Effects of Aerobic Exercise Training on Estrogen Metabolism in Premenopausal Women: A Randomized Controlled Trial Free

Despite the convincing epidemiologic evidence identifying physical inactivity as a breast cancer risk factor (1), the underlying biological mechanisms mediating the association are not well understood. Cumulative lifetime exposure to sex steroids, particularly estradiol, is thought to play an important role in breast cancer risk (2, 3), and physical activity has been suggested to alter levels of sex steroid hormones in both premenopausal (4, 5) and postmenopausal women (6). Estrogen metabolites, 2-hydroxyestrone (2-OHE1) and 16α-hydroxyestrone (16α-OHE1), have been identified as biomarkers of interest in the research aimed at understanding the mechanisms by which physical activity exerts its protective effects against breast cancer.

The first step in estrogen metabolism is the transformation of estradiol to estrone by oxidation. Estrone is further metabolized to produce two main metabolites, 2-OHE1 and 16α-OHE1, by hydroxylation via competitive pathways, so an increase in one metabolite occurs at the expense of the other (7, 8). Other metabolites have been identified, such as 4-hydroxyestrone and 2-methoxy-estradiol, but their concentrations are small by comparison and their actions are under investigation (9). 16α-OHE1 is estrogenic, whereas 2-OHE1 is nonestrogenic (10). Higher 2-OHE1 levels are suggested to have a protective effect, along with a higher 2-OHE1 to 16α-OHE1 ratio, whereas higher 16α-OHE1 levels or a lower 2-OHE1 to 16α-OHE1 ratio are associated with an increased breast cancer risk (9, 11). In addition to the proposed enhanced cellular proliferation due to stimulation of intracellular estrogen receptors, additional mechanisms for cancer development have also been proposed, such as the production of quinone derivatives of metabolites, which may cause oxidative damage to DNA (9).

An association between these estrogen metabolites and breast cancer incidence has been supported by some (8, 12-16) but not all (17, 18) case-control studies, and the evidence from prospective cohort studies suggests a nonsignificant reduced risk of breast cancer in women with higher levels of 2-OHE1 to 16α-OHE1 ratio, especially for premenopausal women (11, 19, 20). Physical activity has been suggested to alter estrogen metabolism. Until recently, this idea was based on three small studies, which indicated higher 2-OHE1 levels in more active young women, especially those who develop menstrual dysfunction associated with exercise (21-23). Two recent cross-sectional studies suggest that higher self-reported physical activity is associated with an increase in levels of 2-OHE1 and 2-OHE1 to 16α-OHE1 ratio; however, this association may have been influenced by body composition (24, 25). To date, only one study has examined the effects of an aerobic exercise training intervention on estrogen metabolism using randomized controlled trial methodology, but it was confined to postmenopausal women (26). The purpose of the present study was to determine the effects of a 12-week aerobic exercise training program on estrogen metabolites in previously sedentary or recreationally active premenopausal women. It was hypothesized that aerobic exercise training would cause an increase in 2-OHE1 and the 2-OHE1 to 16α-OHE1 ratio.

Participants were recruited from the University of Alberta (Edmonton, Alberta, Canada) and surrounding community. This study was approved by the Research Ethics Board of the University of Alberta, and all participants provided written informed consent before participation. The eligibility criteria were the following: (a) female; (b) Caucasian; (c) sedentary or recreationally active (engaging in ≤20 min of vigorous intensity exercise three or more times per week in the past 6 months with no history of significant aerobic training in the past year) and “average” aerobic fitness, determined as maximal oxygen consumption (VO_2max) <40 mL/kg/min; (d) 20 to 35 years of age; (e) self-reported regular menstrual cycles (cycle 24-36 days long and at least 10 cycles in the previous 12 months); (f) normal or overweight body mass index (BMI) of 18 to 29.9; (g) no use of pharmacologic contraceptives (past 6 months); (h) no use of tobacco products (past 12 months); (i) not vegetarian; (j) no self-reported endocrine condition (thyroid or liver disease or diabetes); (k) no use of medication that might interfere with hormonal status (i.e., antidepressants or antibiotics); and (l) free of musculoskeletal conditions that would prevent participation in an aerobic exercise program.

Demographic information, medical history, and reproductive history were collected via self-administered questionnaire at baseline.

At baseline, participants' body mass and height were measured in light clothing without shoes to the nearest 0.1 kg and 0.5 cm, respectively. Waist and hip circumference were measured to the nearest 0.1 cm with an inelastic tape at the narrowest part of the torso and the maximal part of the buttocks, respectively. These measures were also completed at 6 to 7 weeks (midpoint) and at 12 weeks (end of study) from randomization. Body composition was assessed by dual-energy X-ray absorptiometry to determine percentage body fat, fat mass in kilogram, and lean body mass in kilogram at baseline and end of study.

Aerobic fitness was determined with a VO_2max test using an incremental graded exercise on a stationary bike (Monark, Varberg, Sweden). Participants were asked to cycle at a self-selected constant cadence (60-80 rpm). Resistance was increased by 30 W every 2 min until the participant achieved a respiratory exchange ratio of 1.0 and then increased by 30 W every minute until (a) volitional fatigue or (b) a drop in cadence with an increase in resistance (>10 rpm for 30 s). Ventilatory gas exchange was measured throughout the test using indirect calorimetry (True One, Parvo Medics, Sandy, UT). Heart rate was measured continuously using a heart rate monitor (Polar USA, Woodbury, NY). All maximal aerobic fitness tests were done by the same tester using a standardized protocol. Criteria for reaching VO_2max were a plateau in oxygen consumption (<100 mL/min) during exercise with increasing power output and/or respiratory exchange ratio of ≥1.1. “Average” aerobic fitness level based on population values for women ages 20 to 40 years was set at VO_2max ≤40 mL/kg/min (27). Those who scored above this value were excluded from the study. All participants completed another exercise test at the end of the study, whereas those randomized to the exercise intervention also completed an additional test at midpoint to optimize the exercise prescription.

Usual dietary intake was assessed using a 3-day diet record in the follicular phase (days 1-5) of the first and fourth menstrual cycle. Data were entered into a database modified for use in a Canadian population (Food Processor II Nutrient Analysis Program, ESHA Research, Salem, OR). Energy and macronutrient (e.g., protein, fat, and carbohydrate) intake along with selected dietary factors that have been associated with estrogen metabolism (i.e., Brassica vegetables, soy, and fiber intake) were determined.

Usual physical activity was assessed using the Godin Leisure Time Exercise Questionnaire (28). At the time of screening, participants were asked about “strenuous,” “moderate,” and “mild” physical activity in the past 7 days. This was repeated in the last week of the intervention.

Following completion of baseline measurements and one baseline menstrual cycle, participants were randomly assigned to either the “exercise” (12-week aerobic activity program) or “control” (usual lifestyle) group using a computer-generated random numbers list (StatMate, version 1.01, 1998) in a 1:1 ratio. A permuted block design was used to generate the allocation sequence. Randomization was stratified on BMI (<25 or ≥25).

Participants randomized to the intervention arm began the exercise program in the early follicular phase of the next menstrual cycle (days 1-5). The aerobic exercise training intervention was a 12-week individualized, progressive, moderate-to-vigorous intensity, supervised, aerobic training program aimed at improving aerobic fitness, measured as improvement in VO_2max. Intensity of the exercise was individualized for each participant and determined from the power output (watts) at ventilatory equivalents for oxygen (V_E/VO₂) and carbon dioxide (V_E/VCO₂) at the baseline and midpoint aerobic fitness tests (29). This method of exercise prescription uses metabolic variables as the basis for training intensity to minimize training at different relative intensities between participants. All sessions included an additional warm up (5-10 min) and cool down (5 min) on the equipment used. Participants were also encouraged to do general static stretching following a training bout.

Specifically, the intervention was divided into three blocks. (a) Weeks 1 to 4. Participants did three sessions per week of base aerobic training progressing from 20 to 40 min on a stationary bike (Lifestyle Fitness, 9500HR, Life Fitness, Franklin Park, IL). Intensity was based on wattage corresponding to ∼25% higher than the power output at the V_E/VO₂, determined during the graded exercise test. (b) Weeks 5 to 8. Participants did four sessions per week. Two sessions were base aerobic training sessions for 30 to 45 min as described previously. Two additional interval sessions were completed: interval 1, two 10-min intervals at a power output corresponding to the power output at V_E/VCO₂, with 10 min of easy cycling between the intervals, and interval 2, intervals at a power output equivalent to that which elicited VO_2max during the graded exercise test, for 30 s followed by 30 s of easy pedaling, building from two sets of 10 intervals to one set of 20 intervals. (c) Weeks 9 to 12. Participants did four sessions per week, with two base aerobic training sessions for 30 to 45 min and two interval sessions similar to weeks 5 to 8. Exercise prescription was updated based on the midpoint VO_2max test. Interval 2 was changed to 2 min at a power output corresponding to VO_2max and 3 min of easy pedaling, repeated four times a session.

Participants in the control group were asked to maintain their usual activity levels for the duration of the study. Following the control cycle, the first day of the next menstrual cycle was used as the reference start date for participants in the control group. On completion of the 12-week postintervention measurement, participants were given guidance for starting an individualized exercise program and access to the fitness facility for 4 weeks.

For analysis of estrogen metabolites, 2-OHE1 and 16α-OHE1, first morning urine samples were collected between 06:30 a.m. and 11:00 a.m. following a 10-h water-only fast between day 20 and day 22 of the menstrual cycle (i.e., luteal phase) over four consecutive menstrual cycles (one baseline cycle and three intervention cycles). Participants were instructed not to engage in physical activity (beyond activities of daily living) for 24 h before urine sampling. Urine collection was completed at the participants' home using sterile containers and brought within 2 h to the laboratory at the University of Alberta, where it was stored at 4°C to 8°C and processed within 4 h of being received. To prevent the oxidation of metabolites, ascorbic acid (1 mg/mL) was added to urine before being aliquoted and stored at −70°C.

First morning fasted saliva samples were collected during the midluteal phase of each of the baseline and fourth menstrual cycles to allow for determination of average midluteal progesterone levels. Participants were instructed to passively drool into a provided tube on days 19, 21, and 22 of the menstrual cycle. The days of collection were determined by the length of the previous menstrual cycle. All samples were stored in a home freezer before transport on ice to the University of Alberta, where all samples were stored at −70°C.

2-OHE1 and 16α-OHE1 were measured using ELISA kits (Estramet, Immuna Care Corp., Bethlehem, PA; ref. 30). Because most urinary estrogen metabolites are found in the glucuronide conjugate form, removal of the sugar moiety is required to allow for recognition by the monoclonal antibodies. Samples were incubated for 2 h in deconjugating enzyme and then neutralized. Assay incubation time was 3 h at room temperature. The assay was kinetically read using a Molecular Devices Spectra Thermo Max microplate reader (Molecular Devices, Sunnyvale, CA), and the data were analyzed using SoftMax Application software (version 2.35; Molecular Devices). Validity and reproducibility of the ELISA kits have been previously shown by comparison with gas chromatography-mass spectrometry (30). All samples, standards, and controls were assayed in triplicate. Samples were initially assayed at a 1:4 dilution because of high estradiol concentration in premenopausal women. Samples were reassayed if the coefficient of variation was >10%, but reassay of additional samples for within-person batching was not included. Reassay was required for 19 samples. Samples that were too concentrated or dilute, using standard curve for reference, were reassayed at 1:8 or 1:2 dilution, respectively. The 2-OHE1 and 16α-OHE1 urinary levels were standardized to total urinary creatinine (ng/mg Cr). Creatinine was measured by colorimetric microplate assay (Oxford Biomedical Research, Oxford, MA). Intraassay coefficient of variations for 2-OHE1, 16α-OHE1, and creatinine was 4.3%, 4.9%, and 3.7%, respectively. Interassay coefficient of variation for the kit controls was 5.8% for 2-OHE1 and 4.9% for 16α-OHE1. The limit of detection was 0.15 ng/mL for 2-OHE1 and 0.05 ng/mL for 16α-OHE1. Total estrogen metabolite concentration is the combined amounts of 2-OHE1 and 16α-OHE1.

Salivary progesterone was measured using a competitive enzyme immunoassay (Salimetrics, State College, PA). All progesterone samples were analyzed in duplicate. The mean of the duplicate measurements was assigned as the sample value. The intraassay and interassay coefficient of variation for kit controls was 7.8% and 6.2%, respectively. Ovulatory status of participants was confirmed by self-report data of cycle length and midluteal salivary progesterone (i.e., average of all luteal phase samples between 0.11 and 0.2 ng/mL; ref. 31). The limit for detection was 0.005 ng/mL.

A clinically important difference in estrogen metabolites has not been identified and there is little reliable literature that outlines the expected achievable changes in estrogen metabolism with exercise training. Based on an increase in 2-OHE1 to 16α-OHE1 ratio in a dietary intervention trial in premenopausal women (32) and taking into account a possible 10% dropout rate, it was estimated that 16 participants per group were needed to detect a 16% difference in 2-OHE1 to 16α-OHE1 ratio using an independent t test with a power of 0.80 and a two-tailed α of 0.05.

Data were analyzed with Statistical Package for the Social Sciences version 12.0 software (SPSS, Inc., Evanston, IL). Distributions were examined for skewness and outliers. The assumption of normality was not met for one 2-OHE1 value (cycle 1, skewness = 2.6) and three measures of 16α-OHE1 (cycles 1, 2, and 4, skewness = 3.5, 2.4, and 2.0, respectively). Therefore, estrogen metabolite data were log transformed for analyses and presented as geometric means. Planned comparisons between the two groups were analyzed for changes in the primary time points of interest (cycles 1-4) using independent t tests and secondary exploratory analysis for changes over all time points (all four cycles) using repeated measures ANOVA. Associations between aerobic fitness/body composition and estrogen metabolites were analyzed by Pearson correlations at baseline and for change scores. Mean value and SE or 95% confidence intervals (95% CI) are reported for continuous variables, and number and percentage are reported for categorical variables. Last-observation-carried-forward analysis was used for missing data.

Flow of the participants through the trial is presented in Fig. 1. Overall, 17 women were randomized to the exercise intervention and 15 women were randomized to the control group. All women returned to complete the end of study testing.

Table 1 presents the baseline participant characteristics. The groups were balanced on all demographic, anthropometric, aerobic fitness, and reproductive measures. No participant had a first-degree relative with breast cancer.

Participants attended an average of 40 of 44 (91%) prescribed exercise sessions, ranging from 64% to 100%. Fourteen of 17 (82%) participants completed at least 80% of the prescribed exercise sessions. The control group reported no change in the amount of physical activity from baseline.

Three participants in the exercise group experienced an adverse event compared with one adverse event in the control group. The adverse events in the exercise group were influenza (n = 1), broken wrist (n = 1), and car accident (n = 1); none of which were deemed to be associated with the intervention. The adverse event in the control group was mononucleosis (n = 1).

The effect of the aerobic exercise training intervention on aerobic fitness and anthropometrics is shown in Table 2. The exercise group increased VO_2max by 4.6 mL/kg/min or 0.27 L/min (14% increase), whereas the control group decreased by 1.0 mL/kg/min or 0.06 L/min [3% decrease; mean group change = 5.6 mL/kg/min (3.7-7.2); P < 0.001].

The aerobic exercise training intervention had no significant effect on body weight, BMI, waist and hip circumference, or waist-to-hip ratio (WHR). Compared with the control group, however, the exercise intervention group lost fat mass [mean group difference = −1.2 kg (−2.2 to −0.2); P = 0.018] and gained lean mass [mean group difference = 0.9 kg (0.2-1.6); P = 0.009]. No significant change in energy intake or macronutrients was seen in either group across the course of the intervention, and there was no difference between groups for the number of participants who reported Brassica vegetable or soy intake at baseline and postintervention (data not shown).

At baseline, no difference in 2-OHE1, 16α-OHE1, 2-OHE1 to 16α-OHE1 ratio, or total estrogen metabolite concentration was seen between the exercise and control group (Table 3). Baseline 2-OHE1 to 16α-OHE1 ratio was associated with percentage body fat (r = −0.40; P = 0.044), and an association with WHR approached significance (r = −0.34; P = 0.055; Table 4). There was no association between baseline body composition, physical activity, aerobic fitness, and 2-OHE1. A positive association between overall body fat and 16α-OHE1 approached significance (r = 0.33; P = 0.067 for percentage body fat). At baseline, one participant had high values of both 2-OHE1 and 16α-OHE1 (137.9 and 77.1 ng/mL/mg Cr, respectively); however, excluding her from the data did not substantially change the results.

The aerobic exercise training intervention had no effect on levels of 2-OHE1 and 16α-OHE1 or 2-OHE1 to 16α-OHE1 ratio (Table 3). No difference in 2-OHE1, 16α-OHE1, or 2-OHE1 to 16α-OHE1 ratio was seen between groups over the course of the intervention shown by a nonsignificant group by time interaction for 2-OHE1 (F = 0.11; P = 0.741), 16α-OHE1 (F = 0.64; P = 0.430), 2-OHE1 to 16α-OHE1 ratio (F = 0.34; P = 0.567), and total estrogen metabolite concentration (F = 0.33; P = 0.804).

Improved VO_2max was not associated with changes in estrogen metabolites levels (Table 5). However, a positive association between increased total lean body mass and 2-OHE1 to 16α-OHE1 ratio was found (r = 0.43; P = 0.015).

The majority of participants (i.e., 27 of 32) maintained consistent menstrual cycle length within the reference range (24-36 days; ref. 33) and had midluteal progesterone values between 0.11 and 0.2 ng/mL (i.e., 29 of 32), which are associated with a positive ovulatory status (data not shown; ref. 31).

To date, there are no published randomized controlled trials examining the effects of an aerobic exercise training intervention versus usual lifestyle control group on the urinary excretion of estrogen metabolites 2-OHE1 and 16α-OHE1 or their ratio, 2-OHE1 to 16α-OHE1, in premenopausal women. Our aerobic exercise training program resulted in significant improvements in aerobic fitness, body fat mass, and lean body mass, but no significant effects on 2-OHE1, 16α-OHE1, or 2-OHE1 to 16α-OHE1 ratio were observed. We did note, however, that an increase in lean body mass was associated with a favorable change in 2-OHE1 to 16α-OHE1 ratio. Overall, these data do not support the hypothesis that an aerobic exercise training program favorably alters the 2-OHE1 to 16α-OHE1 ratio in a manner that could explain the reduced breast cancer risk observed with physical activity.

The primary finding of the present study is that 12 weeks of aerobic exercise training, sufficient to induce a meaningful improvement in VO_2max and modest body composition changes, had no significant effects on selected markers of estrogen metabolism. This finding is consistent with two previous studies that examined the effects of a physical activity intervention on levels of 2-OHE1, 16α-OHE1, or their ratio, 2-OHE1 to 16α-OHE1 (26, 34), but differs from several observational studies that suggest that physical activity is associated with estrogen metabolite levels (21-24). Atkinson et al. (26) found no difference in estrogen metabolite levels following a randomized controlled trial of a 12-month moderate intensity (60-70% of maximal heart rate for 45 min) aerobic intervention in 170 postmenopausal women, in which the exercise group had a 12% improvement in VO_2max (35). Pasagian-Macaulay et al. (34) also found no difference in 2-OHE1 to 16α-OHE1 ratio between the intervention and control groups during a 20-week group-based weight loss lifestyle intervention in older premenopausal women (ages 44-55 years), in which the intervention group lost body mass, reduced dietary fat intake, and increased self-reported physical activity level by 400 kcal/wk (mainly through walking). Together, the results from the present study and the previous two intervention studies (26, 34) suggest that exercise and physical activity interventions may not have any significant effect on estrogen metabolism in women.

It is possible that the associations reported in observational studies may be spurious. In the three early studies (21-23), sample sizes were small (i.e., five to seven participants per group), menstrual function was not standardized (resulting in comparisons between individuals with and without menstrual dysfunction and women with different reproductive ages), self-reported physical activity measures were used (i.e., swim distance, running mileage, and participation in a varsity-rowing program), and the studies used an older version of RIA (21, 22) or administration of a labeled tracer (23) to measure estrogen metabolites, which are not as valid as the newer solid-phase enzyme immunoassay (30, 36). Methodologic improvements are evident in three more recent studies (24, 25, 37), namely the use of larger sample sizes, standardized menstrual status, and analysis of estrogen metabolites using a newer solid-phase enzyme immunoassay. However, all have used BMI as a measure of adiposity, which is problematic, especially for more athletic populations. Two studies (24, 25) used self-reports of physical activity levels, which suffer from several methodologic issues, particularly overreporting of frequency, duration, and intensity (38). One study (37) used an objective measure of chronic exercise, VO_2max.

An association between estrogen metabolites and baseline body composition was observed in our study, which is consistent with some (22, 23, 25, 39-41) but not all previous studies (42). At baseline, an inverse association between 2-OHE1 to 16α-OHE1 ratio and percentage body fat was noted. This finding is consistent with the lower 2-OHE1 levels reported in obese individuals compared with those of normal body mass (41) and in women with a higher BMI and higher body fat regardless of aerobic fitness level (37). In addition, higher 2-hydroxylation was observed in athletes who were leaner (22, 23) and in anorexic women compared with normal body mass and obese women (40). However, Jernstrom et al. (42) found no association between 2-OHE1, 16α-OHE1, or 2-OHE1 to 16α-OHE1 ratio and height, body mass, or BMI in a cross-sectional study of premenopausal women.

An interaction between body composition and physical activity in relation to estrogen metabolites has also been suggested. Matthews et al. (25) found that, in women who reported low levels of physical activity, those with a higher BMI (>25.0) had lower 2-OHE1 to 16α-OHE1 ratio than those with a lower BMI (<25.0). However, women with a higher BMI (>25.0) who reported being physically active maintained a higher 2-OHE1 to 16α-OHE1 ratio, consistent with women with a lower BMI (<25.0) who also reported being physically active. A similar finding by Bentz et al. (24) showed a positive association between MET-hours per day of physical activity, 2-OHE1 levels, and 2-OHE1 to 16α-OHE1 ratio, independent of BMI, suggesting that overweight women may also benefit more from physical activity in terms of positive alterations in estrogen metabolism. However, when women were split into overweight (BMI, >25) and normal weight (BMI, <25), the association between MET-hours per day and estrogen metabolites remained only for the overweight group. The authors discuss the issues surrounding the measurement of body composition using BMI and that several of those in the higher BMI group were athletic and likely did not have high levels of adiposity, which may account for their findings (24).

Whereas body weight did not change in either group over the course of the present intervention, the exercise group had a significant increase in lean body mass, with a decrease in total fat mass and a corresponding decrease in percentage body fat. Whereas a change in body fat was not associated with a change in 2-OHE1 to 16α-OHE1 ratio, a positive association with change in total lean mass was noted. An association between body fat and 2-OHE1 to 16α-OHE1 ratio is consistent across much of the literature (22, 23, 25, 39-41); however, an association with increases in lean body mass is more novel. Atkinson et al. (26) also showed a positive association between lean body mass and 2-OHE1 in the exercise group. Lean body mass is a major player in insulin-stimulated glucose uptake and storage and, therefore, overall energy balance through a variety of signaling pathways. Several of these signaling pathways are metabolic hormones that have been proposed to affect cancer risk (1). It is possible that lean body mass may affect estrogen metabolism through an interaction with a variety of metabolic hormones and growth factors; however, such an association is speculative and the mechanisms have not been identified.

It is important to acknowledge the strengths and limitations of this trial. The strengths of this trial include the randomized controlled design, validated measurement of VO_2max, individualized exercise programs based on metabolic measures of exercise intensity, supervised aerobic exercise training intervention, a high adherence rate to the intervention, standardized measurement of urinary estrogen metabolites across four consecutive menstrual cycles, body composition measures by dual-energy X-ray absorptiometry, and no loss to follow up.

The limitations of this trial include difficulty standardizing urine sample collection across the menstrual cycle, less invasive measurement of menstrual dysfunction, failure to control for the possible effect of caffeine on estrogen metabolites (43), variability of estrogen metabolites both within an individual and across individuals, a 12-week exercise intervention that does not mimic a habitual pattern of physical activity, and a small sample size. Due to the variability of the measures of estrogen metabolites, a larger sample may be needed to observe meaningful group effects if rigorous methods surrounding urine sampling and intervention delivery can be maintained. However, in the present study, the direction of changes in 2-OHE1, 16α-OHE1, and 2-OHE1 to 16α-OHE1 ratio is uniformly moving in an unfavorable direction based on the literature on estrogen metabolism and physical activity, similar to the findings in a study of 170 postmenopausal women by Atkinson et al. (26). This suggests that a larger sample size, while perhaps alleviating the effect of the large variability of these measures, may not change the direction of the present findings.

Future investigations should focus on the overall association between body composition and estrogen metabolism and on the effect of changes in body composition on changes in estrogen metabolism. The effect of physical activity could be considered further as part of an intervention to induce negative energy balance and/or change body composition. The findings of the present study also suggest that a resistance exercise intervention to increase lean body mass may alter estrogen metabolism. The use of objective measures of physical activity in large observational studies and interventions at different life stages should also be undertaken to follow up on recently observed associations. In addition, further examination of other estrogen metabolites, such as the more genotoxic 4-hyroxyestrone, may be warranted once analysis becomes more feasible.

In conclusion, a 12-week aerobic exercise intervention in premenopausal women, which resulted in significant improvements in aerobic fitness and body composition, did not significantly change urinary excretion of 2-OHE1, 16α-OHE1, or 2-OHE1 to 16α-OHE1 ratio, proposed biomarkers of breast cancer risk. Baseline body fat was favorably associated with estrogen metabolite levels, and a positive association between an increase in lean body mass and higher 2-OHE1 to 16α-OHE1 ratio was observed.

We thank the participants for their time and Lisa Workman, Susan Goruk, Arne Van Aerde, Margie McNeely, and Dana Wilkinson for their invaluable assistance.