In this study of incident primary prostate cancer cases and clinic-based controls, all of whom were men with normal PSA and DRE results, independent effects of folate intake and MTHFR genotype on prostate cancer risk were not observed. We did observe that total lifetime alcohol consumption increased prostate cancer risk two-fold among men in the highest consumption tertile. Further, consumption of more than five alcoholic drinks per week increased prostate risk over two-fold among men with low folate intake and over four-fold among men with the CC genotype. Average weekly alcohol consumption was also associated with increased prostate cancer risk when folate intake, MTHFR C677T genotype, and their multiplicative interaction were accounted for. We observed no gene-environment interaction between the MTHFR C677T polymorphism and folate intake.
Our findings for the effects of folate intake on prostate cancer risk are consistent with five case-control studies (Vlajinac et al., 1997
; Weinstein et al., 2003
; Johansson et al., 2008
; Zhang et al., 2009
; Collin et al., 2010b
) and three cohort studies (Stevens et al., 2006
; Weinstein et al., 2006
; Beilby et al., 2010
). Our results are inconsistent with two case-control studies reporting decreased risk with higher dietary folate intake (Pelucchi et al., 2005
; Shannon et al., 2009
), and a prospective study and randomized clinical trial both finding increased risk with higher intakes (Hultdin et al., 2005
; Figueiredo et al., 2009
). Inconsistent findings from existing studies may be in part due to failure to take interactions and stage and/or grade of prostate cancer at diagnosis into account. Increasing evidence supports a dual role for folate intake in prostate carcinogenesis wherein both high and low intakes may have deleterious effects at different stages in tumor progression (Stevens et al., 2006
; Shannon et al., 2009
; Collin et al., 2010a
; Bistulfi et al., 2011
The distribution of MTHFR
genotypes among both cases and controls in our study was similar to the estimated population distribution of approximately 40, 50, and 10% prevalence for the CC, CT, and TT genotypes, respectively (Lathrop Stern et al., 2000
). Our results for the CT and TT MTHFR
genotypes were null, although the ORs below one that we observed are consistent with one meta-analysis (Bai et al., 2009
), while another shows no association between the MTHFR
C677T polymorphism and prostate cancer risk (Collin et al., 2009
Although this is a relatively small study that lacked power to detect most interactions, our finding of an over two-fold increase in prostate cancer risk for the interaction of low folate and high alcohol intake is consistent biologically with prostate carcinogenesis resulting from folate deficiency (Bistulfi et al., 2010
). Other epidemiologic evidence for a folate-alcohol interaction is conflicting (Weinstein et al., 2003
; Pelucchi et al., 2005
; Stevens et al., 2006
; Shannon et al., 2009
). One study has examined a MTHFR
-folate genotype interaction in prostate cancer, finding that the C677T polymorphism increases risk at high plasma folate levels (Van Guelpen et al., 2006
). Our finding of an over four-fold increase in prostate cancer risk associated with alcohol consumption among men with the CC MTHFR
genotype indicates a potentially strong gene-environment interaction between these two factors that has not been investigated in any other study we are aware of.
Strengths of this study include rigorous exclusion criteria and confirmation of all controls as having normal age-specific PSA levels. This decreased selection bias and ensured that the control group was not contaminated with latent prostate cancer. All men visited the same urologists in a single hospital and resided in the same geographic catchment area. Because all participants were recruited and completed the questionnaire prior to knowledge of disease status, differential recall error, and interviewer bias were avoided in this study. The food frequency questionnaire was highly detailed, reproducible, and valid, and was designed to represent average diet two years prior to study enrollment. The comprehensive statistical analysis of folate, B vitamins, and methionine, using energy-adjustment and three types of input variables also reduced the possibility of exposure misclassification. Questioning about average alcohol intake two years prior (relatively recent) reduced the possibility of recall error (Middleton Fillmore et al., 2009
), and the comprehensive indicators of alcohol consumption reduced the possibility of exposure misclassification.
Dietary folate consumption among our study population was measured prior to 1998, when folate fortification of white wheat flour became mandatory in Canada and the United States. Dietary folate consumption and folate status of our study population are likely lower than that of the more recent general Canadian and US populations (Baily et al., 2010
; Colapinto et al., 2011
). Our results are relevant since assessing lower folate consumption in these countries may be more difficult now because of fortification. However, low variation in our observed range of folate intakes may have limited our ability to detect both main effects and interactions with folate, since folate intake in the first and third tertiles differed by only 38 μg in our study.
In summary, we found that alcohol consumption is associated with increased prostate cancer risk, with higher risk among men with low folate intake and among men with the CC MTHFR genotype. Taking into account potential combined effects and interactions is important for future studies, and may in part explain inconsistent findings to date. Future work should also capture a greater range of folate intakes to determine the true role of folate in prostate carcinogenesis, and whether this is role is dualistic in nature, to inform public policy regarding folate fortification.
Conflict of interest statement
The authors declare that this research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.