Physical Activity and Ovarian Cancer Risk: the European Prospective Investigation into Cancer and Nutrition

Ovarian cancer is the fourth most frequent cause of cancer death in women from Northern and Western Europe, Australia, and North America (1). Physical activity has been proposed to influence ovarian cancer risk most likely by affecting body fat stores and levels of various endogenous hormones (2, 3). To date, nine prospective (4-12) and eight case-control (13-20) studies have examined the association between physical activity and ovarian cancer risk with inconsistent results, predominantly indicating a decreased risk or no effect. Given these equivocal findings, we examined the association between various types of physical activity and ovarian cancer risk overall and by histologic subtype in the European Prospective Investigation into Cancer and Nutrition (EPIC).

EPIC is a multicenter prospective cohort study designed primarily to investigate the association between nutrition and cancer. The EPIC cohort consists of 23 subcohorts in 10 European countries. The design, study population, baseline data collection, and follow-up methods have been previously described in detail (21).

Physical activity data (22) were obtained using in-person interviews or via a self-administered standardized questionnaire and included information on the frequency and duration of past-year nonoccupational physical activity, specifically household activities (housework, home repair, gardening, stair climbing) and recreational activities (walking, cycling, sports). 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). To derive an estimate of total physical activity, household and recreational activities were combined in metabolic equivalent (MET)-hours per week (23), divided into quartiles (low, medium, high, and very high), and cross-classified with the categories of occupational activity. A more detailed description of the assignment of MET values and derivation of combined variables has been given in recent EPIC-reports (24, 25).

The present study was based on data from n = 325,780 women from which we a priori excluded women with prevalent cancer at any site at baseline examination, those who had undergone bilateral oophorectomy, those with missing dietary or physical activity questionnaire data, and those who were in the top or bottom 1% of the ratio of energy intake to estimated energy requirement (calculated from age and body weight) to reduce the effect on the analysis of implausible extreme values. Women from the Norwegian cohort and the cohort from Umeå, Sweden, were also excluded because of lack of standardized data on physical activity. The analytic cohort for this study therefore consisted of 274,740 women (35% premenopausal at baseline examination) from nine countries.

By June 2007, 929 ovarian cancer cases had been reported by the EPIC centers based on information on complete follow-up data up until December 2004 or December 2005 in most of the centers. After application of the aforementioned exclusion criteria and exclusion of cases with nonepithelial, metastatic, or in situ ovarian tumors, 731 cases of epithelial ovarian tumors (primary, malignant) remained available for this analysis: 335 were serous tumors, 70 mucinous, 69 endometrioid, 27 clear cell, 6 transitional cell or undifferentiated, and 182 not otherwise specified or could not be classified; histology information was missing for 42 cases.

We estimated hazard ratios and 95% confidence intervals using Cox regression models, with age as the underlying time variable, to evaluate the association between type of physical activity and risk of ovarian cancer. The models were stratified by age at recruitment and study center, and adjusted for other potentially confounding factors, namely, education, body mass index, age at menarche, parity, menopausal status, unilateral oophorectomy, and use of oral contraceptives. Trend tests were estimated by assigning a score from 1 to 4 to quartile or category physical activity levels. All tests of statistical significance were two-sided and P < 0.05 was considered statistically significant. The statistical analyses were done using SAS version 9.1.3 (SAS Institute).

The average duration of follow-up for the 274,740 cohort members was 9.3 (SD ± 2.0) years, yielding a total of 2,551,816 person-years (Table 1). The median age at baseline was 51.5 years, and the median (min, max) age at diagnosis for the cases was 61.1 (31.3, 86.4) years.

In crude and adjusted models, neither total physical activity nor any individual type of activity was associated with risk of ovarian cancer (Table 2). Most hazard ratio estimates suggested a slightly increased risk, but they were not statistically significant and there were no dose-response trends. Mutual adjustment for recreational, household, and occupational activity did not materially affect the relative risk estimates. Results were similar in subgroups defined by body mass index categories, menopausal status, or hormone replacement therapy use among postmenopausal women (data not shown). The analysis by histologic subtypes revealed no significant associations for serous, mucinous, or endometrioid tumors (data not shown).

The results from this large European prospective study, including 731 epithelial ovarian cancer cases, do not provide support for a protective effect of physical activity on ovarian cancer risk. These results are consistent with findings from half of the previous prospective (4, 5, 8, 10, 11) and two of seven case-control studies (13, 15). In studies that reported on an inverse association between physical activity and ovarian cancer (6, 7, 14, 16-20), the risk reduction ranged between 10% and 67% (highest versus lowest level of activity), and in a meta-analysis of recreational activity and ovarian cancer the risk reduction was estimated as 19%, although when restricted to cohort studies this association did not reach statistical significance (16). The EPIC questionnaire has been found to rank participants satisfactorily (26, 27), and previous analyses of physical activity in EPIC have found inverse associations with breast and colon cancers (24, 25). However, it is possible that measurement error and misclassification associated with questionnaire measurement of physical activity in the present study may have biased the estimates toward the null. In addition, any association with ovarian cancer risk is likely to be modest and, thus, may be difficult to detect with questionnaire measurement. Subgroup analyses in studies with larger sample sizes or more detailed physical activity measures may be required to investigate this association more fully.

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.