Energy availability discriminates clinical menstrual status in exercising women

To assemble the largest data set possible, we combined data from two studies, one performed at the University of Toronto and one performed at The Pennsylvania State University. Both studies were designed to examine menstrual disturbances and alterations in energy balance in exercising premenopausal women and used very similar methodological approaches. The current study includes data from 91 exercising women in whom assessments of menstrual status and EA were performed. Measurements of the variables of interest, i.e., menstrual status, body composition, EI, EEE, and demographics were conducted by the same investigators at both sites using similar methods. The usefulness of EA (assessed with conventional methods) to discriminate menstrual status was compared to laboratory measures of energy status (fasting TT₃, REE, and ratio of actual to predicted REE [REE/pREE]) [5].

Ninety-one participants were recruited by fliers posted on campus and in the surrounding community, newspaper advertisements, and classroom announcements targeting exercising women for a study on women’s health. Initial eligibility criteria included: 1) no history of or current serious medical conditions; 2) no current clinical diagnosis of an eating or psychiatric disorder based on self-report or an interview with a clinical psychologist or licensed clinical social worker; 3) age 18–35 years; 4) non-smoking; 5) no medication use including contraceptives that would alter metabolic or reproductive hormone concentrations; 6) ≥2 hrs/wk of exercise training; and, 7) no history of or clinical diagnosis of polycystic ovarian syndrome (PCOS) and/or not having a free androgen index (FAI), calculated as (total testosterone (nmol/L)/sex hormone binding globulin (SHBG)(nmol/L))*100), > 6 in participants in which these hormones were measured. An FAI greater than 6 has been reported to be consistent with hyperandrogenemia [25] and represents values greater than three standard deviations in our reference sample (n = 37) which consisted of healthy premenopausal exercising women (18–35 years) with documented ovulatory menstrual status [1].

Participants were informed of the purpose, procedures, and potential risks of participation in the study before signing an informed consent approved by either the Human Ethics Board at the University of Toronto or the Institutional Review Board at the Pennsylvania State University. Height and weight were measured, and participants completed questionnaires to assess demographics, medical history, exercise history, menstrual history, eating behaviors, bone health, and mental health. A physical exam was performed on all participants by an on-site clinician to determine overall health and check for physical symptoms of PCOS such as acne or hirsutism. In most participants a fasting blood draw was analyzed for a complete blood count (n = 57), basic chemistry panel (n = 35), and an endocrine panel (n = 59). The latter included measures of LH, follicle stimulating hormone, thyroid stimulating hormone, thyroxine, prolactin, dihydroepiandrosterone (Quest Diagnostics, Pittsburgh, PA), total testosterone, and SHBG to rule out illness or endocrine or metabolic disease for most participants. Participants met with a General Clinical Research Center (GCRC) registered dietitian or trained laboratory personnel to receive instructions for completion of 3-day diet logs (2 weekdays and 1 weekend day). Additionally, DXA scans were performed to assess body composition.

Peak aerobic capacity (VO_2peak) was measured on a treadmill by indirect calorimetry using an on-line MedGraphics Modular VO₂ System (St Paul, MN) or SensorMedics Vmax metabolic cart (Yorba Linda, Calif., USA) during the study period using methods that have previously been published [2,7].

REE was measured on a single occasion during the study period for all participants. REE was determined by indirect calorimetry using a ventilated hood system (SensorMedics Vmax Series, Yorba Linda, CA, USA). Participants arrived between 0600 and 0900 in a 12 hour post absorptive state within 90 minutes after awakening, having not exercised or ingested caffeine within 24 hours and having not consumed food within 12 hours. In our laboratory, the coefficient of variation for REE measurements, established in 81 participants aged 18–35 years who underwent 2 repeated measurements about 2 weeks apart, was 5.3% ± 0.7%. We compared laboratory assessed REE with a predicted REE (pREE) using the Harris-Benedict equation to estimate how much each individual’s measured REE deviated from the pREE [5]. A lower measured to predicted REE ratio has been associated with a high drive for thinness and with menstrual disturbances in exercising women [5].

Total body weight was measured by a digital scale each week for 4 weeks in the laboratory to the nearest 0.01 kg wearing t-shirt and gym shorts. The mean of these measurements is presented. BMI was calculated as a ratio of weight to height (kg/m²). Height was measured to the nearest 1.0 cm without shoes. Body composition, including percent body fat, fat mass (FM), fat free mass (FFM), and LBM was analyzed by a certified technician using DXA on one of three machines. The majority of participants were scanned on either a GE Lunar Prodigy DXA scanner (n = 57) (GE Lunar Corporation, Madison, WI, enCORE 2002 software, version 6.50.069) or a GE Lunar iDXA scanner (n = 26) (General Electric Lunar Corporation, Madison, WI, enCORE 2008 software version 12.10.113). Remaining participants were scanned on a Hologic QDR4500 DXA scanner (n = 8) (Hologic Inc., Bedford, MA). Consistent with the International Society of Clinical Densitometry guidelines, cross calibration studies were performed to remove systematic bias between the systems. For the cross calibration study between the Lunar Prodigy and Lunar iDXA, fourteen participants were scanned in triplicate on both machines. The majority (n = 8) were scanned on both machines within 5 days while approximately one month lapsed between scans for some participants (n = 6). The values for body composition obtained on each scanner were found to be highly correlated (r = 0.97 BF%, r = 0.99 FM, and r = 0.93 LBM) with no significant difference between the sample mean values. For the cross calibration study between the Hologic QDR4500W and the Lunar iDXA, 32 participants were scanned in duplicate on both machines on the same day. Equations were derived using simple linear regression to remove biases, and body composition values obtained from both the Lunar Prodigy and the Hologic QDR-4500 W were calibrated to the Lunar iDXA.

The classification of menstrual status was based on self-reported menstrual histories, menstrual calendars used to chart menstrual symptoms i.e., cramps, bleeding, spotting, discharge, etc., and daily measurements of urinary estrone-1-glucoronide (E1G), pregnanediol glucuronide (PdG), and mid-cycle LH profiles. Participants who self-reported eumenorrheic menstrual status, defined as regular menstrual cycle intervals of 26–35 days [1], were monitored for 1-3 menstrual cycles. Participants who self-reported no menses within the past 3 months or 6 or fewer menses within the past year collected first morning urine samples beginning on an arbitrary day for 28 days.

Ovulatory status was confirmed by the presence of a urinary LH peak, identified as a peak concentration above 25 mIU/ml occurring after a mid-cycle E1G peak greater than 35 ng/ml, and followed by a peak luteal phase PdG concentration above 5 μg/ml in participants who exhibited menstrual cycles of 26–35 days [1]. Luteal phase defects were confirmed when the luteal phase was either less than 10 days (short) or when the sum of the 3 day mid luteal peak PdG (sum of mid luteal peak PdG ± 1 day) was less than 10 μg/ml and when the PdG peak concentration was below 5 μg/ml but greater than 2.5 μg/ml in participants who exhibited menstrual cycles of 26–35 days (inadequate) [1]. Anovulatory cycles were confirmed as cycles in which a minimal increase in EIG was observed concomitantly with a failure of LH to rise at midcycle and when a luteal phase exhibited no increase in PdG concentration above 2.5 μg/ml in participants who exhibited menstrual cycles of 26–35 days [1]. Oligomenorrhea was confirmed if menses occurred at intervals of 36–90 days and if participants self-reported 6 or less menstrual cycles in the last year prior to the study. Lastly, functional hypothalamic amenorrhea was assessed by confirming a negative pregnancy test, no menses in the past 90 days, and chronically suppressed E1G and PdG profiles [1].

After consideration of menstrual status, participants were grouped as follows: exercising eumenorrheic (regular menstrual cycle intervals of 26–35 days) (ExEumen, n = 41), exercising oligomenorrheic (inconsistent and long menstrual cycle intervals of 36–90 days) (ExOligo, n = 20), and exercising amenorrheic (no menses for a minimum of 90 days prior to the study and for the duration of the study period) (ExAmen, n = 30), i.e., based on “clinical” menstrual status. The eumenorrheic group was further divided according to detailed characterizations of ovulatory status i.e., based on “subclinical” menstrual status as follows: a) exercising ovulatory (ExOv, n = 20): consistently ovulatory cycles for the duration of the 1–3 menstrual cycle collection period, b) exercising with inconsistent presentations of subtle menstrual disturbances (ExIncon, n = 13): various inconsistent combinations of ovulatory, luteal phase defects, and anovulatory cycles from cycle to cycle for the duration of the 1–3 menstrual cycle collection period, and c) exercising anovulatory (ExAnov, n = 8): consistently anovulatory menstrual cycles for the duration of the 1–3 menstrual cycle collection period.

Measures of EI were calculated from 3-day diet logs completed during the study period. Participants were provided with a food scale (ECKO Kitchen Scale) and/or food amounts packet. The packet contained diagrams illustrating container sizes, cuts of meat, and various circles and squares which are typically used when estimating portion size for foods. Participants were encouraged to use these scaled diagrams as a guide for describing dimensions and sizes. Also, included in the packet was a sample page of an accurately completed diet record provided as a reference. Participants were asked to record all foods and beverages consumed on 2 weekdays and 1 weekend day. EI was assessed for 3 days instead of 7 as a shorter 3-day diet log has been shown to reduce participant burden and improve compliance [26]. All 3 days of the diet logs were used. Registered dietitians and/or trained laboratory personnel instructed each participant on how to record EI and then later reviewed diet logs with participants for completeness and accuracy. The nutrient data from the 3-day diet logs were coded and analyzed using Nutritionist Pro (Version 3.1, Axxya Systems, Stafford, TX) or the Nutrition Data System for Research (NDSR 2008 Version; University of Minnesota; Minneapolis, MN).

Participants completed exercise logs where all purposeful exercise sessions greater than 10 minutes in duration with a heart rate above 90 beats per minute were recorded for a 7 day period. The 7-day exercise logs were completed the same week as the 3-day diet logs, and all 7 days of the exercise logs were used. Purposeful exercise included activities such as elliptical, running, or strength training, but not daily living activities such as house cleaning or walking a dog. Participants were asked to wear a Polar S610 or RS400 heart rate monitor during all of their purposeful exercise sessions during the 7 day period. Energy expended during these purposeful exercise sessions was measured using the OwnCal feature of the Polar S610 or RS400 heart rate monitors (Polar Electro Oy, Kempele, Finland) [27]. The OwnCal feature has been validated for calculating EEE from heart rate in University age females running, cycling, and rowing at various submaximal intensities [27]. This feature uses body weight, height, age, gender, VO_2peak, individual maximum heart rate (obtained from VO_2peak test or calculated using 220-age), individual heart rate in a sitting position, and heart rate during exercise to derive kilocalories from energy expenditure. Actual VO_2peak values were input into the heart rate monitors to compute EEE. The Polar S601 and RS400 hear rate monitors include rest in their estimation of energy expenditure. To estimate only EEE, we subtracted measured REE (kilocalories/min) from the Polar heart rate monitors estimation of energy expenditure. For the few (<20%) purposeful exercise sessions in which participants did not wear the heart rate monitors, the Ainsworth et al. [28] compendiums of physical activities were used to determine the appropriate metabolic equivalent (MET) level for the exercise performed. To calculate the energy expended during the exercise session, the MET level was multiplied by the duration (min) of the exercise session. The MET value includes a resting component. To estimate only EEE, we therefore subtracted measured REE (kilocalories/min) from this value. Determination of MET levels from exercise logs from both experimental sites was made by the same member of the research team.

EA was operationally defined as EI minus EEE normalized to kilograms of LBM (EA = (EI–EEE)/LBM) [14] and calculated using the data described above for EI, EEE, and LBM. We calculated EA using the averages of each participant’s values for EI, EEE on workout days only, and the LBM from the DXA scan. EEE represents only those calories attributable to exercise in that the estimate of the calories expended for REE throughout the duration of purposeful exercise sessions was subtracted from the estimate of EEE using the Polar heart rate monitors and or physical activity logs.

All urine samples were corrected for specific gravity using a hand refractometer (NSG Precision Cells) to account for hydration status [29] which has been reported to perform as well as creatinine correction for adjusting urinary hormone concentrations. Microtiter plate competitive enzyme immunoassays were used to measure the daily values for urinary metabolites E1G and PdG as previously reported [1,2]. Urinary LH was measured in samples during the ovulatory phase of the menstrual cycle as determined by visual confirmation of a pre-ovulatory rise in E1G followed by a sustained increase in PdG. LH was determined using a coat-a-count immunoradiometric assay (Siemens Healthcare Diagnostics, Deerfield, IL). The sensitivity of the LH assay was 0.15 mIU/ml and the intra-assay and inter-assay coefficients of variation were 1.6% and 7.1%, respectively.

Blood was collected after an overnight fast before 1000 hr once during the study period. Participants were asked to lie supine for at least 15 minutes after which a blood sample was obtained via venipuncture. Samples were allowed to clot for at least 30 minutes at room temperature and then spun in a centrifuge at 4° Celsius for 15 minutes at 3000 rpm whereafter serum was transferred into appropriately labeled 1.5 mL microtubules and stored at −80° Celsius until analysis. Serum samples were assayed in duplicate for TT₃ using an Immulite chemiluminescent assay (First Generation Immulite 1000, Siemens, Deerfield, IL). The sensitivity of the assay was 35 ng/dL and the intra-assay and inter-assay coefficients of variation were 10.3% and 13.3%, respectively. To determine FAI for screening purposes, total testosterone was measured using a radioimmunoassay kit (Siemens, Los Angeles, CA) through competitive immunoassay. The sensitivity of the assay was 0.14 nmol/L (4.0 ng/dl) and the intra-assay and inter-assay coefficients of variation were 6.4% and 7.5%, respectively. SHBG was assayed in duplicate using a chemiluminescence analyzer (First Generation Immulite 1000, Siemens, Deerfield, IL) through competitive immunoassay. The sensitivity of the assay was 0.2 nmol/L (5.76 ng/dl) and the intra-assay and inter-assay coefficients of variation were 6.4% and 8.7%, respectively.

Using expected differences (3.2 kcal/kg body weight) and standard deviations (8.0) from De Souza et al. [2] who demonstrated significant differences in EA in exercising women with subtle menstrual cycle disturbances (ovulatory vs. luteal phase defects), a sample size of 80 participants provides adequate power (1-β = 0.80) to detect significant differences in EA among menstrual groups. All variables were tested for outliers and normality using box plot analyses and Kolmogorov-Smirnova tests of normality, respectively. Extreme outliers represented a value more than 3 times the inter-quartile range (Q3-Q1) from the upper (Q3) or lower (Q1) quartile. Data points identified as extreme outliers were removed from subsequent analyses: BMI (n = 1), LBM (n = 3), VO_2peak (n = 2), EEE (n = 1) and TT₃ concentrations (n = 1). A one-way analysis of variance (ANOVA) was performed to examine differences in EA among subclinical menstrual groups (ExOv, ExIncon, ExAnov, ExOligo, ExAmen). Post hoc testing to reveal where significant differences occurred was performed using t-tests with Bonferroni correction, where P < 0.0125 was considered significant. When differences were compared between clinical menstrual groups i.e., ExEumen vs. ExAmen, an independent t-test was used. Chi-square analyses were performed to examine the frequency of menstrual cycle disturbances at or above, and below an EA of 30 kcal/kg LBM. Data are reported as means ± SEM, and p ≤ 0.05 was considered statistically significant. All data were analyzed using SPSS for Windows (version 18; Chicago, Ill., USA).