A Prospective Study of Dietary Folate Intake and Risk of Colorectal Cancer: Modification by Caffeine Intake and Cigarette Smoking

Evidence is accumulating that folate deficiency is implicated in carcinogenesis, particularly in rapidly proliferative tissues, such as the colorectal mucosa (1, 2). Folate is an essential cofactor for the de novo biosynthesis of nucleotides needed for DNA synthesis and plays a crucial role in DNA methylation, stability and integrity, and repair (1). Folate deficiency may cause defective DNA repair (3) and chromosomal fragile site expression (4), leading to chromosomal breaks (5) and micronucleus formation (4), thus conceivably increasing the risk of colorectal cancer.

Caffeine and theophylline found in coffee and tea have been shown to enhance fragile site expression induced by folate restriction (6, 7). MacGregor et al. (8) reported that folate deficiency causes cytogenetic damage in mice and that caffeine acts synergistically with inadequate folate status to augment this damage. Additionally, consumption of coffee and tea has been directly related to chromosome fragility (9) and chromosomal damage (10) in peripheral lymphocytes and erythrocytes.

Cigarette smoking may increase folate requirements by interfering with folate utilization and/or metabolism (11, 12). It has consistently been shown that smokers have lower circulating folate concentrations than nonsmokers (9, 12, 13). Smoking has been related to higher frequency of cells with chromosome aberrations (9), elevated micronucleus frequency (10), and increased chromosome fragility (14, 15).

Given the widespread consumption of caffeinated beverages, including coffee and tea (16), and high prevalence of cigarette smoking, knowledge about the interplay between these factors and folate intake in predicting the risk of colorectal cancer is of great public health importance. Therefore, in the present study, we analyzed data from a large population-based prospective cohort of Swedish women—a population with a relatively low intake of folate (fortification of cereals and flours are not mandatory) and with the highest per capita caffeine intake in the world (16)—to examine whether the association between dietary folate intake (i.e., folate from food sources) and risk of colorectal cancer is modified by intake of caffeine (in coffee and tea) and smoking.

The Swedish Mammography Cohort was established between 1987 and 1990 when 66,651 women in Uppsala and Västmanland counties, central Sweden, ages 40 to 75 years (74% of the source population), completed a mailed questionnaire about their diet, weight, height, and educational level. A new questionnaire sent to all surviving participants in 1997 also elicited information on physical activity, cigarette smoking, family history of colorectal cancer, and the use of dietary supplements, aspirin, and hormone replacement therapy.

Women were excluded at baseline if they had an erroneous National Registration Number, reported an extreme energy intake (i.e., three SDs below or above the mean value for log-transformed energy), or if they had a cancer diagnosis (except nonmelanoma skin cancer) before recruitment. After these exclusions, 61,433 women were included and comprised the final population for analysis. The ethical committees at the Karolinska Institutet (Stockholm, Sweden) and the Uppsala University Hospital (Uppsala, Sweden) approved this study.

At baseline, participants completed a self-administered food-frequency questionnaire that sought information on the average frequency of consumption of 67 common foods during the past 6 months. For each participant, we computed intake of nutrients by multiplying the consumption frequency of each food by the nutrient content in age-specific servings. We obtained data on food composition from the Swedish National Food Administration database (17). All nutrient values were energy-adjusted by using the residual approach (18).

To evaluate the validity of the food-frequency questionnaire, we compared nutrient intake estimated from the food-frequency questionnaire in a subgroup of 129 randomly chosen women from the cohort with their mean intake from four 1-week (3-4 months apart) diet records. Pearson correlation coefficients between the two instruments were 0.5 for dietary folate, 0.6 for coffee, 0.8 for tea, and 0.9 for alcohol. The mean daily dietary folate intake, as estimated from the diet records, was 211 μg (SD = 61 μg).

We identified incident colorectal cancer cases that occurred in the cohort by linkage with the National Swedish Cancer Registry and the Regional Cancer Registry in the study area. These registries are estimated to be 98% complete (19). Colon cancers were defined as those occurring above the peritoneal delineation of the abdominal cavity, and rectal cancers were those occurring below this delineation. Proximal colon cancers were defined as tumors occurring from the cecum through the splenic flexure, and distal colon cancers were defined as tumors in the descending and sigmoid colon. Ascertainment of deaths and the date when a participant moved out from the study area were accomplished through matching with the Swedish Death Registry and the Swedish Population Registry.

Each woman accrued follow-up time from the date of her entry to the cohort until the first of any of following: the date of a diagnosis of colon or rectal cancer, the date of death, the date of migration out of the study area, or June 30, 2004. We categorized women into quintiles according to dietary folate intake and computed incidence rates by dividing the number of incident cases by person-years of follow-up in each quintile. The rate ratios (RR) were computed as the rate in a particular quintile of dietary folate intake divided by that in the lowest quintile. The data conformed to the proportional hazards assumptions, and we used Cox proportional hazards models with age in days as the underlying time variable to estimate RR with 95% confidence intervals (CI). We also applied a restricted cubic spline Cox proportional hazards modeling approach (using five knots) to flexibly model the association of dietary folate intake with risk of colorectal cancer, avoiding linearity assumptions (20). Multivariate models were simultaneously adjusted for age, body mass index, educational level, and intakes of red meat, total energy, saturated fat, methionine, vitamin B-6, β-carotene, calcium, and cereal fiber. Tests for linear trend across increasing categories of dietary folate intake were conducted by using the median value for each category as a continuous variable.

We did analyses stratified by caffeine intake (categorized into tertiles) and smoking (never smoker, ever smoker <10 years, and ever smoker ≥10 years) to examine whether the association of dietary folate intake with risk of colorectal cancer was modified by these factors. Analyses stratified by smoking were based on a subcohort of 38,286 women who responded to the 1997 questionnaire and for whom complete information on smoking status and duration was available. To test for statistical interaction, we entered into the multivariate models cross-product terms for dietary folate intake (as a continuous variable) and caffeine intake and smoking (status and duration) along with the main-effect terms for each. All analyses were conducted with SAS statistical software (version 8.2; SAS Institute, Cary, NC). All reported P values were based on two-sided tests.

Among 61,433 women followed up for a mean of 14.8 years (911,042 person-years), a total of 805 incident colorectal cancer cases occurred, including 547 colon cancers (249 proximal colon, 170 distal colon, and 128 at an unspecified site in the colon) and 252 rectal cancers; six cases were diagnosed with both colon and rectal cancer.

The mean (±SD) energy-adjusted daily dietary folate intake (i.e., folate from food sources) in the cohort, as estimated from the food-frequency questionnaire, was 183 μg (±43.5 μg). Compared with women with a low dietary folate intake, women with higher intakes were slightly older, were less likely to be smokers, were more likely to have a post-secondary education, and had higher intakes of vitamin B-6, β-carotene, calcium, and cereal fiber but had a lower intake of saturated fat (Table 1). Dietary folate intake was not appreciably related to body mass index and intakes of energy, methionine, caffeine, and red meat.

We observed an inverse, albeit not statistically significant, association between dietary folate intake and risk of colorectal cancer (Table 2). The multivariate RR for women in the top quintile of dietary folate intake compared with those in the bottom quintile was 0.80 (95% CI, 0.60-1.06). In analyses by subsite in the colorectum, there was a statistically significant inverse association of dietary folate intake with risk of colon cancer (RR for the highest versus the lowest quintile, 0.61; 95% CI, 0.41-0.91), but not of rectal cancer (RR, 0.93; 95% CI, 0.55-1.56). The decreased risk was somewhat stronger for proximal colon than for distal colon cancer (Table 2).

Restricted cubic spline regression analysis indicated a dose-response relationship between dietary folate intake and risk of colon cancer (Fig. 1). Each daily increment of 100 μg dietary folate intake corresponded to a 34% reduction in colon cancer risk (RR, 0.66; 95% CI, 0.48-0.92). The inverse association of dietary folate intake with colon cancer risk persisted after additional adjustment for consumption of fruits and vegetables (RR for each 100 μg/d increase of dietary folate intake, 0.69; 95% CI, 0.48-0.98) and did not change materially in a secondary analysis (using data from the 1997 follow-up questionnaire) that was further adjusted for family history of colorectal cancer, cigarette smoking, physical activity, and use of aspirin and hormone replacement therapy (RR, 0.67; 95% CI, 0.48-0.92). Also, exclusion of women who reported use of vitamin supplements containing folic acid (i.e., multivitamins, B-vitamins, and separate folic acid supplements) in 1997 yielded virtually the same results (RR, 0.69; 95% CI, 0.49-0.98) as did exclusion of colon cancer cases that occurred within the first 2 years of follow-up (RR, 0.72; 95% CI, 0.51-1.00).

We conducted stratified analyses to assess whether the inverse association of dietary folate intake with colon cancer risk was modified by caffeine intake or by smoking status and duration (Table 3). Because there were limited numbers of colon cancer cases in each category, we used tertiles of dietary folate intake. We found a statistically significant interaction between smoking and dietary folate intake in relation to colon cancer risk (P_interaction = 0.03). Whereas risk of colon cancer decreased with increasing intakes of dietary folate among women who had smoked for 10 years or more during lifetime, women who had never smoked, and already had a lower risk, did not seem to benefit from a higher dietary folate intake. The association between intake of dietary folate and risk of colon cancer did not vary appreciably across levels of caffeine intake (Table 3).

Our results, from a large population-based cohort of Swedish women, corroborate and extend the findings from previous studies, showing an inverse association between folate intake and risk of colon cancer (2). The observed inverse association of dietary folate intake with colon cancer risk is unlikely to be explained by other potentially beneficial constituents in folate-rich fruits and vegetables or by healthy behaviors related to such foods because the association remained after adjustment for consumption of fruits and vegetables.

Animal data have suggested that caffeine might act synergistically with folate deficiency to cause cytogenetic damage (8). On this basis, we hypothesized that an association between folate intake and risk of colon cancer might vary according to caffeine intake. In this study, however, we found no apparent modification of risk by caffeine intake. This lack of modification may be due to several reasons. For example, the folate intake in our study population may be sufficient to prevent folate deficiency or that only very high caffeine intake has a detrimental effect. Also, the animal data may not be applicable to humans.

In the current study, we noted a stronger inverse association between dietary folate intake and colon cancer risk among smokers than among never smokers. An interaction between pack-years smoked and folate intake was observed for lung cancer in a case-control study of past smokers (21). Furthermore, a recent case-control study (22) reported an interaction between smoking and polymorphisms in genes encoding enzymes involved in the folate-metabolic pathway in relation to risk of bladder cancer.

The major strengths of this study include its population-based nature, large sample size and large number of colorectal cancer cases, and practically complete follow-up through the Swedish Cancer registries (19). The prospective design precluded recall and selection biases. In Sweden, mandatory fortification of cereal-grain products with folic acid has not been introduced and use of vitamin supplements is less common than in the United States. Thus, we had the opportunity to evaluate dietary folate intake in relation to colorectal cancer risk in a population with a relatively low folate intake. The average daily intake of dietary folate was 211 μg among 129 participants who kept 28-day diet records. As comparison, the U.S. Recommended Dietary Allowance of 400 μg/d folate seems to be met by a large percentage of the U.S. adult population (23). It should be noted that the range of dietary folate intake in our study population was fairly narrow; yet, there was still a significant dose-response relationship between dietary folate intake and colon cancer risk.

The limitations of this study merit consideration. First, because diet was assessed through a self-administered food-frequency questionnaire and because we did not have information on the use of vitamin supplements at baseline, misclassification of total folate intake is of concern. Misclassification of folate intake, however, would be mostly random and tend to dilute any true association and, thus, could not explain our findings. In addition, exclusion of women who used vitamin supplements in 1997 did not materially change our results. Because of the observational design of this study, confounding because of unmeasured or imperfectly measured confounding variables cannot be ruled out.

In summary, findings from this large population-based prospective cohort study of women provide further epidemiologic evidence that increasing intake of folate may reduce the incidence of colon cancer. Results from this study also suggest that smokers might benefit most from a high folate intake. This finding is novel and requires confirmation in other studies.