We found that alcohol drinking was associated with the likelihood of promoter methylation in postmenopausal breast cancer for two of the three genes that we examined, positively for E - cadherin and negatively for p16. In addition, drinking pattern appeared to affect methylation; associations were similar for drinks per drinking day for both E - cadherin methylation and p16 methylation after adjusting for total intake. The association of alcohol consumption with E - cadherin was limited to ER-negative tumors, and there was evidence that associations tended to be stronger for alcohol consumption at younger ages for p16 in postmenopausal breast cancer. RAR-β2 methylation did not differ by alcohol consumption.
Frequencies of promoter methylation for p16
genes in our sample were similar to previous reports; E-cadherin
promoter methylation frequency was somewhat lower than has been reported previously (Li et al., 2006
; Parrella et al., 2004
; Shinozaki et al., 2005
). This variation may be related to characteristics of our sample, sample size in this or other studies or it may depend on the sensitivity of the MSP assay and differences in MSP assay design. In our study, we used the same assay conditions for each tumor DNA sample and positive and negative internal controls; our MSP analysis was reliable.
There are few studies examining a possible association between alcohol consumption and promoter methylation in cancers (Kraunz et al., 2006
; Marsit et al., 2006
; Puri et al., 2005
; Slattery et al., 2006
; van Engeland et al., 2003
; Zhu et al., 2003
), and only one of breast cancer (Zhu et al., 2003
). In a case control study of 304 African-American cases and 305 controls, there was a trend toward greater risk of breast cancer with hypermethylation of the ER
α gene associated with alcohol consumption (>0.5 drinks/day), but the confidence interval was wide and included the null (Zhu et al., 2003
). In studies of other cancer sites, Marsit et al (Marsit et al., 2006
) found increased likelihood of SRRP1
hypermethylation with any lifetime alcohol exposure for head and neck squamous cell carcinoma; and van Engeland et al (van Engeland et al., 2003
) found a marginally positive association between promoter methylation of at least one out of six genes, and low folate/high alcohol intake for colorectal cancer. Other studies did not find such associations (Hasegawa et al., 2002
; Kraunz et al., 2006
; Puri et al., 2005
; Slattery et al., 2006
). However, most of these studies did not evaluate drinking patterns or lifetime total alcohol consumption, and none of them examined alcohol intake at different lifetime periods. Our results indicate that alcohol consumption may affect promoter methylation of genes at earlier age periods (before age 40) among postmenopausal breast cancer patients, also providing evidence that DNA methylation alteration is an early event in carcinogenesis, at least for these genes.
In our study, we found an inverse association between the likelihood of p16
promoter methylation and lifetime alcohol consumption among postmenopausal women with breast cancer; the associations with alcohol were stronger for drinking before the age of 40. Our results were different from two previous studies examining p16
methylation of head and neck squamous cell cancers (Kraunz et al., 2006
; Puri et al., 2005
), which found no difference in prevalence by alcohol intake. A recent animal study also observed that alcohol consumption did not affect p16
promoter methylation in mice aged either 18 months or 4 months (Sauer et al., 2010
). This discrepancy might be partly due to differences in pathways at different cancer sites. It may also be that we were able to detect a difference in our study because of the extensive data regarding alcohol consumption history which we had in our study. However, the biological mechanisms underlying the observed inverse associations are not known.
In an in vitro
study of MCF-7 cells, ethanol down-regulated the expression of E-cadherin, and increase cell invasion and migration in human breast cancer cell lines (Meng et al., 2000
), and it was proposed that the alcohol consumption may directly decrease the expression of E-cadherin through an epigenetic mechanism. Meanwhile, the ErbB2/Her2/Neu, a member of the epidermal growth factor receptor (EGFR) family, is over-expressed in 20–30% of invasive breast tumors and is associated with poor prognosis. Previous studies reported that the status of ErbB2 expression determine a cellular response to ethanol exposure and high expression of ErbB2 enhances an ethanol-mediated migration and invasion of breast cancer cells in vitro
(Ke et al., 2006
; Ma et al., 2003
). DNA methylation of CDH13
, coding for H-cadherin (a new member of the cadherin superfamily), has been found to be more prevalent in Her2/neu-positive breast tumors (Fiegl et al., 2006
). Therefore, it is plausible to postulate that increased EGFR signaling may decrease the expression of E-cadherin through aberrant DNA methylation following the alcohol consumption in breast tumors. No previous studies have evaluated the association of alcohol intake and E-cadherin
promoter methylation in breast tumors; a study of head and neck squamous cell carcinoma showed a borderline association of E-cadherin
promoter methylation with increased years of drinking (Hasegawa et al., 2002
). We found increased prevalence of E-cadherin
promoter methylation in breast tumors from postmenopausal drinkers compared with never drinkers. These findings suggest that even the relatively low consumption levels of the study participants may be sufficient to induce aberrant promoter methylation of E-cadherin
. We found that the association between alcohol consumption and E-cadherin
promoter methylation was limited to ER-negative breast tumors. However, the biological mechanisms underlying the observed differences by menopausal status and ER status are not known. Further studies would be needed to elucidate a mechanism.
To our knowledge, our study is the first to examine the association between RAR-β2 methylation and alcohol consumption in breast tumor, and we found no associations between RAR-β2 methylation with alcohol consumption. In addition to the time period of alcohol consumption, the way alcohol is consumed may also affect methylation. Drinks per drinking day was associated with increased likelihood of E- cadherin and decreased likelihood of p16 promoter methylation, even after adjusting for total alcohol intake. Drinking intensity could affect the biological effects of alcohol. There may be important differences between, for example, drinking seven drinks per week as one drink per day, or as seven drinks on a single day each week. In our study, drinking intensity contributed additional information in explaining the difference in likelihood of methylation of these two genes. However, it is important to note that drinking quantity and intensity were highly correlated so that is difficult to separate these different components of drinking behavior.
As in any study of this kind, the strengths and weakness of the study need to be taken into account when considering the findings. Strengths of this study include the population-based study design and relatively large sample size, leading to more stable risk estimates. Nevertheless, the statistical power for examining of subgroups remained limited due to the low frequencies of the promoter methylation, limiting our ability to identify weak associations. Another strength of this study is the detailed information collected regarding lifetime alcohol consumption. The CLDH used in the study has been shown to have high test-retest reliability for estimates of lifetime alcohol consumption (Russell et al., 1997
). While recall bias may be a concern for case control study, there is evidence that there is not much bias in recall of alcohol in case control studies of breast cancer (Friedenreich et al., 1991
; Giovannucci et al., 1993
). Further, it is unlikely that biased recall of alcohol intake would be related to gene promoter methylation, and thus would not differentially affect the case-case comparisons. Among the limitations, the lack of response among cases and controls has the potential for selection bias. However, it is unlikely that there were difference in participation of cases by methylation status; case-case comparisons would not be affected by this bias. A further concern was that we were unable to obtain paraffin-embedded breast tumor tissue for 21.4% of cases. Compared to those for whom we were unable to obtain tissue, those cases with breast tumor tissue were slightly younger at diagnosis and had a higher TNM stage of breast tumor. They were similar in terms of tumor size, histological grade, nuclear grade, ER and PR status. Further, both age and tumor stage were unrelated to methylation of these genes in this population (Tao et al., 2008
); and so selection bias is not a likely explanation for our findings, particularly for the case-case comparisons. Finally, there are concerns with limitations on the outcome measurement of methylation. We examined methylation for three genes that known to be commonly methylated in breast cancer and are known to be significant in three pathways important in breast carcinogenesis. However, clearly we are somewhat limited by the study of a small number of genes. Expansion of our findings to a larger number of genes and genome-wide scale will be important. Further, the methodology used in this study was limited to examination of a single CpG island in the promoter regions. It is assumed that these single regions are sentinels for gene silencing and methylation of other CpG islands, especially in tumors, but it is possible that in some women, these genes are hypermethylated in CpG sequences that we did not study. Finally, in order to assess promoter methylation in this study we used real time MSP that increases the specificity of the MSP by interrogating more than one CpG (Shames et al., 2007
). More specifically, we used a fluorescence based version of the MSP technique due to its increased throughput by eliminating the need of gel electrophoresis (Eads et al., 2000
). This method has been found to be 10 times more sensitive than the classic MSP method, able to detect methylated sequences from an excess of 10,000-fold unmethylated alleles (Eads et al., 2000
). Moreover, due to its increased sensitivity, it can use very small amounts of inferior quality DNA, making it amenable to detect methylation patterns in samples with possible degraded DNA yields and contamination with normal cells, such as archived tumor blocks (Eads et al., 2000
). While we are aware of the limitations of this technique given its qualitative nature compared to other quantitative methods such as pyrosequencing, and the fact that it interrogates a limited number of CpG sites, we have followed stringent quality control criteria to ensure confidence in results. Moreover, recent findings show that results from MSP are highly correlated with other methods. By using the highly specific real time MSP, it is likely that our results would be reproduced by other methods (Lee et al., 2008
In summary, we found an association of alcohol consumption and of drinking pattern with increased promoter methylation of E- cadherin and decreased promoter methylation of p16 genes in postmenopausal breast cancer. The associations were stronger for consumption at younger ages. Thus, our study provides evidence that the observed association of alcohol drinking and breast cancer risk may be related at least in part to alterations in methylation pathways. These findings are important in providing more data to support an etiologic role for alcohol consumption in breast carcinogenesis and for suggesting potential mechanisms for prevention and treatment.