Diet quality and survival after ovarian cancer: results from the Women's Health Initiative

Abstract

Background: Survival after an ovarian cancer diagnosis is poor. Given the high mortality in these patients, efforts to identify modifiable lifestyle behaviors that could influence survival are needed. Earlier evidence suggests a protective role for vegetables, but no prior studies have evaluated overall dietary quality and ovarian cancer survival. The purpose of this analysis was to evaluate the role of prediagnosis diet quality in ovarian cancer survival.

Methods: We identified 636 centrally adjudicated cases of ovarian cancer within the Women's Health Initiative Observational Study or Clinical Trials of 161808 postmenopausal women followed from 1995 to 2012. Dietary quality was assessed for the Healthy Eating Index (2005) using a food frequency questionnaire, covariables were obtained from standardized questionnaires, and adiposity was measured by clinic-based measurements of height, weight, and waist circumference. The association between diet quality and mortality was analyzed using Cox proportional hazards regression, adjusted for potential confounders, and stratified by waist circumference, physical activity level, and diabetes status. Tests of statistical significance were two-sided.

Results: Overall, higher diet quality was associated with lower all-cause mortality after ovarian cancer (hazard ratio [HR] for highest vs lowest tertile = 0.73; 95% confidence interval [CI] = 0.55 to 0.97, P(trend) = .03). The effect was strongest among women with waist circumference of 88 cm or less and with no history of diabetes (HR = 0.73, 95% CI = 0.54 to 0.98). Physical activity level did not modify the association between diet quality and survival.

Conclusion: Our results suggest that overall higher prediagnosis diet quality may protect against mortality after ovarian cancer.

Figure 1.

Association between Healthy Eating Index (HEI)–2005 diet quality score, in tertiles, and all-cause mortality among 636 ovarian cancer survivors, stratified by waist circumference, diabetes, and moderate-to-strenuous recreational physical activity. Cox proportional hazards regression models were adjusted for age at diagnosis, stage at diagnosis, race/ethnicity (non-Hispanic white, black, Hispanic, other/unknown), diabetes, physical activity (0, > 0 to < 9, 9 to < 15, ≥ 15 metabolic equivalent of tasks–hr/week), total energy intake (quintiles), waist circumference (≤ 88, > 88cm), family history of ovarian cancer (no, yes, unknown), and clinical trial arms. HEI = Healthy Eating Index; MET = metabolic equivalent of tasks.