Nutrient-Wide Association Study of 92 Foods and Nutrients and Breast Cancer Risk

Alicia K. Heath; David C. Muller; Piet A. van den Brandt; Nikos Papadimitriou; Elena Critselis; Marc Gunter; Paolo Vineis; Elisabete Weiderpass; Guy Fagherazzi; Heiner Boeing; Pietro Ferrari; Anja Olsen; Anne Tjønneland; Patrick Arveux; Marie-Christine Boutron-Ruault; Francesca Romana Mancini; Tilman Kühn; Renée Turzanski-Fortner; Matthias B. Schulze; Anna Karakatsani; Paschalis Thriskos; Antonia Trichopoulou; Giovanna Masala; Paolo Contiero; Fulvio Ricceri; Salvatore Panico; Bas Bueno-de-Mesquita; Marije F. Bakker; Carla H. van Gils; Karina Standahl Olsen; Guri Skeie; Cristina Lasheras; Antonio Agudo; Miguel Rodríguez-Barranco; Maria-José Sánchez; Pilar Amiano; María-Dolores Chirlaque; Aurelio Barricarte; Isabel Drake; Ulrika Ericson; Ingegerd Johansson; Anna Winkvist; Tim Key; Heinz Freisling; Mathilde His; Inge Huybrechts; Sofia Christakoudi; Merete Ellingjord-Dale; Elio Riboli; Konstantinos K. Tsilidis; Ioanna Tzoulaki

Disclosures

Breast Cancer Res. 2020;22(5) 

In This Article

Methods

This NWAS involved investigation of intakes of 92 foods and nutrients (for which data were available) in relation to breast cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study, and calculation of the associated False Discovery Rate (FDR) to select dietary factors to evaluate in the independent replication cohort, the Netherlands Cohort Study (NLCS).

Study Populations and Ascertainment of Breast Cancer Cases

EPIC. The EPIC study includes 521,330 men and women aged 25 to 70 years at recruitment, which occurred between 1992 and 2000.[14] Participants were from 23 centres in 10 European countries (Denmark, France, Germany, Greece, Italy, Norway, Spain, Sweden, the Netherlands, and the UK) and completed questionnaires on diet, lifestyle, and medical history. Informed consent was provided by all participants, and ethical approval for the study was provided by the internal review board of the International Agency for Research on Cancer and from local ethics committees in each participating country.

Women without a pre-baseline diagnosis of cancer were eligible for inclusion in these analyses; those who did not complete dietary or lifestyle questionnaires or with missing data on relevant confounders were excluded.

Incident breast cancers were identified through population-based cancer registries or active follow-up, and mortality data were obtained from cancer or mortality registries.[14] Breast cancers were classified as ICD-10 code C50. Information on oestrogen receptor (ER) and progesterone receptor (PR) status of the tumours was provided by each centre on the basis of pathology reports; this information was not available for all cases, particularly during the early years of follow-up.

NLCS. The NLCS includes 120,852 participants, of whom 62,573 are women, aged 55 to 69 years when recruited in 1986 from the general population in 204 municipalities in the Netherlands with computerised population registries.[15] At recruitment, participants completed a self-administered questionnaire on dietary habits, lifestyle factors, medical history, family history of cancer, and other risk factors for cancer. The NLCS was approved by the institutional review boards of the Nederlandse Organisatie voor Toegepast Natuurwetenschappelijk Onderzoek (TNO) Quality of Life research institute (Zeist, Netherlands) and Maastricht University (Maastricht, Netherlands).

For efficiency, a case-cohort approach was used for questionnaire processing and follow-up. Breast cancer cases were identified from the entire cohort, but accumulated person-years at risk in the entire cohort were estimated from a subcohort of 2589 women who were randomly sampled from the cohort immediately after recruitment. For cases and members of the subcohort, we excluded women with a prevalent cancer other than non-melanoma skin cancer at recruitment, as well as those with incomplete or inconsistent dietary data or missing confounder data.

Incident breast cancer cases were identified by record linkage to the Netherlands Cancer Registry and the Dutch National Pathology Registry.

Dietary Assessment

In the EPIC study, the diet of participants was assessed at enrolment using validated country-specific or study centre-specific dietary questionnaires or food records.[14,16,17] The EPIC Nutrient Database was used to calculate standardised nutrient intakes for the 10 countries.[18] All foods and nutrients in the centralised EPIC database that were available in most countries (at least eight out of ten countries; 92 dietary factors) were selected for analysis.

Information on dietary intake in the NLCS was collected at recruitment using a 150-item semi-quantitative food frequency questionnaire that estimated the average frequency and amounts of foods and beverages habitually consumed in the previous 12 months. The food frequency questionnaire has been validated and tested for reproducibility.[19,20] Nutrient intakes were calculated by multiplying the frequency of intake by the nutrient content of specified portions based on the Dutch food composition table.[21]

Statistical Analysis

Cox regression models with age as the time scale were fitted to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for risk of breast cancer as a function of each food or nutrient. In the EPIC study, age at recruitment was the entry time, and age at cancer diagnosis (except non-melanoma skin cancer), death, emigration, or last follow-up, whichever occurred first, was the exit time. In the NLCS, the total person-years at risk were estimated from the subcohort, and Prentice-weighted Cox proportional hazards models with robust standard error estimates were used to account for the case-cohort design.[22] Intakes of foods and nutrients were adjusted for energy intake using the residual method[23] and standardised (by subtracting the sample mean and dividing by the sample standard deviation (SD)) prior to modelling. HRs were estimated for a one SD increment in intake. All models were stratified by age at recruitment (5-year groups) and study centre (EPIC only) and adjusted by energy intake, history of diabetes (yes/no), educational attainment (none/primary school, technical/professional school, secondary school, longer education), smoking status (never, former, current), body mass index (BMI) (< 20, [20, 23), [23, 25), [25, 30), [30, 35), ≥ 35 kg/m2), physical activity (EPIC: Cambridge index (inactive, moderately inactive, moderately active, active);[24] NLCS, non-occupational physical activity (≤ 30, > 30–60, > 60–90, > 90 min/day)), menopausal status at baseline (EPIC only: postmenopausal versus pre- and perimenopausal), menopausal status by BMI interaction (EPIC only), age at menopause (NLCS only), age at menarche, the interaction of parous (yes/no) and age at first pregnancy, and family history of breast cancer in mother or sister/s (NLCS only). Adjustment for factors involving menopausal status was not necessary in the replication analyses in the NLCS since all women in the NLCS were postmenopausal at baseline. We used the Benjamini-Hochberg approach to control the FDR at 0.05.[10] The set of foods/nutrients satisfying this FDR (variables with q value < 0.05) within EPIC were carried forward for replication in the NLCS.

We performed the NWAS overall, as well as separately by premenopausal/postmenopausal status at baseline. Associations with breast cancer for the identified foods and nutrients in the EPIC study were also assessed by ER and PR expression in tumours for the 60% of EPIC cases and 46% of NLCS cases for whom receptor status data were available.

All analyses were performed in R version 3.6.1.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....