ketoandrogens needs to be defined.

To explore the normal physiology, we limited the reference group to women within 35 days of their LMP, with no identifiable factors that would have affected their ovarian or adrenal function. The observed midcycle increases in androstenedione and testosterone levels reflect their ovarian biosynthesis during follicular development and after ovulation, as previously reported.[12] The stability of 11KA and 11KT levels across the cycle is consistent with these steroids being downstream products of the adrenal precursors, 11-hydroxyandrostenedione and 11-hydroxytestosterone, with little or no ovarian contribution.[13] That most women had levels of DHT and 11KDHT less than the LOD might reflect their intracrine physiology, with their production and metabolism occurring within the tissues in which they act as local modulators of androgen action, but with the serum levels resulting from unregulated spillover from the tissues into the circulation.[14]

A surge of interest has occurred in the 11-oxygenated C19 steroids. The circulating concentrations of these steroids have been reported to be greater than those of classic androgens in women with premature adrenarche,[15] congenital adrenal hyperplasia,[16] and PCOS.[17] Yoshida et al.[26] reported substantial circulating steroid heterogeneity in women with PCOS, with some having elevated 11-oxygenated C19 steroids and others having only elevated classic androgens.

The approximate decline in testosterone levels from 18 to 39 years of 25% in the present study was more modest than previously reported,[18,19] most likely reflecting the greater specificity of LC-MS/MS than direct RIA, with less cross-reaction between testosterone and other steroids such as DHEA.[20] A study of C19 steroid levels[21] reported lower testosterone, androstenedione, and DHEA levels in 100 premenopausal women compared with 100 postmenopausal women but no associations between age and 11KA or 11KT levels. The present study, which found age-associated reductions in the median levels of all C19 steroids, included 84 to 277 women in each premenopausal 5-year age group. In addition, the analyses were adjusted for BMI and menstrual cycle phase. The enzymatic pathways for the production of testosterone from androstenedione and of 11KT from 11KA are the same, and consistent with this, the ratios of the levels of testosterone to androstenedione and 11KT to 11KA in our study were ~6:1. Similarly, the ratios of 11KA to androstenedione and 11KT to testosterone were both ~4:1, again reflecting the same enzymatic pathways. These ratios did not change with age, such that because both testosterone and androstenedione declined with age, both 11KT and 11KA would also have declined with age. Previous studies have reported median circulating testosterone levels measured using RIA in young women in the order of 1 nmol/L.[18,19] In our study, the median testosterone level of 0.34 nmol/L was similar to that reported by O'Reilly et al..[17] Also using LC-MS/MS, Nanba et al.[21] reported a median testosterone level in premenopausal women of 1.04 nmol/L, similar to that seen with RIA,[19] indicating variations between LC-MS/MS assays. They also reported lower median values for 11KT and 11KA than those found in the present study.[21] This would have affected their analyses of the associations with age.[21]

The inverse association between DHEA and BMI has been reported by others, with a study of monozygotic twins reporting an inverse association between DHEA and the percentage of body fat.[22,23] Previous studies of testosterone measured using RIA have reported positive associations between testosterone and BMI.[24,25] In contrast, small studies of testosterone measured using LC-MS/MS have found no, or weak, associations between testosterone and BMI.[17,24] Both direct[17] and inverse[26] associations have been reported for 11KA and BMI in studies that included only small numbers of healthy women.

The strengths of the present study included recruitment of a community-based sample of non–healthcare-seeking women, the measurement of steroids using LC-MS/MS, and the exclusion of women with irregular cycles and factors known to influence endogenous sex steroid levels. None of the participants had undergone a physical examination; therefore, the presence of PCOS might have been undetected. However, all participants were required to have a regular menstrual cycle, which makes the likelihood of the inclusion of women with PCOS low. The participants' height and weight were self-reported; however, a high level of agreement has been shown for BMI calculated from self-reported and physician-measured height and weight in women of reproductive age.[27–29] The study participants were predominantly Caucasian/white, reflective of the Australian population, such that our findings should not be immediately generalized to women of other ethnicities.

The present study has provided a comprehensive summary of the circulating levels of the major classic and nonclassic androgens in women of reproductive age, including menstrual cycle variation and associations with BMI and age. It highlights the need for age-specific reference ranges for the interpretation of androgen levels in premenopausal women. Although testosterone, DHEA, and androstenedione have been considered primary biomarkers of androgen excess,[17,30] with no level below which a woman can be said to be androgen deficient,[6,7] studies are needed to determine whether 11-oxygenated C19 steroids might serve as informative markers of androgen excess or insufficiency.