Folate Intake and Risk of Breast Cancer by Estrogen and Progesterone Receptor Status in a Swedish Cohort

Folate is a water-soluble B vitamin that has been postulated to be associated with cancer risk. Biological reactions requiring folate, known as one-carbon metabolism, include DNA synthesis and DNA methylation (1, 2). An inadequate folate intake may predispose individuals to cancer through aberrations in these processes. In fact, epidemiologic studies have suggested that a low folate intake or status may increase the risk of certain cancers, and the evidence is most convincing for gastrointestinal cancers (3, 4). Epidemiologic studies of folate intake in relation to risk of breast cancer have been inconclusive. Whereas retrospective case-control studies generally have shown an inverse association between folate intake and breast cancer risk, the majority of prospective studies have found no overall association (5, 6). However, some prospective studies (7-12) have suggested that an adequate folate intake may be important in the prevention of breast cancer specifically among women with high alcohol consumption.

The relation between folate intake and breast cancer may vary by hormone receptor status of the breast tumor (13). However, only four prospective studies have reported findings for folate and breast cancer risk by hormone receptor status and the results are mixed (14-17).

To evaluate further the potential role of folate intake in the development of breast cancer, we analyzed data from a population-based cohort of Swedish women. We examined whether the association varied by estrogen receptor (ER) and progesterone receptor (PR) status of the breast tumor, alcohol consumption, or menopausal status. Whereas folic acid (synthetic form of folate) fortification of grain products was mandated in the United States in 1998 (18), no such fortification program has been initiated in Sweden, which allows an analysis that is not influenced by fortification.

The Swedish Mammography Cohort was established in 1987 to 1989 in Västmanland County and in 1988 to 1990 in Uppsala County in central Sweden. All women born between 1917 and 1948 in Västmanland County and between 1914 and 1948 in Uppsala County received a mailed invitation to be screened by mammography. Enclosed with this invitation was a six-page questionnaire that elicited information on diet, body size, reproductive factors, family history of breast cancer, and others; a completed questionnaire was obtained from 66,651 women, representing 74% of the source population. In the late autumn of 1997, all cohort members who were still alive and residing in the study area received a new questionnaire that was expanded to include about 350 items concerning diet and other lifestyle factors (including smoking); 39,227 (70%) women completed the second questionnaire.

From the baseline cohort of 66,651 women, we excluded women with an incorrect or a missing national registration number as well as those lacking date on the questionnaire, date of moving out of the study area, or date of death. After additional exclusion of women with implausible values for total energy intake (3 SDs from the mean value for log_e-transformed energy intake) and those with a cancer diagnosis (except nonmelanoma skin cancer) before baseline, the analytic cohort consisted of 61,433 women. Among these women, 26,403 were premenopausal, 2,574 were perimenopausal, and 32,456 were postmenopausal. For analyses using information from the second questionnaire, 36,664 women were eligible after excluding those with implausible energy intake on the second dietary questionnaire as well as those who had been diagnosed with cancer between baseline and January 1, 1998. The study was approved by the ethics committees at the Uppsala University Hospital and the Karolinska Institutet.

A food frequency questionnaire (FFQ) with 67 and 96 food items was used to assess diet at baseline and in 1997, respectively. In these questionnaires, women were asked to report how often, on average, they had consumed each food item during the previous 6 months (baseline FFQ) or the previous year (1997 FFQ). The questionnaires had eight mutually exclusive predefined categories for frequency of consumption, ranging from “never/seldom” to “≥3 times per day” (1997 FFQ) or “≥4 times per day” (baseline FFQ). Dietary folate intake was calculated by multiplying the frequency of consumption of each food item by its folate content per age-specific serving using food composition values obtained from the Swedish National Food Administration Database (19). The age-specific serving sizes were based on mean values obtained from 213 randomly selected women from the study area who weighed and recorded their food intake for an average of 27.8 days. The 1997 questionnaire also asked for information on the use of dietary supplements, including specific folic acid supplements, B-vitamin supplements, and multivitamins. We calculated total folate intake by summing intake of folate from foods and dietary supplements.

The validity of the baseline dietary questionnaire was assessed previously by comparing responses from the FFQ with responses from four 1-week dietary records (3-4 months apart) among 129 women randomly chosen from the cohort.³

We ascertained histologically confirmed incident cases of invasive breast cancer by linkage with the national and regional Swedish cancer registers. The completeness of cancer follow-up was estimated to be almost 100% (21). Information on ER and PR status of breast tumors was obtained by reviewing pathology laboratory work logs stored at Uppsala University Hospital (from 1987 to 1994) and by linkage with the clinical database (the Quality Register) at the Regional Oncology Centre in Uppsala (from January 1992 to December 2007), which was based on the patients' original medical records. ER and PR status was evaluated by using an Abbott immunoassay until 1997 and an immunohistochemical method thereafter. Cases with ≥0.1 fmol/μg cytosol DNA were considered hormone receptor positive when using the Abbott immunoassay. By the immunohistochemical method, cases were considered as receptor positive when the percentage of positive cells was ≥10% and receptor-negative when the percentage of positive cells was <10%. The Department of Pathology and Cytology at Uppsala University Hospital and Västerås Central Hospital were involved in this evaluation. Information on dates of death for deceased participants was obtained from the Swedish Death Registry.

Person-time of follow-up was calculated from the date of enrollment until the date of breast cancer diagnosis, death from any cause, or December 31, 2007, whichever occurred first. In analyses of ER/PR status, for women in Västmanland County, person-time of follow-up was counted from January 1998 because routine evaluation of ER and PR status was implemented in Västmanland County first in 1997. For analyses using data from the second questionnaire, the follow-up of all women began in January 1998. Folate intake was adjusted for total energy using the residual method (22) and categorized into quintiles.

To account for changes in diet during follow-up and to better represent long-term dietary intake, we used a cumulative average approach (23). Specifically, breast cancer incidence from baseline to 1997 was related to estimated folate intake at baseline, and breast cancer incidence from 1998 to December 2007 was related to the average folate intake at baseline and in 1997; for women who did not complete the second questionnaire, data from the baseline questionnaire were used for the whole follow-up. We also related folate intake in 1997 to breast cancer incidence from 1998 to December 2007.

Cox proportional hazards models (24) were used to estimate rate ratios (RR) and 95% confidence intervals (95% CI) for the association between folate intake and breast cancer risk. To control as finely as possible for age and calendar time, and possible two-way interactions between these two timescales, we stratified the models by age in months at start of follow-up and year of enrollment. In multivariate models, we further adjusted for education (primary school, high school, university) and potential risk factors for breast cancer, including body mass index (<18.5, 18.5-24.9, 25-29.9, ≥30 kg/m²), height (in cm), parity (nulliparous, 1-2, ≥3), age at first birth (nulliparous, <26, 26-30, ≥31 years), age at menarche (≤12, 13, ≥14 years), age at menopause (<51, ≥51 years), use of oral contraceptives (ever/never), use of postmenopausal hormones (ever/never), family history of breast cancer (yes/no), and intakes of alcohol (nondrinkers, <3.4, 3.4-9.9, ≥10.0 g/d) and total energy (kcal/d). We tested the proportional hazards assumption using the likelihood ratio test and found no departure from the assumption.

Because alcohol intake may modify the relation between folate intake and breast cancer (5), we conducted analyses stratified by alcohol intake (nondrinkers, 0.1-9.9, ≥10.0 g/d). We also performed analyses by menopausal status. To test for trend, we assigned the median value to each quintile of folate intake and treated this value as a continuous variable in the model. All statistical analyses were conducted using SAS version 9.1 (SAS Institute). All P values were two-sided.

In this cohort of Swedish women, the mean ± SD daily intake of dietary folate at baseline was 234 ± 50 μg, ranging from 177 ± 18 μg in the lowest quintile to 315 ± 49 μg in the highest quintile. Baseline distributions of risk factors for breast cancer by quintiles of dietary folate intake are presented in Table 1. Compared with women with a low folate intake, those with higher intakes were slightly older and more likely to have a postsecondary education and to have used oral contraceptives and postmenopausal hormones. Other characteristics did not vary appreciably across quintiles of folate intake.

Among 61,433 women followed-up for 1,071,164 person-years (mean, 17.4 years), a total of 2,952 incident cases of invasive breast cancer were diagnosed. Information on ER and PR status was available for 2,062 cases (information about ER/PR status was available for women in Västmanland County first in 1997). Among them, 1,286 (62.4%) cases were ER+/PR+, 417 (20.2%) were ER+/PR-, 266 (12.9%) were ER-/PR-, and 93 (4.5%) ER−/PR+.

The association between long-term dietary folate intake (assessed at baseline and in 1997) and incidence of breast cancer are shown in Table 2. We observed no association between folate intake and risk of total breast cancer or ER+/PR+ or ER-/PR- tumors. However, there was an inverse association between folate intake and risk of ER+/PR- breast cancer (P_trend = 0.01). The association between folate intake and breast cancer was not modified by alcohol intake (Table 3). In analyses stratified by menopausal status, the multivariate RRs of total breast cancer for an increment of 100 μg/d folate intake were 1.04 (95% CI, 0.93-1.17) among premenopausal women (n = 1,244 cases) and 0.97 (95% CI, 0.88-1.07) among postmenopausal women (n = 1,584 cases). The corresponding RRs for ER+/PR- breast tumors were 0.87 (95% CI, 0.65-1.16) for premenopausal women (n = 209 cases) and 0.71 (95% CI, 0.53-0.95) for postmenopausal women (n = 185 cases). Excluding all cases diagnosed during the first 3 years of follow-up did not change the results materially.

We used information from the second questionnaire to examine the relation between total folate intake (from foods and supplements combined) and breast cancer risk (information on supplement use was not available at baseline). In 1997, 24.5% of participants reported use of multivitamins and 1.0% used folic acid supplements. During 346,162 person-years of follow-up, from January 1998 to December 2007, we ascertained 1,008 incident invasive breast cancer cases. The multivariate RR comparing the highest quintile of total folate intake (mean intake, 552 μg/d) with the lowest quintile (mean intake, 192 μg/d) was 1.06 (95% CI, 0.87-1.28) for total breast cancer and 0.78 (95% CI, 0.53-1.15) for ER+/PR− tumors. Total folate intake was not associated with risk of ER+/PR+ or ER−/PR− tumors.

In this large population-based prospective cohort of Swedish women, we found no association between folate intake and total breast cancer risk. However, there was an inverse association between folate intake and risk of developing ER+/PR- breast cancer. The results did not vary appreciably by alcohol intake or menopausal status.

Our findings are consistent with recent meta-analyses of prospective studies in which no overall association between folate intake or blood folate levels and risk of breast cancer was observed (5, 6). Since those meta-analyses, three additional prospective studies on folate and breast cancer risk have been published (15, 17, 25). Among those, a statistically significant inverse association between dietary and total folate intake and breast cancer risk was found in a cohort of 11,699 postmenopausal Swedish women (including 392 cases) followed up for 9.5 years (25). No association between dietary or total folate intake and risk of breast cancer was observed in the Nurses' Health Study II cohort of 90,663 premenopausal women (including 1,032 cases; ref. 17) or in a nested case-control study (848 cases and 848 controls; ages ≥45 years) within the Women's Health Study (15). However, in the Women's Health Study, plasma folate concentrations were significantly positively associated with the risk of premenopausal breast cancer (15). The inconsistent findings may be due to different validity of the dietary measurement. The validity of estimated folate intake was quite high in the Swedish Malmö Diet and Cancer cohort in which an inverse relation between folate intake and breast cancer risk was observed (25).

Two previous prospective studies observed an inverse association between folate intake and breast cancer risk specifically among women with moderate to high alcohol consumption (>14 g/d; refs. 7, 8). Other prospective studies have suggested that adequate folate intakes may attenuate the excess risk of breast cancer associated with high alcohol consumption (9-12). In the present study, the association between folate intake and breast cancer risk did not vary by alcohol consumption. However, because alcohol consumption in our Swedish population was low (alcohol intake in the highest category was ≥10 g/d; median intake in that category was 12.4 g/d), we were not able to evaluate the association among women with higher alcohol consumption.

Folate plays an important role in DNA methylation (1). Hence, low folate status may alter DNA methylation and thereby influence gene expression and DNA integrity and stability (1). Zhu et al. (13) hypothesized that methyl-deficient diets (low in folate and methionine) may be associated primarily with breast cancer with methylated ER gene CpG island. ER CpG island methylation has been related to lack of ER gene expression in ER- breast tumors (26, 27). Findings from our study do not support the hypothesis that folate intake is associated with risk of ER- breast cancer. Instead, we observed an inverse association between folate intake and risk of ER+/PR− tumors. This finding was unexpected and may be due to chance.

Four previous prospective studies have examined whether the association between folate and breast cancer varies according to ER and PR status of the breast tumor, and the results are inconclusive. In the Nurses' Health Study, folate intake was inversely associated with the risk of developing ER- but not ER+ tumors (14). In the Iowa Women's Health Study (16), there was no overall relation between folate intake and ER- or ER+ tumors; however, an increased risk of ER- tumors was observed among women who had a low total folate intake and a high alcohol intake. In contrast, in the Women's Health Study, plasma folate concentrations were positively associated with risk of developing ER+ or PR+ breast tumors (15). No association between folate intake and ER- breast cancer was found in the Nurses' Health Study II (17).

This study has several strengths, including the prospective and population-based design, a large sample size, detailed information on diet, and information on hormone receptor status. The prospective design precluded recall bias and the almost complete follow-up of the study population through linkage with computerized population-based registers of cancer and deaths minimizes the concern that our results have been affected by differential loss to follow-up. Another strength is that exposure data were collected from participants at two time points. Repeated assessments of diet were used in our main analyses, which provide a better measure of long-term intake than a single measurement of diet at baseline.

Our study is limited by the fact that dietary intake was assessed with a self-administered FFQ, which will inevitably lead to some error in the measurement of folate intake. Because of the prospective design, any misclassification due to measurement error is likely to be nondifferential and would tend to attenuate the associations. We cannot exclude that the lack of association between folate intake and total breast cancer risk was due to misclassification. However, we have observed previously inverse relations between dietary folate intake and risk of ovarian (28), pancreatic (29), and colorectal (30) cancers in this study cohort, suggesting that our assessment of dietary folate is sufficiently accurate to detect true relationships. Furthermore, by updating and averaging the repeated measurements of diet, we reduced within-person fluctuations and took into account changes over time, which should improve validity. Another limitation is the relatively low folate intake in this study population. Thus, we may have overlooked an association between folate intake and total breast cancer if only relatively high folate intakes are associated with risk of breast cancer. Finally, although we adjusted for major known risk factors for breast cancer, we cannot exclude the possibility of residual confounding from these variables or from other uncontrolled dietary or nondietary factors. Nevertheless, the similar results in the age-adjusted and multivariate-adjusted models argue against residual confounding.

In conclusion, results from this prospective cohort study of Swedish women do not support an association between folate intake and total breast cancer risk but suggest that folate intake may be inversely associated with ER+/PR- tumors. Further studies are needed to elucidate the possible association between folate and hormone receptor-defined breast cancer.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.