Folate Intake and Risk of Breast Cancer Characterized by Hormone Receptor Status

A majority of breast tumors express the estrogen receptor (ER) gene and synthesize ER protein, but about one third of breast tumors lack ER protein, rarely respond to hormone therapy, and are associated with a poorer prognosis (1). Gene expression can be modulated by epigenetic mechanisms; alterations in DNA methylation patterns are frequently seen in tumor cells, with wide areas of hypomethylation along the genome accompanied by localized areas of hypermethylation at specific sites, such as cytosine-guanine-rich areas, termed CpG islands (2-5). Lack of ER gene transcription, as seen in ER⁻ breast tumors, may be due to extensive methylation of the ER gene CpG island in the 5′ regulatory region and first exon (6-11). Although hypermethylation of the CpG island was found to be involved in silencing progesterone receptor (PR) gene expression (9), reactivation of PR gene expression was independent of demethylation (12), suggesting that methylation may play a smaller role in regulating PR gene expression.

Folate in the form of methyltetrahydrofolate participates in regeneration of methionine by providing a methyl group to homocysteine to form methionine and eventually S-adenosylmethionine (13). S-adenosylmethionine can serve as a methyl donor for DNA methylation (14). Low folate status may thus alter DNA methylation and thereby influence gene stability and expression (14). Thus, as hypothesized by Zhu and Williams (15), low folate status may be associated primarily with ER gene–negative breast tumors. Folate also functions as a coenzyme in the synthesis of purines and thymidylate for DNA. Diminished folate status may result in misincorporation of uracil into DNA, leading to chromosome breaks and disruption of DNA repair (14, 16, 17).

Several epidemiologic investigations (18-27) including three large prospective cohorts, the Nurses' Health Study (18, 19), the Canadian National Breast Screening Study (20), the Iowa Women's Health Study (21), suggest that adequate folate intake or blood levels may be important in the prevention of breast cancer, particularly among women who regularly consume alcohol (18-22, 26). Alcohol is a known folate antagonist (28, 29) and thus could plausibly increase the requirement for folate intake. Because of limited progress in the identification of risk factors for ER⁻ breast cancer, we conducted a prospective analysis to evaluate folate intake in relation to breast cancer risk according to ER status in the Nurses' Health Study.

The Nurses' Health Study was established in 1976 when 121,700 female registered nurses ages 30 to 55 years living in 11 states completed a mailed questionnaire about their medical history and lifestyle. Follow-up questionnaires have been sent to cohort members biennially to update health-related information and to ascertain newly diagnosed diseases including breast cancer. Dietary information was collected using the food frequency questionnaires in 1980, 1984, 1986, 1990, 1994, and 1998. The study was approved by the Use of Human Subjects in Research Committee at the Brigham and Women's Hospital.

The questionnaires assessed the average consumption over the past year of a specific amount of each food and allowed nine responses, ranging from “never” to “six or more times per day.” Nutrient intake was calculated by multiplying the frequency response by the nutrient content of the specified portion sizes. Total alcohol intake was the sum of the alcohol content contributed from beer, wine, and liquor, assuming 12.8 g of ethanol for 360 mL (12 oz) of beer, 11.0 g for 120 mL (4 oz) of wine, and 14.0 g for 45 mL (1.5 oz) of liquor. Duration, brand, and type of multivitamin supplement use were updated biennially. A comprehensive database on the content of vitamins of the multivitamin preparations was developed.

The validity and reliability of the food frequency questionnaires used in the Nurses' Health Study have been described in detail elsewhere (30-33). In a sample of 188 participants, the correlation coefficients between folate intake calculated from the 1980 dietary questionnaire and erythrocyte folate concentrations in 1987 were 0.55 for total folate and 0.38 for dietary folate (34). In a sample of 712 participants who provided blood specimens during 1989 to 1990 and served as control subjects in a nested case-control study of breast cancer, the correlation coefficients between folate intakes calculated from the 1990 food frequency questionnaire and plasma levels were 0.55 for total folate and 0.35 for dietary folate (19). In addition, in a sample of 173 participants, alcohol intake calculated from the 1984 dietary questionnaire was highly correlated with the 1980 four 1-week diet records (r = 0.84) and plasma high-density lipoprotein cholesterol levels, which are known to be influenced by alcohol (r = 0.40; ref. 35).

Incident cases of breast cancer were first identified by self-report on each biennial questionnaire. Deaths in the cohort were identified by reports from family members, the postal service, and a search of the National Death Index, and we estimate that 98% of all deaths were identified (36). Women (or their next of kin if they had died) who reported breast cancer were asked for permission to obtain hospital records and pathology reports relevant to diagnosis. Self-reported breast cancer cases were confirmed by medical personnel who reviewed the records. Information on ER and PR status was extracted from the records. A total of 4,422 invasive cases of breast cancer were identified from 1980 to 2000 among 88,744 women who completed the baseline food frequency questionnaire, had total energy intake within the range of 500 to 3,500 kcal/d, and were free of cancer in 1980. Of these cases, 2,812 (63.6%) were ER⁺, 985 (22.3%) were ER⁻, 2,256 (51.0%) were PR⁺, and 1,361 (30.8%) were PR⁻; ER status was not available for 625 cases (14.1%) and PR status was not available for 805 cases (18.2%).

Person-years for each participant were calculated from the date of returning the 1980 questionnaire to the date of diagnosis of breast cancer, death, or May 31, 2000, whichever came first. Folate intake was adjusted for total energy using the residual method (33) and categorized into quintiles. The cumulative average intakes of folate and alcohol from all available questionnaires were used to best represent long-term intakes (37). In this analysis, we modeled the incidence of breast cancer in relation to the cumulative average of dietary intake from all available dietary questionnaires up to the start of each 2-year follow-up interval (37). For example, the incidence of breast cancer during the 1980 through 1984 time period was related to the dietary information from the 1980 questionnaire, and the incidence of breast cancer during the 1984 through 1986 time period was related to the average intake from the 1980 and 1984 questionnaires. The incidence rate was calculated by dividing the number of cases by person-years of follow-up in each category. The relative risk (RR) was calculated as the rate in a specific category divided by that in the reference category, with adjustment for 5-year age categories by the Mantel-Haenszel method (38). In multivariable analysis using pooled logistic regression with 2-year time increments (39, 40), we simultaneously adjusted for age (5-year age categories), period of follow-up (ten 2-year periods), age at menarche (≤12, 13, or ≥14 years), age at first birth (≤24, 25-29, or ≥30 years), parity (0, 1 or 2, 3, or 4, or ≥5 births), history of breast cancer in mother or a sister (yes or no), history of benign breast disease (yes or no), alcohol intake (0, 0.1-4.9, 5.0-14.9, 15.0-29.9, or ≥30 g/d), body mass index at age 18 years (<20, ≥20 to <22, ≥22 to <24, ≥24 to <27, or ≥27 kg/m²), weight change from age 18 years to present (loss >2.0, loss or gain of 2.0, gain 2.1-5.0, gain 5.1-10.0, gain 10.1-20.0, gain 20.1-25.0, or gain >25.0 kg), height in inches (continuous), age at menopause (<45, 45-49, 50-54, or ≥55 years), menopausal status (premenopausal, postmenopausal, or uncertain), and postmenopausal hormone use (premenopausal, uncertain, never, past <5 years, past ≥5 years, current <5 years, current ≥5 years). Information on age, weight change from age 18 years to present, age at menopause, menopausal status, postmenopausal hormone use, alcohol intake, family history of breast cancer, and history of benign breast disease was updated every 2 to 4 years. For all RRs, we calculated 95% confidence intervals (95% CI).

In additional analyses, we stratified by alcohol intake (<15 and ≥15 g/d, approximately one drink) because alcohol intake has been shown to modify the association between folate and breast cancer risk in our earlier studies (18, 19). All analyses were conducted according to ER or PR status of breast tumors. Tests for trend were done by using the median value of each category of the cumulative average nutrient intake as a continuous variable. A test for the difference in the estimates of folate intake (median values for each quintile as a continuous variable) for ER⁻ versus ER⁺ breast tumors was conducted by using the squared T statistic, which has a χ² distribution with one degree of freedom. All P values were two tailed.

Tables 1 and 2 present RRs and 95% CIs for intakes of total folate (from foods and supplements) and dietary folate (from foods only) by ER or PR status of breast tumors. The RRs from age-adjusted analyses were similar to those from multivariable analyses with additional adjustments for other risk factors for breast cancer. Higher intake of total folate was associated with significantly lower risk of developing ER⁻ but not ER⁺ breast cancer (Table 1). The multivariable RRs and 95% CIs comparing the highest with the lowest quintile of total folate intake were 0.81 (0.66-0.99; P_trend = 0.03) for ER⁻ tumors and 1.00 (0.89-1.14; P_trend = 0.83) for ER⁺ tumors. The difference in results for ER⁻ versus ER⁺ breast tumors was marginally significant for total folate intake (P = 0.07). However, dietary folate intake was not associated with risk of developing ER⁻ breast tumors but was associated with significant but small increased risk of developing ER⁺ breast tumors. The multivariable RRs and 95% CIs comparing the highest to the lowest quintile of dietary folate intake were 0.97 (0.79-1.18; P_trend = 0.34) for ER⁻ tumors and 1.15 (1.01-1.30; P_trend = 0.05) for ER⁺ tumors. Excluding folate supplement users in the analysis for dietary folate intake did not appreciably change the results (data not shown). Neither total nor dietary intakes of folate were associated with risk of developing either PR⁻ or PR⁺ breast tumors (Table 2).

When we further examined the associations between folate intake and breast cancer risk stratified by alcohol intake, we found that the inverse association between total folate intake and ER⁻ breast cancer was mainly present among women consuming at least 15 g/d of alcohol (Table 3). The multivariable RRs and 95% CIs of ER⁻ tumors comparing the highest to the lowest quintile of total folate intake were 0.46 (0.25-0.86) among women consuming at least 15 g/d of alcohol and 0.88 (0.71-1.10) among women consuming less than 15 g/d of alcohol. The difference in estimates for ER⁻ versus ER⁺ breast tumors among women consuming at least 15 g/d of alcohol was marginally significant for total folate (P = 0.07). Dietary folate intake was also associated with marginally significant lower risk of developing ER⁻ breast cancer among women consuming at least 15 g/d of alcohol (multivariable RR, 0.66; 95% CI, 0.36-1.21; P_trend = 0.06; top versus bottom quintile). However, dietary folate intake was associated with marginally significant higher risk of developing ER⁺ breast cancer among women consuming <15 g/d of alcohol (multivariable RR, 1.16; 95% CI, 1.02-1.33; P_trend = 0.04; top versus bottom quintile). Neither total nor dietary intakes of folate were associated with risk of developing either PR⁻ or PR⁺ breast tumors even according to alcohol consumption (Table 4).

In this large prospective study, higher total folate intake was associated with significantly lower risk of developing ER⁻ breast cancer among women. The inverse association between total folate intake and risk of ER⁻ breast cancer was particularly strong among women who regularly consumed alcohol. The prospective design in this study minimizes the possibility that our findings are due to methodologic biases. Because comprehensive risk factors for breast cancer are available in this study and controlling for them had minimal effect on the RRs, our results are also unlikely to be explained by residual confounding by these factors.

Folate plays an important role in DNA methylation (14). ER CpG island methylation has been associated with lack of ER gene expression in breast cancer cell lines and primary breast tumors (6-11). In addition, absence of ER gene expression in ER⁻ breast cancer cells has been associated with increased mRNA level and activity of DNA methyltransferase, an enzyme that methylates cytosine to 5-methylcytosine (7). Upon treatment with demethylating agents, the ER CpG island in ER⁻ breast cancer cell lines is partially demethylated; such changes are correlated with reexpression of the ER gene and production of ER protein (8). In animal studies, folate/methyl-deficient diets increase the activity of DNA methyltransferase (41, 42). Furthermore, in a case-control study among African American women, lower folate intake was associated with 2-fold higher risk of breast tumors with a methylated ERα gene but not those with an unmethylated ERα gene (43). Our finding that total folate intake was primarily associated with lower risk of developing ER⁻ breast tumors is consistent with these experimental and epidemiologic data and supports the hypothesis that low folate status promotes breast carcinogenesis at least in part through its influence on DNA methylation. Thus, adequate folate intake may be primarily important for preventing ER⁻ breast tumors.

In contrast to total folate, data suggest a small positive association between dietary folate intake and risk of developing ER⁺ breast cancer, which was largely limited to women who consumed <15 g/d of alcohol. The difference in findings between total folate and dietary folate according to ER status is intriguing and merits further investigations. This small positive association associated with intake of dietary folate is unlikely to be explained by folate because this was not observed for intake of total folate, which contains folic acid supplements with much higher bioavailability (44). Alternative explanations are that dietary folate may merely serve as a marker for other factors that are correlated with dietary folate intake or it was a result of chance.

Several previous investigations including the Nurses' Health Study cohort (18, 19), the Canadian National Breast Screening Study cohort (20), the Iowa Women's Health Study cohort (21), and two case-control studies in Italy and Switzerland (22, 26) have found that the inverse association between folate intake or plasma levels and breast cancer risk was strong among women consuming alcohol intake. However, the American Cancer Society Cancer Prevention Study II Nutrition Cohort has reported no association between folate intake and risk of breast cancer even among women who consumed alcohol (45). Alcohol may perturb folate metabolism through reducing intestinal absorption, increasing renal excretion (29), inhibiting methionine synthase in the liver (46, 47), directly destroying folate (48), or interacting with tetrahydrofolate (46, 49). In the Iowa Women's Health Study, the only published cohort study that has examined the interaction of folate intake, alcohol, and risk of breast cancer by hormone receptor status, there was no appreciable association with risk of ER⁺ tumors at all levels of folate and alcohol, but there was an elevated risk of ER⁻ tumors among women who had low intake of total folate and high intake of alcohol (50). We also observed a similar pattern for the effects of total folate intake stratified by alcohol according to ER status. Both studies found that the association between folate intake and breast cancer risk did not seem to differ according to PR status, suggesting the effects of folate may be more specific for ER. The consistence in findings from two large cohorts suggests such findings are unlikely to be explained by chance.

A potential limitation of this study is that data on the presence or absence of hormone receptors were determined from laboratories affiliated with hospitals where breast cancer cases were diagnosed and not from a single reference laboratory. However, the measurement of hormone receptors has been standardized clinically. Due to the use of a prospective design, the measurement error associated with the use of hormone receptor data from various hospitals is likely to be nondifferential and would tend to weaken the associations. Another potential limitation of this study is that ER status was not available for 14.1% of cases and PR status was not available for 18.2% of cases. However, folate intake before diagnosis is unlikely to be the indication for the availability of hormone receptor status. This small proportion of cases without hormone receptor data is thus unlikely to affect the associations appreciably.

In summary, our findings support the hypothesis that higher folate intake reduces the risk of developing ER⁻ breast cancer. Ensuring adequate folate intake seems particularly important for women at higher risk of breast cancer due to alcohol consumption.