Participants

Participants included in this study were part of the IronFEMME project, which received ethical clearance from the Research Ethics Committee of the Universidad Politécnica de Madrid. The purpose of IronFEMME was to determine the influence of sex hormones on iron metabolism and muscle damage, hence, the present study is a secondary analysis that was carried out after the trial was completed. This trial was registered at clinicaltrials.gov. To be included in the IronFEMME study, participants were required to meet the following criteria: (i) healthy adult females between 18 and 40 years; (ii) regular MCs (defined as normally occurring MCs from 21 to 35 days in length) [19] at least 6 months prior to the study; (iii) or using monophasic combined OC pills for at least 6 months prior to the study; (iv) no regular consumption of medication or nutritional supplements; (v) non-smokers; (vi) non-pregnant or oophorectomized; and (vii) participating in endurance training between 3 and 12 h per week. By using blood samples collected as part of the IronFEMME study, the present trial was designed as a secondary analysis, for which the inclusion criteria were further narrowed beyond those determined for the IronFEMME project. These additional criteria were (i) age between 20 and 32 years; (ii) not taking collagen supplements, calcium, or any substance that interferes/participates in bone metabolism; (iii) not having suffered any bone fracture for at least one year prior to the start of the study; and (iv) participating in endurance training involving running (i.e. long distance running, trail running, triathlon) between 3 and 12 h per week (see Table 1 for training volume). Therefore, the study sample was limited to eight eumenorrheic females and eight monophasic OC users (see Table 1 for participants’ characteristics and training volume). All participants were informed of the study procedures (i.e. for the present study on bone (re)modelling) and risks prior to participation and written informed consent was obtained from each subject prior to inclusion. Participants also agreed to the use of their data for other scientific purposes a posteriori, which applies to the present study.

Table 1 Participant characteristics show as mean ± SDMC and OC Cycle Monitoring

MC monitoring was based on the three-step methodology (see Peinado et al. [20] for detailed protocol). The theoretical MC phases were predicted by a gynecologist using the calendar-based counting method, based on records of the length of each participant's last six MCs. Secondly, a urine-based predictor kit (Ellatest, Alicante, Spain) was used to identify the LH surge and subsequent ovulation. Participants collected their mid-morning urine (always at the same time of day) starting three to five days before the estimated LFP testing day until the test result was positive. A participant was excluded from the trial if a positive LH test result was not obtained in three MCs, was as they were considered to have anovulatory MCs, and if the progesterone concentration in the MLP was lower than 16 nmol/L. Finally, all phases were confirmed by serum sex hormone analysis taken on study days prior to the exercise bout. The EFP was characterised by lower levels of 17β-oestradiol and progesterone. The LFP was characterised by higher 17β-oestradiol concentrations than in the EFP and MLP and higher progesterone concentrations than in the EFP, but lower than 6.36 nmol/L. The MLP was characterised by a progesterone concentration greater than 16 nmol/L.

OC users took their active hormone pill daily for 21 days during the active pill-taking phase, followed by a 7-day withdrawal phase (pill without hormonal content). The mean duration of the OC use was 4.09 ± 2.93 years (mean ± SD). The brands and dosages of exogenous sex hormones in the monophasic combined OC preparations used by these participants were as follows: Yasmin® (n = 2): 0.03 mg ethinyl oestradiol and 3 mg drospirenone; Linelle® (n = 2): 0.02 mg ethinyl oestradiol and 0.1 mg levonorgestrel; Sibilla® (n = 2): 0.03 mg ethinyl oestradiol and 2 mg dienogest (n = 2); Edelsin® (n = 1): 0.035 mg ethinyl oestradiol and 25 mg norgestimate; and Yasminelle® (n = 1): 0.02 mg ethinyl oestradiol and 3 mg drospirenone.

Experimental Overview

Eumenorrheic participants came to the laboratory on four occasions (Fig. 1), the first for a maximal incremental treadmill test and the following three times to perform the intervallic running test in each phase of the MC phases (EFP, LFP and MLP). The EFP testing session took place on day 4 ± 1 of the MC. The LPF testing session took place 2 days prior to predicted ovulation, on the day 12 ± 2 of the MC; predicted ovulation was based on previous cycles in which ovulation was confirmed. If ovulation did not occur within 2 days of the predicted LPF testing session, the testing session was deemed invalid. The MLP took place on the day 21 ± 3 of the MC. OC users came to the laboratory on 3 occasions, the first for the maximal incremental test and the following 2 times to carry out the intervallic test in the WP (day 6 ± 1) and APP (day 22 ± 5) of the OC cycle. The first session consisted of participants screening, while on the following sessions participants performed the interval running test in each of the MC and OC cycle phases in a randomized and counterbalanced manner. In the eumenorrheic group, the order of performance of the intervallic tests was randomized according to the phases of the MC as follows: EFP-LFP-MLP; LFP-MLP-EFP; MLP-EFP-LFP; LFP-EFP-MLP, and EFP-MLP-LFP. For the group of OC users, the randomization was: WP-APP and APP-WP.

Fig. 1

Diagram of the study design considering the mean length of the participants' menstrual cycles (30 ± 3 days) and the mean day of LH positive (day 12 ± 2) for eumenorrheic females and participants’ oral contraceptive (OC) cycles. Intervallic trials days expressed as mean (black boxes) ± standard deviation (grey boxes) for the early-follicular phase (EFP; 4 ± 1 days), late-follicular phase (LFP; 12 ± 2 days) and mid-luteal phase (21 ± 3 days) for eumenorrheic females and for withdrawal phase (WP; 6 ± 1 days) and active pill-taking phase (APP; 22 ± 5 days) for OC users. The variables measured pre- and post-exercise in blood samples were 17-βoestradiol, progesterone, P1NP and β-CTX1

On day of screening, volunteers attended the laboratory between 8:00 a.m. and 10:00 a.m. in a resting and fasting state: during the EFP in the eumenorrheic group and day 4–7 of the WP in the OC users. Baseline antecubital venous blood samples were collected for complete blood count, biochemical, and hormonal analysis. After collecting the blood sample, a total body DXA was performed. This screening session was completed with an incremental running exercise to exhaustion on a computerised treadmill (H/P/COSMOS 3PW 4.0, H/P/Cosmos Sports & Medical, Nussdorf-Traunstein, Germany) to determine their maximal oxygen uptake. Expired gases were measured breath-by-breath with a Jaeger Oxycon Pro gas analyser (Erich Jaeger, Viasys Healthcare, Friedberg, Germany). This incremental maximal protocol began with a 3 min warm-up at 6 km/h followed by the incremental test in which the initial speed was set at 8 km/h and then increased by 0.2 km/h every 12 s until exhaustion. Prior to the maximal aerobic test and all the intervallic running tests all participants were instructed to refrain from alcohol, caffeine, and any intense physical activity or sport 24 h before to visit the laboratory.

Intervallic Running Protocol

After the screening day in which the maximal incremental treadmill test was performed with the objective of determining maximal aerobic speed (vVO2peak), interval running tests were performed based upon the obtained values. This intervallic protocol consisted of a 5-min warm-up at 60% of the vVO2peak followed by eight bouts of 3 min at 85% of the vVO2peak with 90-s recovery at 30% of the vVO2peak between bouts. Finally, a 5-min cooling down was performed at 30% of vVO2peak. The intervallic tests were performed in a maximum of two consecutive MCs or two consecutive OC cycles. This protocol was designed for the IronFEMME project with the aim of stimulating IL-6 production, resulting in the subsequent elevation of hepcidin 3 h after exercise. [21] However, this protocol differs in characteristics with respect to those that have been used to study the bone (re)modelling markers response to exercise, which are typically continuous protocols (60–120 min) and intensity between at 65–75% VO2max [22].

Blood Collection

Blood samples were taken between 8 and 10 am to avoid diurnal variability of biochemical parameters [11]. Intervallic tests were always performed between 8 a.m. and 10 a.m. as well, and the time window was reduced to 1 h between tests in the different phases of the MC and OC to reduce the intra-participant variability of the results. Two samples (at rest and immediately post-exercise) were drawn from each participant at each MC and OC phase, from an antecubital vein while they were seated to determine the bone (re)modelling marker and sex hormone concentrations. All venous blood samples were obtained using a 21-gauge (0.8 mm × 19 mm, Terumo®) needle. Blood samples for serum variables were collected in a 9 mL Z serum separator clot activator tubes (Vacuette®) and allowed to clot at room temperature for 60 min. They were then centrifuged for 10 min at 1610 g to obtain the serum (supernatant), divided into 600 μL aliquots, and stored at − 80°C.

Blood Analysis

17-β-oestradiol, progesterone, P1NP and β-CTX-1 were analysed in serum by electrochemiluminescent immunoassay using Roche Diagnostics reagents in a Cobas e411 Elecsys automated analyser (Roche Diagnostics GmbH, Mannheim, Germany) in the Spanish National Centre of Sport Medicine (Madrid, Spain). Inter-assay and intra-assay CV were: 1.8 and 2.4% at 57.2 ng·ml−1 level for P1NP; were 2.1 and 2.8% at 0.403 ng·ml−1 level for β-CTX; 11.9% and 8.5% at 93.3 pg·ml−1 and 6.8% and 4.7% at 166 pg·ml−1 for 17β-oestradiol; and 23.1% and 11.8% at 0.7 ng·ml−1 and 5.2% and 2.5% at 9.48 ng·ml−1 for progesterone.

Corrections for Plasma Volume Changes

Plasma volume changes (ΔPV) can affect the interpretation of biochemical measurements in blood. In the current study the Dill and Costill equation was used for calculation of the % ΔPV using changes in serum total albumin levels post-exercise in each subject, given their correlation with % ΔPV [23]. The following equations [23] for P1NP and β-CTX-1 corrections were used:

$$\% \Delta }\,\,\left( }} \right)$$

$$}\,}_}} \, = \,\left( }\,}_}} } \right)*\left( }} \right)$$

Sex hormone concentrations were not corrected because, although part of the increase in post-exercise circulating hormone concentrations was a result of a decrease in plasma volume, the biological action of these hormones is of greater interest and the concentration of a hormone determines its effect [24].

Nutritional Recommendations

In order to ensure that nutrient intake was not a confounding factor in our results, a nutritionist prescribed the breakfast meal, and participants replicated the same breakfast at least 2h prior to the intervallic tests in all the MC and OC phases before the different blood draws. Nutritional recommendations were standardised 48 h prior and 24 h following the running protocol (for diet composition see Supplementary Material 1).

Statistical Analysis

Normality tests were performed using the Shapiro–Wilk test. Data for non-normally distributed variables were log-transformed for analysis [25].

Participant characteristics were analysed using independent samples t-tests. To explore our objectives, mean concentrations of bone (re)modelling markers and sex hormones were compared between MC phases (EFP vs LFP vs MLP) and OC cycle phases (WP vs APP) using the mixed linear model to analyse repeated measures. The phases and time were set as fixed effects (both intra-subject), and subjects were set as random effect. Comparing hormonal profiles, the mixed linear model analysis was also performed, conducting a separate analysis for each of the following comparisons: EFP vs WP, EFP vs APP, LFP vs WP, LFP vs APP, MLP vs WP, and MLP vs APP. In this case, the ovarian hormonal profile (inter-subject) and time (intra-subject) were set as fixed effects, and subjects were set as random effects. Bonferroni’s post hoc test was applied to pairwise comparisons when the main effect was significant (p < 0.05). The effects sizes are reported as partial eta squared (η2p) whose interpretation is 0.01 = small, 0.06 = moderate, 0.14 = large effect. For significant post hoc comparisons Cohen’s d was used and interpreted based upon the following criteria: 0.2 = small, 0.5 = medium, 0.8 = large effect [26]. Data are presented as mean ± 1SD.