Physical Activity and Endogenous Sex Hormone Levels in Postmenopausal Women: a Cross-Sectional Study in the Prospect-EPIC Cohort

A recently published systematic review of 19 cohort studies and 29 case-control studies reported strong evidence for a protective effect of physical activity on postmenopausal breast cancer risk (1). For premenopausal breast cancer, however, evidence for an association with physical activity was weaker (1). Another recent review, examining the impact of type and dose of activity on breast cancer risk (2), found stronger protective effects for recreational activity compared with occupational and household activities. Also for vigorous activities somewhat stronger protective effects were observed.

Several possible biological mechanisms have been proposed through which physical activity may reduce breast cancer risk, i.e. endogenous sex hormones (estrogens and androgens), growth factors, energy balance, weight, and immune function (3-5).

The evidence that estrogens contribute to breast cancer risk is strong and widely accepted. A reanalysis of nine prospective studies showed that a relatively high level of endogenous estrogens in postmenopausal women is associated with a 2- to 3-fold increased risk of breast cancer (6). Postmenopausal women with elevated levels of androgens or dehydroepiandrosterone (DHEA) also have increased risk of developing breast cancer, even after adjustment for estrogens (6, 7).

Physical activity may lower postmenopausal sex hormone levels (both estrogens and androgens; refs. 8-13). This might be largely explained by preventing weight gain because after menopause the principal source of estrogen production, peripheral aromatization of androstenedione, is fat tissue (14, 15). Prevention of weight gain, however, might not be the only pathway as some studies still observe an effect of physical activity on hormone levels after adjustment for body mass index (BMI; refs. 8, 11-13). Physical activity might also influence postmenopausal sex hormones by increasing levels of sex hormone binding globulin (SHBG), resulting in lower amounts of unbound (free), active estrogens and androgens in circulation (8, 11, 16).

DHEA is a natural steroid produced from cholesterol in the adrenal cortex. DHEA and its sulphate form (DHEAS) can be understood as prohormones for the sex hormones as they are converted to androgens and estrogens in peripheral tissues (17, 18). DHEAS may be viewed as a buffer and reservoir for these hormones, which is reflected in the relatively high levels of DHEAS in blood (18). Physical activity might also influence sex hormone levels by lowering the levels of this prohormone.

The objective of the current study was to investigate the association between usual physical activity and circulating levels of estrogens, androgens, DHEAS, and SHBG in postmenopausal women. Furthermore, the separate effect of cycling, sporting, and occupational physical activity on these hormones was investigated.

This study included women participating in Prospect-EPIC, one of the two Dutch cohorts participating in the European Prospective Investigation into Cancer and Nutrition (EPIC). EPIC is a multicenter cohort study with 10 participating European countries. The Prospect-EPIC study consists of 17,357 women, ages 49 to 69 y. The rationale and design of both EPIC and Prospect-EPIC have been described in detail elsewhere (19, 20). In brief, women residing in Utrecht or its vicinity were recruited from 1993 to 1997 through a regional breast cancer screening program. The regional program is part of the national Dutch cancer screening program.

At time of enrolment, participants were asked to fill in two questionnaires. A general questionnaire was used to gather information on demographic, reproductive, and lifestyle factors, and past and current morbidity. To determine regular dietary intake in the year prior to enrollment, an extensive self-administered food frequency questionnaire was used containing 178 food items, which was validated and described in more detail by Ocké et al. (21, 22). In addition, anthropometric measurements were taken, and a 30 mL nonfasting blood sample was donated between 8 a.m. and 4 p.m. Within 24 h, samples of 4 mL serum, 9 mL citrate plasma, 2 mL WBC, and 2 mL RBC were fractionated into 0.5 mL aliquots, and stored at −196°C.

Among all Prospect-EPIC participants, a 10% random sample (n = 1,736) was taken for laboratory measurements. Sex hormones were measured in postmenopausal women only. For the present study, we included only those women who were postmenopausal at recruitment, had donated a plasma sample, and did not use hormone replacement therapy or oral contraceptives at the time of the blood donation. Women were considered postmenopausal if they reported (a) a natural menopause (no menstrual periods for at least 12 mo after spontaneous cessation of their menses), (b) a bilateral ovariectomy, (c) a hysterectomy with one or both ovaries retained and were ≥54 y if smokers, or (d) a hysterectomy with one or both ovaries retained and were ≥56 y if nonsmokers. Finally, women were only included if they answered all questions of the physical activity questionnaire. The present study included 806 postmenopausal women.

A self-administered questionnaire was used to obtain information on duration and types of physical activity during the year preceding study recruitment. The validated Cambridge Physical Activity Index described by Wareham et al. (23) was calculated by combining occupational physical activity with time spent on cycling and sporting in summer and winter. Finally, women were divided into four physical activity categories (inactive, moderately inactive, moderately active, and active; Table 1).

Levels of estrone, estradiol, androstenedione, DHEAS, testosterone, and SHBG were measured in plasma using commercially available double-antibody RIA kits (Diagnostic System Laboratories Inc.). The following kits were used: DSL-8700 for estrone, DSL-39100 for estradiol, DSL-4200 for androstenedione, DSL-2700 for DHEAS, DSL-4100 for testosterone, and DSL-6300 for SHBG. The kits that we used for measurements of estradiol, estrone, androstenedione, and testosterone levels were validated by Rinaldi et al. (24), and they showed a high relative validity in terms of ranking postmenopausal women by relative hormone levels.

The intra-assay coefficients of variation were 5.6%, 3.9%, 4.3%, 5.2%, 7.7%, and 3.0% for estrone, estradiol, androstenedione, DHEAS, testosterone, and SHBG, respectively. The interassay coefficients of variation were 11.1%, 4.1%, 6.3%, 5.3%, 8.1%, and 4.0%, respectively. Free estradiol and free testosterone were calculated using the measured values for estradiol or testosterone, SHBG, and an assumed constant for albumin (25, 26). Although technically SHBG is not a hormone, for reasons of convenience it will be referred to as such.

Means and SD (for normally distributed variables), medians and ranges (for the not-normally distributed characteristics) or frequencies (for categorical variables) of baseline characteristics were presented. Concentrations of hormones that were not normally distributed were logarithmically transformed for the analysis. For each physical activity category (inactive, moderately inactive, moderately active, and active) the geometric mean and the 95% confidence interval was calculated for each hormone. Next, general linear models were used to examine the association between usual physical activity and sex hormone levels, adjusted for potential confounders. Potential confounders were age at enrollment (y), smoking (pack-years), alcohol intake (g/d), energy intake (Kcal/d), age at menarche (y), age at menopause (y), number of full-term pregnancies, age at first full-term pregnancy (y), education (primary, technical/professional, secondary school, university), past oral contraceptive use (ever, never), and past hormone replacement therapy use (ever, never). The two variables “number of full-term pregnancies” and “age at first full-term pregnancy” were combined into one new variable with the categories: nulliparous, age <23 y at first full-term pregnancy, age 23-26 y at first full-term pregnancy, and age >26 y at first full-term pregnancy. The potential confounders that changed the association between physical activity and sex hormones compared with the univariate analysis were included in the multivariate models.

BMI, as a marker of body fat, is thought to be an intermediate in the association between usual physical activity and circulating sex hormone levels. Therefore, we did not include BMI in the primary models. To further investigate whether physical activity had any effect on hormone levels independent of changes in BMI, we repeated all analyses including BMI in the models. All statistical analyses were done using SPSS 14.0 software.

Table 2 shows characteristics of the included women stratified by the level of physical activity. This table shows that smoking (7.2 pack-years versus 1.2 pack-years), young age (≤11 years) at menarche (15.4% versus 9.8%), nulliparity (18.5% versus 10.8%), >31 years of age at first pregnancy (24.6% versus 16.3%), surgical menopause (32.3% versus 25.1%), and primary education as highest education (43.1% versus 28.8%) were more prevalent in inactive women than in active women. Active women drank more alcohol (4.4 g/day versus 0.37 g/day), used oral contraceptives more often (49.2% versus 38.5%), had a shorter time since menopause (11.6 years versus 15.2 years), and more often had a university degree (15.2% versus 6.1%).

Table 3 shows the geometric mean (with corresponding 95% confidence intervals) of the hormone levels for all women. In Table 4,, the unadjusted and adjusted geometric mean hormone levels and corresponding 95% confidence intervals for each hormone according to the level of usual physical activity are shown. We found an inverse association between usual physical activity and estradiol levels (free: inactive, 0.26 pg/mL; active, 0.23 pg/mL; P-trend = 0.045; total: inactive, 8.8 pg/mL; active, 8.0 pg/mL; P-trend = 0.08). A positive association was found between usual physical activity and SHBG (active, 15.1 nmol/L; inactive, 19.3 nmol/L; P-trend = 0.05) and between usual physical activity and DHEAS (inactive, 352.4 ng/mL; active, 460.3 ng/mL; P-trend = 0.01). After adjustment for BMI, the association between physical activity and (free) estradiol attenuated as well as the association between physical activity and SHBG. The positive association with DHEAS, however, remained statistically significant.

Table 5 shows the geometric mean hormone levels and 95% confidence intervals for each hormone according to the separate types of physical activity (cycling, sporting, and occupational) that together compose the Cambridge Physical Activity Index, adjusted for confounders excluding BMI. Inverse associations were observed between sporting and estradiol levels [free: 0 metabolic equivalent (MET)-hours/week, 0.26 pg/mL; >12 MET-hours/week, 0.22 pg/mL; P-trend = 0.03; total: 0 MET-hours/week, 8.7 pg/mL; >12 MET-hours/week, 7.7 pg/mL; P-trend = 0.09]. A positive association was observed between sporting and SHBG (0 MET-hours/week, 16.3 nmol/L; >12 MET-hours/week, 20.5 nmol/L; P-trend = 0.03). A weak positive association was observed between cycling and androstenedione (<6 MET-hours/week, 0.44 ng/mL; >33 MET-hours/week, 0.49 ng/mL; P = 0.08). A weak inverse association was observed between occupational physical activity and estradiol. No associations were observed between occupational physical activity and any of the other hormones.

When the different types of physical activity, shown in Table 5, were also adjusted for BMI, the associations between sporting and free estradiol (0 MET-hours/week, 0.26 pg/mL; >12 MET-hours/week, 0.23 pg/mL; P-trend = 0.14) and between sporting and SHBG (0 MET-hours/week, 16.4 nmol/L; >12 MET-hours/week, 19.3 nmol/L; P-trend = 0.11) attenuated. The association between cycling and androstenedione (<6 MET-hours/week, 0.43 ng/mL; >33 MET-hrs/week, 0.50 ng/mL; P = 0.04) was still present (data not shown in table).

In this study, we observed that a high level of usual physical activity was associated with lower estradiol levels and higher SHBG levels. Both hormone levels were influenced most by vigorous physical activity (sporting). Furthermore, our results suggest that BMI plays a role in the mechanism between physical activity and sex hormone levels. We also found that high levels of usual physical activity were related to high levels of DHEAS.

Most of the earlier published studies also reported an inverse association with estrogens (6, 8, 11-13) and a positive association with SHBG (8, 11). No association between physical activity and these hormones were found by the studies of Bjornerem et al. (27) and Madigan et al. (9). The latter study, however, did find a significant inverse association between nonrecreational activity and estrone (9). Verkasalo et al. found somewhat elevated estradiol concentrations in women with the highest level of vigorous exercise and did not find any association with SHBG (28).

Few studies have investigated the relationship between physical activity and androgens (testosterone and androstenedione) in postmenopausal women (8-10, 13), and all have reported an inverse association. In our study, however, we did not observe any association between usual physical activity and androstenedione/testosterone.

The positive association between physical activity and DHEAS in our study did not support the hypothesis that physical activity might decrease breast cancer risk by lowering levels of this prohormone. Only three other studies investigated the effect of physical activity on circulating levels of DHEAS in postmenopausal women, with mixed results (10, 13, 29). In a 12-month exercise trial among 170 postmenopausal women, McTiernan et al. (11) did not observe any effect of the exercise intervention on DHEAS. In women who lost >2% body fat, however, exercisers had greater (nonsignificant) 12-month changes in DHEAS compared with controls (−21.8% versus +3.3%, respectively), which is in contrast to our findings. In the recent Penn Ovarian Aging Study (13), no association between physical activity and DHEAS in any menopausal transition stage was found. The study of Straub et al. (29), reported a positive association between physical activity and plasma levels of DHEAS in women with peripheral obesity. The results of this study should be interpreted with caution because these are not adjusted for potential confounders. Thus, the effect of physical activity on DHEAS levels is unclear and further data are required to unravel the role of DHEAS in the relation between physical activity and breast cancer risk.

Differences among studies on the association between sex hormone levels and physical activity might be explained by the fact that there is no general definition for low or high level of physical activity; studies have used different cutoff points, which complicated comparisons. For example, if we used the same cutoff points as in the EPIC-Norfolk study (8), almost all women in our study (96%) would have been categorized into the highest physical activity groups. Furthermore, the age of participating postmenopausal women could also explain differences among studies. In younger postmenopausal women, sex hormone production is in a transition phase from ovary production to fat tissue production and in older postmenopausal women production is only present in fat tissue. Therefore, relations between physical activity and hormone levels may be better detectable in older than in younger postmenopausal women, when assuming that the effect largely acts through prevention of weight gain. The mean age of postmenopausal women in our study was lower than in most other studies (8, 9, 27). We did an age-stratified analysis (≤60 years; >60 years). We did not find any association between usual physical activity and sex hormones in women age <60 years, whereas in older postmenopausal women the inverse association was significant (e.g. estradiol: inactive, 8.7 pg/mL; active, 7.2 pg/mL; P-trend = 0.04).

Little is known about duration and intensity of physical activity in relation to sex hormone levels. In our study, we investigated three types of physical activity, namely, cycling, sporting, and occupational, which together make up the Cambridge Physical Activity Index. For occupational physical activity, we only observed a weak association with estradiol levels. The association was probably weak because only few women were employed (n = 335), mainly in sitting or standing jobs. In the Netherlands, cycling is integrated in daily life and often not considered as a type of sport. Therefore, cycling is difficult to measure and prone to misclassification, which might explain the lack of effect in this study. In our study, sporting was responsible for the main effect of physical activity on sex hormone levels. Although recall of sporting is less prone to misclassification, it might also be the case that vigorous physical activity, such as sporting, is needed to induce an effect on sex hormone levels. Our results on type and intensity of physical activity are in accordance with the results of a recent review on physical activity and breast cancer risk (2).

Some strengths and limitations of the present study should be considered. Physical activity was assessed by self-administered questionnaires instead of, for example, a structured interview or accelerometers. Furthermore, physical activity was only measured at one point in time, which may not be the etiologically most relevant period. Another limitation of the physical activity data is the lack of information on the duration and frequency of occupational activity. These aspects of our physical activity assessment might have caused random misclassification of the level of physical activity and subsequently dilution of the effect. Despite these limitations, Wareham et al.(23) showed a satisfactory repeatability and validity for the Cambridge Physical Activity Index derived from the same physical activity questions and concluded that it is useful for ranking participants in terms of their physical activity in large epidemiologic studies. Furthermore, the Cambridge Physical Activity Index is clearly associated with a reduced risk of mortality and cardiovascular disease (30).

The strengths of our study are the large (population-based) sample size, which enables us to analyze the separate effects of different types of physical activity on sex hormone levels. A further strength is the availability of extensive information on potential confounders.

In conclusion, our results are in accordance with a protective effect of physical activity on breast cancer via a decrease in estradiol levels and an increase in SHBG levels. The effect was stronger in older postmenopausal women. Vigorous physical activity (sporting) was most strongly related to sex hormone levels. Finally, our results suggest that BMI could play a role in the mechanism between physical activity and sex hormone levels.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.