The Athlete Biological Passport (ABP) was adopted by the World Anti-Doping Agency (WADA) to longitudinally monitor steroid biomarkers (WADA, 2021). While the urinary module of the ABP has proven effective for male athletes, its utility in female athletes is more limited. This is primarily due to lower baseline androgen concentrations, greater hormonal fluctuations associated with the menstrual cycle, and the use of hormonal contraceptives (HC). These factors introduce significant variability to the steroid profile, thereby widening the tolerance thresholds in the ABP and reducing its sensitivity (Elings Knutsson et al., 2021; Schulze et al., 2021; Ekström et al., 2023).

We have recently demonstrated that female Olympic athletes exhibit lower urinary androgen levels compared to sedentary controls even though serum androgen levels were comparable. Furthermore, the urinary androgen levels negatively correlated with training volume (Eklund et al., 2021). Similar findings were reported by Timon et al., showing lower urine steroid profile values in trained male cyclists than sedentary controls (Timon et al., 2008a). Since a possible eliminations path of androgens may occur via sweat (Thieme et al., 2003), we hypothesized that the increased sweating during intense training over time, may contribute to the reduction in urinary androgen concentrations in highly trained athletes.

Initially, the ABP focused solely on the urinary steroid profile, but it has since been expanded to include serum testosterone (T) (WADA, 2023). Androstenedione (A4) was also introduced to form the ratio T/A4 which is considered more stable and sensitive than T alone and less affected by menstrual cycle and hormonal contraceptives (Salamin et al., 2022b; Knutsson et al., 2023). While various confounding factors have been extensively studied for the urinary model, they remain less explored in the serum-based model. The effect of training/exercise on T and A4 levels in women is not clear. Evidence supports an increase in T after acute exercise although this normalizes to baseline or lower within hours (Enea et al., 2011; Hackney and Willett, 2020). Long-term exercise studies have shown more variability, most likely attributed to different study designs (Enea et al., 2011; Swain et al., 2022).

Dried blood spots (DBS) collected via capillary sampling have been implemented as a complementary matrix in anti-doping testing, mainly in situations where venous blood collection is challenging (Thevis et al., 2016). Even though routine T and A4 is analyzed in serum today, DBS may be employed for blood steroids profiling provided that all comparisons are conducted within the same matrix (Salamin et al., 2021; Salamin et al., 2022a). Studies have shown that analysis of T and A4 in DBS is reliable to detect micro-dosing of T in healthy females (Desai et al., 2024). In light of its minimally invasive nature and suitability for remote or home-based sampling, DBS was selected as the preferred matrix for this study.

The aim of the present study was to investigate the urinary steroid profile and estimated sweat production during a non-training day and a training day in female athletes training on average 2–3 h per day compared to controls training less than 2 h per week. In addition, for the athlete’s group, we aimed to study the acute effect of exercise on T and A4 concentrations when measured in DBS.

To our knowledge this is the first study investigating the urinary steroid profile in 48 and 24 h urine collection, in female athletes compared to untrained controls and in the athletes between a training day and a non-training day. Our hypothesis was that the athletes would eliminate less urinary steroids than the controls in agreement with results from our previous study (Eklund et al., 2021). However, the total amount of steroids excreted in urine showed only significantly lower concentrations of A and 5αAdiol in the athletes. The reason why only 5α metabolites showed a difference is not known. It could be a random finding, or it may be that A is the predominant metabolite excreted via sweat (Thieme et al., 2003), and subsequently primarily affected. As expected, the estimated sweat production was higher during the training day (day 2) compared to day 1 among the athletes. There was, however, a large inter-individual variation in the estimated perspiration, which could be due to the various training activities the participants were engaged in. Moreover, the total sweat production calculation may be limited when considering an entire day, especially considering the estimation error of sweat loss in cooler environments mentioned by Cheuvront et al. (2002).

The reason for not seeing similar differences on the urinary steroid profile between female athletes and controls as in Eklund et al. (2021), despite power analyses, could be trait differences in the study participants. In our previous study, female Olympic athletes were included, i.e., individuals training at an extreme level with the goal of competing in upcoming Olympic Games. Even though the inclusion criteria for the athletes in this study were to exercise 2–3 h per day, there was no criteria for competing or the intensity of the training nor information regarding how well-trained the study subjects were. Subsequently, the Olympic athletes included in our previous study were likely more well-trained than the athletes included in the present study. Therefore, there is a possibility that there was not a high enough training difference between the groups in this study to observe a difference in all urinary steroids.

There was no clear acute effect of training on the urinary levels. As the urine samples were not taken at specific time points after the training session (varied from collected just after finishing the training session to up to 2 h and 15 min after) it was not possible to assess an acute effect of the urinary steroids. Other training studies have shown a direct decrease in urinary steroids that normalizes within hours (Timon Andrada et al., 2007; Timon et al., 2008b; Maynar et al., 2010).

The acute effect on circulatory T and A4 on the other hand could be evaluated as all participants collected DBS directly before and after their training session. The lack of increase in serum T further supports the notion that training intensity might not have been high enough. It has been shown that the training-intensity threshold for an androgen increase in serum may depend on fitness level. It has been reported that non-athletes significantly responded as early as 40% maximal oxygen consumption (VO₂ max), whereas athletes did not respond below 90% maximal effort (Sato et al., 2016).

In women, serum T has been shown to increase directly after acute exercise and thereafter returning to resting values or lower within hours (Enea et al., 2011; Hackney and Willett, 2020). Whereas effects of long-term training on serum androgens have shown both elevated and reduces levels (Enea et al., 2011; Swain et al., 2022), these conflicting results are likely due to variations in study design. Also, training load with inadequate recovery time can influence the hormonal response to exercise. The strength of our study design is that the participants performed their training sessions after 2 days recovery period.

We found that T in DBS remained stable throughout the study, whereas significant changes were observed for A4 and the T/A4 ratio. However, these changes were not seen directly between pre-training (DBS sample 2) and post-training (DBS sample 3). Instead, the changes in A4 and the T/A4 ratio were observed between pre-training and follow-up and post-training and follow-up, respectively. The follow-up DBS samples were collected in the morning the day after the study participants performed their regular training regime. The pre-and post-training samples were collected at different time-points during the day (differing from in the forenoon to late evening) depending on when the study participant performed their regular exercise routine.

As the sample time during the day differs between the follow-up samples and the pre- and post-training samples, diurnal effects on A4 concentrations may explain our results. Both A4 and T are known to be higher in morning samples compared to samples taken later during the day (Burger, 2002; Davison and Bell, 2006). The stability of T, in contrast to the variability observed in A4 between DBS time points, may speculatively be explained by differences in diurnal variation.

Unfortunately, serum A4 has not been investigated to the same extent as serum T in relation to exercise in women. Further studies are needed to determine how serum ABP markers are influenced by intensive exercise in female athletes both acute and more long-term effects. In addition, physical activity performed under stress (e.g., competition vs. training) may also affect steroid disposition (Mullen et al., 2020), complicating the design of human exercise studies.

The T and A4 DBS concentrations were in same range as noted in previous DBS studies conducted in healthy young women (Salamin et al., 2022b; Desai et al., 2024), except two DBS samples, where surprisingly, T in male ranges was quantified. From the T administration study, Salamin et al. (2022b) concluded that high capillary testosterone concentration in DBS is highly likely to reflect a direct contact with exogenous testosterone gel. In our study, no T was administered, however it may be possible that the participants encounter T gel from a T-using partner, gym equipment, etc., resulting in T residues on the skin. Both the measured supra-physiological T were from DBS collected at home, and the T levels were normal in the adjacent samples, further supporting an exogenous contamination rather than a pathological explanation.

A limitation with the study design is the heterogenicity in regard to type of training as the participants did their regular- and not a standardized training session. Furthermore, for the urinary- and DBS sampling we could not consider the phase of the menstrual cycle. During the ovulation phase of a normal menstrual cycle there is a small increase in serum A4 as well as T lasting for about 24–48 h (Handelsman et al., 2018; Knutsson et al., 2023). Furthermore, urinary androgens, especially E, fluctuate during the menstrual cycle with a mid-cycle peak (Schulze et al., 2021). However, considering that the blood variation in T and A4, seen in some participants, was not consistent, as well as the sample collection of T and A4 in DBS being within 24 h for all four samples and within 2–3 h for sample 2 and 3, we find it unlikely that the hormonal variations in blood could be explained by menstrual cycle phase. However, as the urinary androgens were collected on 2 consecutive days, menstrual cycle phase could potentially have influenced the comparison of urinary androgens between athletes and controls. Additionally, HC use was allowed. Even so, the frequency of HC use was comparable between groups. Since doping tests are collected randomly, we believe that results presented here are relevant in an anti-doping context. Regarding sweat volume calculations, because sample collection spanned more than a year, environmental variation may have influenced fluid loss; however, this impact is reduced through intraindividual comparisons and paired statistical analyses.

We found that the urine steroids A and 5αAdiol were lower in the athletes than sedentary controls. Furthermore, in the athletes, T in DBS was not affected by training. No change was observed in A4 directly before and after training although a slight increase was observed between pre-training and the sample taken in the morning the day after the training session. The T/A4 ratio slightly decreased between directly post-training and the day after the training.

These findings suggest that T, A4 and the T/A4 ratio are not acutely affected by training. As the sample time during the day differed for the pre-, post-training and follow-up samples these minor changes in A4 and the T/A4 ratio may be due to diurnal variations and not training dependent effects. These results may be of interest when interpreting results of the ABP.