Serum androgen levels in elite female athletes.

ORIGINAL E n d o c r i n e

ARTICLE R e s e a r c h

Serum Androgen Levels in Elite Female Athletes Stéphane Bermon, Pierre Yves Garnier, Angelica Lindén Hirschberg, Neil Robinson, Sylvain Giraud, Raul Nicoli, Norbert Baume, Martial Saugy, Patrick Fénichel, Stephen J. Bruce, Hugues Henry, Gabriel Dollé, and Martin Ritzen International Association of Athletics Federations Medical and Anti-Doping Department and Commission (S.B., P.Y.G., M.S., G.D.), 98000 Monaco; Laboratoire Motricité Humaine Education Sport Santé (S.B.), Nice Sophia Antipolis University, 06107 Nice, France; and Monaco Institute of Sports Medicine and Surgery (S.B.), 98000 Monaco; Department of Women’s and Children’s Health (A.L.H., M.R.), Karolinska Institutet and University Hospital, SE-141 86 Stockholm, Sweden; Swiss Laboratory for Doping Analyses (N.R., S.G., R.N., N.B., M.S.), University Center of Legal Medicine, Geneva and Lausanne, and Centre Hospitalier Universitaire Vaudois and University of Lausanne, 1005 Lausanne, Switzerland; Department of Reproductive Endocrinology, and INSERM Unité 1065 (P.F.), Hôpital l’Archet, University Hospital of Nice, 06-003 Nice, France; Department of Clinical Chemistry (S.J.B., H.H.), Centre Hospitalier Universitaire, University Hospital of Lausanne, Vaudois, 1011 Lausanne, Switzerland

Objective: Prior to the implementation of the blood steroidal module of the Athlete Biological Passport, we measured the serum androgen levels among a large population of high-level female athletes as well as the prevalence of biochemical hyperandrogenism and some disorders of sex development (DSD). Methods and Results: In 849 elite female athletes, serum T, dehydroepiandrosterone sulphate, androstenedione, SHBG, and gonadotrophins were measured by liquid chromatography-mass spectrometry high resolution or immunoassay. Free T was calculated. The sampling hour, age, and type of athletic event only had a small influence on T concentration, whereas ethnicity had not. Among the 85.5% that did not use oral contraceptives, 168 of 717 athletes were oligo- or amenorrhoic. The oral contraceptive users showed the lowest serum androgen and gonadotrophin and the highest SHBG concentrations. After having removed five doped athletes and five DSD women from our population, median T and free T values were close to those reported in sedentary young women. The 99th percentile for T concentration was calculated at 3.08 nmol/L, which is below the 10 nmol/L threshold used for competition eligibility of hyperandrogenic women with normal androgen sensitivity. Prevalence of hyperandrogenic 46 XY DSD in our athletic population is approximately 7 per 1000, which is 140 times higher than expected in the general population. Conclusion: This is the first study to establish normative serum androgens values in elite female athletes, while taking into account the possible influence of menstrual status, oral contraceptive use, type of athletic event, and ethnicity. These findings should help to develop the blood steroidal module of the Athlete Biological Passport and to refine more evidence-based fair policies and recommendations concerning hyperandrogenism in female athletes. (J Clin Endocrinol Metab 99: 4328 – 4335, 2014)

ince its very early stages, the fight against doping in sports has mostly consisted of detecting prohibited substances in urine or blood samples collected from ath-

S

letes. More recently the Athlete Biological Passport (ABP) was proposed as an alternative means to drug testing (1). Indeed, there is an ongoing shift in the drug-testing par-

ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received February 9, 2014. Accepted July 25, 2014. First Published Online August 19, 2014

Abbreviations: A4, androstenedione; ABP, Athlete Biological Passport; DHEAS, dehydroepiandrosterone sulfate; DSD, disorders of sex development; FT, free T; IAAF, International Association of Athletics Federations; IOC, International Olympic Committee; OC, oral contraceptive; P, percentile; PCOS, polycystic ovary syndrome; WADA, World Anti-Doping Agency.

4328

jcem.endojournals.org

J Clin Endocrinol Metab, November 2014, 99(11):4328 – 4335

doi: 10.1210/jc.2014-1391

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 20 November 2014. at 08:17 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/jc.2014-1391

adigm, from the direct identification of banned substances in the athlete’s biological samples to the detection of abnormalities in biomarkers that potentially indicate a doping offense. A departure of an athlete’s values compared with his or her historical data, for these selected biomarkers, indicates either doping or a potential medical condition that deserves closer health examination. This Bayesian statistical approach is of particular interest when athletes use either substances similar to those produced by the human body, such as erythropoietin, T, and GH or previously undetectable synthetic molecules or methods that are used to illegally enhance athletic performance. The hematological module of the ABP is now successfully implemented by many national and international sports governing bodies (2– 4). Although it does not totally eliminate blood doping, it acts as an effective deterrent on doping in sports (5). The two other modules, still to be developed and scientifically validated, are the steroidal (blood and urine) and the endocrine modules. The ABP endocrine module mainly concerns the indirect detection of doping with GH, IGF-1, and related compounds, whereas the steroidal module aims to detect direct and indirect forms of doping with anabolic agents including the endogenous ones. Androgens (in particular T) show ergogenic effects in both genders (6), but one could expect that these performance-enhancing effects are more spectacular when these molecules are given (or produced) to (by) female athletes. The only available source of documentation that clearly illustrates the effect of exogenous androgens on physical performance in female athletes are the now disclosed documents from experiments performed by sports scientists in the former German Democratic Republic (7). These scientists concluded after the 1972 Olympic Games in Munich that “the effects of the treatment with androgenic hormones were so spectacular, particularly in female athletes in strength dependent events, that few competitors not using the drugs had a chance of winning.” The sports governing bodies, and later on the World Anti-Doping Agency (WADA), consequently banned all use of anabolic/androgenic agents. The highly increased endogenous productions of androgens demonstrated by some female athletes, as well as their virilized phenotype, have recently been advertised in sports and the scientific communities (8, 9). Thus, in 2011 and 2012, the International Association of Athletics Federations (IAAF) and the International Olympic Committee (IOC) respectively published regulations and recommendations governing eligibility of women with hyperandrogenism to compete in women’s competition (10 –12). Unfortunately, and to the best of our knowledge, there are neither available data on serum androgen levels


4329

nor reliable statistics on the so-called hyperandrogenism among a large and high-level female athletes’ population. This lack of information should be overcome, should the scientific and sports governing bodies want to develop, validate, and implement a blood steroidal module of the ABP in the future. Therefore, the primary purpose of this study was to measure the serum androgen levels among a large population of high-level female athletes, including different types of sport events. The secondary purpose of this study was to estimate the occurrence of hyperandrogenism among a high-level female athlete population.

Materials and Methods This project was initiated by three main partners: the IAAF, the WADA-accredited laboratory in Lausanne, Switzerland, and WADA. The aim was to test all athletes taking part in the 2011 IAAF World Championships in Daegu (South Korea), as an initial step in creating a blood passport for each athlete. As a second step, reference values for the endocrine and steroidal modules of the ABP would be set up. All logistical and methodological (including sampling and preanalytical) aspects of this project have been extensively described and published elsewhere (12). The project comprised a study of 1833 high-level athletes from 200 countries, competing at the highest level in 26 different track and field events. Eight hundred fifty-five females competed in 23 different events.

Subjects Of the 855 female athletes, a single blood sample was obtained from 849 women from 163 countries. Four athletes were not included because of missing or incomplete data. As for all athletes competing in international competitions in athletics, these World Championships’ athletes were asked to sign an acknowledgment and agreement form by which they agreed to be bound and in compliance with the antidoping rules and by which they consented to the processing of their personal information for antidoping purposes (13) and processing of pooled and anonymous information for antidoping research purpose. Although institutional review board approval was not requested, this first step in creating an ABP for androgens to be used in future antidoping activities was fully approved by the WADA. The present report is simply a compilation of data from this project and a necessary first step in formulating references for future use of the ABP. We therefore consider the approval signed by each athlete fully compliant with good ethical standards.

Measures Athletes were required to go to the doping-control station at any time during their stay, provided it was at least 24 hours after their arrival in Daegu and no less than 2 hours after intense exercise. These requirements were imposed to limit the effects of jet lag and of intense exercise on the results. After registration, athletes were required to sit for at least 10 minutes in the waiting room equipped with televisions and computers with Internet access. Prior to undergoing blood sampling, all female athletes were asked to answer a short questionnaire (available in six different languages) to collect individual information about the date


4330

Bermon et al

Androgen Levels in Elite Female Athletes

of the first day of the last menstruation, the use of medicines or oral contraceptives (OCs), and the date of their arrival in the Daegu time zone (see Supplemental Data). The regular use of OCs was confirmed by the measurement of SHBG concentration and a value over 86 nmol/L (14) in addition to a declaration of OC use were two necessary conditions. The sampling hour was nonnormally distributed ranging from 7:05 AM to 10:20 PM. Blood samples were collected between the August 18 and September 1, 2011, and serum was frozen at ⫺20°C until assay. All analyses for endocrine parameters were conducted within less than 25 working days in the Lausanne WADA-accredited laboratory. Serum tubes (Becton Dickinson AG Vacutainer SST II Advance) were thawed at approximately 4°C overnight before analyses. Serum SHBG, LH, and FSH concentrations were determined with an Immulite 2000 (Siemens Healthcare Diagnostics SA), whereas serum human chorionic gonadotropin and dehydroepiandrosterone sulfate (DHEAS) were determined with an Immulite 2000 XPi (Siemens Healthcare Diagnostics SA). Altogether the immunoassays were using 250 ␮L of serum sample, including the dead volume. Measurement kit reagents, equipments, adjustors, and protocols were provided by the manufacturer. All samples, adjustors, and quality controls were run with the same batch lot numbers. Precisions were expressed by the coefficients of variation (in percentages) calculated for the intra- and interday determined on different quality control levels analyzed three times a day at least (quality controls were also provided by the manufactures of the kits; see Supplemental Data). Finally serum T, androstenedione (A4), and 17-hydroxyprogesterone (data not shown) were measured by liquid chromatography-mass spectrometry at high resolution (Q-Exactive; ThermoFisher Scientific; see Supplemental Data). Free T (FT) was calculated by using the Sodergard formula (15), with a standard average albumin concentration of 4.3 g/dL.

Statistical analyses Data distributions were assessed for normality using visual inspection, calculation of skewness and kurtosis, and the Kolmogorov Smirnov test. Correlation between data was analyzed with the Spearman’s rank correlation. To test for the possible

Table 1.


effects of the type of athletic event on hormonal parameters, all female athletes were assigned to one of six groups: throwing, jumping, sprinting, combined events, middle distance running, and long distance running. The throwing group consisted of the following events: discus throw, javelin throw, hammer throw, and shot put. The jumping group consisted of the following events: high jump, long jump, triple jump, and pole vault. The sprinting group consisted of the following events: 100 m, 4 ⫻ 100 m, 110-m hurdles, 200 m, 400 m, 4 ⫻ 400 m, and 400 m hurdles. The combined event group consisted of heptathlon. The middle distance running group event consisted of the 800 m and 1500 m, whereas the long distance running group gathered 3000 m steeple chase, 5000 m, 10000 m, marathon, and 20 km race walking. To test for the possible effects of ethnicity on hormonal parameters, all female athletes were assigned to one of five groups: black, Middle East and Arabics, Hispanics, Asians and Pacific Islanders, and Caucasians. This determination was done by using information such as the place of birth and physical traits. The effects of the type of athletic event, menstrual status, and ethnicity were tested with a one-way ANOVA on log-transformed age and endocrine parameters across the considered groups and Tukey honestly significant difference (Spjotvoll/Stoline) post hoc test when appropriate. Differences between OC users and nonusers were assessed with a nonpaired t test or a Mann-Whitney test. A value of P ⬍ .01 was considered as statistically significant. All data analyses were performed using Statistica version 7.1 (StatSoft).

Results The age and the serum T and DHEAS concentrations and the calculated FT are presented in Table 1. Five athletes who were suspected to be doped prior to the championships but for whom this could not be established were finally tested positive for androgenic or anabolic steroid drugs. After medical investigations, five female athletes

Age and Androgenic Parameters in the Studied Population

Age, y T, nmol/L DHEAS, ␮mol/L A4, nmol/L SHBG, nmol/L FT, pmol/L LH, IU/L FSH, IU/L After removal of 10 athletes with confirmed DSD and/or doping Age, y T, nmol/L DHEAS, ␮mol/L A4, nmol/L SHBG, nmol/L FT, pmol/L LH, IU/L FSH, IU/L

n

Median

P25-P75

Minimum

Maximum

849 849 849 849 849 849 849 849

26.0 0.69 4.21 3.32 61.00 8.20 3.53 4.12

23.0 –29.0 0.50 – 0.93 2.82–5.86 2.51– 4.40 43.40 – 83.70 5.34 –12.18 1.84 – 6.24 2.55–5.67

16.0 0.01 0.40 0.47 5.66 0.12 0.10 0.10

47.0 29.30 15.40 18.85 573.00 469.28 88.40 65.90

839 839 839 839 839 839 839 839

25.0 0.69 4.23 3.32 61.40 8.06 3.53 4.10

23.0 –29.0 0.50 – 0.91 2.80 –5.86 2.49 – 4.40 43.70 – 84.20 5.31–11.97 1.81– 6.25 2.51–5.64

16.0 0.01 0.40 0.47 5.66 0.12 0.10 0.10

47.0 11.90 15.40 18.85 573.00 242.70 88.40 65.90

Data are presented as median [25th percentile (P25) to 75th percentile (P75)], minimum, and maximum.


doi: 10.1210/jc.2014-1391

Table 2.


4331

Age and Androgenic Parameters in the Ten Athletes Diagnosed as Either DSD or Doping (see Table 1)

Status

n

Age, y

T, nmol/L

DHEAS, ␮mol/L

A4, nmol/L

SHBG, nmol/L

FT, pmol/L

LH, IU/L

FSH, UI/L

DSD Doping Z value P value

4 6

20.0 (19 –29) 30.5 (20 –31) ⫺1.95 .05

18.3 (15.6 –29.3) 10.0 (3.6 –15.6) 2.35 .02

3.7 (3.7–3.9) 5.0 (2.8 –11.7) ⫺0.85 .39

2.7 (2.3– 4.6) 3.8 (2.2– 6.7) ⫺0.73 .46

39.7 (33.6 –58.3) 40.7 (19.5– 62.0) ⫺0.21 .83

350.7 (331.3–390.0) 154.5 (94.6 –209.5) 2.56 .01

5.5 (2.8 –12.3) 3.2 (0.8 – 8.5) 1.28 .20

12.2 (8.2–25.1) 6.5 (2.0 –18.6) 1.71 .09

Data are presented as median (minimum–maximum). A Mann-Whitney test was used to test for differences between DSD and doping athletes. The DSD female athlete who was convicted of doping is classified as doping.

were later diagnosed with a hyperandrogenic 46 XY disorder of sex development (DSD) condition. Moreover, one athlete among the five females diagnosed with DSD subsequently tested positive for an anabolic steroid drug. These 10 athletes were then removed from our database and final statistical analyses were performed. Their age and results for their androgenic parameters are presented in Table 2. Nine of the 839 female athletes (after exclusion of the 10 who were doped or had a diagnosis of hyperandrogenic DSD) showed a T concentration greater than 3 nmol/L. Six of these nine athletes, plus three others (a total number of nine, ie, 1.1% of our population) showed a FT value above 40 pmol/L. None of these nine was an OC user, but all were either oligomenorrhoic or amenorrhoic. In non-OC users, the sampling hour was inversely correlated to T (␳ ⫽ ⫺0.22; P ⬍ .001) and A4 (␳ ⫽ ⫺0.20; P ⬍ .001) concentrations and to FT (␳ ⫽ ⫺0.17; P ⬍ .001), but significantly linked to neither SHBG nor DHEAS concentrations. Between 8:00 AM and 8:00 PM, T and A4 mean concentrations decreased by 21% and 18%, respectively. The athletes’ age was inversely correlated to A4 concentration (␳ ⫽ ⫺0.11; P ⬍ .01) and DHEAS (␳ ⫽ ⫺0.24; P ⬍ .001) concentrations. T, DHEAS, and SHBG concentrations but not FT differed significantly between the types of athletic event (Table 3). Throwers, sprinters, and to a lesser extent jumpers showed higher levels of androgenic hormones than longTable 3.

distance runners. Female athletes involved in sprinting events were significantly (P ⬍ .001) younger than the ones involved in throwing, jumping, and long-distance running events (Table 3). Questionnaire analysis revealed that 717 of 839 female athletes (85.5%) declared not using OCs. Among these 717 subjects, 168 females (23.5%) were either oligomenorrhoic or amenorrhoic (more than 30 d since the first day of their last menstruation) (Table 4). Among this subgroup, three athletes were pregnant, proven by 5–7 weeks of amenorrhea associated with an increased serum ␤-human chorionic gonadotrophin (between 2581 and 40536 IU/L). Three hundred eighteen subjects (44.2%) reported having the first day of their last menstruation between 0 and 14 days before blood sampling, whereas 231 (32.1%) between 15 and 28 days ago. We assumed that these two last subgroups are composed of athletes in the follicular and luteal phases, respectively. Oligomenorrhoic and amenorrhoic athletes showed the lowest T, DHEAS, and A4 concentrations but the highest SHBG concentrations (Table 4). Menstrual status had a significant effect on neither LH and FSH concentrations nor FT (Table 4). Non-OC users showed significantly (P ⬍ .001) higher DHEAS, A4, FT, LH, and FSH values than OC users. SHBG was significantly higher in OC users than in nonusers (Table 5). No significant difference in androgen levels was found between the different ethnic groups (Table

Age and Androgenic Parameters of Women Competing in the Different Groups of Athletic Events

Group

n

Age, y

Throwing Jumping Sprinting Combined events Middle distance Long distance Fisher’s Fd P value All groups

106 130 335 28 99 141

27.0 (24.0 –30.0)a 26.5 (23.0 –29.0)a 24.0 (22.0 –27.0) 26.0 (25.0 –28.5) 25.0 (23.0 –29.0) 27.0 (24.0 –31.0)a 14.59 ⬍.001 25.0 (23.0 –29.0)

839

OC, OC/NOC, %

Testosterone, nmol/L

DHEAS, ␮mol/L

Androstenedione, nmol/L

SHBG, nmol/L

FT, pmol/L

17/89 (19.1%) 16/114 (11.1%) 55/280 (19.6%) 6/22 (27.3%) 14/85 (16.5%) 14/127 (11.0%)

0.74 (0.59 –1.01)b 0.68 (0.48 – 0.90) 0.73 (0.52– 0.98)c 0.65 (0.47– 0.87) 0.64 (0.51– 0.90) 0.55 (0.41– 0.74) 5.55 ⬍.001 0.69 (0.50 – 0.91)

4.44 (3.12–5.84)b 4.16 (2.85–5.78)b 4.61 (3.18 – 6.43)b 3.91 (2.80 – 4.70) 4.13 (2.71–5.62) 3.20 (2.04 – 4.64) 8.31 ⬍.001 4.23 (2.80 –5.86)

3.45 (2.46 – 4.57) 3.40 (2.59 – 4.68) 3.41 (2.54 – 4.37) 3.08 (2.21– 4.09) 3.39 (2.71– 4.88) 3.03 (2.16 –3.94) 2.10 .06 3.32 (2.49 – 4.40)

47.4 (31.6 – 87.8) 62.8 (48.3–79.4) 64.2 (45.3– 87.3) 73.0 (52.4 –115.0) 60.4 (44.1– 80.0) 59.6 (42.2–77.7) 2.93 .01 61.4 (43.7– 84.2)

9.64 (6.44 –15.41) 7.39 (5.42–12.10) 8.63 (5.38 –12.35) 6.27 (4.20 – 8.84) 8.05 (5.44 –12.86) 6.79 (4.37–9.36) 1.45 .20 8.06 (5.31–11.97)

122/839 (14.5%)

Abbreviations: NOC, no OC medication and the percentage of athletes taking OC. Data are presented as median (25th percentile to 75th percentile). a

Different from sprinting, P ⬍ .001,

b

Different from long distance, P ⬍ .01.

c

Different from long distance, P ⬍ .001.

d

A one-way ANOVA was used to test for significant differences of log-transformed age and endocrine parameters across the considered groups.


4332

Bermon et al

Table 4. Menstrual Status Follicular phase Luteal phase Oligo- or amenorrhea Fisher’s Fc P value



Endocrine Parameters in Athletes Not Using OC, in Different Phases of the Menstrual Cycle n

T, nmol/L

DHEAS, ␮mol/L

A4, nmol/L

SHBG, nmol/L

FT, pmol/L

LH, IU/L

FSH, IU/L

318

0.76 (0.59 –1.04)a

6.34 (5.29 –7.79)a,b

3.90 (3.14 –5.13)a,b

53.4 (37.2–71.2)a

8.38 (5.74 –12.26)

4.01 (2.15–7.15)

4.16 (2.73–5.74)

231 168

a

a

59.9 (43.7–75.3) 68.9 (51.9 – 88.3)

9.39 (6.59 –13.25) 7.90 (5.49 –11.80)

3.72 (2.11– 6.32) 3.44 (1.98 –5.59)

4.27 (2.94 –5.95) 4.23 (2.81–5.66)

19.5 ⬍.001

0.0833 .92

2.04 .13

0.19 .83

0.73 (0.52– 0.96) 0.54 (0.40 – 0.79)

3.80 (3.34 – 4.26) 2.00 (1.59 –2.49)

17.7 ⬍.001

1349 ⬍.001

a

3.60 (2.74 – 4.47) 2.64 (1.90 –3.71)

a

47.4 ⬍.001

Data are presented as median (25th percentile to 75th percentile). a

Different from oligo- or amenorrhea, P ⬍ .001.

b

Different from luteal, P ⬍ .001.

c

A one-way ANOVA was used to test for significant differences of log-transformed endocrine parameters across the considered groups.

6). There was no difference between any groups (event, ethnicity, OC users) in the mean clock times when the blood was drawn.

Discussion To the best of our knowledge, this study is the first to report serum androgen concentrations in such a large number of high-level female athletes. Blood testing prior to the IAAF World Championships was a part of the ABP program and mandatory for all these track and field specialists. Evolution from a population basis to a subject basis as the number of individual test results increases is the rationale of the current urine steroidal passport, which has been implemented by the WADA. To establish the ABP, it is necessary to collect data from a population to set the reference ranges, allowing then the application of the individual follow-up. Because this campaign was organized with a view to the future development of the blood ABP steroidal module through the Bayesian approach, it was important that all athletes were tested through the collection of small serum volumes. The fulfillments of this important condition led to various logistic and preanalytical challenges that are explained in detail elsewhere (12). From a statistical point of view and regardless of the predictive model to be used in the future for the steroidal module of the ABP, it was important to avoid the most important confounding factors when establishing refer-

ence values in serum sex hormones concentration in female athletes. Doping and some forms of DSD are likely to be the two most important confounding factors when hyperandrogenism in female athletes is considered (16). On the basis of clinical features (masculine body build, hirsutism, abnormal external genitalia observed during the antidoping procedures, secondary investigations performed in specialized centers), urinary test results (urinary steroid profiles, T to epitestosterone ratio, screening for banned anabolic doping agents, isotope ratio mass spectrometry to detect illicit T use), and hormonal blood results obtained from antidoping campaigns, it was possible to suspect and confirm 10 cases of doping and/or DSD among the female population studied. All this information was obtained from an ongoing process, starting several months before and continuing after these World Championships, as a part of IAAF antidoping rules and regulations governing eligibility of females with hyperandrogenism to compete in women’s competition (10). After having removed these 10 women from our study group (Table 1), nine subjects still showed T concentrations above 3 nmol/L and three above 10 nmol/L. Although six of these nine subjects came from a geographical area in which doping in female athletes is not rare, it was not possible to prove any antidoping rules violations or DSD. This is a limitation of this study. One could speculate that high-level female athletes would demonstrate higher T and FT values than their sedentary counterparts (17). Although the present study was not designed to test this hypothesis and did not include

Table 5.

Age and Endocrine Parameters of Athletes Using or Not Using OCs

Group

n

Age, y

T, nmol/L

DHEAS, ␮mol/L

A4, nmol/L

SHBG, nmol/L

FT, pmol/L

LH, IU/L

FSH, IU/L

OC NOC Z value P value

69 717

25.9 (23–29) 25.0 (23–29) 0.08 .93

0.62 (0.46 – 0.83) 0.71 (0.52– 0.94) ⫺2.15 .03

3.18 (2.10 – 4.56) 4.40 (2.90 – 6.00) ⫺4.42 ⬍.001

2.22 (1.76 –2.91) 3.56 (2.65– 4.57) ⫺8.05 ⬍.001

155.0 (118.0 –206.0) 59.0 (41.9 –77.6) 12.53 ⬍.001

3.26 (2.47– 4.42) 8.67 (5.89 –12.61) ⫺10.53 ⬍.001

1.22 (0.14 –2.99) 3.76 (2.11– 6.60) ⫺7.35 ⬍.001

1.84 (0.56 – 4.31) 4.21 (2.79 –5.77) ⫺6.00 ⬍.001

Abbgreviation: NOC, non-OC users. Data are presented as median (25th percentile to 75th percentile). A Mann-Whitney test was used to test for differences between OC users and nonusers.


doi: 10.1210/jc.2014-1391

Table 6.


4333

Age and Androgenic Parameters in the Different Ethnic Groups

Group

n

Age, y

T, nmol/L

DHEAS, ␮mol/L

A4, nmol/L

SHBG, nmol/L

FT, pmol/L

Blacks Middle East and Arabics Hispanics Asians and Pacific Islanders Caucasians Fisher’s Fc P value All groups

240 30

25.0 (23.0 –28.0)a 25.5 (23.0 –28.0)

0.74 (0.52– 0.96) 0.69 (0.51– 0.80)

3.88 (2.51–5.25) 4.10 (2.38 –5.54)

3.19 (2.42– 4.20) 3.38 (2.59 – 4.19)

66.8 (47.8 – 87.2) 51.5 (31.9 – 80.0)

8.20 (5.46 –11.79) 10.41 (5.01–12.65)

43 110

25.0 (23.0 –28.0) 23.0 (21.0 –26.0)a

0.66 (0.43– 0.90) 0.70 (0.48 – 0.85)

4.75 (3.31–5.97) 4.40 (2.90 –5.92)

3.14 (1.84 –3.78) 3.68 (2.63– 4.68)

54.0 (40.0 – 86.2) 58.1 (37.8 –73.8)b

8.70 (4.04 –11.69) 8.33 (5.87–13.12)

416

26.0 (24.0 –30.0) 13.78 ⬍.001 25.0 (23.0 –29.0)

0.66 (0.48 – 0.92) 1.52 .19 0.69 (0.50 – 0.91)

4.42 (3.08 – 6.16) 2.93 .02 4.23 (2.80 –5.86)

3.37 (2.52– 4.51) 1.83 .12 3.32 (2.49 – 4.40)

60.9 (44.3– 84.0) 4.85 ⬍.001 61.4 (43.7– 84.2)

7.53 (5.02–11.81) 1.94 .10 1.09 (0.69 –1.79)

839

Data are presented as median (25th percentile to 75th percentile). a

Different from Caucasians, P ⬍ .001.

b

Different from blacks, P ⬍ .01.

c

A one-way ANOVA was used to test for significant differences of log-transformed age and endocrine parameters across the considered groups.

such a control group, it appears that this hypothesis is not confirmed. Indeed, Haring et al (18), also using liquid chromatography-tandem mass spectrometry, reported even slightly higher T and FT concentrations (median 1.03 nmol/L and 12 pmol/L, respectively) in a large population of healthy and untreated 20- to 29-year-old women, from the northeastern area of Germany. OC intake, which is known to decrease T concentration, was an exclusion criterion in the study from Haring et al (18). This could explain the slight difference observed between our two populations measured with the same technique. Our study in elite female athletes did not confirm the lower androgen levels in black women compared with Caucasian women reported by some authors (19). The DHEAS concentration observed in our population is very similar to the one reported in age-matched healthy sedentary women (mean 4.72 ␮mol/L; percentile (P) 10 3.00 ␮mol/L; P90 9.46 ␮mol/L) (20). However, in our study, 19 female athletes showed a DHEAS concentration greater than 10 ␮mol/L, whereas 82 subjects showed DHEAS concentrations greater than 7.8 ␮mol/L, with this last value being considered as a valid threshold for biological hyperandrogenism (21). Although the present study was not designed to diagnose polycystic ovary syndrome (PCOS), one may suspect, as previously reported by Hagmar et al (22), that several of our athletes experienced this medical condition. Our data confirmed that higher T and FT concentrations are measured in the early morning (23, 24). It is also possible that this effect of the circadian rhythm on T concentration was influenced by a jet-lag phenomenon experienced by our athletes. Indeed, according to our questionnaire results, the average interval between the time of their arrival in the Daegu time zone and the sampling date was 2.4 days. Although many national teams and individuals managed to stay in surrounding countries or areas before their arrival in Daegu, to minimize the jet-lag effect on athletic performance, a residual effect of the trans-

meridian travel on the hormonal parameters cannot be ruled out. Female athletes involved in long-distance running consistently showed lower T and DHEAS concentrations when compared with athletes involved in events requiring strength, power, and speed, but the differences were moderate. Although still controversial, this finding is reported in many other studies (16). In our athletic population, we confirmed the inverse relationship between DHEAS, A4 concentrations, and age (25). In the population studied, 14.5% of the athletes declared using OC. Although published statistics on OC use in high-level athletes are scarce, this percentage is lower than the 47% reported by Hagmar el al (22) in female Swedish Olympians. Our database revealed that a large majority of women coming from developing countries (widely represented in athletics) were not using OCs. Inappropriate reporting from the athlete side because of a lack of understanding of the questionnaire or because not considering OCs as medicines is likely because only 56.5% of our female population who declared using OC showed high SHBG concentration associated with OC use. However, our results confirm that regular OC use is significantly associated with lower androgen concentrations and higher SHBG concentrations (26, 27). This androgen-lowering effect of OCs could also explain our low percentage of use in athletes preferring a higher androgen action. While establishing reference values in serum sex hormones concentration in female athletes, prior to a possible future implementation of a blood steroidal module of the ABP, it is important to note that ethnicity, as defined in our work, does not seem to be a confounding factor. In recent decades, the participation in competitive sports of athletes with DSD and their sometimes associated hyperandrogenism has proved to be a matter of controversy (28). This controversy was recently reactivated when the IAAF and IOC officially released their new pol-


4334

Bermon et al


icies and recommendations regarding the eligibility of hyperandrogenic females to compete in women’s competitions. A hot point of discussion here concerned an independent expert medical panel recommending that the athlete is not eligible to compete in women’s competition if she has normal androgen sensitivity and serum T levels above the lower normal male range (10 nmol/L). This arbitrary definition was chosen in the absence of normative statistics of androgen levels in a high-level athlete female population. In our present cohort, the 99th percentile for T concentration is calculated at 3.08 nmol/L. It is close to the 2.78 nmol/L threshold proposed as one of the criteria for the diagnosis of PCOS (29) and the 3.0 nmol/L cutoff proposed by others (30) to detect hyperandrogenism. This finding should clearly help to refine more evidence-based fair policies and recommendations concerning hyperandrogenism in female athletes. The lack of definitive research linking female hyperandrogenism and sporting performance is problematic and represents another central point of the controversy (9, 31). With the exception of data extracted from doping programs in female athletes in the former German Democratic Republic (7), there is no clear scientific evidence proving that a high level of T is a significant determinant of performance in female sports. A randomized placebo-controlled study would be unethical. However, a publication by Rickenlund et al (32) studying athletes active in endurance sports reported that the hyperandrogenic subgroup (T concentration 1.9 ⫾ 0.2 nmol/L) showed a more anabolic body composition, a higher total bone mineral density (BMD), and upper to lower fat mass ratio as well as the highest maximal oxygen uptake and performance values in general than did oligomenorrheic or amenorrheic athletes with normal androgen levels (1.1–1.2 ⫾ 0.4 nmol/L). Similarly, Hagmar et al (22) reported an overrepresentation of polycystic ovaries in female Olympic athletes (37% vs 20% in the general population). This PCOS subgroup showed a higher T concentration and free androgen index than those observed for regularly menstruating or nonPCOS Olympian athletes. This last recruitment bias supports the assumption that there is an ergogenic effect of T in high-level female athletes. Congenital adrenal hyperplasia is a possible cause of virilization in our elite female athletes. Among our studied population, none of the 13 athletes with a 17-hydroxyprogesterone serum concentration above 8 nmol/L showed increased T or A4 (data not presented), ruling out the possibility of an untreated 21hydroxylase deficiency, the most common form of congenital adrenal hyperplasia, as a cause of the high T levels. Within the female population studied, six subjects (five plus one who was previously declared eligible to compete) were later identified with a hyperandrogenic 46 XY DSD


(four with 5␣-reductase deficiency and two with partial androgen insensitivity). Thus, the calculated prevalence of this type of medical condition within this cohort of high]level female athlete is 7.1 per 1000. Because, in this study, screening for SRY was not performed (we looked only for women that were hyperandrogenic), it is possible that the actual prevalence of 46 XY was even higher. Morel et al (33) estimated a 46 XY DSD occurrence rate of 1 in 20 000 in the general population. Although the occurrence may change from one ethnic group to another, our reported prevalence remains approximately 140 times higher than expected in the general population. Moreover, our results concur with previous report (34) and the DSD (carrying the SRY gene) prevalence observed during the 1992 (7.5 per 1000) and the 1996 (2.67 per 1000) Olympic Games (28). Our 46 XY DSD athletes tended to be younger and showed a higher T concentration than the doped female athletes. Within this small subgroup, LH did not help to discriminate between DSD and doped athletes. This unique study showed for the first time that the androgenic parameters measured in a large sample of high-level female athletes were close to those observed in a healthy young population. From an ABP perspective, our data confirm that T, FT, and A4 concentrations are weakly linked to the time of sampling and that A4 and DHEAS concentrations are inversely linked to the athlete’s age. Oligo- and amenorrhea as well as OC intake leads to lower androgen levels. It also appears that there is no need for normative statistics for serum androgen levels according to the type of athletic events or ethnic origin. Like PCOS, hyperandrogenism secondary to a DSD is much more frequent in a population of high-level female athletes than in the general population. This important recruitment bias is, in our opinion, an indirect evidence for performance-enhancing effects of hyperandrogenic DSD conditions and their associated high T concentration in female athletes, but we cannot exclude that the Y chromosome in some unknown way may bring an advantage to female athletes.

Acknowledgments We acknowledge the organizational and logistical support provided by the local organizing committee and the Korean AntiDoping Agency. Special recognition is due to the staff from the Doping Control Center, Korea Institute of Science and Technology, Antidoping Center Moscow, the Swiss Laboratory for Doping Analyses, and the Centre Hospitalier Universitaire Vaudois, who did the laboratory work. We also acknowledge Siemens Healthcare Diagnostics SA (Zürich, Switzerland) for providing the hormone tests and Ms Sharon Bertholet-Loubert and Susanna Verdesca for processing the data and checking the manuscript.


doi: 10.1210/jc.2014-1391

Address all correspondence and requests for reprints to: Stéphane Bermon, MD, PhD, Head, Monaco Institute of Sports Medicine and Surgery-Exercise Physiology, 11 Avenue d’Ostende, Monaco, Monaco 98000. E-mail: [email protected]. This work was supported by the International Association of Athletics Federations, the World Anti-Doping Agency, and the Anti-Doping Switzerland. Disclosure Summary: The authors have nothing to declare.

References 1. Sottas PE, Robinson N, Rabin O, Saugy M. The athlete biological passport. Clin Chem. 2011;57:969 –976. 2. Sottas PE, Vernec A. Current implementation and future of the Athlete Biological Passport. Bioanalysis. 2012;4:1645–1652. 3. Zorzoli M, Rossi F. Case studies on ESA-doping as revealed by the Biological Passport. Drug Test Anal. 2012;4:854 – 858. 4. Schumacher YO, d’Onofrio G. Scientific expertise and the Athlete Biological Passport: 3 years of experience. Clin Chem. 2012;58: 979 –985. 5. Banfi G. Limits and pitfalls of Athlete’s Biological Passport. Clin Chem Lab Med. 2011;49:1417–1421. 6. Mooradian AD, Morley JE, Korenman SG. Biological actions of androgens. Endocr Rev. 1987;8:1–28. 7. Franke WW, Berendonk B. Hormonal doping and androgenization of athletes: a secret program of the German Democratic Republic government. Clin Chem. 1997;43:1262–1279. 8. Xavier NA, McGill JB. Hyperandrogenism and intersex controversies in women’s olympics. J Clin Endocrinol Metab. 2012;97:3902– 3907. 9. Bermon S, Ritzén M, Hirschberg AL, Murray TH. Are the new policies on hyperandrogenism in elite female athletes really out of bounds? Response to “Out of bounds? A critique of the new policies on hyperandrogenism in elite female athletes.” Am J Bioeth. 2013; 13:63– 65. 10. International Association of Athletics Federation. IAAF regulations governing eligibility of females with hyperandrogenism to compete in women’s competitions. 2011. http://www.iaaf.org/mm/document/ aboutiaaf/publications/05/98/78/20110430054216_httppostedfile_haregulations(final)-appendices-amg-30.04.2011_24299.pdf. 11. International Olympic Committee. 2012 IOC regulations on female hyperandrogenism: games of theXXXOlympiad in London, 2012. http://www.olympic.org/documents/commissions_ pdffiles/medical_commission/2012-06-22-ioc-regulations-onfemale-hyperandrogenism-eng.pdf. 12. Robinson N, Dollé G, Garnier PY, Saugy M. 2011 IAAF World Championships in Daegu: blood tests for all athletes in the framework of the Athlete Biological Passport. Bioanalysis. 2012;4:1633– 1643. 13. International Association of Athletics Federation. IAAF athletes’ guide and advisory notes. Use of personal information. 2013. http:// www.iaaf.org/download/download?filename⫽9c266b70-2f6f-44ceb2cd-111069ed89ed.pdf&urlslug⫽Use%20of%20personal%20 information. 14. Zimmerman Y, Eijkemans MJ, Coelingh Bennink HJ, Blankenstein MA, Fauser BC. The effect of combined oral contraception on testosterone levels in healthy women: a systematic review and metaanalysis. Hum Reprod Update. 2014;20:76 –105. 15. Sodergard R, Backstrom T, Shanbhag V, Carstensen H. Calculation of free and bound fractions of testosterone and estradiol-17␤ to


16. 17.

18.

19.

20.

21.

22.

23. 24.

25. 26.

27.

28. 29.

30.

31.

32.

33.

34.

4335

human plasma proteins at body temperature. J Steroid Biochem. 1982;16:801– 810. Wood RI, Stanton SJ. Testosterone and sport: current perspectives. Horm Behav. 2012;61:147–155. Enea C, Boisseau N, Fargeas-Gluck MA, Diaz V, Dugué B. Circulating androgens in women: exercise-induced changes. Sports Med. 2011;41:1–15. Haring R, Hannemann A, John U, et al. Age-specific reference ranges for serum testosterone and androstenedione concentrations in women measured by liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab. 2012;97:408 – 415. Spencer JB, Klein M, Kumar A, Azziz R. The age-associated decline of androgens in reproductive age and menopausal Black and White women. J Clin Endocrinol Metab. 2007;92:4730 – 4733. Davison SL, Bell R, Donath S, Montalto JG, Davis SR. Androgen levels in adult females: changes with age, menopause, and oophorectomy. J Clin Endocrinol Metab. 2005;90:3847–3853. Carmina E, Rosato F, Jannì A, Rizzo M, Longo RA. Extensive clinical experience: relative prevalence of different androgen excess disorders in 950 women referred because of clinical hyperandrogenism. J Clin Endocrinol Metab. 2006;91:2– 6. Hagmar M, Berglund B, Brismar K, Hirschberg AL. Hyperandrogenism may explain reproductive dysfunction in female Olympic athletes. Med Sci Sports Exerc. 2009;41:1241–1248. Dabbs JM Jr. Salivary testosterone measurements: reliability across hours, days, and weeks. Physiol Behav. 1990;48:83– 86. Dabbs JM Jr, La Rue D. Salivary testosterone measurements among women: relative magnitude of circadian and menstrual cycles. Horm Res. 1991;35:182–184. Kroboth PD, Salek FS, Pittenger AL, Fabian TJ, Frye RF. DHEA and DHEA-S: a review. J Clin Pharmacol. 1999;39:327–348. Wiegratz I, Kutschera E, Lee JH, et al. Effect of four different oral contraceptives on various sex hormones and serum-binding globulins. Contraception. 2003;67:25–32. Jung-Hoffmann C, Heidt F, Kuhl H. Effect of two oral contraceptives containing 30 ␮g ethinylestradiol and 75 ␮g gestodene or 150 ␮g desogestrel upon various hormonal parameters. Contraception. 1988;38:593– 603. Ritchie R, Reynard J, Lewis T. Intersex and the Olympics. J R Soc Med. 2008;101:395–399. The Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Working Group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod. 2004;19:41– 44. Al Kindi MK, Al Essry FS, Al Essry FS, Mula-Abed W-AS. Validity of serum testosterone, free androgen index, and calculated free testosterone in women with suspected hyperandrogenism. Oman Med J. 2012;27:471– 474. Karkazis K, Jordan-Young R, Davis G, Camporesi S. Out of bounds? A critique of the new policies on hyperandrogenism in elite female athletes. Am J Bioeth. 2012;12:3–16. Rickenlund A, Carlström K, Ekblom B, Brismar TB, von Schoultz B, Hirschberg AL. Hyperandrogenicity is an alternative mechanism underlying oligomenorrhea and amenorrhea in female athletes and may improve physical performance. Fertil Steril. 2003;79:947–955. Morel Y, Rey R, Teinturier C, et al. Aetiological diagnosis of male sex ambiguity: a collaborative study. Eur J Pediatr. 2002;161: 49 –59. Fénichel P, Paris F, Philibert P, et al. Molecular diagnosis of 5␣reductase deficiency in 4 elite young female athletes through hormonal screening for hyperandrogenism. J Clin Endocrinol Metab. 2013;98:E1055–E1059.


Bone mineral density levels in elite female athletes.

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

Sleep-hygiene Education improves Sleep Indices in Elite Female Athletes.

Prevalence of stress urinary incontinence in elite female endurance athletes.

Kinesio taping and jump performance in elite female track and field athletes and jump performance in elite female track and field athletes.

Vaccination in elite athletes.

Coercing women athletes with high androgen levels into medical interventions.

Do Elite Athletes Live Longer? A Systematic Review of Mortality and Longevity in Elite Athletes.

Injuries to elite wheelchair athletes.

Countermovement Jump Performance with Increased Training Loads in Elite Female Rugby Athletes.

Energy availability and the female athlete triad in elite endurance athletes.

Different training patterns and bone mineral density of the femoral shaft in elite, female athletes.

[Adherence to a mediterranean diet and sport performance in a elite female athletes futsal population].

Self-reported versus diagnosed stress fractures in norwegian female elite athletes.

Sedentary behaviour and adiposity in elite athletes.

Predictors of airway hyperresponsiveness in elite athletes.

Injuries in elite Taekwondo Poomsae athletes.

Reduced thymic output in elite athletes.

Cardiovascular health in former elite male athletes.

Reduced mortality in former elite endurance athletes.

LA Size in Former Elite Athletes.

Patterns of performance development in elite athletes.

Elite female athletes' ventilatory compensation to decreased inspired O2 during the wingate test.

Serum zinc is associated with plasma leptin and Cu-Zn SOD in elite male basketball athletes.