Folate, Vitamin B₆, Multivitamin Supplements, and Colorectal Cancer Risk in Women

Abstract

The authors evaluated associations between intakes of folate and vitamin B₆ and colorectal cancer risk among women enrolled in a randomized trial on aspirin and vitamin E in disease prevention. At baseline (1992–1995), 37,916 US women aged ≥45 years who were free of cancer and cardiovascular disease provided dietary information. During an average of 10.1 years of follow-up (through February 20, 2004), 220 colorectal adenocarcinoma cases were documented. Total folate and vitamin B₆ intakes were not significantly associated with the risk of colorectal cancer. However, dietary intakes of folate and vitamin B₆ were significantly inversely associated with colorectal cancer risk among women who were not taking supplements containing folate and vitamin B₆. Multivariable relative risks among women in the highest quintiles of intake versus the lowest were 1.16 (95% confidence interval (CI): 0.76, 1.79) for total folate, 1.14 (95% CI: 0.77, 1.69) for total vitamin B₆, 0.46 (95% CI: 0.26, 0.81) for dietary folate, and 0.69 (95% CI: 0.41, 1.15) for dietary vitamin B₆. The use of multivitamin supplements was not related to colorectal cancer risk. These findings suggest that higher dietary intakes of folate and vitamin B₆ may reduce the risk of colorectal cancer in women. An alternative explanation is that other factors related to dietary intakes of folate and vitamin B₆ account for the inverse associations.

Because of their roles in the maintenance of intracellular normal DNA synthesis and methylation (1–3), folate and vitamin B₆ have been hypothesized to be associated with reduced risk of colorectal cancer. Folate participates in regeneration of methionine, which is converted into S-adenosylmethionine, a methyl donor for DNA methylation (1, 4). Thus, folate deficiency can reduce the availability of S-adenosylmethionine for DNA methylation (1, 2) and may thereby influence gene expression (2). Aberrant DNA methylation patterns are frequently seen in colorectal tumors, with wide areas of hypomethylation along the genome accompanied by regional hypermethylation at specific sites, particularly areas rich in cytosine-guanine (termed “CpG islands”) (5). Folate and vitamin B₆ function as coenzymes in the synthesis of purines and thymidylate for DNA. Low levels of these vitamins may result in misincorporation of uracil into DNA, leading to chromosome breaks and disruption of DNA repair (2, 6, 7). In addition, recent data from cell culture and animal studies have suggested that vitamin B₆ may prevent the development of colon cancer through several other mechanisms, including the suppression of cell proliferation, oxidative stress, nitric oxide synthesis, and angiogenesis (8).

Although findings from prospective cohort studies and case-control studies generally suggest an inverse association between folate intake and risk of colorectal cancer, the data are not entirely consistent (9–28). In addition, data relating vitamin B₆ intake to the risk of colorectal cancer are limited. Thus, we conducted a prospective analysis of folate and vitamin B₆ intakes and the risk of colorectal cancer in a large cohort of women.

MATERIALS AND METHODS

Study cohort

The Women's Health Study was established in 1992 when 39,876 female US health professionals aged 45 years or older who were free of cancer and cardiovascular disease at baseline were enrolled in a randomized, double-blind, placebo-controlled clinical trial evaluating the benefits and risks of low-dose aspirin and vitamin E in the primary prevention of cancer (other than nonmelanoma skin cancer) and cardiovascular disease (29). Upon enrollment in the study, all participants completed a baseline questionnaire about their medical history and lifestyle characteristics, including potential risk factors for colorectal cancer. The current analysis was restricted to 37,916 women after exclusion of those who did not provide dietary information, had implausible total energy intakes (<600 kcal/day or >3,500 kcal/day), did not provide information on use of multivitamin supplements, or had newly diagnosed nonadenocarcinoma colorectal cancer.

Dietary assessment

At baseline, 39,345 (98.7 percent) women in the Women's Health Study completed a 131-item food frequency questionnaire, a format that has been used in the Nurses' Health Study. The questionnaire assessed 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 size. The food frequency questionnaire included a section on current use of multivitamin supplements; this section was completed by 96.6 percent of the women. Participants were asked about the exact brand and type of multivitamins they used and how many times they took multivitamins per week (≤2, 3–5, 6–9, or ≥10). Intakes of folate and vitamin B₆ from supplements were estimated by linking this information with a comprehensive database on the folate and vitamin B₆ content of multivitamin preparations that was created by the Department of Nutrition at the Harvard School of Public Health. Information about the status (never, past, or current) and duration (0–1, 2–4, 5–9, 10–14, 15–19, or ≥20 years) of multivitamin supplement use was collected in the baseline enrollment questionnaire.

The validity and reliability of the food frequency questionnaire has been assessed in the Nurses' Health Study (30–33), which has many characteristics similar to those of the Women's Health Study. In a sample of 188 participants from the Nurses' Health Study, the correlation coefficients for correlation between folate intakes calculated from the 1980 dietary questionnaire and erythrocyte folate concentrations in 1987 were 0.55 for total folate (folate from both foods and supplements) and 0.38 for dietary folate (folate from foods only) (34). In a sample of 712 participants who provided blood specimens during 1989–1990 and served as controls in a nested case-control study of breast cancer within the Nurses' Health Study, the correlation coefficients for correlation between average intakes of folate and vitamin B₆ calculated from the 1980, 1984, 1986, and 1990 food frequency questionnaires and plasma levels were 0.49 for total folate and 0.52 for total vitamin B₆ (35). The corresponding correlation coefficients for dietary folate and dietary vitamin B₆ after exclusion of supplement users were 0.33 and 0.25, respectively (35).

Ascertainment of colorectal cancer cases

Every 6 months during the first year of follow-up and then annually thereafter, participants were sent questionnaires asking about newly diagnosed diseases, including colon or rectal cancer, and other clinical endpoints. Deaths of participants were identified through reports from family members, postal authorities, and a search of the National Death Index. To date, follow-up rates for morbidity and mortality are 95 percent and 97 percent, respectively. For participants who reported a diagnosis of colorectal cancer and for those who were deceased, we sought medical records and other relevant information. This information was reviewed by an Endpoints Committee consisting of physicians for confirmation of medical diagnoses. Medical record review confirmed 96 percent of the self-reports.

Statistical analysis

For nutrient analysis, women were categorized into quintiles of nutrient intake, with adjustment for total energy by means of the residual method to reduce measurement errors due to general over- or underreporting of food items (33). For the analysis by tumor location (proximal colon, distal colon, and rectum), women were categorized into tertiles because of the modest numbers of cases.

We compared the distributions of baseline data on risk factors for colorectal cancer by quintiles of total folate and total vitamin B₆ intake to assess their potential for confounding. Person-years were calculated for each participant, ranging from the date of randomization to the date of confirmed cancer diagnosis, death, or February 20, 2004, whichever occurred first. Cox proportional hazards regression models were used to calculate relative risks and 95 percent confidence intervals. We first estimated the relative risks according to category of nutrient intake, with adjustment for age (in years) and randomized treatment assignment (aspirin vs. placebo, vitamin E vs. placebo). We further performed a multivariate analysis that additionally adjusted for known or potential risk factors for colorectal cancer, including body mass index (weight (kg)/height (m)²; <23, ≥23–<25, ≥25–<27, ≥27–<30, or ≥30); family history of colorectal cancer in a first-degree relative (yes, no); history of self-reported colon polyps at baseline (yes, no); physical activity (kcal/week, in quartiles); smoking status (never, past, or current); red meat intake (servings/day, in tertiles); alcohol consumption (0, >0–<15, or ≥15 g/day); total energy intake (kcal/day, in quintiles); menopausal status (premenopausal, postmenopausal, or uncertain/unknown); baseline postmenopausal hormone use (never, past, or current); and aspirin use before the trial (yes, no).

In additional analyses, we excluded incident colorectal cancer cases diagnosed within the first 2 years of follow-up, with further adjustment for the use of colonoscopy or sigmoidoscopy for regular screening (yes, no), which participants were asked about in the 12-month follow-up questionnaire. We conducted analyses of total and dietary folate intake stratified by level of alcohol consumption (<75th percentile (<5.0 g/day) or ≥75th percentile (≥5.0 g/day)) to evaluate whether the association between folate intake and colorectal cancer risk might exist among persons who consumed more alcohol. Because alcohol is a known folate antagonist (36, 37), women who consume alcohol may have a higher requirement for folate. We conducted tests for interaction between folate and alcohol intake in relation to colorectal cancer risk by using the median value for each quintile of folate intake to form a continuous variable, an indicator variable for alcohol intake (<5 g/day vs. ≥5 g/day), and an interaction term for the product of these two variables. We used the Wald test to assess the statistical significance of the multiplicative interaction term. Because multivitamins are major sources of supplemental folate and vitamin B₆, we also evaluated the relation between use of multivitamin supplements at baseline and risk of colorectal cancer. Use of multivitamin supplements was categorized according to status (never, past, or current use), frequency of current use (none, five or fewer times per week, or more than five times per week), and duration of past or current use (<2, 2–4, 5–9, or ≥10 years). We performed tests for trend using the median value for each nutrient category to form a continuous variable. All p values were two-sided.

RESULTS

The median duration of follow-up was 10.1 years between 1992 and February 20, 2004. During follow-up, 220 incident cases of invasive colorectal cancer were confirmed, including 89 cases of proximal colon cancer, 82 cases of distal colon cancer, 44 cases of rectal cancer, and five cases for which the site in the colon was not specified. Median baseline intakes were 348 μg/day for total folate, 307 μg/day for dietary folate, 2.3 mg/day for total vitamin B₆, and 2.0 mg/day for dietary vitamin B₆. Fifty-seven percent of the women had taken multivitamin supplements in the past, and 29 percent were taking multivitamin supplements currently.

Table 1 presents the distributions of baseline data on risk factors for colorectal cancer according to quintiles of total intake of folate and vitamin B₆. Women who consumed more total folate and total vitamin B₆ tended to be current users of multivitamins; 78 percent of women in the highest category of total folate intake and 75 percent of women in the highest category of total vitamin B₆ intake were current users of multivitamin supplements. Women who consumed more total folate and total vitamin B₆ tended to be older, leaner, more physically active, and more likely to currently use postmenopausal hormones and aspirin before the trial, but they were less likely to be current smokers and to report a family history of colorectal cancer in a first-degree relative. Women with greater total folate and vitamin B₆ intakes also tended to undergo colonoscopy or sigmoidoscopy for screening or symptoms, to consume less alcohol and red meat, and to consume more dietary fiber. Personal histories of colon polyps did not appear to differ significantly according to intakes of total folate and total vitamin B₆.

While intakes of total folate and total vitamin B₆ were not significantly associated with the risk of colorectal cancer, an inverse association was observed for intakes of dietary folate and dietary vitamin B₆ after exclusion of users of multivitamins or other supplements containing folate and vitamin B₆ (table 2). The multivariable relative risks comparing the highest quintile with the lowest were 0.46 (95 percent confidence interval (CI): 0.26, 0.81) for dietary folate (p for trend = 0.02) and 0.69 (95 percent CI: 0.41, 1.15) for dietary vitamin B₆ (p for trend = 0.047). Additional adjustment for dietary fiber intake slightly attenuated the associations; the multivariable relative risks were 0.53 (95 percent CI: 0.28, 1.01) for dietary folate (p for trend = 0.12) and 0.74 (95 percent CI: 0.42, 1.32) for dietary vitamin B₆ (p for trend = 0.14). In addition, an apparent threshold effect of dietary folate and vitamin B₆ was observed; a reduced risk of colorectal cancer was observed in the second and fifth quintiles of dietary folate and in the higher three quintiles of dietary vitamin B₆ (table 2). The multivariable relative risk comparing women in the higher three quintiles of dietary vitamin B₆ with those in the lower two quintiles was 0.66 (95 percent CI: 0.47, 0.93). When dietary intakes of folate and vitamin B₆ were evaluated jointly, the multivariable relative risks were 0.42 (95 percent CI: 0.21, 0.86) for women in the highest quintiles of both dietary folate and vitamin B₆ and 0.60 (95 percent CI: 0.38, 0.96) for women in other quintiles of both, as compared with those in the lowest quintiles of both. There was no significant association between methionine intake and the risk of colorectal cancer (table 2).

The associations between intakes of either total or dietary folate and the risk of colorectal cancer did not appear to differ according to level of alcohol intake; the multivariable p values for interaction were 0.40 for total folate and 0.58 for dietary folate. However, the number of cases in this analysis was limited.

The use of multivitamin supplements at baseline was not significantly associated with the risk of colorectal cancer; the multivariable relative risks were 0.94 (95 percent CI: 0.64, 1.37) for past users and 1.07 (95 percent CI: 0.72, 1.61) for current users (table 3). The use of multivitamin supplements for 10 or more years or six or more times per week was also not significantly associated with the risk of colorectal cancer.

In an additional analysis, we excluded cases diagnosed within the first 2 years of follow-up to address the potential bias that women might have changed their diets because of preclinical symptoms of colorectal cancer; the associations with folate, vitamin B₆, and multivitamin supplements did not change appreciably (data not shown). In addition, in the analysis by tumor location, there were no clear patterns for intakes of dietary folate and dietary vitamin B₆ or for the use of multivitamin supplements. However, positive associations between total folate and total vitamin B₆ intakes and the risk of distal colon cancer were observed. The multivariable relative risks comparing the highest tertile of total folate intake with the lowest were 0.75 (95 percent CI: 0.45, 1.24) for proximal colon cancer, 2.06 (95 percent CI: 1.11, 3.82) for distal colon cancer, and 0.94 (95 percent CI: 0.43, 2.03) for rectal cancer. The comparable multivariable relative risks comparing the highest tertile of total vitamin B₆ intake with the lowest were 0.81 (95 percent CI: 0.49, 1.36) for proximal colon cancer, 1.67 (95 percent CI: 0.95, 2.93) for distal colon cancer, and 0.82 (95 percent CI: 0.41, 1.64) for rectal cancer.

DISCUSSION

In this large cohort of women, we observed no association between total folate and vitamin B₆ intakes and the risk of colorectal cancer. However, we found an inverse association between dietary intakes of folate and vitamin B₆ and the risk of colorectal cancer among women who did not take these supplements.

Because of the prospective design, recall or selection biases are unlikely to explain our findings in this study, and the high follow-up rates minimize the concern that differential rates of follow-up affected our results. Symptoms of colorectal cancer might have caused some women to increase their intake of vitamins before clinical diagnosis. However, the similar results observed after exclusion of colorectal cancer cases diagnosed within the first 2 years of follow-up do not support this explanation. Although confounding by unknown variables cannot be excluded, it seems unlikely, because adjustment for a number of potential risk factors for colorectal cancer had a minimal effect on the relative risks. Estimates of nutrient intake calculated from the food frequency questionnaire are subject to measurement error. Such error is likely to have been nondifferential because of our prospective design and may have resulted in attenuation of risk estimates. Since we assessed intakes of nutrients only at baseline, measurement error due to random within-person variation may have been inevitable. Finally, because the number of events was relatively modest, we had limited statistical power for stratified analyses and analyses of tumor locations.

Although the available data are not totally consistent, the findings from prospective cohort studies that have evaluated the relation between folate intake and risk of colorectal cancer suggest an inverse association (9–18), particularly among persons with high alcohol consumption. In the Nurses' Health Study, women with folate intakes of more than 400 μg/day were found to have a significantly lower (31 percent) risk of colon cancer than women with intakes of less than 200 μg/day (11). Furthermore, women who had been taking folic acid-containing multivitamins for 15 or more years were 75 percent less likely to develop colon cancer than were women who did not take multivitamins (11). Two (12, 38) of three (10, 12, 38) prospective studies have also suggested that high blood levels of folate are related to reduced risk of colorectal cancer. A number of investigators have conducted case-control studies of folate intake and colorectal cancer risk (19–28); most have found a lower risk of colorectal cancer associated with higher folate intake (19–22, 25–27). In comparison with some other North American cohorts (39), the level of alcohol consumption in the Women's Health Study cohort was lower, which may have limited our power to detect a difference in the folate-colorectal cancer association by level of alcohol intake.

Few studies have assessed the relation between vitamin B₆ and colorectal cancer risk. In the Nurses' Health Study, plasma vitamin B₆ concentration was inversely associated with the risks of colorectal cancer and adenoma (40). In the Iowa Women's Health Study, a higher vitamin B₆ intake was not independently associated with the risk of colon cancer, but it was significantly inversely associated with proximal colon cancer when combined with a high folate intake (15). In several case-control studies, investigators reported a lower risk of colorectal cancer associated with a higher vitamin B₆ intake (19, 24, 26). However, in two other studies, no overall associations were observed (28, 41).

Our findings suggesting a possible inverse relation of colorectal cancer risk with intakes of dietary folate and dietary vitamin B₆ rather than total folate and total vitamin B₆ are intriguing. Such findings are consistent with results from a recent meta-analysis of seven cohort studies and nine case-control studies on folate intake and colorectal cancer risk, which found a stronger inverse association for dietary folate than for total folate (42). However, these findings for folate seem to conflict with the fact that folic acid used in supplements is in the form of monoglutamate, which bypasses deconjugation for intestinal absorption and is thus more bioavailable than dietary folate (43). One possible explanation is that dietary nutrient intakes (intakes from foods only) calculated from the food frequency questionnaire may be more likely to reflect participants' long-term intakes than their total nutrient intakes (from foods and supplements), since dietary intakes may have been consistent over long periods of time whereas widespread use of vitamin supplements is relatively recent. Because of cancer's long latency period, it is highly plausible that colorectal cancer may be related to long-term or remote dietary exposure. In the Nurses' Health Study, a protective effect of multivitamin supplement use on colon cancer risk was mostly seen among women who had used supplements for 15 or more years (11). Because of the small number of cases, we were unable to evaluate the risk of colorectal cancer among participants who had used multivitamin supplements for 15 or more years.

Another possible explanation is that folate and vitamin B₆ may reduce the risk of colorectal cancer among women whose diets are low in these vitamins. Our observation of inverse associations for dietary folate and dietary vitamin B₆ among women not taking these supplements appears to support such an explanation. It is also consistent with the apparent threshold effect observed for dietary folate and dietary vitamin B₆ (i.e., low intake increases risk but incremental intake above the threshold level may add minimum benefits). Because vitamin supplements are typically taken in excess of dietary intake, their removal from the analysis would strengthen the association between dietary folate or dietary vitamin B₆ intake and the risk of colorectal cancer.

An alternative explanation for the present observations, however, is that dietary intake of folate or vitamin B₆ may merely serve as a marker for other constituents of foods that are rich in folate or vitamin B₆ (such as fiber or other potentially protective constituents) or lifestyle factors related to risk of colorectal cancer. In our analysis, additionally controlling for dietary fiber slightly attenuated the associations for dietary folate. Null associations between total intakes of folate and vitamin B₆ and colorectal cancer risk might also be a result of surveillance bias, because of increased use of colonoscopy or sigmoidoscopy procedures among supplement users. However, this explanation is not supported by the unchanged results from the analysis excluding colorectal cancer cases diagnosed within the first 2 years of follow-up, in which the use of colonoscopy or sigmoidoscopy was included in multivariable models. In a subgroup analysis of tumor locations, we also observed an increased risk associated with total intakes of folate and vitamin B₆ for distal colon cancer but not for proximal colon and rectal cancer. However, because we had a limited number of cases in this analysis, these findings should be interpreted with caution.

Although both the Women's Health Study and the Nurses' Health Study consisted of female health professionals, participants in the Women's Health Study were enrolled much later (1992 vs. 1976) and had healthier diets. More than half of the follow-up in the Women's Health Study took place after mandatory fortification of grain products with folic acid in the United States (44). Approximately half of participants in the Women's Health Study consumed at least 400 μg/day of total folate (Dietary Reference Intake of folate for adults (45)) at baseline. In contrast, only 29 percent of participants in the Nurses' Health Study consumed this amount of total folate at baseline (11). In addition, alcohol consumption in the Nurses' Health Study (39) was almost twice as high as that in the Women's Health Study. These factors may have contributed to the discrepancy in results from these two studies.

The data from the present study support recommendations to improve folate and vitamin B₆ intakes from dietary sources for colorectal cancer prevention. However, the null data for multivitamin supplements alone cannot be definitely viewed as suggesting no role for multivitamin supplements in reducing risk, since some other studies have reported an inverse association between multivitamin supplement use and colorectal cancer risk, especially among long-term users (9, 11, 25). On the other hand, the null findings on use of multivitamin supplements argue against their playing a major role in colorectal cancer prevention.

In summary, the findings from this large prospective cohort study of women suggest that higher dietary intakes of folate and vitamin B₆ may reduce the risk of colorectal cancer in women. However, an alternative explanation is that other factors related to dietary intakes of folate and vitamin B₆ may account for the inverse associations.

This study was supported by research grants CA-47988 and HL-43851 from the National Institutes of Health. Dr. Shumin M. Zhang was supported in part by National Cancer Institute Career Development Award CA096619.

The authors acknowledge the contributions of the entire staff of the Women's Health Study under the leadership of David Gordon, as well as Mary Breen, Susan Burt, Marilyn Chown, Georgina Friedenberg, Inge Judge, Jean MacFadyean, Geneva McNair, David Potter, Claire Ridge, and Harriet Samuelson. The authors also acknowledge the Endpoints Committee of the Women's Health Study (Dr. Wendy Y. Chen) and Seetha Medabalimi for her assistance with the manuscript.

Conflict of interest: none declared.

References