Pre- and Post-Diagnosis Diet Quality and Ovarian Cancer Survival

Ovarian cancer is associated with significant morbidity and mortality; in 2019, it accounted for an estimated 22,530 new cases and 13,980 deaths in the United States (1). Most women are diagnosed with advanced disease and 5-year survival is poor. Following diagnosis, women often want to know if there is anything they can do to improve their prognosis. Diet, particularly diet quality, is a potentially modifiable aspect of lifestyle that could influence cancer outcomes (2).

Evidence from the breast and colorectal cancer literature indicates that diet quality may modify the progression and prognosis of these cancers (3, 4). Only one study has comprehensively investigated this association among women with ovarian cancer (5). In that article published in 2014, Thomson and colleagues (5) reported that higher diet quality, assessed before diagnosis using the Healthy Eating Index (HEI-2005), was associated with significantly lower all-cause mortality among 636 women with ovarian cancer. However, a limitation of the analyses, noted by the authors, was the lack of data on diet quality or diet-quality change after diagnosis. Our goal was to answer an important question for women diagnosed with ovarian cancer, namely “would improving the quality of my diet increase my chances of survival?”

To address this, we analyzed data from the Ovarian cancer Prognosis And Lifestyle (OPAL) study, a prospective study of Australian women, ages 18–79 years, newly diagnosed with ovarian cancer (6). Information about diet pre-diagnosis (n = 650, collected as soon as possible after recruitment) and post-diagnosis (n = 503 women who had a complete response to primary treatment, collected 12 months post-diagnosis) was obtained using a validated food frequency questionnaire. All women provided signed informed consent and the study was approved by all relevant Human Research Ethics Committees. We assessed diet quality using five indices: The HEI-2010, Alternative Healthy Eating Index (AHEI-2010), World Cancer Research Fund (WCRF) index, American Cancer Society (ACS) guidelines, and the Australian Dietary Guideline Index (DGI; ref. 2). Diet-quality scores were divided into tertiles with tertile 3 being the highest diet quality and tertile 1 the lowest. Clinical data were abstracted from women's medical records, surgical and pathology report and information about disease recurrence, vital status and cause of death were collected annually from medical records.

Survival time was calculated from the date a woman started primary treatment (surgery or neoadjuvant chemotherapy) for analyses of pre-diagnosis diet or the end of primary treatment date (for analyses of post-diagnosis diet) to the earliest date of study exit due to death from ovarian cancer or last follow-up, and was left truncated to the date the woman completed the baseline (pre-diagnosis analysis) or 12 month (post-diagnosis analysis) questionnaire. Progression-free survival (PFS) time was computed from the date a woman started primary treatment to the first date a woman received treatment for disease progression or last known to be progression free.

Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI). Directed acyclic graphs were used to identify potential confounders of the association between the pre-and post-diagnosis diet-quality indices and survival. We present models adjusted for age at diagnosis, log energy intake, smoking status, BMI and stratified by diabetes and physical activity for pre-diagnosis analyses; and age, log energy intake, smoking status at 12 months, physical activity level at 12 months and FIGO stage, and stratified by physical activity at 12 months for post-diagnosis analyses. We also stratified by stage of disease (I/II vs. III/IV). To examine diet change, we classified diet-quality score as low or high at each time point using the baseline median score. Women were classified into four groups according to whether their diet quality was always low, always high, increased from low to high, or decreased from high to low. All analysis was performed using SAS version 9.4 (SAS Institute Inc.).

We found no association between higher pre-diagnostic diet-quality scores and cancer-specific survival (Table 1). Results were similar when we considered PFS or overall survival as the end-point. There was no evidence of effect modification by BMI, diabetes or physical activity. When we repeated our analysis replicating the methods of Thomson and colleagues (ref. 5; same covariates, diet quality index, and tertile cutoff points), our results were also null. It is unclear why the association seen in Thomson's study differs from our study but, compared with Thomson's study, our cohort of women was younger and had higher HEI-2005 scores; we also had relatively few deaths in the lowest HEI-2005 group.

Unlike Thomson and colleagues (5), we were also able to analyze the relation with diet quality after diagnosis; we restricted this analysis to women who had a complete response to primary treatment as we felt diet was less likely to help women whose cancer did not respond to chemotherapy. Again, we found no association between higher diet-quality scores and improved survival, rather HR for women with higher diet quality were consistently greater than 1.0 (Table 1). Excluding women with ≥3 symptoms that might affect their dietary intake (e.g., nausea/vomiting) made no material difference to effect estimates, nor was there any evidence that results varied by stage of disease. Inclusion of a six or 12-month lag between the measurement of diet quality and start of follow-up (to reduce reverse-causality where a woman's current health might affect her diet) showed similarly null results. When we compared change in diet quality from pre- to post-diagnosis there was no evidence of better survival among women who had either improved their diet quality or continued to consume a high-quality diet (Table 2).

Identifying modifiable factors for improving the prognosis of women with ovarian cancer is a high priority given the very poor outcomes that many women experience; however, in this population of Australian women, we found no evidence that women consuming a diet with a high-quality score, assessed before and/or 12 months after their diagnosis of ovarian cancer, had better survival. Larger studies and randomized controlled trials are required to ascertain more definitely whether improving diet quality after a diagnosis of ovarian cancer would improve outcomes. In the meantime, it would seem prudent to encourage women to follow current dietary guidelines for cancer survivors.

M. Friedlander reports grants and personal fees from AstraZeneca and Novartis, as well as personal fees from GlaxoSmithKline, Takeda, Merck Sharp & Dohme, and Lilly outside the submitted work. A. DeFazio reports grants from National Health and Medical Research Council of Australia during the conduct of the study and AstraZeneca outside the submitted work. P.M. Webb reports grants from National Health and Medical Research Council of Australia during the conduct of the study and AstraZeneca outside the submitted work. No disclosures were reported by the other authors.

R.M. Al Ramadhani: Formal analysis, methodology, writing-review and editing. C.M. Nagle: Conceptualization, formal analysis, supervision, methodology, writing-review and editing. T.I. Ibiebele: Formal analysis, methodology, writing-review and editing. P. Grant: Resources, writing-review and editing. M. Friedlander: Resources, data curation, writing-review and editing. A. DeFazio: Resources, data curation, writing-review and editing. P.M. Webb: Resources, supervision, funding acquisition, methodology, writing-review and editing.

The OPAL Study was funded by the National Health and Medical Research Council (NHMRC) of Australia (GNT1025142 and GNT1120431). C.M. Nagle and T.I. Ibiebele were funded by a program grant from the NHMRC (GNT1073898), and P.M. Webb was funded by a fellowship from the NHMRC (GNT1173346). We acknowledge the OPAL Study team and all the clinicians and participating institutions who helped make this study possible (see opalstudy.qimrberghofer.edu.au for a complete list). We also thank consumer representatives Karen Livingstone, Hélène O'Neill, and Merran Williams as well as all the women who took part.