Urinary steroid profile in relation to the menstrual cycle

Abstract The interpretation of the steroidal module of the Athlete Biological Passport (ABP) in female athletes is complex due to the large variation of the endogenous urinary steroids. The menstrual cycle seems to be one of the largest confounders of the steroid profile. The duration of the different phases in the menstrual cycle differs between women and is difficult to predict only by counting days after menstruation. Here, we have determined the follicle, ovulation, and luteal phases, by assessing the menstrual hormones in serum samples collected from 17 healthy women with regular menses. Urine samples were collected three times per week during two consecutive cycles to measure the urinary steroid concentrations used in the ABP. The metabolite that was mostly affected by the menstrual phases was epitestosterone (E), where the median concentration was 133% higher in the ovulation phase compared to the follicle phase (p < 0.0001). The women with a large coefficient of variation (CV) in their first cycle also had a large CV in their second cycle and vice versa. The inter‐individual difference was extensive with a range of 11%–230% difference between the lowest and the highest T/E ratio during a cycle. In conclusion, E and ratios with E as denominator are problematic biomarkers for doping in female athletes. The timing of the sample collection in the menstrual cycle will have a large influence on the steroid profile. The results of this study highlight the need to find additional biomarkers for T doping in females.

Here, we have determined the follicle, ovulation, and luteal phases, by assessing the menstrual hormones in serum samples collected from 17 healthy women with regular menses. Urine samples were collected three times per week during two consecutive cycles to measure the urinary steroid concentrations used in the ABP.
The metabolite that was mostly affected by the menstrual phases was epitestosterone (E), where the median concentration was 133% higher in the ovulation phase compared to the follicle phase (p < 0.0001). The women with a large coefficient of variation (CV) in their first cycle also had a large CV in their second cycle and vice versa. The inter-individual difference was extensive with a range of 11%-230% difference between the lowest and the highest T/E ratio during a cycle.
In conclusion, E and ratios with E as denominator are problematic biomarkers for doping in female athletes. The timing of the sample collection in the menstrual cycle will have a large influence on the steroid profile. The results of this study highlight the need to find additional biomarkers for T doping in females. Five ratios (T/E, A/Etio, 5αAdiol/E, 5αAdiol/5βAdiol, and A/T) are longitudinally monitored in a urinary module of the athlete biological passport (ABP). 1 The intra-subject variations of all the urinary steroid metabolites and the ratios are larger in females than in males 2 making the Abbreviations: 5α-Adiol, 5α-androstane-3α,17β-diol; 5β-Adiol, 5β-androstane-3α,17β-diol; A, androsterone; ABP, athlete biological passport; EAAS, endogenous anabolic androgenic steroid; E, epitestosterone; E2, estradiol; Etio, etiocholanolone; GC-MS/MS, gas chromatography tandem mass-spectrometry; LC-MS/MS, liquid chromatography tandem mass-spectrometry; LH, luteinizing hormone; P, progesterone; T, testosterone; WADA, World Anti-Doping Agency.
interpretation of female samples more challenging. It has been shown that T/E fluctuates during the menstrual cycle, 3,4 which was further investigated in six 5 and nine 6 women, respectively. E was higher in the luteal phase, and hence, the T/E ratio decreased throughout the menstrual cycle. 5 However, in both these studies, the menstrual phases were interpreted based on the number of days after the first menstrual day. As this "day method" often do not corroborate with the actual menstrual phase, it will be more accurate to identify the biological phases of the menstrual cycle by hormonal analyses. 7 A cycle starts with menstruation for 3-7 days, followed by a pituitary release of follicle stimulating hormone (FSH) and luteinizing hormones (LH) to promote development of follicles. The formed Graafian follicle secretes estradiol (E2), which peak just before mid-cycle and stimulate a rise of LH leading to rupture of the Graafian follicle and the development into corpus luteum which secretes progesterone (P) in the later part of the cycle. 8 The reason why urinary E increases in the later phase of the cycle is not known but may be associated with the increase of LH and P in the ovulation and luteal phase, respectively. We have seen that hormonal contraceptives (HC) mediated E changes exhibit strong correlation with the gonadotropins to a much larger extent than the other ABP metabolites, 9 whereas no studies have investigated the correlation between urinary E and the other female hormones P and E2.
Here, we aim to study if the concentrations of the urinary steroid metabolites differ between the different biological phases of the menstrual cycle. Moreover, the present study will provide an understanding if the intra-individual fluctuation of the urinary steroid ABP-ratios is similar between two menstrual cycles.

| Study population
Seventeen healthy women with regular menses not using hormones were included. Urine and blood samples were collected 3 times/week and 1 time/week, respectively, for two consecutive cycles; see Figure 1. The mean age and body mass index (BMI) for the subjects at inclusion were 33.2 ± 6.5 years and 21.6 ± 1.7 kg/m 2 . During the study, the mean number of bleeding days was 7, and cycle length was 24.5 days. For more details, see Mullen et al. 10 Serum levels of E2, P, and gonadotropins were determined by radioimmunoassay using a commercial kit from Diagnostica, Basel,

| Urinary steroid analyses
The conjugated steroids were hydrolyzed using β-glucuronidase and method for doping agents in urine by gas chromatography-triple quadrupole mass spectrometry. 12,13 The limit of detection (LOD) and limit of quantification (LOQ) for T were 0.4 ng/ml and 0.88 ng/ml; E 0.4 ng/ml and 0.56 ng/ml; 5αAdiol 0.9 ng/ml and 3.4 ng/ml; 5βAdiol 1.1 ng/ml and 4.1 ng/ml; A 4 ng/ml and 57 ng/ml and Etio 4 ng/ml and 60 ng/ml, respectively.

| Instrumentation
The steroids were chromatographically separated on a HP1 column The specific gravity of the samples was analyzed by a digital refractometer (UG-1 from Atago).
All urinary steroid concentrations are expressed as the unconjugated plus the glucuronide conjugated fraction. The effect of the urine dilution was adjusted for by normalizing the concentrations to a specific gravity of 1.020.

| Analysis of confounding factors
According to WADA TD2018EAAS, 14     No other confounding factors were found in the urine samples.

| Correlations between serum hormones and urinary steroid metabolites
In order to further study the urinary steroid profile in relation to the menstrual hormones, Spearman correlation analyses were conducted using all the data points from both the cycles (n = 114) (

| Intra-subject variations of ABP ratios between two menstrual cycles
In Figure 3, the individual T/E ratio during two consecutives menstrual cycles is presented. The fluctuation pattern of the T/E ratio between the two cycles is very similar. The same relationship between two cycles was also found for the other ABP ratios ( Figure S1). The median CVs for T and E and each ABP-ratio are shown in Table 2 In agreement with our previous pilot study of six women, 5 urinary E concentration was highest in the later part of the cycle, mimicking the hormonal fluctuation profile of E2 during the menstrual cycle. In fact, here, we show for the first time that urinary E strongly correlates with serum E2, through the whole menstrual cycle as well as in the different menstrual phases. E has been shown to exert anti-androgenic properties in animal studies, and it is possible that E amplifies the estrogenic effects. 18,19 This theory is supported by the marked increase in urinary levels of E (but not the other ABP metabolites) observed during pregnancy. 20 Interestingly, E was the urinary metabolite correlating with P serum levels to a much larger extent than the other urinary steroids.
This corroborates with the hypothesis that E, like P, originates from pregnenolone. 21 The biosynthesis of E is not fully elucidated, but it is believed to occur via the intermediate 5-androstene-3β,17α-diol metabolized by CYP17A1. 21 This is supported by an association between CYP17A1 single nucleotide polymorphism and E concentrations 22 and the fact that co-regulation of E and 5-androstene-3β,-17α-diol are down-regulated in same fashion post T administration in males. 23 However, the reason for the absence of urinary E and serum P correlation in the luteal phase when the highest amounts of P (and pregnenolone) is produced 24 and is unknown.
Hormonal contraceptives have been shown to be an important confounding factor on the urinary steroid profile in female athletes. 9,23,25 During the follicular phase, the urinary E concentration was suppressed in all the women in the study (Figure 2a). The suppression of E was almost as profound as seen in women taking hormonal contraceptives. 9,23,25 Hormonal contraceptives are not one of the confounding factors that the laboratories have to analyze and report according to TD2018EAAS. 14 The findings of this study support the decision not to include HC in the technical document, since it will be very difficult to distinguish between an atypical steroid passport finding due to hormonal contraceptive intake and the follicular phase of the menstrual cycle.
We noted that the urinary concentrations of T and its metabolites, that is, 5αAdiol, 5βAdiol, Etio, and A, differed between the menstrual phases, being highest in the ovulatory phase. In the ovulatory phase, some studies have reported an increase in serum T, 26 which may explain the increase in the urinary rate of T and its metabolites.
The dissimilarity in excretion pattern between E and T further indicates different synthetic and/or elimination pathways. All steroids are excreted mainly as glucuronides, 27 except E which is also subject to sulphate conjugations. 28 Note: Every participant has collected 9-14 urine samples per cycle depending on the length of the cycle (see Figure 1). Participants with T concentrations below LOD (0.4 ng/ml) were excluded in the calculations including T.
polymorphism is plausibly larger than the impact of the menstrual cycle on the urinary steroid levels, these six volunteers were excluded from calculations including testosterone concentrations.
In this study, we found EtG concentrations of >5 μg/ml in 5% of the samples. When excluding these samples from the calculations, the level of significance of the differences between the menstrual phases did not change. Hence, we believe that the EtG only had minor impact on the results in this study. Nine of the samples with reported EtG concentrations were from one individual (p 17) who had T concentrations < LOD and most likely did not have an effect of the results. All the samples that may have been confounded by ethanol are marked with white dots, or squares around the dots in the figures.
Notably, previous studies have shown that EtG concentrations <5 μg/ml also can exert effect on the urinary steroid profile and that 5αAdiol is sometimes also affected. 15 In addition to the large intra-individual variation seen in E concentrations in women, another gender dilemma is that after T administration, urinary E excretion is not down-regulated in women in contrary to men, regardless of UGT2B17 genotype. 33,34 Consequently, T administration in women is not necessarily associated with elevated T/E. 35,36 In addition to monitoring the urinary T/E ratio, future testing programs should consider monitoring serum T and/or other serum markers, 37 which may improve the chance to identify women doping with T. 35 In conclusion, we have investigated the extent of the variation of urinary steroids during the menstrual cycle. The steroid that is most affected by the hormonal fluctuations during the cycle is E, which is on average 133% higher during ovulation compared to the luteal phase. We show that the variation is fairly constant during two menstrual cycles, but the inter-individual variation is large with some women showing hardly any variation, while others have extensive variation throughout the cycle. A more sensitive biomarker for T doping in females than E, for example, serum T is desirable.