Physical activity and risk of breast cancer overall and by hormone receptor status: The European prospective investigation into cancer and nutrition
Physical activity is associated with reduced risks of invasive breast cancer. However, whether this holds true for breast cancer subtypes defined by the estrogen receptor (ER) and the progesterone receptor (PR) status is controversial. The study included 257,805 women from the multinational EPIC-cohort study with detailed information on occupational, recreational and household physical activity and important cofactors assessed at baseline. During 11.6 years of median follow-up, 8,034 incident invasive breast cancer cases were identified. Data on ER, PR and combined ER/PR expression were available for 6,007 (67.6%), 4,814 (54.2%) and 4,798 (53.9%) cases, respectively. Adjusted hazard ratios (HR) were estimated by proportional hazards models. Breast cancer risk was inversely associated with moderate and high levels of total physical activity (HR = 0.92, 95% confidence interval (CI): 0.86–0.99, HR = 0.87, 95%-CI: 0.79–0.97, respectively; p-trend = 0.002), compared to the lowest quartile. Among women diagnosed with breast cancer after age 50, the largest risk reduction was found with highest activity (HR = 0.86, 95%-CI: 0.77–0.97), whereas for cancers diagnosed before age 50 strongest associations were found for moderate total physical activity (HR = 0.78, 95%-CI: 0.64–0.94). Analyses by hormone receptor status suggested differential associations for total physical activity (p-heterogeneity = 0.04), with a somewhat stronger inverse relationship for ER+/PR+ breast tumors, primarily driven by PR+ tumors (p-heterogeneity < 0.01). Household physical activity was inversely associated with ER–/PR– tumors. The results of this largest prospective study on the protective effects of physical activity indicate that moderate and high physical activity are associated with modest decreased breast cancer risk. Heterogeneities by receptor status indicate hormone-related mechanisms.
Approximately 73 studies have been conducted world-wide on breast cancer risk and physical activity, including about 33 cohort studies.1 On average, a 25% reduction in breast cancer risk has been observed amongst physically active women in comparison to the least active women.1 The World Cancer Research Fund classified the evidence for the inverse association of breast cancer with physical activity in postmenopausal women as probable. In premenopausal women the evidence is more limited, with a suggestion of a somewhat weaker association.2 Overall, despite such intensive focus inconsistencies still remain.
Breast cancer is a heterogeneous disease, with estrogen receptor (ER) and progesterone receptor (PR) status being one of the markers for tumor subtypes.3 Differences have been observed in the etiology, treatment and prognosis of ER and/or PR-positive and -negative breast cancers.4, 5 To date, the relation of physical activity with the risk of hormone-receptor positive and negative breast cancers has been investigated in varying detail in 19 epidemiological studies.6–24 Of the 10 prospective studies,7, 11, 13, 15–19, 21, 24 with a maximum of 4,677 breast cancer cases,21 only 7 prospective studies provided results on ER and PR receptor status combined.7, 11, 13, 17, 19, 21, 24 Study results varied from null associations for any hormone receptor type,17–19 protective effects for hormone receptor positive and negative carcinomas,15, 16 inverse associations for ER− and ER−/PR− but not for ER+/PR+ or ER+/PR−,11, 24 to protective effects only for ER+/PR+,13 only for ER+/PR−,7 or only for ER+.21 Based on this diversity of results, overall risk associations of physical activity with breast cancer subtypes remain unclear.
This analysis from the European Prospective Investigation into Cancer and Nutrition (EPIC) has a cohort of 257,805 women with a total of 8,034 invasive breast cancers diagnosed during a median follow-up interval of 11.6 years. Compared to an earlier publication from this cohort, the database included further 5 years of follow-up, during which over 4,600 additional cases occurred, as well as data on tumor subtypes by receptor status.25 Thus, besides analyses for total breast cancer, a major objective was to contribute with this large prospective study to the discussion whether the inverse association of physical activity with breast cancer risk was valid for hormone receptor defined breast cancer subtypes while incorporating several aspects of physical activity such as type (occupational, household, recreational and total).
Material and Methods
The European Prospective Investigation into Cancer and Nutrition (EPIC) is a multicenter prospective cohort originally established to investigate the associations between dietary, lifestyle, genetic and environmental factors and risk of specific cancers and other chronic diseases. The design and baseline data collection methods have been previously described.26 There were 366,521 women enrolled between 1992 and 2000 in 23 regional or national centers in 10 European countries (Denmark, France, Germany, Greece, Italy, Norway, Spain, Sweden, The Netherlands and United Kingdom). These participants were recruited from the general population from defined areas in each country with some exceptions: women who were members of a health insurance scheme for state school employees in France; women attending breast cancer screening in Utrecht, the Netherlands; blood donors in some Italian and Spanish centers, and a high number of vegans and vegetarians in the Oxford “Health conscious” cohort. Women were mostly aged between 35 and 70 years at enrollment and provided written informed consent at the time they completed the standardized baseline questionnaires on diet, reproductive characteristics, socio-demographic variables, lifestyle and medical history. Study subjects were then invited to a center to provide a blood sample. Approval for this study was obtained from the ethical review boards of the International Agency for Research on Cancer and from local institutions in the participating countries.
The present study is based on data from 345,158 participants after excluding a priori women with prevalent cancer at any site apart from nonmelanoma skin cancers at baseline examination. Further, we excluded the study centers Malmo (Sweden), Granada and Murcia (Spain) (n = 26,091) as for these subcohorts receptor information could not be provided, and we also excluded women with missing nondietary questionnaire data or missing or nonstandardized physical activity questionnaire data, comprising the full Norwegian subcohort (n = 35,890) and the Umeå center (Sweden, n = 12,513). Thus, the analytical cohort for this study consisted of 257,805 women from eight countries.
End points and ascertainment of cases
In most countries (Denmark, Italy, The Netherlands, Spain and the United Kingdom), incident breast cancer cases were identified through a linkage with population-based cancer registries. For these centers, follow-up was completed as follows: December 2004 (Asturias), December 2006 (Florence, Varese, Ragusa, Naples and San Sebastian), December 2007 (Navarra, Oxford, Bilthoven, Aarhus and Copenhagen), January 2008 (Utrecht), June 2008 (Cambridge) and December 2008 (Turin). Active follow-up of study participants and next-of-kin, as well as of social security records, and cancer and pathology registries was used in France, Germany and Greece. For centers using active follow-up, end of follow-up was the last known contact, date of diagnosis, or date of death. Mortality data were coded according to the 10th Revision of the International Statistical Classification of Diseases, Injuries and Causes of Death (ICD-10) and cancer incidence data were coded according to the International Classification of Diseases for Oncology (ICD-O-2). This analysis included 8,034 invasive (primary, malignant) breast cancer cases.
Breast cancer receptor-status information
Information on participants' receptor status, as well as the laboratory methods and quantification descriptions used to determine receptor status was provided by 20 EPIC centers. To standardize the quantification of receptor status received from the EPIC centers, the following criteria for a positive receptor status were used: ≥10% cells stained, any ‘plus-system’ description, ≥20 fmol/mg, an Allred score of ≥3, an immunoreactive score (IRS) ≥2, or an H-score ≥10.27 The estrogen receptor status, the progesterone receptor status and the combined status of the tumor could be assessed for 6,007 (74.8%), 4,814 (59.9%) and 4,798 (59.7%) cases, respectively.
Physical activity data
A description of the physical activity ascertainment, validity and reproducibility used in the EPIC study has been published in detail elsewhere.28–30 Physical activity data were obtained either by face-to-face interviews (Germany, Greece, Spain) or were self-administered (Denmark, France, Italy, the Netherlands and UK) using a standardized questionnaire in all centers included in this analysis. Data on current occupational activity included employment status and the level of physical activity done at work (nonworker, sedentary, standing, manual, heavy manual and unknown). Housewives were categorized as nonworkers. In the Danish centers, the questionnaire focused on type of work activity done within the last year, and participants who did not answer this question were categorized as nonworking.
The number of hours per week of nonoccupational physical activities during the past year were collected in all centers and was comprised household activities, including housework, home repair (do-it-yourself activities), gardening and stair climbing, and recreational activities including walking, cycling and sports combined as done in winter and summer separately. Since the intensity of recreational and household activities was not directly recorded, a metabolic equivalent (MET) value was assigned to each reported activity according to the Compendium of Physical Activities.31 A MET is defined as the ratio of work metabolic rate to a standard metabolic rate of 1.0 kcal (4.184 kJ)/kg/hr, one MET is considered a resting metabolic rate obtained during quiet sitting. The MET values assigned to the nonoccupational data were: 3.0 for walking, 6.0 for cycling, 4.0 for gardening, 6.0 for sports, 4.5 for home repair (do-it-yourself work), 3.0 for housework and 8.0 for stair climbing. These mean MET values were obtained by estimating the average of all comparable activities in the Compendium. The mean number of hours per week of summer and winter household and recreational activities were estimated and then multiplied by the appropriate MET values to obtain MET-hours/week of activity.
Housework (including child and older adult care), home repair, gardening, and stair climbing were combined to obtain an overall estimate of household activity. Walking (including walking to work, shopping and leisure time), cycling (including cycling to work, shopping and leisure time) and sports activities were combined to derive overall recreational physical activity. Household and recreational activities in MET-hours/week were combined and cohort participants classified according to EPIC-wide quartiles of total nonoccupational physical activity.
Two indices of total physical activity, combining occupational and non-occupational activities, have been developed previously. Both indices categorize individuals into inactive, moderately inactive, moderately active and active groups. The “Cambridge index” cross-classifies occupational activity with hours spent doing cycling and sports.30, 32 The second index, the “total physical activity index”, cross-classifies all household and recreational activity combined with occupational activity.29, 33
Associations between physical activity and the risk of breast cancer overall and subtypes were evaluated using Cox proportional hazards models to estimate hazard ratios (HR) and 95% confidence intervals (CI). Age was used as the primary dependent time variable. Women were considered at risk from the time at recruitment until breast cancer diagnosis or censoring (age at death, loss to follow up, end of follow up, or diagnosis of other cancer entities). All models were stratified by age in one-year categories and by study center to control for potential variability between centers and prevent violations of the proportional-hazards assumption.
Continuous variables of physical activity (e.g., MET-hours/week of activity) were categorized into quartiles using cut-points based on the overall cohort. Trend tests for categorical variables were estimated on scores (1–4) applied to the categories of the physical activity variables and entered as a continuous term in the regression models. For continuous physical activity variables, trend tests were based on mean values calculated within the quartiles. To test hazard ratios for overall significance, p-values for Wald χ2 were computed with degrees of freedom equal to the number of categories minus one. The fully adjusted multivariable model was stratified by age and center, and adjusted for the following potential confounders as assessed at baseline: body mass index (BMI, <25, ≥25-<30, ≥30 kg/m2), age at menarche (≤11, 12–14, >14 years, missing), age at first birth (<20, 20–30, >30 years, nulliparous, missing), breastfeeding (yes, no, missing), oral contraceptive use (ever, never, missing), menopausal status (premenopausal, perimenopausal, postmenopausal), age at menopause (<43, 43–46, 47–49, 50–51, 52–53, ≥54, missing), number of full-term pregnancies (0, 1, 2, 3, 4, missing), use of hormone replacement therapy (ever, never, missing), smoking status (never, former, current, unknown), alcohol consumption as grams of ethanol per day (<1.5, 1.5–9, 10–19, 20–30, >30 g/d, missing), education (none/primary school, technical/professional school, secondary school, university, not specified). Missing values (generally <2%) were accounted for by creating an extra category for each covariable. Where applicable, models were mutually adjusted for other types of physical activity (i.e., occupational, household or recreational). All potential confounders were retained in the multivariable models, as the exclusion of single or multiple factors did not result in more precise estimates for the effects of physical activity, thus, there was no advantage in using more parsimonious models.34
Heterogeneity between tumor subgroups defined by ER receptor status only, PR receptor status only, and for the combined classification of ER and PR-receptor status and the associations of the physical activity variables were assessed using the data augmentation method as described by Lunn and McNeil.35 Women who developed the competing breast cancer subtype or were missing receptor status were censored at the time of occurrence.
The proportional hazards assumptions were verified by including interaction terms between exposure and time/age and comparing the interaction model with the model without the interaction terms by means of a likelihood ratio test. In all cases, for the Cox regression model and the stratified Cox Lunn McNeil model, the proportional hazards assumptions were not violated.
Since BMI is closely associated with physical activity, we examined the multivariable models with and without adjustment for BMI as well as stratified by BMI subgroups (<25, ≥25-<30, ≥30 kg/m2). We also examined the association of physical activity and breast cancer risk in subgroups defined by age at diagnosis (≤50, > 50 years of age). As information on menopausal status was only available for the time of recruitment but not for the time of diagnosis, age at diagnosis was used as proxy for the menopausal status at time of breast cancer diagnosis. For the subgroup analysis of women diagnosed before the age 50, we applied right censoring at age 50 for all women. Left censoring for all women was performed for the subgroup analysis for breast cancer cases diagnosed at older ages.
All analyses were performed using SAS Statistical Software, version 9, and all statistical tests were two-sided.
The analytic cohort of 257,805 women was followed for a sum of 2,782,367 person-years (Table 1). The mean age at recruitment into this cohort was 51.3 years (median: 51.8 years). A total of 8,034 invasive breast cancer cases were diagnosed during the average period of 10.8 years (median, 11.6 years) of follow up. Of these cases 936 were diagnosed before or at 50 years of age and 7,098 were older than 50 years at diagnosis.
Table 1. Size of the EPIC cohort for the analyses of physical activity and breast cancer, by country
More physically active women had a lower education and higher BMI level than less active women; however, smoking habits, age at menarche, age at menopause and age at first birth were comparable (Table 2). In comparison to inactive women, active women had more children, had a higher frequency of breastfeeding and used less oral contraceptives and hormone replacement therapy.
Table 2. Demographic and lifestyle characteristics at the time of recruitment among 257,805 study participants by total physical activity categories, the EPIC cohort study
For breast cancer risk overall (Table 3), the multivariable adjusted models showed no association with occupational activity. For household activity alone, as well as for household and recreational activity combined, statistically significant inverse associations with overall breast cancer risk were observed. The multivariable risk estimate for the highest quartile of nonoccupational physical activity (>123 MET-hours/week) was 0.87 (95% CI 0.81–0.94, p-trend < 0.001) when compared to the lowest quartile (≤50.5 MET-hours/week). For the Cambridge index, moderately active and active study participants had HRs of 0.93 (95% CI 0.87–1.00) and 0.94 (95% CI 0.87–1.01), respectively, with a tendency towards a significant trend test. For total physical activity a statistically significant inverse association (p-trend < 0.01) was observed. Physically active study participants had a HR of 0.87 (95% CI 0.79–0.97), and those moderately active had a HR of 0.92 (95% CI 0.86–0.99) as compared with the physically inactive participants. In general, for all types of physical activity that showed significant risk reductions for the highest activity group also moderate physical activity yielded significant risk reductions. Overall, we reported only models that were fully adjusted for BMI because models with and without adjustment yielded almost identical results. In additional analyses, no evidence for effect modification of associations for BMI was found for any of the analyses in this publication (see Supporting Information Tables S1–S6).
Table 3. Physical activity and risk of breast cancer overall and by age at diagnosis, the EPIC cohort study
In analyses by age at diagnosis (Table 3), protective effects were observed for both groups with some indications for variations in shape of dose-response. For women diagnosed with breast cancer after age of 50 years, the strongest inverse association was observed for the highest activity level (HR = 0.88, 95% CI 0.81–0.95 and p-trend = 0.001 for combined recreational and household activity; HR = 0.86, 95% CI 0.77–0.97 and p-trend = 0.01 for total physical activity), compared to the lowest level, indicating a monotone inverse association. For breast cancers diagnosed in younger women (≤50 years) strongest inverse associations were found for moderately active women (HR = 0.73, 95% CI 0.60–0.89 and p-trend = 0.19 for combined recreational and household activity; HR = 0.78, 95% CI 0.64–0.94 and p-trend = 0.07 for total physical activity), indicating more to a U-shaped function. When restricting the group of women aged less than 50 at diagnosis to those who were premenopausal at baseline recruitment (n = 796) similar results were obtained (data not shown).
Of the 4,798 cases with known combined receptor status, 2,943 (61.3%) were ER+/PR+, 875 (18.2%) were ER+/PR−, 808 (16.8%) were ER−/PR− and 172 (3.6%) were ER−/PR+ (Table 4). The latter subgroup was not considered as an outcome category due to small numbers and ER−/PR+ cases were censored at the time of event. With regard to breast tumor subtypes, Cox regression models did not show an inverse association of occupational activity with any of the distinct endpoint analyses (Table 4). The heterogeneity test comparing PR+ with PR- tumors yielded a p-value of 0.033 (detailed results by single receptor outcome are presented in the supplementary table S7). For combined recreational and household physical activity, the strongest association was observed for ER+/PR+ tumors, with the lowest HR of 0.84 (95% CI 0.74-0.96) for the active ER+/PR+ group compared to the inactive ER+/PR+ group (p-trend = 0.02). For the other breast cancer subtypes, all activity levels above inactivity resulted in HRs below unity, with none of them reaching statistical significance. The overall test for heterogeneity between the hormone-receptor status subgroups was non-significant (p-value = 0.73). The same was true for the ER and PR specific testing. A strong inverse association was found for household activity for ER−/PR− tumors, with a hazard ratio of 0.62 (95% CI 0.47–0.81, p-trend = 0.01) for the active group compared with the inactive group. The heterogeneity tests resulted in p-values of 0.06 for the combined groups, of 0.02 for the ER status, and of 0.19 for the PR status.
Table 4. Physical activity and risk of breast cancer by estrogen and progesterone receptor status, the EPIC cohort study
For total physical activity (both indices), a stronger inverse association was found with hormone receptor positive (ER+/PR+) breast tumors (Table 4), whereas no association with ER+/PR- and with ER-/PR- was observed. The tests of overall heterogeneity were 0.04 (Total index) and 0.06 (Cambridge index). Further analyses showed that primarily the PR+ status (p-heterogeneity = 0.01 for both indices) rather than the ER+ status (p-heterogeneity = 0.23 and 0.35, respectively) dominated the heterogeneity between the ER/PR subgroups. For consistency reasons within table 4, the results of the heterogeneity tests reported in the last but two column for the ER status only and the last column for the PR status only were restricted to the 4,626 breast cancer cases with available information on the combined receptor status for ER+/PR+, ER+/PR−, and ER−/PR− tumors. However, as there were a total of 6,007 women with known ER status, and 4,814 women with known PR status we also performed the heterogeneity tests for ER status only and PR status only based on the more complete datasets. These sensitivity analyses yielded almost identical results (data not shown). Furthermore, there were no differences in the distribution of confounders for the breast cancer populations with and without information on estrogen and progesterone receptor status (data not shown).
In this large European prospective cohort study of 8,034 incident, invasive breast cancer cases we confirmed the inverse association between total physical activity and risk of breast cancer. Association of moderate total physical activity resulted in significant relative risk reductions of 8% compared with the inactive group, and the highest activity group showed a statistically significant risk reduction of 13%. Results for household and recreational activity combined were similar, while, by contrast, no associations of breast cancer with occupational activity were observed. Moderate total physical activity showed a somewhat stronger risk reduction for cancers diagnosed before age of 50, as compared with cancers diagnosed at an older age. In analyses by tumor subtypes, total physical activity levels showed a modest inverse relationship with risk of hormone receptor positive (ER+/PR+) tumors and especially PR+ tumors, whereas no such association was seen for the other tumor subtypes. For household physical activity an inverse association with ER−/PR− tumors was observed.
Strengths of our study are its large size (this is the largest prospective study to date on physical activity assessing risk of breast cancer overall, and by tumor subtypes), the Europe-wide coverage of the study population, and the availability of data on a wide range of other risk factors for breast cancer and on tumor subtypes. By contrast, a major limitation is that self-reported physical activity assessment one year prior to study entry may not represent long-term activity. Also, while all types of activity were assessed in this study at the time of recruitment, there was no information on the duration and frequency of occupational activity. This precluded estimating a quantitative sum of all types of activity for total activity in MET-hr/week, and resulted in categorizing individuals into inactive, moderately inactive, moderately active and active groups after cross-classifying occupational and non-occupational activity. A further limitation could be that we only looked at ER and PR status. At the time of this study, additional information on HER2 status to determine breast cancer subtypes into more detailed molecular subclassifications could not be completed because of insufficient information on HER2. However, as the routine assessment of HER2 is relatively more recent than ER and PR assessment, future cohort analyses will be able to include HER2.
For total breast cancer, the observed risk reduction of 13% in the EPIC cohort is lower than that reported by most of the other studies: A recent systematic review summarizing 33 previous prospective studies has shown a 20% average risk reduction amongst physically active women as compared to the least active women.1 The weaker relative risk estimates in our study may reflect a dilution effect due to not fully capturing long term activity habits. The inverse dose-response relationship suggested by our study, with both moderate (8% relative risk reduction) and high physical activity levels (13%) being inversely related to breast cancer risk, particularly seen for ER+/PR+, is in line with findings from a systematic review.36 In our study, the limit of being classified as at least moderately active translates into, for example, 2 hr per day of vigorous physical activity (MET = 6) or three hours per day of moderate intensity physical activity (MET = 4), when all recreational and household activities are combined. There are very few studies to compare this categorization with as many studies assessed recreational physical activity only. However, for recreational activities, our limit of 24 MET*hr/week for moderate activity refers, for example, to 4 hr of cycling per week. This is comparable to other studies (see Table 3 in Monninkhof et al.36).
For hormone-receptor defined breast tumors, previous studies have yielded inconsistent results and direct comparisons between investigations are challenging due to the diversity in physical activity definitions and reference periods, the differences in study populations examined, and the large proportions of cases with missing receptor status. Only two previous cohort studies formally tested for heterogeneity of risk associations between the subtypes, as we did.13, 24 Our observation of a modest inverse risk association primarily for ER+/PR+ and PR+ carcinomas is in line with findings from the largest case-control study on this topic (n = 3,147 cases with combined receptor status information, and 6,569 controls), which showed significant inverse associations especially of recent leisure time physical activity with risk of ER+/PR+ tumors, but not for any other subgroup, and the strongest risk associations for PR+ tumors.22 Two other investigations also reported significant relations for ER+/PR+ only.13, 23 However, the rest of the studies found either null associations by tumor receptor type9, 17–19 or protective effects of exercise for both hormone receptor positive and negative carcinomas.6, 8, 12, 16, 21 Two studies reported results in clear contrast to our findings. The California Teachers Study11 found inverse associations of recreational long-term physical activity for ER− and ER−/PR− cancer risk but not for ER+/PR+ or ER+/PR− risks. However, in this study moderate and strenuous recreational activities were analyzed separately, and walking and cycling for locomotion were not assessed. Thus the association with overall recreational activity as well as household and total physical activity is unclear. Similar but nonsignificant results were observed in another study.24 In that investigation, physical activity assessment was restricted to one question on the weekly frequency of total vigorous activity during the past year. The inconsistency of the results from other studies may be partly due to limited power.
We investigated whether associations with breast cancer risk differed by type of physical activity. Occupational activity was not related to breast cancer risk overall. However, nonoccupational (recreational and household) was associated with a reduced risk of overall breast cancer risk and ER+/PR+ disease. The result for overall breast cancer risk is in agreement with the most recent comprehensive review of published studies that concluded that the greatest risk reductions were found for recreational and household activities (average 21%), followed by walking/cycling (18%).1 In our study, the inverse association of household activities with breast cancer risk was more pronounced in the ER-/PR- subtype. To our knowledge, no other study has provided data to compare this result with. Assessing household activity is problematic as household tasks comprise many different, often unstructured and intermittent activities. Besides the possibility that our findings reflect true associations, it can still be speculated that results for household are particularly susceptible to residual confounding and do not point to some etiologic specificities of household activities. Women who are active in the household domain may also differ in even more variables than we or previous authors could adjust for.
Recent reviews have observed that risk reductions of overall breast cancer from physical activity were somewhat stronger among postmenopausal women than in premenopausal women, hypothetically due to major physiologic changes in women at menopause.1, 2 However, this summarized evidence is limited by the fact that studies, particularly cohort studies, vary in the time when the menopausal status is considered. For example, in our study menopausal status was only known at the time of recruitment but not at the time of diagnosis, with the latter being of more interest for etiologic research. Thus, we used age at diagnosis as a proxy for menopausal status at diagnosis. Some of our results for total as well as for subtypes of physical activity stratified by age at diagnosis are in agreement with these reviews if only statistical significance of the highest activity category and trend tests are considered. However, the absence of statistical significance of some of the results for cancer diagnosed at a younger age might be explained by more limited power compared to the larger group of women diagnosed after age 50. In contrast to these reviews, we even observed slightly greater relative risk reductions for breast cancer diagnosed before or at the age of 50, as compared with cancers diagnosed at older ages, especially for moderate physical activity.
The exact biological mechanisms for associations of physical activity with breast cancer risk overall and potentially differential associations on tumor subtypes remain unclear. Our results on tumor subtypes suggest that in older women who are physically active the risk of invasive breast cancer may be reduced at least in part via steroid hormonal pathways. Epidemiological studies have found direct associations between postmenopausal endogenous levels of steroid hormones and breast cancer overall,37 and particularly for ER+/PR+ cancer.38 Postmenopausal women who are physically active in comparison to inactive women may have lower concentrations of serum estrone, estradiol and androgens39, 40 and increases in sex hormone-binding globulin.41 Substantial epidemiologic, clinical and experimental evidence have clearly established a late-stage growth promoting effect of estrogens especially on estrogen sensitive tumors,42 and this may be reflected by the observed association of physical activity with ER+/PR+ tumors. Physical activity has been thought to alter menstrual characteristics by delaying menarche and, in women of reproductive age, by affecting ovulatory cycle.43 However this is observed in scenarios of extreme physical activity and our study showed risk reductions with moderate physical activity. Other proposed mechanisms include, for example, favorable effects on the immune system, an increase in antioxidant enzymes, and reduction of chronic inflammation.44
Our results suggest that PR may be important in the mode of action of physical activity on breast carcinogenesis. Growing evidence suggests that progesterone plays a key role in the regulation of cell proliferation and differentiation in the mammary gland. It has been demonstrated that, either in ER+ or ER− cell context, progestin induced proliferation and regulated proteases activity and metastasis through PR ability to activate c-Src-dependent signaling pathways.45 The expression of PR is up-regulated by estrogen stimulation,46 hence PR expression might be affected by exercise via reduced estrogen levels.
In conclusion, this large prospective study conducted in a heterogeneous population of Europeans provides further epidemiologic evidence that physical activity is associated with a modestly decreased breast cancer risk. Risk reductions of 13% for invasive breast cancer among the physically active women compared with inactive women were observed. The associations were largely confined to ER+/PR+ and PR+ tumors, which are the most common breast cancer subtypes. An important finding of the study is that inverse associations of breast cancer with both moderate and high physical activity levels were observed. Unlike many risk factors for breast cancers, physical activity is an exposure that can be modified. Thus, it is worthy of consideration for cancer prevention programs to note that these levels of activity are achievable by most of the at risk population, and that already some changes in physical activity behavior could have a positive influence on breast cancer incidence.
The authors would like to thank Ms. Jutta Schmitt and Ms. Jutta Kneisel for their assistance during the collection of hormone receptor status data, Ms. Sina Silberberger for helping with the literature, and they thank all the EPIC cohort participants for their contributions to data collection at baseline recruitment and during follow-up.