Long-term adherence to healthful and unhealthful plant-based diets and breast cancer risk overall and by hormone receptor and histologic subtypes among postmenopausal females

Abstract

Background: Epidemiological studies assessing the influence of vegetarian diets on breast cancer (BC) risk have produced inconsistent results. Few studies have assessed how the incremental decrease in animal foods and the quality of plant foods are linked with BC.

Objectives: Disentangle the influence of plant-based diet quality on BC risk between postmenopausal females.

Methods: Total of 65,574 participants from the E3N (Etude Epidémiologique auprès de femmes de la Mutuelle Générale de l'Education Nationale) cohort were followed from 1993-2014. Incident BC cases were confirmed through pathological reports and classified into subtypes. Cumulative average scores for healthful (hPDI) and unhealthful (uPDI) plant-based diet indices were developed using self-reported dietary intakes at baseline (1993) and follow-up (2005) and divided into quintiles. Cox proportional hazards models were used to estimate adjusted HR and 95% CI.

Results: During a mean follow-up of 21 y, 3968 incident postmenopausal BC cases were identified. There was a nonlinear association between adherence to hPDI and BC risk (P_nonlinear < 0.01). Compared to participants with low adherence to hPDI, those with high adherence had a lower BC risk [HR_{Q3 compared withQ1} (95% CI): 0.79 (0.71, 0.87) and HR_{Q4 compared with Q1} (95% CI): 0.78 (0.70, 0.86)]. In contrast, higher adherence to unhealthful was associated with a linear increase in BC risk [P_nonlinear = 0.18; HR_{Q5 compared with Q1} (95% CI): 1.20 (1.08, 1.33); P_trend < 0.01]. Associations were similar according to BC subtypes (P_{heterogeneity} > 0.05 for all).

Conclusions: Long-term adherence to healthful plant foods with some intake of unhealthy plant and animal foods may reduce BC risk with an optimal risk reduction in the moderate intake range. Adherence to an unhealthful plant-based diet may increase BC risk. These results emphasize the importance of the quality of plant foods for cancer prevention. This trial was registered at clinicaltrials.gov (NCT03285230).

Keywords: breast cancer; dietary score; estrogen receptor; plant-based diet quality; progesterone receptor; prospective study.

FIGURE 1

Associations of the healthful plant-based diet index with BC fitted with restricted cubic splines (N = 65,574; 5 knots placed at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles), P_nonlinear < 0.01. Risk estimates were adjusted for age (as the time scale), educational level, physical activity, smoking status, family history of BC, breastfeeding, age at menarche, age at first full-term birth, past history of benign breast disease, ever use of the contraceptive pill, ever use of menopausal hormone therapy, mammography in the last follow-up cycle, body mass index, energy intake, and alcohol (model stratified by birth cohort). The solid line represents the hazard ratio, and the dashed lines the lower and upper 95% confidence interval. BC, breast cancer.

FIGURE 2

Associations of the unhealthful plant-based diet index with BC fitted with restricted cubic splines (N = 65,574; 5 knots placed at the 5th, 27.5th, 50th, 72.5th, and 95th percentiles), P_nonlinear = 0.18. Risk estimates were adjusted for age (as the time scale), educational level, physical activity, smoking status, family history of BC, breastfeeding, age at menarche, age at first full-term birth, past history of benign breast disease, ever use of the contraceptive pill, ever use of menopausal hormone therapy, mammography in the last follow-up cycle, body mass index, energy intake, and alcohol (model stratified by birth cohort). The solid line represents the hazard ratio, and the dashed lines the lower and upper 95% confidence interval. BC, breast cancer.

FIGURE 3

The healthful and unhealthful plant-based diet indices and breast cancer risk, overall and subtypes, E3N (Etude Epidémiologique auprès de femmes de la Mutuelle Générale de l'Education Nationale) cohort (N = 65,574). Hazard ratios (Model 3)¹ for the highest (Q5) compared with the lowest (Q1) quintiles was presented in the figure. CI, confidence interval; ER, estrogen receptor; HR, hazard ratio; PR, progesterone receptor. ¹HR adjusted for age (as the time scale), educational level, physical activity, smoking status, family history of breast cancer, breastfeeding, age at menarche, age at first full-term birth, past history of benign breast disease, ever use of the contraceptive pill, ever use of menopausal hormone therapy, mammography in last follow-up cycle, body mass index, energy intake, and alcohol (model stratified by birth cohort).