Since 2006, numerous microRNAs have been reported to be hypermethylated in various cancer types (8
is a tumor suppressor gene that targets Cdk6, an oncogene involved in progression past the G1
/S-phase checkpoint, the loss of which has been shown to increase cellular proliferation in vitro
) and is possibly involved in cellular differentiation (11
). Therefore, its transcriptional repression via DNA promoter methylation could have consequences with respect to carcinogenesis. Recent studies have identified miR-137
promoter methylation in several types of solid tumors (8
), including OSCC (8
). In the present study, we have observed promoter methylation of miR-137
in 21.2% of oral rinse samples taken from SCCHN patients, a significantly higher proportion than observed in oral rinse samples from controls. At present, this is the only study of miR-137
methylation status to include patients with cancers of the pharynx and larynx and, although others have detected promoter methylation in oral rinse samples, this marks the first time that microRNA promoter methylation has been evaluated as a biomarker of SCCHN using an oral rinse collection method.
The proportion of samples exhibiting miR-137
promoter methylation was particularly high among patients with cancers of the oral cavity (37.8%), and lower for pharyngeal (13.5%) and laryngeal (8.0%) cancers. This discrepancy by cancer site could be a result of local variations in quantity and quality of exposure to carcinogens, with the oral cavity more probably to be directly and highly exposed to alcohol and tobacco than the pharynx and larynx. Differences in cancer biology and/or sample collection method could also contribute to the observed differences. The frequency of promoter methylation among cancers of the oral cavity is somewhat lower than that found in a small study conducted by Kozaki et al.
), where promoter methylation was observed in 7/11 (63.6%) tumor tissue from patients with OSCC. In the subset of our patients where tissue from oral cavity cancers was available, we still observed a methylation frequency of 16%. Differences between our results and that of the Kozaki study could be due to random chance (P
0.13 for comparison of frequencies between our study and the Kozaki study using a two-sample test of proportions) or to variations in study populations resulting in genetic, behavioral and exposure diversity leading to differences in frequencies of miR-137
methylation. The present study includes predominately Caucasian American subjects compared with the Japanese population included in the Kozaki study. Another possible reason for the interstudy variation could be due to the study design and laboratory methods; the present study used methylation-specific polymerase chain reaction analysis of DNA isolated from FFPE tumor tissue, whereas Kozaki employed combined bisulfite restriction analysis (36
) on DNA from fresh-frozen tissue.
It is possible that DNA derived from tumor tissue is more frequently methylated than DNA derived from oral rinse as a consequence of the carcinogenetic process. In an attempt to address this issue, we have compared methylation in tumor tissue and in mouthwash in a subset of cases and found that mouthwash has a high specificity (81.1%); therefore, patients with a positive methylation test in the saliva are very likely to have methylation in their cancer tissue.
We observed miR-137
promoter methylation in 3.0% of oral rinse samples from cancer-free subjects. Others have reported hypermethylation of other genes in benign upper aerodigestive tract mucosa (37
), including CDKN2A
, a tumor suppressor gene also involved control of the G1
/S-phase checkpoint, in oral tissue of smokers (38
). miR-137 promoter methylation has also been reported in ‘normal’ adjacent colonic tissue of colorectal cancer patients (10
). This is the first report to evaluate miR-137
promoter methylation status in cancer-free subjects and to identify methylation status in benign mucosa of the upper aerodigestive tract.
All three methylation-positive control subjects were identified as an ever-user of either tobacco, alcohol or both. This supports the notion that environmental exposures may play a role in miR-137 hypermethylation in non-cancerous tissue. These epigenetic field defects could potentially predispose such individuals to SCCHN, and therefore, the relationship between smoking, drinking and miR-137 promoter methylation in upper aerodigestive tract mucosa of cancer-free individuals should be examined further in future studies with larger sample sizes.
Using detailed information regarding smoking and drinking history, as well as other demographic and behavioral exposures, we were able to examine potential risk factors associated with miR-137
promoter methylation among SCCHN patients. The most striking finding in this regard is the observed increase in risk for miR-137
promoter methylation among women. The result was confirmed by the assessment of methylation in tumor tissue, thus making the possibility of a gender bias in mouthwash collection unlikely. Presently, there are mixed reports on the role of sex in DNA promoter methylation. Female gender has been positively associated with DNA methylation for some genes, including MTAP
in gastric cancer (40
in colorectal cancer (41
) and CDH1
in lung cancer tissue (42
), but protective for others, such as RASSF1A
in lung cancer (42
) and methylation latent trait in bladder cancer based on a 16 gene panel (45
). Experimental evidence suggests that risk of hypermethylation for certain genes may vary in a tissue-specific gender-dependent manner, as do the transcriptional targets, based on the effects of sex hormones on epigenetic states and differential distribution of sex hormone receptors (46
). In experimental models, murine studies have demonstrated higher frequencies of methylation of certain genes following estrogen administration (47
). In transgenic breast cancer mouse models, estrogen increases epigenetic inactivation of genes involved in cell cycle control and apoptosis with a dose-response effect (49
). Previous studies on sex hormone receptor distribution in SCCHN report that 51–79% of tumors are estrogen receptor positive and 42–49% are progesterone receptor positive (50
), with no apparent variation by site. Post hoc
analysis of our SCCHN study subjects reveals no significant difference in tumor site distribution for miR-137 promoter methylation by gender.
We also observed an inverse association of BMI with miR-137 methylation status in cases, although there is presently little to no support for this in the literature. This relationship warrants further examination in future studies to identify the role of BMI in miR-137 promoter methylation and whether such an effect generalizes to aberrant promoter methylation of other genes.
Recent work in the literature has begun to focus on epigenetic environment interactions, particularly with regard to DNA methylation. Here, we report a borderline positive linear trend between smoking duration and miR-137
promoter methylation. Although the environmental risk factors for promoter methylation and their corresponding mechanisms are poorly understood at present, several studies have reported an association between hypermethylation and cigarette smoking in lung (42
), prostate (53
) and bladder cancer tissue (54
). Others have also found similar dose–response relationships between smoking and promoter methylation (37
). This suggests that one mechanism through which chronic cigarette smoking contributes to development of head and neck cancer is through stimulation of aberrant promoter methylation of tumor suppressor genes, possibly including miR-137
. Although there were no significant associations of miR-137
promoter methylation with smoking or alcohol, the moderate sample size of this study limits our power to identify small effects. Rather, our finding of a borderline dose–response trend adds support, albeit mild, to the growing body of evidence that smoking is capable of inducing epigenetic alterations in addition to somatic mutations and chromosomal breaks. At present, no other study has evaluated the effects of environmental exposures specifically on microRNA promoter methylation.
Also, worthy of further discussion is the borderline association with wearing dentures. Although studies have found no apparent relationship between general denture use and oral cancer (55
), wearing of poorly fit or malfunctioning dentures has recently been associated with oral and pharyngeal cancer (57
). Improperly fitting dentures can result in chronic irritation of the surrounding mucosa. Halogenated pyrimidines that stem from damage caused by reactive oxygen species as a result of the chronic inflammatory process mimic 5-methylcytosine and have been shown to stimulate DNA (cytosine-5-)-methyltransferase 1-mediated CpG methylation (58
), suggesting a potential role of inflammation in the induction of aberrant promoter methylation. Further inquiry into the association with promoter hypermethylation in oral mucosa is required to elucidate the role, if any, of denture use.
Strengths of this study include the high quality data collection and exposure details, which allowed us to carefully examine the relationship between potential environmental exposures and miR-137 promoter methylation. Complete data eliminated some potential biases, whereas the detailed exposure data was conducive for quantitative assessment of smoking and drinking duration and intensity. The use of oral rinse as a collection media allowed for non-differential sample ascertainment from SCCHN patients and cancer-free control subjects, reducing the possibility of sample collection bias. However, statistical power may be an issue due to the moderate sample size of this study. It is therefore conceivable that the null findings of associations with alcohol and tobacco exposures are due to a lack of power rather than a true lack of effect. It is also possible that this study understates the prevalence of miR-137 promoter methylation as a result of misclassification bias stemming from the collection method, due to differential contact of tumor cells with the oral rinse media among cases. This may be particularly evident in smaller tumors or tumors that are not in or adjacent to the oral cavity, such as some pharyngeal and most laryngeal tumors.
Overall, the results of this study suggest that promoter methylation of miR-137 is a relatively common event in SCCHN. Although it has a low sensitivity (21.2%), miR-137 promoter methylation may have utility as a biomarker in DNA methylation panels, particularly given the good specificity (97.0%) and its presence in early stage tumors. Additionally, we have shed some light on environmental and personal risk factors associated with aberrant methylation of miR-137, particularly female gender. Due to the involvement of miR-137 in cell cycle control and differentiation pathways, subsequent studies should further evaluate promoter methylation as a prognostic biomarker of SCCHN. Also, future avenues of research should be aimed at correlation of miR-137 promoter methylation in mouthwash with tissue samples, overall and by specific tumor site, and at quantitative assessment of miR-137 promoter methylation in tumor tissue, with the final aim of developing etiologic and prognostic markers that could be used on large populations in a public health setting.