Study population and design
Case subjects were recruited from September 2019 to March 2020 from patients admitted to the Ho Chi Minh City Oncology Hospital, Ho Chi Minh City, Vietnam. Inclusion criteria were Vietnamese female subjects aged from 25 years with primary and histologically confirmed breast cancer diagnosed at Ho Chi Minh City Oncology Hospital. Women were excluded if they had a prior history of any cancers. Controls for the present study were randomly selected from the cohort of the Vietnam Osteoporosis Study. According to their medical records follow-up in 6 years in the Vietnam Osteoporosis Study, these control participants had no cancer at the recruitment. The study’s rationale and protocol have been described elsewhere [11]. This project is a longitudinal cohort study which has been collecting roughly 4200 random citizens aged 18 years and above in Ho Chi Minh City and surrounding rural areas. The controls were frequently matched by a 5-year age group to the cases on a 1:1 ratio. For each age bracket, a sample of control candidates of the required size was generated randomly using the base R sample function [12].
Ho Chi Minh City contributes significantly to Vietnam's economy, making up about 27% of the country's overall budget and 23% of the gross domestic product in 2020 [13]. With those great employment opportunities, Ho Chi Minh City has become the most attractive destination for the migrant labor force in Vietnam. It is estimated that there were 200 immigrants per 1,000 people in Ho Chi Minh City, coming from the Mekong River Delta, the North Central, and Central Coastal areas [14].
All women who participated in this study provided written informed consent. The study was approved by the Ethics Committee of Vietnam Ministry of Health.
Data collection
Detailed information on dietary habits and other factors (sociodemographic characteristics, weight, height, menstrual and reproductive history, physical activity, and family history of breast cancer) among the participants was elicited by face-to-face interviews using a structured questionnaire conducted by trained interviewers.
The sociodemographic characteristics were age and education level. The menstrual and reproductive history included information about the age atmenarche, age at first birth, parity, menopausal status, benign breast disease, and oral contraceptive use. Physical activity was assessed using the Global physical activity questionnaire [15]. The participants were measured for height and weight using an electronic portable, wall-mounted stadiometer (Seca Model 769; Seca Corp, CA, USA) while not wearing shoes, ornaments, hats, or heavy layer clothing. For the case group, we retrieved the information on the participant’s weight before the operation from the medical record. Body mass index (BMI) was calculated by dividing weight in kilograms by height in meters squared (kg/m2). The BMI was classified into four categories (< 18.5, 18.5–23, 23-27.5, > 27.5 kg/m2) based on the cut-off for Asian of the World Health Organization [16].
Dietary intakes were assessed using a validated 76-item food frequency questionnaire (FFQ). A commonly used portion size was specified for each food (e.g., slice, glass, bowl, or unit, such as one apple or banana). Food photographs with usual intake portions were provided to help participants estimate and record the amounts of food consumed. The development and validation of the FFQ among Vietnamese in Ho Chi Minh City were described elsewhere [17]. Food items were categorized into food groups based on their similar nutrient contents, including: dark green vegetables (amaranth, swamp cabbage, mustard green, malabar nighshare, crown daisy, Chinese leek, broccoli), red/orange vegetables (tomato, carrot, pumpkin squash, green pepper), legumes (French bean), starchy vegetables (white potato, sweet potato, Chinese yam), other vegetables (cabbage, wax gourd, cucumber, gourd, bitter gourd, mushroom), fruits (dragon fruit, banana, papaya, pomelo, longan, orange, water melon, pear, grape, guava, apple), red meat (pork lean, pork upper leg, pork medium fat, pork rib, pork lower leg, beef), white meat (chicken, duck), innard, seafood (fish, shrimp, squid), egg (hen egg, duck egg), soybean products (fried tofu, raw tofu), tea, dairy beverages (fresh milk, milk powder, yogurt, condensed milk, soybean milk), soft drink, fruit juice (fruit shake juice, lemon juice, orange juice, coconut juice with kernel), and coffee (coffee, instant coffee, instant coffee with milk and sugar). The period of reference in assessing diet factors was three years before diagnosis for cases and three years before being interviewed for controls.
Statistical analysis
Characteristics between cases and controls were compared using the t-test or Mann-Whitney U test for continuous variables, and the chi-square test for categorical variables. Quartiles of food group intakes were defined based on the distribution among the control group. Conditional logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) of each quartile, using the lowest quartile group as the reference. The relationships between different food groups and the risk of breast cancer were further examined after adjusting for various potential confounding variables using multivariate logistic regression models. The following potential confounding factors were included in the models: age (continuous), education level (secondary school and lower, high school and higher), body mass index (BMI, < 18.5, 18.5–23, 23-27.5, > 27.5 kg/m2), age at menarche (continuous), age at first birth (continuous), parity (nulliparous, 1, 2, ≥ 3 births), menopausal status (yes/no), history of benign breast disease (yes/no), oral contraceptive use (ever, yes/no), physical activity (≥ 3 times/week, yes/no), family history of breast cancer (yes/no).
All P values are two-sided and statistical significance was determined at the p < 0.05 level. All data analyses were performed by using R 4.1.2 [12].