Serum complexed and free prostate specific antigen levels are lower in female elite athletes in comparison to control women

Introduction

Prostate specific antigen (PSA) is a well-known and clinically useful biomarker of prostate adenocarcinoma¹. PSA circulates in blood of males as a complex with alpha 1 antichymotrypsin (cPSA) (approx. 80% of total) or as free, non-complexed PSA (fPSA) (approx. 20% of total)^2,3. It has now been well-documented that PSA is also produced by many female tissues, including breast, periurethral, salivary and thyroid tissues, and by many tumors^4,5. The PSA gene is up-regulated by androgens and progestins in breast and other female tissues, as well as in model systems such as breast carcinoma cell lines^6–12. Serum PSA in women fluctuates during the menstrual cycle, and these changes are attributed to up-regulation by progesterone during the luteal phase^13–15.

PSA also circulates in female serum as complexed and free PSA¹⁶, but its concentrations are exceedingly low (around 1pg/mL), precluding accurate determination with third generation PSA assays^17,18. In some circumstances, such as in women with hyperandrogenic syndromes, including polycystic ovarian syndrome (PCOS) and hirsutism, it has been shown that total PSA in female serum is elevated and this finding may be used as an aid to disease diagnosis^19–25.

Recently, fifth generation PSA assays have been developed by many groups, allowing accurate PSA determinations in the low fg/mL range^26–31. These assays can quantify both complexed and free PSA in serum and urine of females and allow examination of the possible role of female PSA in healthy and disease states. We recently confirmed the diagnostic value of cPSA and fPSA in women with PCOS by using one of these assays³².

It has previously been speculated that female elite athletes may have higher circulating androgen levels than sedentary age-matched females, and that hyperandrogenic syndromes such as PCOS, congenital adrenal hyperplasia and 46XY disorders of sex development are more common in elite athletes^33–37.

In this paper we speculated that serum cPSA and fPSA, due to their in-vivo up-regulation by androgens and progestins, as seen with female-to-male transsexuals treated with testosterone^38,39, may represent an integrated index of total androgenic/progestational activity in female tissues. We thus examine here, the levels of serum complexed and free PSA in 106 female elite athletes and 114 age-matched sedentary controls, along with levels of estrogens, androgens and progesterone. The observed differences in serum cPSA and fPSA between athletes and controls were examined in the context of oral contraceptive use and ovulatory cycles.

Materials and methods

Results

The included excel file contains all anthopometric and biochemical data used to perform statistical analyses. Significant differences between athletes and controls, irrespective of oral contraceptive use, were noted for the following variables: total BMD, spine BMD, lean mass (all parameters were elevated in athletes; p<0.001) and fat percentage (decreased in athletes; p<0.001).

Initial examination of our data revealed the significant effect of oral contraceptives on hormonal and PSA measurements. For this reason, we stratified both athlete and control groups by hormonal measurements and PSA, according to oral contraceptive use (data shown in Table 1 and Table 2). E1 and E2 levels were significantly reduced in athletes (p=0.003 and 0.004, respectively). Other measurements were not significantly different between athletes and controls that were not taking oral contraceptives, but there was a trend for athletes to have higher DHEA (p=0.095) and Delta 5 (p=0.084). In Table 2 it is shown that both cPSA and fPSA were significantly lower in athletes in comparison to control subjects, irrespective of hormonal contraceptive use. Median cPSA was 776 fg/mL in athletes and 1249 fg/mL in controls (p=0.003). Median fPSA was 70 fg/mL in athletes and 169.5 fg/mL in controls (p=0.013).

It is shown in Table 3 that in athletes, the use of oral contraceptives significantly increased both cPSA (from 776 to 1812 fg/mL; (p=0.046) and fPSA from 70 to 216 fg/mL (p=0.009). In control subjects, we noticed similar trends (from 1249 fg/mL to 2002 fg/mL for cPSA and from 169.5 fg/mL to 220 fg/mL for fPSA) but the differences were not significant. The elevation of serum PSA with oral contraceptive use in women has been reported before⁹.

The distribution of all clinical and demographic variables between athletes and controls, stratified by oral contraceptive use, is shown in Supplementary Figure 1. The distribution of all hormonal and PSA measurements in controls and athletes, stratified by oral contraceptive use, is shown in Supplementary Figure 2 and Supplementary Figure 3.

We then examined the correlation between cPSA and fPSA in serum of both athletes and controls. The Spearman correlation coefficients (rs) were highly significant for both groups of subjects (Figure 1). For athletes, rs=0.75 (p < 0.001) and for controls, rs=0.64 (p < 0.001). We also examined the correlation between the two PSA variables and all other clinical, demographic and hormonal variables in subjects not taking oral contraceptives (Supplementary Table 1). The only statistically significant correlations (p < 0.01; all positive) were between cPSA and fPSA and age, and cPSA with Testo, Delta4 and Delta5 in the athletes group, and between cPSA and fPSA with DHEA, Testo, Delta4 and Delta5 in the control group. The correlations between all of the available parameters in the control and athlete groups not taking oral contraceptives are further depicted in Figure 2.

Figure 1. Serum fPSA vs. cPSA in athletes and controls. Spearman correlations (p-values) are shown.

Figure 2. Spearman correlation plots among all variables in athletes (lower diagonal) and controls (upper diagonal) for women not taking hormonal contraceptives.

The pairs for which ovals are displayed indicate a Spearman correlation that is significantly different from zero, with p-value < 0.01. See also Table 1 for parameter description.

We also examined whether cPSA and fPSA correlated with progesterone. We found no correlation, in either the control or athlete groups, as stratified by hormonal contraceptive use (Supplementary Figure 4). We further dichotomously classified progesterone as being ≥ 5.3 ng/mL (indicative of luteal phase) or < 5.3 ng/mL in athletes and controls not using oral contraceptives, and examined any association with cPSA or fPSA. We found no association (Supplementary Figure 5).

Discussion

PSA is used widely for diagnosis and monitoring of prostatic adenocarcinoma in males¹. Quantification of PSA in female serum is problematic, due to its very low concentrations^4,5. Newly developed fifth generation PSA assays demonstrate enough sensitivity for quantifying both complexed PSA and free PSA in serum of normal women²⁷. We have recently shown that median cPSA concentration in normal women is around 900 fg/mL, while fPSA is about ten times lower (70 fg/mL). In women with PCOS, serum PSA concentration is elevated by about 3-fold³².

It is well-known that the PSA gene is up-regulated by androgens^7–12. Previous studies have shown that women who are taking androgenic steroids to change sex (e.g. young female-to-male transsexuals) demonstrate high elevations of serum PSA and urine PSA^37,38. We have also shown previously that women who are taking oral contraceptives upregulate their PSA at both the tissue and the serum level⁹. Based on this data, we hypothesized that serum PSA concentration may represent a novel integrated index of androgenic stimulation in normal women.

It has been suggested that elite athletes more frequently suffer from various hyperandrogenic syndromes such as PCOS, congenital adrenal hyperplasia and 46XY disorders of sex development^33–37. In this paper, we examine the possible differences in serum PSA between Olympic elite athletes and sedentary control women. For this, we measured both cPSA and fPSA, as well as a panel of androgens, estrogens, and progesterone in serum.

We have shown that in athletes, E1 and E2 concentrations are significantly reduced, whereas levels of DHEA and Delta 5 tend to be slightly elevated. Moreover, we identified for the first time a significant decrease in serum cPSA and fPSA in elite athletes, in comparison to the control group. We have also confirmed the previous finding⁹ that oral contraceptive use increases both cPSA and fPSA in serum of athletes (but the differences were not significant in the control group, although the same trend was observed).

Recently, we speculated that serum and urine PSA in female athletes could be used to test for doping with androgenic steroids⁴². In this work, we did not examine this possibility due to lack of samples from provenly doping athletes. However, now that serum PSA in women can be reliably quantified, this parameter could be considered for inclusion into the athlete’s biological passport, as suggested by Sottas et al⁴³.

Our finding that serum E1 and E2 are lower in athletes or exercising individuals in comparison to controls is supported by previous literature^43–45. An explanation for this phenomenon may be related to menstrual cycle disturbances induced by strenuous exercise and stress during the competition. Such findings have prompted some to speculate that the lower incidence of breast cancer in individuals who exercise may be due to reduced estrogen and androgen levels^43–45. Previously, we also established a connection between tumoral or nipple aspirate PSA and breast cancer prognosis^46–49. Clearly, the interconnections between exercise, androgen, estrogen and serum PSA levels and breast cancer need to be better defined.

It has been reported before that serum PSA fluctuates with the menstrual cycle, likely due to up-regulation by progesterone in the luteal phase^13–15. We examined the correlation and the association of cPSA and fPSA with serum progesterone, as a continuous and dichotomous variable, respectively. Our results have shown that there was no correlation or association between either cPSA or fPSA and progesterone or with the frequency of ovulatory cycles, (ovulation being defined as progesterone ≥ 5.4 ng/mL in the luteal phase).

While we found that many anthropometric measurements are different between athletes and controls (Table 1), among steroid measurements, only E1, E2, and to a lesser extent DHEA and Delta5 concentrations were different between the two groups. However, we identified a significant difference between both cPSA and fPSA between elite athletes and controls. The differences in cPSA and fPSA levels between these two groups do not seem to be associated with hyperandrogenism or the menstrual cycle. It will be interesting to examine, in the future, what other parameters are responsible for the differential concentrations of cPSA and fPSA in serum of these women. Since PSA in women originates mostly from the breast, differences in breast size, which was one parameter we did not study, could be one possible cause for the different levels of these proteins.

Data availability

Dataset 1: Anthropometric and biochemical data for athletes and controls. The data file contains all available anthropometric and biochemical data for all athletes and controls included in this study, and includes free PSA and complexed PSA. It was used to do statistics and to derive the results of this paper. DOI, 10.5256/f1000research.11821.d168039⁵⁰

Consent

This study was approved by the Regional Ethics Committee of the Karolinska Institutet, Stockholm, Sweden (EPN 2011-1426-3). Written informed consent was obtained from all participants.