Figure. Flowchart of study participants.
Andreeva VA, Touvier M, Kesse-Guyot E, Julia C, Galan P, Hercberg S. B Vitamin and/or ω-3 Fatty Acid Supplementation and CancerAncillary Findings From the Supplementation With Folate, Vitamins B6 and B12, and/or Omega-3 Fatty Acids (SU.FOL.OM3) Randomized Trial. Arch Intern Med. 2012;172(7):540-547. doi:10.1001/archinternmed.2011.1450
Author Affiliations: Nutritional Epidemiology Research Unit, Unité Mixte de Recherche, Institut National de la Santé et de la Recherche Médicale U557, U1125 Institut National de la Recherche Agronomique, Conservatoire National des Arts et Métiers, University of Paris XIII (Drs Andreeva, Touvier, Kesse-Guyot, Julia, Galan, and Hercberg), and Department of Public Health, Avicenne Hospital (Dr Hercberg), Bobigny, France.
Background To advance knowledge about the cancer-chemopreventive potential of individual nutrients, we investigated the effects of B vitamin and/or ω-3 fatty acid supplements on cancer outcomes among survivors of cardiovascular disease.
Methods This was an ancillary study of the Supplementation With Folate, Vitamins B6 and B12 and/or Omega-3 Fatty Acids (SU.FOL.OM3) secondary prevention trial (2003-2009). In all, 2501 individuals aged 45 to 80 years were randomized in a 2 × 2 factorial design to one of the following 4 daily supplementation groups: (1) 5-methyltetrahydrofolate (0.56 mg), pyridoxine hydrochloride (vitamin B6; 3 mg) and cyanocobalamin (vitamin B12; 0.02 mg); (2) eicosapentaenoic and docosahexaenoic acid (600 mg) in a 2:1 ratio; (3) B vitamins and ω-3 fatty acids; or (4) placebo. Overall and sex-specific hazard ratios (HRs) and 95% CIs regarding the cancer outcomes were estimated with Cox proportional hazards models.
Results After 5 years of supplementation, incident cancer was validated in 7.0% of the sample (145 events in men and 29 in women), and death from cancer occurred in 2.3% of the sample. There was no association between cancer outcomes and supplementation with B vitamins (HR, 1.15 [95% CI, 0.85-1.55]) and/or ω-3 fatty acids (HR, 1.17 [95% CI, 0.87-1.58]). There was a statistically significant interaction of treatment by sex, with no effect of treatment on cancer risk among men and increased cancer risk among women for ω-3 fatty acid supplementation (HR, 3.02 [95% CI, 1.33-6.89]).
Conclusion We found no beneficial effects of supplementation with relatively low doses of B vitamins and/or ω-3 fatty acids on cancer outcomes in individuals with prior cardiovascular disease.
Trial Registration isrctn.org Identifier: ISRCTN41926726
Proper nutritional status is considered protective against cancer; however, much is unknown regarding the roles of individual nutrients in different populations.1,2 Cell differentiation and chromosomal stability are modulated through DNA methylation, which uses methyl groups supplied by various nutrients.2,3 Folate deficiency, for example, is considered a potentiator because it could alter DNA methylation, thus disrupting DNA synthesis/repair.2,4,5 The chemopreventive properties of the vitamin B group have been rigorously investigated with respect to colorectal carcinogenesis.4- 7 Reviews and meta-analyses of observational studies suggest beneficial effects regarding colorectal cancer risk6- 9; however, inconsistencies in the findings are common.10,11
Evidence about cancer risk from randomized controlled trials (RCTs) is also equivocal. A small RCT involving individuals with gastritis found significant protective effects of folic acid, 20 mg/d, and vitamin B12, 1 mg/mo, on gastrointestinal cancer.12 A larger RCT in individuals with prior colorectal adenoma reported significantly higher noncolorectal cancer rates in those treated with folic acid, 1 mg/d, for at least 3 years compared with their counterparts in the placebo group, with the difference driven by prostate cancer.13 Combined analyses of the Norwegian Vitamin Trial (NORVIT) and Western Norway B Vitamin Intervention Trial (WENBIT) RCTs revealed a more than 20% higher cancer risk in survivors of ischemic heart disease treated with folic acid (0.8 mg/d) and vitamin B12 (0.4 mg/d) compared with those treated with vitamin B6 (40 mg/d) or placebo.14 However, a recent meta-analysis of large RCTs (including NORVIT and WENBIT) with participants at increased risk for cardiovascular disease (CVD) did not find an effect of folic acid supplementation on cancer incidence (and no heterogeneity by sex).15 Overall, the chemopreventive properties of vitamin B supplementation regarding specific cancer sites and in specific populations are presently unclear.
Long-chain polyunsaturated fatty acids of the ω-3 series represent another class of nutrients that has received attention in the chemoprevention literature.16,17 Via suppression of arachidonic acid–derived eicosanoid biosynthesis, influence on transcription factor activity, and signal transduction, ω-3 fatty acids could restrict tumor cell proliferation by increasing apoptotic potential along the crypt axis and could modulate inflammation and immunity.2,16,17 A small RCT demonstrated protection by ω-3 fatty acids against early genotoxic markers for skin cancer.18 However, RCTs have focused on the cancer-treating rather than cancer-preventive properties of these nutrients. Evidence from epidemiological studies is heterogeneous and their methodologic quality has been questioned.19- 22 Given the insufficient and inconclusive evidence, herein we present secondary analyses of data from the Supplementation With Folate, Vitamins B6 and B12 and/or Omega-3 Fatty Acids (SU.FOL.OM3) RCT assessing the effects of several B vitamins and/or ω-3 fatty acids on cancer outcomes.
The SU.FOL.OM3 RCT was conducted in France from February 1, 2003, through July 1, 2009.23,24 Individuals aged 45 to 80 years who had experienced an acute myocardial infarction, unstable angina, or ischemic stroke within the preceding 12 months were eligible for recruitment (Figure) via a network of 417 physicians. Individuals with a history of noncardiovascular disease (eg, solid cancer and leukemia) and with expected survival of less than 5 years were ineligible.23 The trial's primary outcomes were recurrent myocardial infarction, stroke, and CVD mortality. The design, implementation, and principal findings of the study have been described previously.23,24 Written informed consent was provided by all participants, and the protocol was approved by the respective ethics and information protection committees.24
After stratification by sex, age, prior CVD, and recruitment center, the participants were randomized in a 2 × 2 factorial design to one of the following 4 groups, with supplements given as 2 capsules to be taken once daily: (1) B vitamins 5-methyltetrahydrofolate (0.56 mg), pyridoxine hydrochloride (vitamin B6; 3 mg), and cyanocobalamin (vitamin B12; 0.02 mg); (2) ω-3 fatty acids eicosapentaenoic and docosahexaenoic acid, 600 mg, in a ratio of 2:1; (3) B vitamins and ω-3 fatty acids; or (4) placebo. Details about the supplementation are available elsewhere.24
All health events were reported biannually by the treating physicians and/or the participants. On notification of a suspected major health event, all relevant medical records were solicited. Regarding cancer (the main outcome in this analysis), all reported cases were confirmed by pathology reports. Cancer diagnosis was classified according to the International Statistical Classification of Diseases, 10th Revision. Guided by the trial's steering committee expert decision, we included cancer diagnoses within the following codes or code ranges: C00 to C78, C81 to C97, D03, D09, D45, and D46. We also investigated the effects of the supplementation on cancer mortality, which was adjudicated by 2 independent physician committees blinded to treatment allocation.24
We assessed sociodemographic, behavioral, and clinical characteristics and concentrations of folate, vitamin B12, vitamin B6, homocysteine, cholesterol, triglycerides, creatinine, and fasting glucose. All blood samples were treated and stored and all biomarkers measured according to strict protocol guidelines.24
Whereas synergism between the ω-3 fatty acids and B vitamins was not expected,25 the factorial design necessitated the assessment of interaction. Because these tests did not reveal any effect modification (P = .35), we evaluated the effect of B vitamins (comparing individuals receiving B vitamins alone or combined with ω-3 fatty acids with individuals not receiving B vitamins) and the effect of ω-3 fatty acids (comparing individuals receiving ω-3 fatty acids alone or combined with B vitamins with individuals not receiving ω-3 fatty acids) regarding cancer outcomes (significance level, .05, 2 sided). Baseline characteristics and group comparability were explored with χ2 tests, unpaired t tests, and Wilcoxon rank sum test. We strove to specify our models well, adjusting for the most pertinent covariates to minimize the potential for type I error. Consistent with knowledge about sex differences in cancer incidence, we performed tests for interaction between sex and each supplement type. Because time from the detrimental exposure to a clinically detectable tumor (carcinogenesis) could be measured in decades,1,2 any potential role of the supplements would pertain to cancer progression, not initiation.
For the statistical models, the participants contributed time-at-risk (in years) up to the date of the initial cancer diagnosis, the date of the last returned questionnaire, or July 1, 2009 (the scheduled end of the trial), whichever occurred first. The hazard ratios (HRs) and 95% CIs associated with the effect of group assignment on cancer outcomes were estimated with Cox proportional hazards models. The participants' ages were the time scale; thus, all HR estimates are age adjusted. The model's proportionality assumption was evaluated graphically and was met for assignment to B vitamins (yes/no) and ω-3 fatty acids (yes/no). We conducted analyses using commercially available software (SAS, version 9.1; SAS Institute, Inc) according to the intent-to-treat principle.
Baseline characteristics of the 514 women (20.6%) and 1987 men (79.4%) randomized in the trial are summarized in Table 1 (by supplementation type) and Table 2 (by sex and supplementation type). Treatment groups were balanced except for some variability between the vitamin B supplement groups (yes/no) regarding median eicosapentaenoic acid (P = .01) and triglyceride concentrations (P = .05). Mean (SD) baseline age was 61.3 (9.0) years (mean, 60.9 years among men and 63.2 years among women); at cancer diagnosis, 65.6 (9.0) years (65.1 years among men and 67.4 years among women). The median time between the acute CVD event and randomization was 101 days,24 and median follow-up was 4.7 (interquartile range, 1.5) years.
In total, 174 participants (7.0%) presented with incident primary cancer (not including 14 cases of basal cell carcinoma). Of these events, 145 occurred in men (7.3%) and 29 in women (5.6%). The rates per 1000 observation-years were 13.2 among women and 17.2 among men. Approximately 70% of the cancer incidence occurred in the second half of the trial, and only 2 events occurred during the first year of follow-up. Among men, the distribution of the affected anatomical locations was 50 in the prostate, 22 in the lung/bronchus, 16 in the bladder, 13 in the colon/rectum, and 44 in all other locations. Among women, the respective distribution was 9 in the breast, 4 in the lung, 3 in the colon/rectum, and 13 in all other locations. Cancer mortality occurred in 2.3% of the sample. There were 47 deaths among men (2.4%) and 11 among women (2.1%). Table 3 summarizes the characteristics distribution by cancer status. Compared with noncases, individuals with incident cancer were older (P < .001) and had somewhat lower median vitamin B6 concentrations (9.4 vs 8.9 ng/mL [P = .05]; to convert to nanomoles per liter, multiply by 4.046) and somewhat higher median homocysteine concentrations (1.7 vs 1.9 mg/L [P < .001]; to convert to micromoles per liter, multiply by 7.397).
Results with the full sample and by sex (P = .054 for interaction) are summarized in Table 4. The sex- and age-adjusted full-sample models revealed a lack of effect of the B vitamins on cancer incidence (HR, 1.15 [95% CI, 0.85-1.55]) or cancer mortality (HR, 1.30 [95% CI, 0.77-2.18]). Because randomization to B vitamins (yes/no) had resulted in slight variability between the groups regarding eicosapentaenoic acid and triglyceride concentrations, we conducted sensitivity analyses that further adjusted for these variables and noted that the results remained unchanged (data not tabulated).
Among men, there were 74 cancer cases in the B vitamins group and 71 cases in the comparison group. Among women, there were 20 cases in the B vitamins group and 9 cases in the comparison group. Overall, allocation to B vitamins did not have an effect on cancer incidence or cancer mortality in men or women, although the incidence model in women (adjusted for age, homocysteine and creatinine concentrations, and prior unstable angina) reached borderline significance (HR, 2.18 [95% CI, 0.98-4.85]; P = .06).
Results with the full sample and by sex (P = .02 for interaction) are summarized in Table 4. The sex- and age-adjusted full-sample models revealed a lack of effect of the ω-3 fatty acids on cancer incidence (HR, 1.17 [95% CI, 0.87-1.58]) or cancer mortality (HR, 1.47 [95% CI, 0.87-2.48]). Among men, there were 72 cancer cases in the ω-3 fatty acids group and 73 cases in the comparison group. Allocation to ω-3 fatty acids did not have an effect on overall cancer incidence or cancer mortality in men. Among women, there were 21 cases in the ω-3 fatty acids group and 8 cases in the comparison group. Positive associations were noted between ω-3 fatty acid supplementation and cancer incidence (HR, 3.02 [95% CI, 1.33-6.89]) and mortality (HR, 5.49 [95% CI, 1.18-25.97]).
These ancillary results from the SU.FOL.OM3 trial do not provide evidence of beneficial effects of supplementation with B vitamins and/or ω-3 fatty acids in relatively low doses for 5 years on cancer incidence or mortality among CVD survivors. Our results are consistent with those of the Study of the Effectiveness of Additional Reductions in Cholesterol and Homocysteine (SEARCH) trial, in which daily supplementation with substantially larger doses did not have any effects on cancer incidence after 6.7 years of follow-up.26 That trial used 2 mg/d of folic acid and 1 mg/d of vitamin B12, whereas our doses were 0.56 mg of 5-methyltetrahydrofolate and 0.02 mg of cyanocobalamin. The lack of effects of vitamin B supplementation on cancer incidence is in line with a meta-analysis of large RCTs involving individuals at increased CVD risk who received folic acid supplementation in daily doses ranging from 0.8 to 40.0 mg during a median of 5 years.15 Regarding the lack of beneficial effects of the ω-3 fatty acids on cancer incidence, an argument has been extended that the critical period for dietary exposure to these nutrients may be during childhood or early adulthood.16
In total, 83.3% of the cancer incidence and 81.0% of the cancer mortality occurred in men (who represented 79.4% of the sample); however, neither type of supplementation produced any effects. Among women (about 83% of the women were menopausal), both types of supplementation had a tendency to increase cancer risk; however, these results were derived from very few cases and should be regarded as preliminary. Indeed, the supplements might have acted as potentiators of subclinical dysplasia rather than as cancer initiators. Potential biological mechanisms of the sex-specific effects of B vitamins might be linked to homocysteine concentrations because it has been reported that folate and vitamin B12 explained a higher percentage of the total homocysteine variance in women than in men.27 Sex-specific modulation of tumorigenesis by folic acid has been seen in animal models in which female mice fed diets with normal levels of folic acid had more and larger tumors compared with folic acid–depleted female mice and male mice with depleted and adequate levels of folic acid.28 A crucial period of vulnerability to folic acid supply might occur after tumor initiation, especially in female subjects.28 B vitamins play a role in cell cycle progression2 and might have dual modulatory effects depending on the dose and timing of the supplementation.2,29
In turn, our findings are consistent with a large epidemiological study with postmenopausal women in which fish intake was positively associated with breast cancer risk.30 These associations held only for estrogen receptor–positive breast cancer. Indeed, high intake of polyunsaturated fatty acids might stimulate carcinogenesis by increasing oxidative DNA damage.31 Sex-specific effects of polyunsaturated fatty acids were suggested by a case-control study reporting trends of increased colorectal cancer risk in women and a decreased risk in men (although neither attained statistical significance) associated with intake of total polyunsaturated fatty acids (and to a lesser extent, ω-3 fatty acids).32 Whereas mechanisms underlying the potential sex-specific effects of these nutrients are unclear, they appear to modulate estrogen metabolism.16
The performance of ancillary data analyses of a secondary outcome is a major limitation of the present study. The SU.FOL.OM3 RCT was designed as a secondary CVD prevention trial, and its main results showed that allocation to B vitamin or to ω-3 fatty acid supplementation had no effects on major vascular events.24 In this trial, the 5-year duration revealed some, but likely not all, progression of subclinical dysplasia possibly present at enrollment. Also, our analyses might have been insufficiently powered. Another limitation pertains to the small number of cancer cases, which prevented site-specific analyses and, regarding women, resulted in unstable and equivocal risk estimates. Furthermore, the available data on dietary supplement use outside the trial did not permit an accurate account for such covariates. However, the randomization was successful in balancing the treatment groups. Finally, although data on cancer screening were not collected, we do not have a reason to suspect that the intensity of follow-up varied by sex.
Supplementation-based RCTs with results concerning cancer incidence typically do not combine B vitamins and ω-3 fatty acids. Another distinctive feature of this trial was the relatively low supplementation doses. Another strength pertains to the use of 5-methyltetrahydrofolate, which is the most abundant natural form of folate. Unlike folic acid, 5-methyltetrahydrofolate supplementation is not likely to lead to circulating unmetabolized folic acid, which could mask vitamin B12 deficiency. Furthermore, treatment adherence (defined as taking ≥80% of the assigned supplements) was high, evidenced by self-reports and increased blood concentrations of vitamin B analytes or ω-3 fatty acids at follow-up.24 The lack of folic acid fortification in France bears on the generalizability of our findings. Overall, a replication of the models with larger cohorts of men and women is necessary. In summary, this study does not support dietary use of B vitamins or ω-3 fatty acids for cancer prevention. The preliminary evidence of adverse effects among women necessitates confirmation before firm conclusions could be drawn.
Correspondence: Valentina A. Andreeva, PhD, Nutritional Epidemiology Research Unit, University of Paris XIII, 74 rue Marcel Cachin, Bobigny 93017, France (firstname.lastname@example.org).
Accepted for Publication: November 15, 2011.
Published Online: February 13, 2012. doi:10.1001/archinternmed.2011.1450
Author Contributions: Dr Andreeva had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Galan and Hercberg. Acquisition of data: Galan and Hercberg. Analysis and interpretation of data: Andreeva, Touvier, Kesse-Guyot, and Julia. Drafting of the manuscript: Andreeva. Critical revision of the manuscript for important intellectual content: Touvier, Kesse-Guyot, Julia, Galan, and Hercberg. Statistical analysis: Andreeva, Touvier, and Kesse-Guyot. Obtained funding: Galan and Hercberg. Study supervision: Galan and Hercberg.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grant R02010JJ from the French National Research Agency and by the Ministry of Health, Sodexo, Candia, Unilever, Danone, Roche Laboratories, Merck Eprova AG, and Pierre Fabre Laboratories (SU.FOL.OM3 trial).
Funding for Less Is More: Staff support for topics research funded by grants from the California Health Care Foundation and the Parsemus Foundation.
Role of the Sponsors: The funding organizations had no involvement in the design and conduct of the study; in the collection, management, analysis, and interpretation of the data; or in the preparation, review, and approval of the manuscript.