Dietary Factors and Female Breast Cancer Risk: A Prospective Cohort Study

Breast cancer is the leading cause of cancer in females and has become a major global health priority. This prospective cohort study investigated the association of dietary factors, including food items and dietary habits, with the risk of breast cancer in Korean women. Study participants were women aged 30 years or older, recruited from the National Cancer Center in South Korea between August 2002 and May 2007. They were followed until December 2014 using the Korea Central Cancer Registry to identify breast cancer cases. Among 5046 non-pre-diagnosed cancer participants, 72 breast cancer cases were prospectively identified. Participants with breast cancer had a significantly higher educational level (college or higher: 58.3% vs. 39.5%, p = 0.01), were more likely to have ever smoked (22.2% vs. 7.8%, p < 0.001), and were more likely to have a history of benign breast tumors (10% vs. 4%, p = 0.02) than non-cases. Consumption of grilled meat conferred a significantly higher risk of breast cancer in all women (hazard ratio (HR) 1.77, 95% confidence interval (CI) 1.09–2.85) and in postmenopausal women (HR 3.06, 95% CI 1.31–7.15). High-cholesterol food intake was associated with a higher risk in all women (HR 1.69, 95% CI 1.01–2.82). Irregular meal intake was associated with an elevated risk in all women (HR 2.19, 95% CI 1.20–3.98, p for trend = 0.01) and in premenopausal women (HR 2.35, 95% CI 1.13–4.91, p for trend = 0.03). Our findings suggest that grilled meat and high-cholesterol food intake and irregular eating habits may be associated with a higher risk of breast cancer. Further studies with longer follow-up periods that include information on portion size, hormone receptor status, carcinogen levels in grilled meat, and a classification of foods by source are required.


Introduction
Cancer is the leading cause of death worldwide and has become a major public health problem [1,2]. In Korea, cancer accounted for 28.3% of deaths in 2013 [3]. Of the various cancers, breast cancer is the most frequently diagnosed in women worldwide [1]. In Korea, the incidence of breast cancer has continually increased in the past 20 years [4] and was the second leading cause of female cancer in both incidence and prevalence in 2013. According to the Korean Central Cancer Registry (KCCR), the age-standardized incidence of breast cancer was 45.7 per 100,000 women, and the annual percent change between 1999 and 2013 was 5.9% in women [3].
Diet is considered a modifiable risk factor and accounts for approximately 35% of all cancer causes. Therefore, it is important to identify the dietary risk factors of cancer [5][6][7][8]. The westernization of

Data Collection and Management
At the time of enrollment, all participants were asked to complete self-administered questionnaires regarding socio-demographic characteristics (e.g., age, education, occupation, household income, and marital status), family history of cancer, cigarette and alcohol consumption, regular exercise habits, history of benign breast tumor, and reproductive factors (e.g., age at menarche, menopausal status, age at menopause, type of menopause, postmenopausal hormone use, parity, delivery frequency, lactation, and oral contraceptive use). Height and weight were measured using an InBody 3.0 (Biospace, Seoul, Korea) body composition analyzer or the X-SCAN PLUS II Body Composition Analyzer (Jawon Medical, Gyeongsan, Korea), and body mass index (BMI) was calculated as weight (kg)/height (m 2 ).
The FFQ assessed 16 food groups: cereals, salty vegetables and seafood, light-colored vegetables, green-yellow vegetables, seaweed, fruit, grilled meat, healthy protein foods, dairy foods, bony fish, fried foods, high-cholesterol foods, animal fat-rich foods, sweet foods, fast foods, and caffeinated drinks (see Appendix A). The participants were asked to record the frequency of their current intake of each item according to eight categories: consumed rarely, once a month, two to three times a month, once a week, two to three times a week, four to six times a week, once a day, or more than two times a day. This FFQ has been used in several epidemiological studies on dietary factors related to chronic diseases [30][31][32] and health behaviors [33]. In our previous study, the FFQ was validated as a reference standard in 1401 participants from the source cohort based on a comparison with three-day dietary records. Cross-classification between the two methods of the distribution of subjects by tertiles indicated good agreement, ranging from 38% to 96% depending on the food group and nutrient [30].
Six questionnaires examining dietary habits were also assessed: meal frequency (three times a day, two times a day, irregular); breakfast frequency (always, often, not at all); meal time regularity (always regular, often regular, always irregular); meal speed (slow, average, fast); frequency of overeating (less than once a week, two to three times a week, more than four times a week); and the use of additional spices such as soy sauce, table salt, soybean paste, or hot pepper paste (often, sometimes, never).

Statistical Analysis
The characteristics of the study subjects were compared using Student's t-test, chi-square test, or Fisher's exact test. As the distribution of intake frequencies of each food item was skewed, the FFQ items were reclassified into two categories, low and high consumption, based on the median distribution of non-cases. A Cox proportional hazards model with person-years as the underlying time metric was used to evaluate the hazard ratio (HR) and 95% confidence interval (CI) of breast cancer for each dietary factor, including 16 food groups and six dietary habits. Multivariate Model 1 was adjusted for only significant factors of our study subjects' general characteristics (Table 1), and Model 2 was adjusted for potential confounders according to other breast cancer researches [34,35]. Model 1 of all women was adjusted for age, smoking status (current, ex, non-smoker), education status (elementary school or less, middle school, high school, college or higher), and benign breast tumor history (yes, no). Premenopausal women were adjusted for age and smoking status (current, ex, non-smoker), and postmenopausal women was adjusted for age, education status (elementary school or less, middle school, high school, college or higher), and benign breast tumor history (yes, no). Model 2 was adjusted for age, BMI (<23.0, 23.0 to <25.0, ≥25 kg/m 2 ), family history of breast cancer (yes, no), smoking status (current, ex, non-smoker), alcohol consumption (current, ex, non-drinker), physical activity (yes, no), age at menarche (≤13, 14, 15, ≥16 years), parity (0, 1, 2, ≥3), oral contraceptive use (yes, no), and benign breast tumor history (yes, no) in premenopausal women. In all women, we additionally adjusted for hormone use and menopausal status (premenopausal, postmenopausal hormone users, postmenopausal hormone non-users, unknown menopausal status), and age at menopause (premenopausal, menopause at <46, 46-48, 49-51, ≥52 years, unknown menopausal status). In postmenopausal women, we additionally adjusted for hormone use (yes, no) and age at menopause (<46, 46-48, 49-51, ≥52 years). All statistical analyses were performed using SAS software (version 9.4, SAS Institute, Cary, NC, USA), and the level of significance was set at p < 0.05.

Results
A total of 5046 Korean women were eligible for this analysis. During a mean follow-up of 9.46 years, 72 women were diagnosed with breast cancer. The subjects with breast cancer had a significantly higher educational level in the all group (college or higher: 58.3% vs. 39.5%, p = 0.01) and the postmenopausal group (college or higher: 56.5% vs. 270%, p = 0.04), were more likely to be ever-smokers in the all (22.2% vs. 7.8%, p < 0.001) and premenopausal groups (27.1% vs. 9.6%, p < 0.001), and more frequently had a history of benign breast tumor in the all (9.7% vs. 3.7%, p = 0.02) and postmenopausal groups (21.7% vs. 3.3%, p = 0.001) than did non-cases (Table 1). Table 2 shows the intake frequencies of the 16 food groups based on the short-form FFQ and their association with breast cancer risk in all women. Consumption of grilled meat was associated with a significantly higher risk of breast cancer in the age-adjusted model, and this trend was maintained in multivariate Model 1 (HR 1.66, 95% CI 1.03-2.68) and Model 2 (HR 1.77, 95% CI 1.09-2.85). High-cholesterol food intake was positively associated with a higher risk only in the age-adjusted model (HR 1.71, 95% CI 1.03-2.85) and multivariate Model 2 (HR 1.69, 95% CI 1.01-2.82). Among the premenopausal women, there were no differences between the cases and non-cases (Table 3). In postmenopausal women, grilled meat consumption was associated with an elevated risk of breast cancer in both multivariate Model 1 (HR 2.41, 95% CI 1.05-5.54) and Model 2 (HR 3.06, 95% CI 1.31-7.15) ( Table 4).  (Table 6), but not in postmenopausal women (Table 7).

Discussion
This prospective cohort study investigated the impact of dietary factors on the risk of breast cancer. After adjusting for confounding factors, grilled meat intake was proportionally associated with breast cancer risk in the all and postmenopausal groups, high cholesterol food intake was associated with a higher risk of breast cancer in all women, and meal irregularity was positively associated with breast cancer risk in the all and premenopausal groups.
The higher risk of breast cancer associated with grilled meat consumption is speculated to be due to carcinogenic mutagens such as heterocyclic amines (HCAs) and polycyclic aromatic hydrocarbons (PAHs), which are highly abundant in meat cooked at high temperatures, especially that which is grilled or barbecued [36,37]. HCAs are formed when creatinine, amino acids, and sugars present in meat muscles react at high temperatures [38]. PAHs are produced on or near the surface of meat when meat is cooked directly over an open flame and fat is pyrolyzed or when imperfectly combusted carbon and hydrogen from the fat fall onto hot coals and produce smoke [39]. One of the most common causes of PAH exposure in females is grilled food intake [40]. Although a few studies have shown a non-significant association of grilled meat consumption with breast cancer risk in both all women [41] and postmenopausal women [42], a similar trend to the one observed in this study was identified in several other studies. In case-control studies, grilled meat was associated with a higher risk of breast cancer in all women [43,44]. In postmenopausal women, a prospective study showed a borderline significantly higher breast cancer risk associated with the consumption of grilled/pan-fried/well-done meat [45], and a case-control study also showed a higher risk for total lifetime intake of grilled/barbecued beef, pork, and lamb [46]. The association of grilled meat with a higher risk of breast cancer in postmenopausal women may be due to differences in estrogen metabolism pathways based on menopausal status [41,47]. Before menopause, the ovaries are the major estrogen sites [48], whereas after menopause, adipose tissue plays a crucial role in synthesizing estrogen [47]. However, the mechanisms of the association between menopausal status and breast cancer and the interactions with diet remain unclear [49]. Fried foods also generate PAHs and HCAs [50], but they were not found to be associated with breast cancer risk in our studies. Previous studies showed inconsistent outcomes. One cohort study indicated that pan-fried meat was not significantly associated with breast cancer risk in postmenopausal women [42], although in other case-control studies, fried meat was associated with a higher risk in all women [43,44].
Other byproducts of grilled meat are advanced glycation end products (AGEs), which are created from a Maillard or browning reaction-non-enzymatic glycosylation of reducing sugars interacting with free amino groups of proteins, lipids, or nucleic acid [51,52]. By cross-linking with body proteins or binding with cell surface receptors, AGEs induce oxidative stress and inflammation and are therefore associated with dietary-related chronic diseases, including cancer [51]. N -carboxymethyllysine (CML) AGE is a biologically and chemically well-defined marker for analysis [53]. One study analyzed CML levels in 549 foods and concluded that, although AGEs are naturally present in animal foods, high-temperature cooking methods, especially frying, broiling, grilling, and roasting, produced more AGEs [51]. Some studies have used this database for analysis, and higher dietary CML was associated with pancreatic cancer in men [53], Alzheimer's disease [54], and Barrett's esophagus [55]. However, previous studies regarding breast cancer and AGEs have been limited to AGE receptor status [56,57] and the level of AGEs accumulated in breast tumors [58] to predict breast cancer progression rather than dietary AGE associations.
High-cholesterol food intake was associated with a higher breast cancer risk in the all women in the age-adjusted model and in multivariate Model 2, but this association was not found in multivariate Model 1. The role of dietary cholesterol in breast cancer remains unclear [59,60]. However, several in vivo and in vitro experiments have found that cholesterol may function as a signaling molecule in cancer cells as a cholesterol metabolite, 27-hydroxycholesterol (27HC), was found to act as an antagonist blocking estrogen receptor (ER) activation in the cardiovascular system and also functions as an agonist activating ER+ breast cancer [61]. Therefore, when cholesterol is converted to 27HC, it may induce ER+ breast cancer growth [61]. In several case-control studies, a study primarily consisting of postmenopausal women and two studies of all women showed that excessive cholesterol intake was associated with a higher risk of breast cancer [62][63][64]. In a cohort study, the baseline premenopausal women who remained premenopausal at the time of the last biennial questionnaire before censoring or the end of the follow-up also revealed that dietary cholesterol was associated with a higher risk of breast cancer [65]. However, in several cohort studies, no associations were found in all women [66][67][68] or postmenopausal women [69,70]. When the studies on dietary cholesterol and breast cancer risk were pooled in a meta-analysis, a negative association was found with breast cancer risk in all and premenopausal women [71].
In our study, the animal fat-rich food and fast food groups primarily consisted of processed meat products. Nitrate and nitrite, which are abundant in processed meats, are precursors of N-nitroso compounds (NOCs), which are potential carcinogens [72]. However, our study found no association between these food groups and breast cancer risk. Several previous studies also showed inconsistent trends. Processed meat intake was found to be associated with a higher risk of breast cancer among all women in case-control [41] and cohort studies [72,73], in postmenopausal women in nested case-control studies [74], and in both all women and in all subgroups of pre-and postmenopausal women in a cohort study [19]. By contrast, no association was observed between processed meat intake and the likelihood of breast cancer in all women in case-control [75] and cohort studies [76], in postmenopausal women in cohort studies [42,77], and in both all and pre-and postmenopausal women in cohort studies [78,79].
Light-colored, green-yellow, and salted vegetables and fruits were included in the category of fruits and vegetables. Although these food groups showed non-significant associations with the risk of breast cancer, several studies have indicated significant associations of foods other than salted vegetables. Vegetable intake may decrease the risk of breast cancer, but no association has been identified for fruits [22,80,81]. Similar results were reported in a meta-analysis of Korean studies as light-colored and green-yellow vegetable intake reduced the risk of breast cancer in all women, whereas fruit intake did not [29]. Among postmenopausal women reporting the highest lifetime intake of grilled/barbequed meat, those with the lowest intake of fruit and vegetables were associated with a significantly higher total lifetime risk [46]. However, a Korean case-control study found that the consumption of pickled vegetables was associated with a higher breast cancer risk in all women [81]. Other plant-based foods and soybean and its products were considered healthy protein foods. The intake of soy foods, including total soy products, soybean curd, and soymilk, showed a decreased risk of breast cancer in a Korean meta-analysis [29]. Our study observed no association of seafood, including seaweed and fish in the healthy protein food group and bonefish, with the risk of breast cancer. In a meta-analysis in Korea, a reduced risk of breast cancer was reported for seaweed, but not for fish consumption [29]. However, a case-control study in Korea identified a decreased risk with a combined vegetable-seafood dietary pattern [82].
We found no association between cereal consumption and breast cancer risk. Several cohort studies of postmenopausal women also showed non-significant associations of whole grain [83] and refined grain consumption [83,84], but conflicting results regarding bread and cereal were found in a case-control study [85]. Sweet foods also showed non-significant associations in our study, but other case-control studies reported different results. One study indicated a borderline significantly higher risk of early-stage breast cancer in the group with the highest intake of sweets [86]. Furthermore, another study that classified sources of sweets found that a higher intake of biscuits, sugar, and chocolate as food items and both desserts and sugars as a food group was associated with a higher risk of breast cancer and hypothesized that this was due to reduced diversity in the diet and increased meal frequency [87].
Dairy foods were not associated with breast cancer risk in our study. In a meta-analysis of 12 prospective studies, total dairy food intake was inversely associated with breast cancer, with low-fat dairy intake showing stronger associations in premenopausal women in particular [88]; however, a meta-analysis of Korean studies reported non-significant results [29]. The intake of caffeinated drinks also exhibited non-significant associations in our study. In a 22-year follow-up cohort study, no significant association of caffeinated drink intake with breast cancer risk was identified in all women, but a slightly reduced risk was observed in postmenopausal women [89].
Of the six dietary habits assessed, meal irregularity was associated with a higher risk of breast cancer in all and premenopausal women. However, no previous studies have addressed the direct association between regular eating and breast cancer risk. One study identified an interaction between eating frequency and an inflammatory biomarker, C-reactive protein (CRP), as a putative factor associated with breast cancer. This study showed that a higher eating frequency was inversely proportional to serum CRP levels, suggesting that eating more frequently may lower systemic inflammation and subsequently reduce breast cancer risk [90].
Adding spices to food was not associated with breast cancer risk. In a case-control study, adding table salt was a non-significant factor [75]. Psychological interruption may have affected this result. Subjects already recognize that a higher intake of sodium can increase the risk of chronic diseases, including cancer, and those who are concerned about their health may consequently select a lower frequency in the questionnaire. However, although the HR was not significant, the P for trend was significant in all women after adjusting for covariates. This result indicates that adding less spice may decrease the risk of breast cancer.
Our study has certain strengths: it is a cohort study with a follow-up period of more than nine years. We examined dietary factors at baseline when the subjects did not have breast cancer, so the reports were not subject to selection or recall bias. Additionally, as breast cancer is known to have a long latency, it is important to evaluate dietary patterns prior to the development of disease when environmental exposure may play a strong role [91], indicating that a cohort design would be more meaningful. Additionally, to our knowledge, this is the first prospective study conducted in Korea to address the associations between dietary habits and the risk of breast cancer.
Several limitations should also be mentioned: (1) Our FFQ data cover only baseline information and cannot reflect changes in dietary factors after baseline; (2) Additionally, residual confounding variables may have affected our results. Baseline nutritional status and genetic susceptibility may interact with diet and influence its relationship with breast cancer [25]; however, we could not adjust for these factors; (3) The food dataset included only frequency information, not portion size. Therefore, the intake amount could not be calculated, making it difficult to divide the study subjects into similar groups by the amount of intake of each food item; (4) Although the follow-up period was longer than nine years, there was still an insufficient number of breast cancer cases to infer strong statistical power; (5) The FFQ was limited to 16 items, making it difficult to ensure that all items of certain food groups were measured. Food items in a group may have different associations with breast cancer risk (e.g., red meat versus soy). A potential solution for this problem would be dividing the food group into corresponding food sources to specify their precise influences (e.g., rather than evaluating the consumption frequency of the healthy protein group, it is more appropriate to identify the intake frequency of each food item, such as soybean products, lean meat, fish, and egg whites); (6) Stratification of breast cancer by specific characteristics should be considered in further studies, especially hormone (estrogen and progesterone) receptor and human epidermal growth factor receptor 2 (HER-2) status [25].

Conclusions
In conclusion, this study showed that breast cancer risk was proportionally associated with grilled meat intake in all women and in postmenopausal women, with high-cholesterol food intake in all women, and with irregular eating in all and premenopausal women. Further studies with longer follow-up periods that include information on portion size, hormone receptor status, levels of each carcinogen in grilled meat, and classification of food groups by the source are needed.