This was a cross-sectional controlled laboratory study. With unchanged living and training conditions, subjects reported energy intake (EI) by completing dietary diaries for 7 consecutive days  (Fig. 1). During this period, exercise energy expenditure (EEE) was monitored during all training units. After 7 days, blood samples were drawn and after 1 day of rest (on day 9), body composition was assessed and REE was measured, followed by three performance tests for determining basal performance. At the end of the study, participants completed psychological questionnaires.
All participants needed to sign an informed consent before commencing all protocols for allowing data to be gathered and analyzed anonymously. This research complied with the declaration of Helsinki. National medical ethical approval was acquired before the start of the study (No. 0120–202/2020/5).
Subjects were invited to participate in the study through national cycling and triathlon organizations, professional cycling team’s coaches. The information was also disseminated through faculty’s laboratory, where national best endurance athletes regularly perform various testings.
Subjects were informed of all procedures and were selected based on inclusion criteria, high motivation and compliance.
Energy availability calculation
All procedures were carried during a 9-day period (Fig. 1). Participant EI was measured by completing dietary diaries for 7 consecutive days . All participants received detailed information on how to complete the diary and how to weigh food or measure its quantity with the help of cups and other measuring tools. They were asked to provide photographic evidence of all food and liquid ingested in that time. EI data was analyzed with Foodworks 9 Professional Edition (version 9.0.3973, Xyrix Software, Australia). During this same period EEE was estimated from heart rate using wearable heart rate monitors during all exercise sessions (Polar V800, Polar Electro, Kempele, Finland). EA was calculated as EA = (EI-EEE)/FFM.
To test performance, three different tests were chosen to assess explosive power of lower extremity (Countermovement jump), motor task execution time (agility t-test) and maximal aerobic capacity (incremental aerobic endurance test). The details of warm-up protocol and tests can be found in the Additional file 1.
First, CMJ test was performed using a bilateral force plate system (Type 9260AA, Kistler Instrumente AG, Winterthur, Switzerland) with Kistler MARS software (S2P Ltd., Ljubljana, Slovenia) to acquire ground reaction force. Each subject has performed three to five maximal counter movement jumps before the testing.
Second, to asses motor task execution time, validated modified agility t-test was used, as described by Haj-Sassi, et al. (2011). The time of best repetition (seconds) were used in further analysis.
After 1 h of rest, endurance was measured with the incremental test to exhaustion. Heart rate, ventilatory, and gas data were collected during the incremental test with metabolic cart (V2 mask (Hans Rudolph, USA), K5 (Cosmed, Albano Laziale, Rome, Italy) with Quark 8.1. PC software support) on a cycle ergometer (Cyclus 2, Leipzig, Germany).
On day 8, venous blood samples were drawn in the morning at 9 am in a fasted state to assess complete blood count, ferritin, serum iron (Fe), triiodothyronine (T3), thyroid stimulating hormone (TSH), morning testosterone, fasting insulin, insulin like growth factor 1 (IGF-1) and 9 am cortisol. Blood was collected using standard clinical procedures. Haemoglobin was analysed with Sysmex XN-550 (photometric detection, EDTA tubes), iron with Cobas c501 (colorimetric analysis, serum tubes), ZSH, T3, testosterone, cortisol and ferritin with Cobas e411 (electrochemiluminescence immunoassay, serum tubes). Serum insulin level was analyzed with a double antibody RIA (serum tubes) and for IGF-1 the RIA kit (serum tubes) was used.
Body composition assessment
Body composition was assessed using tetra polar eight point tactile bioelectrical impedance device InBody 720 (Biospace, Seul, South Korea) on day 9. Prior to body composition measurement, participants received instructions how to be adequately hydrated to enable precise measurement of FFM and body fat percentage that were used in further analysis.
Resting energy expenditure assessment
REE was measured with indirect calorimetry (V2 mask (Hans Rudolph, USA), K5 (Cosmed, Albano Laziale, Rome, Italy) with Quark 8.1. PC software support) based on the Weir equation [14, 15]. The measurement was performed in a thermoneutral environment, in silence, between 6.00 and 9.00 a.m., after 12 h of fasting . It lasted 30 min and the final 20 min were used for REE measurement . During REE measurement, respiratory quotient was monitored since measures under 0.70 or above 1 suggest protocol violations or inaccurate gas measurement . To obtain predicted REE (pREE), a Harris-Benedict equation was used . The mREE/pREE ratio was then calculated for further analysis.
The Three Factor Eating Questionnaire (TFEQ-R18) and Well-being questionnaire were used for psychological assessment [19, 20]. TFEQ-R18 was used to detect early changes in eating behaviors and has three subscales including cognitive restraint, disinhibition and susceptibility to hunger, with higher scores indicating greater eating disturbances in participants. The subscale of interest was cognitive restraint. General well-being was assessed by a simple questionnaire as recommended by Hooper and Mackinnon (1995) including six subjective ratings (fatigue, sleep, stress, muscle soreness, mood and morning erections) on a 1–5 scale. The last item about morning erections was added to the original set as proposed by a study on professional rugby players  (Additional file 2).
All data were analyzed using the IBM SPSS Software for Windows (version 21, SPSS Inc., Armonk, New York, USA). Categorical variables are displayed as numbers and percentages, and numeric variables are presented as means and standard deviations. All numeric variables were first checked for normality of distribution with Shapiro-Wilk’s test. Pearson’s correlation coefficient was computed to assess the relationship between EA and obtained performance, laboratory, body composition and psychological parameters. Based on the EA value, the subjects were later divided into two subgroups (with EA ≥ 30 kcal/kg FFM/day and with EA < 30 kcal/kg FFM/day). The possible differences in performance, blood, anthropometric, body composition, and psychological parameters between those two groups were analyzed using the t-test for independent samples. The significance level was set at p-values < 0.05 for all calculations.