Dietary Vitamin B6 Intake and the Risk of Colorectal Cancer

Vitamin B6, a coenzyme in the folate metabolism pathway, may have anticarcinogenic effects. Laboratory and epidemiologic studies support the hypothesis of its protective effect against colorectal cancer (CRC). The aim of this large Scottish case-control study, including 2,028 hospital-based cases and 2,722 population-based controls, was to investigate the associations between dietary and supplementary intake of vitamin B6 and CRC. Three logistic regression models adjusted for several confounding factors, including energy, folate, and fiber intake, were applied in the whole sample and after age, sex, cancer site, folate, MTHFR C677T (rs1801133), MTHFR A1298C (rs1801131), MTR A2756G (rs1805087), and MTRR A66G (rs1801394) stratification (analysis on genotypes on 1,001 cases and 1,010 controls ≤55 years old). Moderately strong inverse and dose-dependent associations in the whole sample were found between CRC risk and the intake of dietary and total vitamin B6 in all three models [model III: odds ratio (OR), 0.77; 95% confidence interval (95% CI), 0.61-0.98; P for trend = 0.03; OR, 0.86; 95% CI, 0.69-1.07; P for trend = 0.12]. In addition, meta-analyses of published studies showed inverse associations between vitamin B6 and CRC (combined relative risk, 0.81; 95% CI, 0.68-0.96; test for overall effect P = 0.01; combined odds ratio, 0.67; 95% CI, 0.60-0.75; test for overall effect P < 0.00001). Analysis within the stratified subgroups showed similar associations apart from a stronger effect among ≤55-year-old individuals. Evidence from larger cohort and experimental studies is now required to confirm and define the anticarcinogenic actions of vitamin B6 and to explore the mechanisms by which this effect is mediated. (Cancer Epidemiol Biomarkers Prev 2008;17(1):171–82)

Vitamin B6 is a water-soluble vitamin present in the human body as pyridoxine, pyridoxal, and pyridoxamine (1, 2). The main dietary sources of vitamin B6 include poultry, fish, meat, legumes, nuts, potatoes, and whole grains (3). The current reference daily intake is 1.4 mg/d for men and 1.2 mg/d for women (4). Vitamin B6 is involved in the folate metabolism pathway, which has been reported to be associated with colorectal carcinogenesis. Vitamin B6 acts as a coenzyme in two different steps in this pathway: in the synthesis of 5,10-methylenetetrahydrofolate (MTHF) and in catabolism of homocysteine to glutathione (5-7). Alcohol intake has also been found to affect several steps of the folate metabolism pathway, resulting possibly in interruption of DNA methylation and influencing the absorption and degradation of vitamin B6 and folate (7, 8).

Few observational studies have investigated the association between vitamin B6 intake and colorectal cancer (CRC; refs. 7, 9-19) and only five of them considered supplement intake in parallel to dietary intake (Table 1A and B; refs. 9, 10, 14, 16, 19). Additionally, three studies have investigated the modification effect of alcohol on vitamin B6 (7, 10, 20), three studies have investigated the modification effect of folate on vitamin B6 (10, 19, 21), and five studies have investigated the interaction effects between polymorphisms of the genes involved in the folate metabolism pathway and vitamin B6 (13, 14, 16, 21, 22).

The objective of this large case-control study was to evaluate the effect of dietary and supplementary intake of vitamin B6 on CRC in the whole sample and after age, sex, and CRC site stratification and to investigate whether vitamin B6 effects are modified by folate and alcohol intakes. Additionally, the genotypic effect of three polymorphic genes [MTHFR C677T (rs1801133), MTHFR A1298C (rs1801131), MTR A2756G (rs1805087), and the MTRR A66G (rs1801394)] involved in the folate metabolism pathway on vitamin B6 and CRC associations was examined. The first two polymorphisms are associated with a lower activity enzyme, whereas the functions of the other two are not clearly established (23). Vitamin B6 does not directly interact with MTHFR, MTR, and MTRR enzymes in the one-carbon metabolism pathway. However, this is a multistep and complicated pathway and different levels of vitamin B6 intake and polymorphisms that affect the enzymatic activity might change the relationship with CRC.

We studied 2,028 cases and 2,722 controls from a case-control study of CRC (Study of Colorectal Cancer in Scotland). We aimed to recruit prospectively all incident cases of adenocarcinoma of colorectum in patients ages 16 to 79 years presenting to surgical units in Scotland. Exclusions were patient death before ascertainment, patient too ill to participate, recurrent cases, or patient unable to give informed consent due to learning difficulties or other medical conditions. We recruited ∼40% of all incident cases in Scotland over the study period. During the same period, controls were drawn at random from a population-based register (community health index) and invited to participate. Participation rates among those approached were ∼58% for cases and ∼57% for controls. Questionnaire completion was sufficient for valid analysis in 82% of cases and 97% of controls recruited. More than 99% of the study participants were white Caucasian (see ref. 24 for further recruitment details). Ethical approval was obtained from the Multicentre Research Ethics Committee for Scotland and relevant Local Research Ethics committees, and all participants provided written informed consent.

Subjects completed one questionnaire with lifestyle and cancer information reporting their status 1 year before diagnosis or recruitment (reference period; ref. 24). Participants also completed a semiquantitative food frequency questionnaire (FFQ) consisted of 150 food items (Scottish Collaborative Group FFQ, version 6.41).⁵

Its validity for ranking macronutrients and micronutrients in younger adults (25) and its main characteristics have been described.⁵ The main sources of vitamin B6 that were included in the questionnaire were beans, legumes, nuts, eggs, meats, fish, breads, cereals, potatoes, and bananas. Participants were also asked to give full details of dietary supplement taken in the reference period. Frequencies of consumption of the specified measures of each food were converted into nutrients using an in-house calculation program based on the weights of these measures and the nutrient composition of representative foods derived from the UK food composition tables (26-32). Nutrient information on supplements was collected from the manufacturer's product information.

Genotyping was undertaken for MTHFR, MTR, and MTRR single nucleotide polymorphisms (rs1801133, rs1801131, rs1805087, and rs1801394) as part of an array-based candidate gene approach. Genotyping of patients ages ≤55 years along with matched controls was undertaken together using the Illumina Infinium I Custom array platform and done by Illumina in San Diego. DNA samples were accurately quantified by PicoGreen and quality controlled before dispatch to San Diego. To avoid potential systematic batch-to-batch variation or bias, samples were anonymized as to affection status and were randomly distributed within plates. Only analysis of the MTHFR, MTR, and MTRR genotypes is considered here as this was a hypothesis-driven project at the outset to test individual candidate genes and enzymatic systems. Data were subject to Illumina quality control procedures and genotypes were discarded if call rates were <99.5%. Genotypes were available for 1,001 cases and 1,010 controls.

The statistical package used was STATA version 10.0 (Stata Corp.). Spearman rank correlation coefficients were calculated to test the correlation between vitamin B6 and other selected micronutrients, individual foods, and food groups. The Pearson χ² test and the t test were used to test the difference between cases and controls in terms of categorical and continuous confounding variables. The Wilcoxon rank-sum test was used to test for differences in energy-adjusted vitamin B6 intake.

Logistic regression models were used to estimate the strength of association between CRC risk and vitamin B6 intake and its main food sources. Participants were divided into quartiles based on the combined distributions of cases and controls (to have the same measure of exposure in both groups). Vitamin B6 intake was adjusted for total energy intake by using the residual method (33). The core statistical model (model I) was corrected for age and sex. Model II was further corrected for folate intake (μg/d, energy adjusted, quartiles). Model III was further adjusted for fiber intake (g/d, energy adjusted, quartiles), alcohol intake (g/d, energy adjusted, continuously), smoking (nonsmoker, current smoker, and former smoker), body mass index (BMI; kg/m², continuously), regular nonsteroidal anti-inflammatory drug (NSAID) intake (yes versus no), family history of cancer (low and medium/high risk), and physical activity (total hours of sports and cycling, four categories). In addition to the whole sample analysis, odds ratios (OR) and 95% confidence intervals (95% CI) were calculated in stratified subgroups according to cancer site (colon cancer and rectal cancer), colon cancer subsite (proximal and distal colon cancer), sex, age (≤55 and >55 years), folate intake (low and high intake), alcohol intake (data not shown), MTHFR C677T genotype (CC, CT, TT), MTHFR A1298C (AA, AC, CC), MTR A2756G (AA, AG/GG), and MTRR A66G (AA, AG, GG). The associations between CRC and each of the four genotypes were tested in unadjusted logistic regression models. Interaction associations were examined by investigating the combined effects of MTHFR C677T, MTHFR A1298C, MTR A2756G, and MTRR A66G genotypes and vitamin B6 intake in the whole sample and stratified by site of cancer. Interaction was tested by deviance of two different nested models: an interactive model and its nested multiplicative one. The referent category used was homozygotes of the wild-type allele being at greatest risk (low dietary vitamin B6 intake).

We identified published cohort and case-control studies [keywords: colorectal neoplasms (MeSH), colon cancer, rectal cancer, colon, rectum, vitamin B6 (MeSH), vitamin B6, and B6] searching MEDLINE. References from these publications were also examined to identify previous studies. The inclusion criteria were as follows: (a) cohort or case-control studies examining the associations between CRC (primary end point) and vitamin B6 intake (providing at least three categories of the exposure), (b) limited to humans and publications in English published until February 2007, and (c) providing relative risks (ORs for the case-control studies) and 95% CI or information allowing us to calculate them. Review Manager software (4.2) was used to do meta-analyses of three published cohort studies (including one nested case-control study) on CRC, three cohort studies on colon cancer, two cohort studies on rectal cancer, six case-control studies (including this current study), three case-control studies on colon cancer, and three case-control studies on rectal cancer to compare high versus low dietary intakes of vitamin B6. Fixed-effect models were adopted when there was no evidence for heterogeneity, which was quantified using a χ² test and I² score.

There were no significant differences between the cases and the controls in terms of age, sex, BMI, daily alcohol intake (energy adjusted), smoking, and area deprivation index. Control individuals had a lower family history risk (P < 0.0005) and reported a significant lower total daily energy, fiber (energy adjusted), and folate intake (P < 0.00005, 0.00005, and 0.00005, respectively), taking NSAIDs (P = 0.005) and supplements (P = 0.0007) regularly more often, and being more physically active than cases (P = 0.018; Table 2). The variant allele frequencies of the four polymorphisms in the control sample were under Hardy-Weinberg equilibrium (MTHFR 677T, 11.6%; MTHFR 1298C, 10.0%; MTR 66G, 19.6%; MTRR 2756G, 3.0%).

The five main food sources of vitamin B6 were boiled or baked potatoes (14.4% of dietary vitamin B6 intake), bananas (5.0%), mixed vegetable dishes (4.3%), breakfast cereals (4.0%), and semiskimmed milk (3.5%). Spearman's correlation coefficients for correlation with vitamin B6 intake were for dietary energy-adjusted nutrient intake (folate, 0.75; fiber, 0.53; vitamin B2, 0.41; and vitamin B12, 0.28) and for food group intake (potatoes, 0.44; mixed vegetable dishes, 0.21; bananas, 0.31; and alcohol, 0.16). All the above correlations were statistically significant (P < 0.00005). Results from the Wilcoxon rank test showed that the consumption of both dietary and total (including supplement intake) vitamin B6 intake significantly differed between cases and controls (P < 0.00005 and 0.0001, respectively).

Table 3 presents three multiple logistic regression models on the relationship between quartiles of dietary and of total vitamin B6 intake and the risk of CRC as OR, 95% CIs, and P for trend for CRC. Intakes of dietary and of total vitamin B6 showed a significant, inverse, and dose-dependent effect on CRC risk in all models (model I: P for trend < 0.0005 and 0.0005; model II: P for trend = 0.002 and 0.013; model III: P for trend = 0.03 and 0.12) with approximately 20% to 30% reduction in risk for those of high versus those of low intake (model I: OR, 0.70; 95% CI, 0.60-0.83; OR, 0.74; 95% CI, 0.63-0.87; model II: OR, 0.71; 95% CI, 0.57-0.89; OR, 0.79; 95% CI, 0.64-0.96; model III: OR, 0.77; 95% CI, 0.61-0.98; OR, 0.86; 95% CI, 0.69-1.07). ORs, 95% CIs, and P for trend for CRC risk were estimated as before for groups stratified by sex, cancer site (colon or rectal), age (≤55 and >55 years), folate intake (≤0.33 and >0.33 mg/d), and alcohol intake (data not shown) for model I, II, and III analysis (Table 3). In general, vitamin B6 (dietary and total) had a similar strong inverse association with both colon, rectal, proximal, and distal cancer (Table 3). In addition, its effect did not vary by level of daily folate intake or daily alcohol intake (data not shown). In contrast, the associations were nonsignificantly stronger among those ages ≤55 years (model III, high versus low quartile of total intake, OR, 0.71; 95% CI, 0.47-1.09; P for trend = 0.11) than among those ages >55 years (model III, high versus low quartile of total intake, OR, 0.94; 95% CI, 0.73-1.21; P for trend = 0.45; Table 3). Furthermore, we explored the associations between CRC and the three main vitamin B6 food sources in our population. ORs (model I) for CRC risk of highest versus lowest quartile intakes were 1.08 (95% CI, 0.91-1.28; P for trend = 0.78) for boiled or baked potatoes, 0.69 (95% CI, 0.59-0.82; P for trend < 0.0005) for bananas, and 0.80 (95% CI, 0.68-0.93; P for trend = 0.018) for mixed vegetable dishes.

None of the polymorphisms was significantly associated with CRC (Table 4). However, we observed a nonsignificant association between CRC and MTR 2756GG (OR, 1.30; 95% CI, 0.79-2.12; P for trend = 0.19; Table 4). Vitamin B6 intake (both dietary and total) was inversely associated in a dose-dependent fashion with CRC independently of the three genotypes of MTHFR C677T, MTHFR A1298C, MTR A2756G, and MTRR A66G (Table 5; Supplementary Table S1). Furthermore, there was no significant evidence for interaction between each polymorphism and vitamin B6 greater than the one expected from a multiplicative model. Results in groups stratified by cancer site (colon and rectal; Supplementary Table S2) were similar. There was an indication of a significant interaction effect between MTHFR A1298C polymorphism and total vitamin B6 intake among rectal cases for both models (P for interaction = 0.04 and 0.02, respectively; Supplementary Table S2). However, this was not an a priori hypothesis and is likely to be a chance finding due to multiple testing.

Supplement intake was significantly associated with the disease status. In addition, intake of >2 mg/d of supplementary vitamin B6 versus no supplementary vitamin B6 intake was nonsignificantly inversely associated with CRC (OR, 0.81; 95% CI, 0.54-1.22). We identified the exact nutrient composition of these dietary supplements and added the supplement nutrients to the estimate of dietary nutrient intake. However, we also examined vitamin B6 effects after excluding the supplement takers (data not shown) and found no difference in the direction and strength of the associations.

The combined effect of high intake of vitamin B6 on CRC in the data set of three cohort studies (including one nested case-control study; refs. 7, 9, 10) and of five case-control studies (13, 15-18) showed a significant inverse association for the cohort studies and a strong significant inverse association for the case-control studies (combined relative risk, 0.81; 95% CI, 0.68-0.96 and combined OR, 0.65; 95% CI, 0.55-0.76, respectively; Fig. 1A; Supplementary Fig. S1A). The combined effect of case-control studies after additionally including the current study (13, 15-18) showed the same decreased risk (combined OR, 0.67; 95% CI, 0.60-0.75; Fig. 1B). For different cancer sites, colon cancer was significantly inversely associated with vitamin B6 intake after combining the results of three cohort (7, 10, 19) and three case-control studies [combined relative risk, 0.73; 95% CI, 0.62-0.87 (Supplementary Fig. S1B) and combined OR, 0.68; 95% CI, 0.60-0.78 (Supplementary Fig. S1D), respectively; refs. 12, 18]. Rectal cancer was significantly inversely associated with vitamin B6 intake after combining the results of two cohort (7, 19) and three case-control studies [combined relative risk, 0.72; 95% CI, 0.54-0.96 (Supplementary Fig. S1C) and combined OR, 0.81; 95% CI, 0.70-0.94 (Supplementary Fig. S1E), respectively; refs. 11, 18].

We found in this large case-control study that patients with CRC reported lower dietary and total intakes of vitamin B6 than controls. These associations persisted after controlling for several confounding factors, including energy, fiber, and folate intake. There was a dose-response relationship with trend tests being highly significant. However, due to the high correlation and the common food sources of vitamin B6 and folate, this inverse association might be triggered by the folate effect on CRC. We tried to separate the vitamin B6 and folate effects by adjusting for folate intake and by examining the vitamin B6 and CRC associations among subjects with low folate intake. The results suggested that the vitamin B6 inverse association was of similar strength when adjusting for folate and among subjects with low and high folate intake (Table 3). In addition, the folate and CRC associations in our population were less strong than the vitamin B6 ones and were further attenuated after fiber intake adjustment (Supplementary Table S3). The vitamin B6 effect was of same direction and magnitude after cancer site stratification but was stronger among females and those ages ≤55 years. After exploring the associations between CRC and the three main vitamin B6 food sources in our population, there was some evidence in favor of an inverse association with intake of bananas and mixed vegetables but less defined than the association of vitamin B6 and CRC. The lack of association with the intake of baked and boiled potatoes (one of the main vitamin B6 sources) might be because potatoes contribute to the glycemic load and this could possibly counter a vitamin B6 benefit.

When we examined the effect of high intake versus low intake of vitamin B6 in combined data sets of cohort and case-control studies, we found statistically significant decreased risks for colorectal (Fig. 1), colon, and rectal cancer (Supplementary Fig. S1). Additionally, individual results from most of the published cohort and case-control studies showed an inverse effect of vitamin B6 intake on CRC risk (Table 1A and B). In the Nurses' Health Study (10), both plasma vitamin B6 and dietary (including supplement) intake were inversely associated with CRC. In the Women's Health Study (9), dietary but not total vitamin B6 intake was associated with a decreased risk of CRC. The two other cohort studies (7, 19) did not include vitamin B6 intake from supplements. In the Swedish Mammography Cohort (7), both colon and rectal cancer were inversely associated with vitamin B6 intake, whereas in the Iowa Women's Health Study (19) only colon cancer was inversely associated with vitamin B6 intake and this relationship was diluted after further adjustment. Of the eight case-control studies (11-18), only two included vitamin B6 intake from supplements (14, 16). Five of them showed an inverse association (11, 12, 15, 17, 18) and three of them showed no association with CRC (Table 1B; refs. 13, 14, 16). However, because most studies of dietary factors of one-carbon metabolism were focused on folate, nonsignificant findings for vitamin B6 could have been omitted from publications. Therefore, the results of the meta-analyses might be subject to publication bias.

It has been proposed that vitamin B6 effects might be modified by the intake of other nutrients, such as alcohol and folate (7), but our data showed no evidence of this. From the three published studies that investigated alcohol and B6 (7, 10, 20), only one found a clear interaction especially among women with high alcohol intake (>30 g/wk; ref. 7). (The mean intakes of daily alcohol intake of the studies that have studied the interaction between vitamin B6 and alcohol are as follows. Slattery et al. (20): men: cases, 37.5 g/d; controls, 29.1 g/d; women: cases, 18.2 g/d; controls, 17.9 g/d; Larsson et al. (7): 21.5 g/d; Wei et al. (10): cases, 7.6 g/d; controls, 6.3 g/d.) All three studies that investigated plasma or dietary folate and B6 (10, 19, 20) failed to show significant interaction.

Two previous meta-analyses (34, 35) have reported an inverse association of the MTHFR 677TT and the lack of statistical significance in our study might be due to limited power. Our data did not support the hypothesis that the vitamin B6 association with CRC is modified by polymorphisms of genes in the folate metabolism pathway (7). Three previous studies reported a lower risk of CRC (16, 21) and adenomas (22) in subjects carrying the MTHFR 677TT genotype and reporting high vitamin B6 intake. Otani et al. (13) also studied the effect of MTHFR A1298G and MTRR A66G and reported an interaction between the MTRR polymorphism and vitamin B6 intake.

There is a substantial body of data supporting the biological plausibility of a protective effect of vitamin B6 on CRC risk. Vitamin B6 plays a key role in the folate metabolism pathway as a coenzyme of the cystathionine β-synthase, which converts homocysteine into cystathionine (23). In addition, its role as a coenzyme in the synthesis of MTHF might be critical for synthesis, repair, and methylation of DNA and inhibition of single and double DNA breaks (7, 36). However, results from a HuGE review on folate-metabolizing polymorphisms and CRC (23) as well as the inconsistent results of the folate effect (37) indicate that the roles of both the polymorphic genes and the involved dietary factors are complicated and need further research. Laboratory studies on mice suggest that high intake of pyridoxine has other anticarcinogenic effects by reducing cell proliferation, oxidative stress, nitric oxide production, and angiogenesis (38, 39) and a cultured human lymphocyte study reported a protective action against chromosomal damage (40). It has been proposed that the inhibition of DNA polymerases and steroid receptors of vitamin B6 may be useful as an adjuvant in cancer chemotherapy (41).

The strengths of our study include a very large sample size, use of a validated FFQ (25), and the identification of vitamin B6 intake from dietary supplements. Many different foods contributed to the intake of the vitamin B6 and so results are not determined by one major food category. Limitations of our study have been previously described (24). Briefly, underrepresentation of cases that were very ill when at presentation might limit external validity of results. Validity studies on nutrient estimates of this FFQ were carried out in younger subjects and we cannot be certain of the degree of validity in this older age group (25, 42). In addition, the FFQ was not validated against biomarkers, which is also a limitation of the study. However, any measurement error would most likely be nondifferential and thus underestimate true relationships. Limitations of case-control studies using FFQs include recall bias, misclassification bias due to imprecise measures of dietary intake, and residual confounding after attempts to control for confounders. However, we attempted to limit these problems by careful adjustment (e.g., fiber intake), adoption of identical study procedures in cases and controls, use of a FFQ that had been validated (25, 43), use of images of portion sizes and careful instructions to improve accuracy of reporting diet, and adoption of a recall period 1 year before diagnosis or recruitment date to reduce recall bias.

In conclusion, we report the findings of the largest case-control study to date investigating the association between CRC and dietary and total vitamin B6 intake. Given the evidence of anticarcinogenic effects from in vitro, animal in vivo studies, and some published epidemiologic studies, we gave this analysis high priority to minimize problems with multiple testing. Our finding of a moderately strong inverse association in models I and II between vitamin B6 intake and CRC risk with a dose-response relationship remained constant and statistically significant after further energy and fiber adjustment and stratification. This association was found for both colon and rectal cancer risk and was stronger in the younger age group of cases. Lack of any observed interaction with folate or alcohol intake or folate pathway polymorphisms may, despite the large sample size, be due to inadequate power to detect these effects. In addition, possible mechanism of action of vitamin B6 outwith its coenzyme function in the folate metabolism pathway needs further investigation. This inverse association is supported by our meta-analysis of all published case-control and cohort studies, which shows consistent evidence. Evidence of association is further strengthened by the report of a similar inverse association between CRC risk and plasma vitamin B6 levels (10). Confirmatory evidence from larger cohort studies and experimental studies is now required to confirm these associations, to explore the mechanisms by which this effect, and to define the anticarcinogenic actions of vitamin B6.

We thank R. Bisset, M. Edwards, L. McGoohan, and G. Barr for recruitment supervision and data management and Dr. N. Anderson for assistance in data interpretation and analysis.