Vitamin B6 and Connections to Colorectal Cancer Risk

Abstract and Commentary

By David Kiefer, MD. Dr. Kiefer is Clinical Instructor, Family Medicine, University of Washington, Seattle; Clinical Assistant Professor of Medicine, University of Arizona, Tucson; and Adjunct Faculty, Bastyr University, Seattle; he reports no financial relationship to this field of study.

Synopsis: The authors reviewed 12 articles representing 13 prospective studies searching for a link between vitamin B6 consumption or blood pyridoxal 5-phosphate (PLP, the main active coenzyme form of B6) and lower rates of colorectal cancer, finding a linear inverse relationship (relative risk 0.52) between blood PLP and colorectal cancer risk.

Source: Larsson SC, et al. Vitamin B6 and risk of colorectal cancer: A meta-analysis of prospective studies. JAMA 2010;303:1077-1083.

The authors of this study did a comprehensive search using MEDLINE and EMBASE for vitamin B6 (pyridoxine), PLP, and colon, rectal, or colorectal cancer studies, including any studies mentioned in the reference sections. The authors' focus was on studies of a prospective design, and ones that involved vitamin B6 intake or PLP blood levels and the aforementioned cancers, as well as those that reported relative risk (RR) values. The search started with 123 articles and was weaned down to 12 articles after all exclusions were applied. Of the 12 articles included in the meta-analysis, eight articles representing nine studies examined vitamin B6 intake, while four examined blood PLP levels.

The nine studies on vitamin B6 intake involved 6,064 cases, and were of a lower quality (6-7 on a 10-point rating system, usually because physical activity was not adjusted for) than the four studies on blood PLP levels, which included 883 cases and 1,424 controls and had an average research quality of 8-9. The overall relative risk of colorectal cancer for the highest vs. lowest quintiles of vitamin B6 intake was 0.90 (95% confidence interval [CI], 0.75-1.07), indicating, essentially, no effect; closer examination of the individual studies showed both positive and inverse relationships that averaged out to no effect when combined together. On the other hand, all four trials on blood PLP levels showed an inverse relationship that was statistically significant in three of four trials, yielding a RR of 0.52 (95% CI, 0.38-0.71).

Other salient points from this meta-analysis include the heterogeneity analysis, the dose-response meta-analysis, and subgroup analyses. Only the vitamin B6 intake studies had statistically significant heterogeneity (P = 0.01), most of which could be accounted for by a large cohort study in the Netherlands1 that used a very small difference in highest to lowest quintiles of vitamin B6 intake compared to the other studies. Interestingly, when this study was removed, the highest vs. lowest vitamin B6 intake RR was 0.80 (95% CI, 0.69-0.92). No differences in dietary vs. total vitamin B6, or geographic region were found, but in the five studies that were age-adjusted, the RR across the quintiles of vitamin B6 intake was 0.79 (95% CI, 0.70-0.90). The dose response for blood PLP appeared linear, with a 100 pmol/mL amount (approximately 2 standard deviations) conferring a RR benefit of 0.51 (95%CI, 0.38-0.69).


It would be nice to know that as a result of this study, we could now definitively tell patients to take vitamin B6 supplements to prevent colon cancer. However, that is not what the study is all about. This meta-analysis brings together a few pieces of the vitamin-cancer puzzle, its prospective design focus avoiding a major pitfall of retrospective analyses ("I'm pretty sure I ate that vitamin B6-enriched food last week..."), while adding an interesting, and hard-to-dispute, laboratory side with the PLP levels.

Vitamin B6 should work, as we know from the proposed mechanism of action outlined in prior work, via the 1-carbon metabolic pathway where it helps to transfer 1-carbon groups during DNA synthesis and methylation. This process is impaired in animals that are vitamin B6-deficient: DNA changes, and increased risk of cancer, would follow. Other research shows that dietary B6 may decrease cell proliferation, oxidative stress, nitric oxide, and angiogenesis, all of which should decrease the development of colorectal cancer.2

The results of this meta-analysis hint at a connection between blood PLP and colorectal cancer, though not vitamin B6 intake, but a few issues complicate any final interpretation of the results. The authors point out three main issues: possible confounders, a potential misclassification problem, and the ever-present problem with pooling studies that have different methodologies. With respect to confounders, most, but not all, of the individual trials controlled for other known risk factors, such as physical activity, smoking, and alcohol and folate intake, though lingering effects from these or unknown confounders are theoretically still a problem. Also, six of the vitamin B6 intake studies only assessed the diet at baseline, whereas serial dietary estimates would have more accurately classified study participants into the correct quintiles. Finally, the heterogeneity resulting from methodological differences of the individual trials, a particular problem for the vitamin B6 intake studies, could have affected the final results.

The authors call for, and rightfully so, a well-designed randomized controlled trial (RCT) to settle the above questions. Yes, in conversations about many aspects of inte- grative health we've heard that request before. Until that RCT happens, however, this meta-analysis provides some food for thought and a few extra data points in a very important topic on the minds of practitioners and patients alike: Do these vitamins really help? We're still not completely sure.


1. de Vogel S, et al. Dietary folate, methionine, riboflavin, and vitamin B-6 and risk of sporadic colorectal cancer. J Nutr 2008;138:2372-2378.

2. Matsubara K, et al. Vitamin B6-mediated suppression of colon tumorigenesis, cell proliferation, and angiogenesis. J Nutr Biochem 2003;14:246-250.