Aberrant DNA methylation of the p16, APC, MGMT, TIMP3 and CDH1 gene promoters in tumours and the surgical margins of patients with oral cavity cancer.

Oral cavity cancer is a type of head and neck squamous cell carcinoma (HNSCC) and contributes to significant morbidity and mortality each year. An epigenetic pathway of transcriptional inactivation for many genes has been described in various cancers, including HNSCC. For our study, we selected genes for which silencing caused by hypermethylation can promote cancer development. In 75 primary HNSCC tumours and paired surgical margins, we investigated the methylation status of the p16, APC, MGMT, TIMP3 and CDH1 gene promoters by methylation-specific PCR after bisulphite treatment. The promoter methylation rates of p16, APC, MGMT, TIMP3 and CDH1 in tumours were 58.67%, 49.33%, 58.67%, 50.67%, and 57.33% and 50.67%, 41.33%, 37.33%, 42.67%, and 25.33% in the surgical margin, respectively. Our observations confirm the presence of epigenetic changes not only in the cancer cells, but also in the surrounding mucosa and represent a basis for further analysis to unravel these complicated issues. Appropriate cancer risk assessment based on epigenetic alterations in surgical margins may influence a patient's diagnosis and cure.

Introduction

Oral cavity cancer is an example of head and neck squamous cell carcinoma (HNSCC) in the head and neck area and can be defined as a malignant tumour derived from squamous epithelial cells that contributes significant morbidity and mortality each year 1. Alcohol and tobacco abuse is recognized as an initiating element in HNSCC, and HPV infection is also defined as a preliminary factor 2-4. Molecular alterations and aberrant signalling pathways in carcinogenesis of the head and neck have been identified 5. As well as genetic abnormalities, epigenetic alterations have also been implicated in cancer. Epigenetic changes are defined as modified gene expression patterns caused by mechanisms that do not concerns the primary DNA sequence. Epigenetic alteration relates to gains and losses of DNA methylation and to histone modifications. Aberrant DNA methylation comprises gene-promoter-specific hypermethylation and its mechanism causing loss of gene expression, and global hypomethylation as a prelude to structural changes in chromosomes, genome instability, and oncogene activation 6. DNA methylation is an encouraging marker for past exposure to carcinogens and future risk of cancer development 7. An epigenetic pathway of transcriptional inactivation for many genes has been described in various cancers 8-11. Other publications address epigenetic changes in HNSCC cancers 10, 12-17, but few have compared the methylation level of tumours vs surgical margins 18-20. Promoter hypermethylation in tissue samples can be detected by many molecular methods 11, 18, 21, 22 including methylation-specific polymerase chain reaction (MSP) 23, 24. In our study, we analysed samples of tumours and paired surgical margins to examine the promoter methylation status of p16, APC, MGMT, TIMP3 and CDH1 genes. All these genes are known to participate in the oncogenic pathway and to show tumour suppressor activities. We also evaluated the association of promoter methylation of these genes with clinicopathological features, habitual factors, metastasis, tumour recurrence rate, and 5-year survival rate of patients.

Materials and Methods

Study population

We studied 75 patients with primary tumours in the oral cavity. There were 50 men and 25 women, with a mean age of 56 ± 11 years. There was no difference in age between men and women (55 ± 12 vs 56 ± 81, p = 0.708). All of the patients were graded by the TNM staging system. Of the 75 patients, 4 (5.3%) were in the T1 stage, 11 (14.7%) were in the T2 stage, 18 (24.0%) were in the T3 stage and 42 (56.0%) were in the T4 stage. There was no statistically significant difference in the T and N stage distributions between men and women. Moreover, 31 (41.3%) were in the N0 stage, 28 (37.3%) were in the N1 stage, 15 (20.0%) were in the N2 stage and 1 was in the N3 stage. Histologic grading was classified as grade 1 (G1, well differentiated), grade 2 (G2, moderately differentiated), and grade 3 (G3, poorly differentiated). In G1 there were 11 (14.7%) subjects and in G2 there were 57 (76.0%), and the rest were in G3.

In all groups, 58 (77.3%) subjects smoke (currently), 57 (76.0%) reported alcohol intake and 30 (40.0%) reported a family cancer episode (first-degree family history of cancer). Females smoked and drank less frequent than males (64.0 % vs 84.0; p<0.05 and 52.0% vs 88.0%; p<0.001). There was no difference in previously reported family cancer episodes between men and women (40% in both groups). Survival rate in the analysed group was 46.7%, with metastasis observed in 13.3% and tumour recurrence in 32%. No patients received preoperative radio- or chemo-therapy. This study was approved by the Institutional Review Board on Medical Ethics of the Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology in Gliwice (Nos. KB/493-15/08 and KB/430-47/13). All patients gave written informed consent.

Tissue samples

Samples were collected from 75 HNSCC patients with a previously untreated squamous cell carcinoma from the oral cavity at the Clinic of Oncological and Reconstructive Surgery, Maria Sklodowska-Curie Memorial Cancer Center and Institute of Oncology, Gliwice, Poland. Samples of surgical margins were obtained pair-wise from the site opposite the tumour and verified by a pathologist as free of cancer cells. Only patients who had tumour-free margins when the primary tumour was resected were included in the study. Tissue samples were quickly frozen in liquid nitrogen and stored at -80°C before DNA extraction.

DNA extraction and bisulphite modification

Genomic DNA was extracted from each tumour sample and corresponding surgical margin (20 mg) using a DNeasy Blood & Tissue Kit (Qiagen, USA) according to the manufacturer's instructions and after tissue homogenization in a FastPrep®-24 instrument using Lysing Matrix A tubes (MP Biomedicals, USA). Total DNA concentration was measured using a ND-1000 spectrophotometer (NanoDrop, USA). The gDNA was bisulphite-treated using the method described by Herman and colleagues 23: 2 μg DNA in a volume of 50 μl was alkali denatured in 2M NaOH and incubated in 10 mM hydroquinone (Sigma, USA) and 3M sodium bisulphite, pH 5.0 (Sigma, USA) for 16 hours at 50°C in a microfuge tube with a mineral oil layer. Modified DNAs were purified using the Wizard DNA Clean-Up System (Promega, USA) and eluted into 50 μl of sterile water. DNA was again treated with 3M NaOH, precipitated with ethanol in 3M sodium acetate, pH 5.2 and resuspended in 20 μl of water.

Methylation-specific PCR (MSP)

DNA methylation patterns in the CpG islands of the p16, APC, MGMT, TIMP3 and CDH1 gene promoters were determined using a methylation-specific PCR (MSP) technique following bisulphite modification of isolated genomic DNA 23. To obtain specific PCR products, two separate PCRs were performed for each sample. “U” primers amplified only unmethylated DNA, while “M” primers amplified only methylated DNA in the regions of p16, APC, MGMT, TIMP3 and CDH1 gene promoters. Each chemically modified DNA sample was amplified with primers “U” and “M”, respectively. Primer sequences of gene promoters for the unmethylated reaction '"U" and for the methylated reaction "M" with thermal cycling conditions and product size are shown in Table 1. A primer set was chosen for the promoters of selected gene sequence regions based on the previously published sequences in Table 1. Primers were synthesized by Genomed (Poland). Amplification of the p16, APC, MGMT, TIMP3 and CDH1 genes was performed in 25 μl of reaction mixture under the following conditions: 12.5 μl PCR Master Mix (Cat No. M7501; Promega, USA), 1.2 μl of each primer at a concentration of 10 μM; 2 μl of modified DNA; and 8.1 μl of sterile water (Eppendorf, Germany). CpGenome Universal Methylated DNA (Chemicon Cat No. S7821) was used as a positive control for methylation, while water was used as a negative control for PCR. To verify the results, we used the EpiTec PCR Control DNA Set (Qiagen, USA, Cat No 59695) consisting of bisulphite-converted methylated and unmethylated human DNA and unconverted unmethylated human DNA. Amplification was performed in a Mastercycler Personal thermocycler (Eppendorf, Germany). PCR products were visualized on a 2% agarose gel (Sigma, USA) with ethidium bromide (Serva, Germany) staining.

Table 1

Primer sequences and amplicon characteristics of analysed genes

Gene	Primer sequences	Annealing temperature	Product size	Reference
p16 (M)	F:TTATTAGAGGGTGGGGCGGATCGC	61ᵒC - 30 s	150 bp	Wang et al. 25
	R:GACCCCGAACCGCGACCGTAA
p16 (U)	F:TTATTAGAGGGTGGGGTGGATTGT	60ᵒC - 30 s	151 bp	Wang et al. 25
	R:CAACCCCAAACCACAACCATAA
APC (M)	F:GAACCAAAACGCTCCCCAT	59ᵒC - 45 s	74 bp	Righini et al. 26
	R:TTATATGTCGGTTACGTGCGTTTATA
APC (U)	F:AAACCAAAACACTCCCCATTC	59ᵒC - 45 s	76 bp	Righini et al. 26
	R:AGTTATATGTTGGTTATGTGTGTTTAT
MGMT (M)	F:TTTCGACGTTCGTAGGTTTTCGC	59ᵒC - 45 s	81 bp	Shilpa et al. 27
	R:GCACTCTTCCGAAAACGAAACG
MGMT (U)	F:TTTGTGTTTTGATGTTTGTAGGTTTTTGT	59ᵒC - 45 s	93 bp	Shilpa et al. 27
	R:AACTCCACACTCTTCCAAAAACAAAACA
TIMP3 (M)	F:GCGTCGGAGGTTAAGGTTGTT	60ᵒC - 30 s	116 bp	Righini et al. 26
	R:CTCTCCAAAATTACCGTACGCG
TIMP3 (U)	F:TGTGTTGGAGGTTAAGGTTGTTTT	59ᵒC - 1 min	122 bp	Righini et al. 26
	R:ACTCTCCAAAATTACCATACACACC
CDH1 (M)	F:TTAGGTTAGAGGGTTATCGCGT	58ᵒC - 1 min	173 bp	Righini et al. 26
	R:TAACTAAAAATTCACCTACCGAC
CDH1 (U)	F:TAATTTTAGGTTAGAGGGTTATTG	58ᵒC - 1 min	173 bp	Righini et al. 26
	R:CACAACCAATCAACAACACA

Statistical analysis

Statistical analysis was performed using STATISTICA 10.0 PL (StatSoft, QUEST, Tulsa, Oklahoma, USA). Statistical significance was set at a p value below 0.1. All tests were two-tailed. Nominal and ordinal data were expressed as percentages, while interval data were expressed as mean value ± standard deviation. Categorical variables were compared using χ2 tests. The consistency of methylation between a tumour and a margin was determined and assessed statistically with the McNemar test. For comparison of quantitative data between males and females, a parametric t-Student test was used. Distribution of variables was evaluated by the Shapiro-Wilk test, and homogeneity of variance was assessed by the Levene test. The assessment of association between clinical status, habitual factors and methylation occurrence was performed with the multivariable backward-stepwise logistic regression. Factors affecting survival, metastasis and tumour recurrence was assessed with the multivariable backward-stepwise logistic regression. Results were presented as odds ratios with confidence interval and statistical significance.

Results

Methylation of the p16, APC, MGMT, TIMP3 and CDH1 gene promoters was detected in tumours and surgical margins. The methylation frequency of the MGMT and CDH1 genes was significantly higher in tumours than in surgical margin tissues (p<0.01). The promoter methylation status of these genes in tumours and surgical margins is summarized in Table 2.

Table 2

Promoter methylation frequency of p16, APC, MGMT, TIMP3 and CDH1 genes in tumour and surgical margin in oral cavity patients

Gene	Tumour			Margin			p-value
	Total	Methylation	Frequency (%)	Total	Methylation	Frequency (%)	P
p16	75	44	58.67	75	38	50.67	0.325
APC	75	37	49.33	75	29	41.33	0.188
MGMT	75	44	58.67	75	28	37.33	< 0.01
TIMP3	75	38	50.67	75	32	42.67	0.326
CDH1	75	43	57.33	75	19	25.33	< 0.001

Moreover, we investigated the clinical and habitual factors associated with methylation occurrence. In the APC gene we observed that positive nodal status conferred a higher methylation rate in matched margin samples (OR=1.66; 95% CI: 0.92-3.01; p<0.1). Habitual factors linked to cigarettes were associated with higher methylation of the APC gene in tumour samples (OR=3.60; 95% CI: 0.97-13.39; p<0.1). MGMT methylation was diminished in tumour samples (OR=0.38; 95% CI: 0.13-1.06; p<0.1) and in margin samples from females (OR=0.23; 95% CI: 0.06-0.84; p<0.05). Patients with a family history of cancer showed more frequently methylated MGMT genes in tumour samples (OR=3.04; 95% CI: 1.08-8.55; p<0.05). We noted that advanced tumour stage was associated with a higher frequency of methylation of this gene in matched surgical margin samples (OR=1.78; 95% CI: 0.98-3.24; p<0.1). However, patients with a history of abusing alcohol showed lower MGMT promoter methylation in the surgical margin (OR=0.13; 95% CI: 0.03-0.56; p<0.01). Analysis of the TIMP3 gene showed that the female gender conferred a higher methylation level in margin samples (OR=2.49; 95% CI: 0.89-6.95; p<0.1) and a positive N stage showed a higher methylation level in tumour samples (OR=1.79; 95% CI: 0.96-3.35; p<0.1). For the CDH1 gene, an increased methylation level was seen in tumour samples from females (OR=2.61; 95% CI: 0.88-7.73; p<0.1) and smoking was associated with higher CDH1 methylation both in tumour (OR=3.13; 95% CI: 0.91-10.80; p<0.1) and margin samples (OR=7.12; 95% CI: 0.87-59.36; p<0.1).

Patients with an advanced T classification were significantly associated with a increased risk of death (OR=9.64; 95% CI: 2.07-44.87; p<0.01). Disease recurrence was significantly related to female gender (OR=0.22; 95% CI: 0.06-0.78; p<0.05) and showed a trend towards a lower recurrence rate. No association was observed between metastasis, recurrence, or survival rate and hypermethylation of any of the genes analysed. Moreover, aberrant promoter hypermethylation of all five genes (CpG Island Methylator Phenotype, CIMP-positive) was found in tumour samples in four cases (5.3%).

Discussion

p16

p16(CDKN2A) is one of the most extensively studied genes in cancer, including epigenetic alterations. Hypermethylation of p16 has been observed in many tumour types e.g., colon, breast, brain, kidney, pancreas, liver 28 and also in HNSCC in several studies 12, 16, 19, 24, 28-31, not only in tumour tissues but also in adjacent healthy mucosa 19, 24, 30, 31. In our study, the promoter region was highly methylated in tumour tissue (58.7%). Moreover, 42.6% of patients showed hyper-methylation of this gene both in the tumour and matched surgical margin. According to our collected literature, the frequencies of hypermethylation in HNSCC tumours vary from 86.8% 31 to 82% 32, 49% 24, 36% 12, and 27% 28 to 20% 19. Other analyses showed a significant increase in promoter hypermethylation in tumours compared to normal control tissue from the resection margin in oral cancer 33-35. A coherent methylation pattern was found in primary tumours and matched metastatic lymph nodes, and also in 65% of patient's plasma 24. Interestingly, some studies showed methylation in patients with premalignant oral lesions and healthy controls 24, 29, 32. p16 hypermethylation showed no association with clinical and demographic features in our study population, as confirmed in other studies 12, 19, 24, 30, 36-38. However, some reports show that promoter methylation of p16 both in tumours and margins may be linked to chronic exposure to carcinogens in alcohol and tobacco 24, 39, 40. Moreover, based on other findings, hypermethylation of p16 was associated with younger age, nodal involvement, distant metastasis, increased recurrence rate and shortened disease-free survival, suggesting it as a candidate prognostic and predictive biomarker in oropharyngeal squamous cell carcinoma 18, 29, 31, 34, 39-41.

APC

Adenomatous Polyposis Coli (APC) is a tumour suppressor gene that, through Wnt signalling, inhibits cell proliferation 42. In the present study, methylation of this gene ranges from 49% in tumour samples to 41% in margin samples. Independent studies also reported hypermethylation of APC in OSCC samples 36, 38, 43, 44, but Esteller et al. did not detect hypermethylation of this gene in head and neck cancers 28. No differences between patients with HNSCC and healthy patients in methylation of the APC gene were observed by Longo et al., and surprisingly, this gene was frequently methylated in control samples 45. Brait et al. reported that APC was methylated in 7% of DNA samples extracted from the plasma of a cancer-free population, and promoter methylation of this gene was not associated with several potential risk factors e.g., age, smoking and alcohol status, family cancer history, diet, and nutrition 46. Other data also showed no correlation between aberrant methylation and clinical features and outcomes, such as survival 36, 38, 44.

MGMT

O6-methylguanine-DNA methyltransferase (MGMT) is involved in the guanine alkylation repair mechanism 47. Expression of this gene varies between tissues and individuals 48. Differential protein expression of MGMT between normal and cancer tissue was also confirmed 47. Aberrant promoter hypermethylation of this gene is often observed in cancers including HNSCC 8, 12, 16, 19, 28, 30, 47, 49. The results in the present study demonstrated that MGMT was statistically significantly hypermethylated in tumours compared to margins (58% versus 37%, p<0.01). Aberrant promoter hypermethylation of the MGMT gene was detected in 73.7% of oral cavity cancers, with a significant difference between cases and controls, by Kordi-Tamandani et al. 17. Martone et al. and Kato et al. observed that MGMT was hypermethylated in 50% and 56% of primary HNSCC tumours, and the results showed association of gene-specific hypermethylation status in tumours and paired surgical margins 19, 30. In our study, population, methylation of the MGMT gene was not associated with age, nodal status, and smoking, as confirmed in other studies 12, 19, 30. However, Paluszczak et al. showed an association between methylation in tumour cells and lymph node involvement, and in turn Taioli et. al observed that reduced disease-free survival and reduced overall survival are associated with hypermethylation of this gene in HNSCC patients 49, 50. Moreover, hypermethylation of MGMT is postulated as a potential prognostic biomarker 19.

TIMP3

The tissue inhibitor of metalloproteinase-3 (TIMP3) may play a significant role in tumour development, growth and metastasis by interaction with metalloproteinases in the extracellular matrix 51. In this study, no statistically significant differences between tumour and margin methylation levels for this gene was observed. Some results showed hypermethylation of TIMP3 in various tumour types 28, 51, 52. Hypermethylation of this gene in HNSCC tumours and saliva samples was also observed 15, 16, 26, 53-55 but no relation between hypermethylation and clinical features was shown 16, 38, 53, 55. Furthermore, this gene was hypermethylated in exfoliated tumour cells in HNSCC patients compared to the healthy control group. Interestingly, the work of Longo et al. was the first study regarding methylation of exfoliated cells obtained from patients with non-invasive techniques, namely cytobrushes 45.

CDH1

Cadherin 1 (CDH1/E-cadherin/E-cad) is related to cell adhesion and regarded as an invasion-suppressor gene 56. Inactivation of this gene by methylation was seen in HNSCC 16, 57. Several studies showed a range of 32% to 61.8% methylation of this gene in tumour tissues of HNSCC patients 17, 33, 35, 58. Our results showed tumour tissue with significantly higher methylation compared to the surgical margin (57% vs 25% p<0.001). Another study supported this finding 34, 35, but other investigators did not show significant differences in promoter methylation between cases of oral cancer and normal control tissue 17, 33. Chang at al. showed hypermethylated promoters of E-cadherin in 64% of oral carcinoma cases, and downregulation of its expression was found to be related to promoter hypermethylation 59. The influence of CDH1 promoter methylation in the invasive progression of HNSCC was observed based on an increased frequency of gene methylation at stages beyond the early tumour stage 60. Oral leucoplakia patients also showed a high percentage of methylation, which can be considered as a diagnostic marker 58. Šupić et al. showed that patients with advanced oral squamous cell carcinoma with E-cad promoter methylation had significantly worse overall survival, and this factor can be proposed as a potential molecular marker for poor survival 36. In other research, no differences between patients with HNSCC and a healthy control population regarding methylation of CDH1 was observed; surprisingly, this gene was frequently methylated in control samples 45. Using human carcinoma and fibroblast cell lines, Youshiura et al. investigated silencing of E-cadherin and postulated hypermethylation as a mechanism of inactivation. Furthermore, demethylating agents can be used in therapies as epigenetic drugs 56.

CpG Island Methylator Phenotype (CIMP) has been reported in cancer and was first described in colorectal cancer as a phenotype that includes methylation of multiple genes 61. CIMP status was also observed and classified in HNSCC and indicated a correlation between environmental factors and CIMP in tumour tissue. Smoking was strong associated with CIMP-positivity compared to CIMP-negativity, and poor survival was associated with CIMP-positivity (five or more methylated genes) 35. In our study, aberrant promoter hypermethylation was found in all five genes (CIMP-positive) in four cases (5.3%) of tumour samples.

Our results based on correlations between aberrant methylation and patient's characteristics indicated that positive nodal status was related to a higher methylation rate of the APC and TIMP3 genes. We also noted that an advanced tumour stage was related to higher rates of methylation of the MGMT gene, supporting a role for gene methylation in the invasive progression of HNSCC. There is increasing evidence that methylation of specific genes is related with tumour biology, such as prognosis and drug response, and is linked with particular tumour histological features 62.

We also observed that the female gender showed a trend towards a lower recurrence rate. Furthermore, female gender conferred an increased methylation level of CDH1 and TIMP3 but with a decreased MGMT gene methylation level. Based on our results, it is unclear why gender influences methylation status. Other studies showed that dietary components can influence gene expression through epigenetic mechanisms 63, 64. Notably, methylation differences related to coffee consumption were observed only in women who never used menopausal hormone therapy (MHT) and suggest that coffee may affect DNA methylation levels in immune cells of the blood 65. It is also suggested that oestrogen receptor alpha (ERα) may regulate gene expression partially via DNA methylation 66.

Furthermore, in our cohort of patients those with a family history of cancer showed more frequently methylated MGMT genes. Some studies indicated that a positive family history of cancer increases the risk of HNSCC cancer 67, 68. A connection between the presence of methylation and a family history of cancer was also shown previously, indicating a shared aetiology such as genetic predisposition 46. It is well known that environmental noise exposure can induce changes in DNA methylation and is connected with many human diseases 69. Our results suggested that habitual factors associated with cigarettes confer higher methylation of the APC and CDH1 genes. However, our patients with a history of abusing alcohol showed lower promoter methylation of the MGMT gene in the surgical margin. Numerous studies describe genes, including MGMT, that are frequently methylated due to smoking 24, 32, 70-74. It is postulated that tobacco-specific nitrosamines prevalent in tobacco cause hypermethylation of genes 73. It is difficult to explain how alcohol protects from methylation of genes involved in the DNA repair process. These results are unclear, but in tumour samples of squamous cell carcinoma of the head and neck there was an effect of alcohol use on gene hypermethylation 24, 60, 75. There is also evidence that epigenetic mechanisms such as DNA methylation play a crucial role in the pathophysiology of alcoholism 76. Multiple studies have shown alcohol-associated changes to DNA methylation which are complex and depend on numerous factors including gender and tissue type 77. Pierini et al. noticed that intensive alcohol consumption is inversely associated with methylation of only one gene out of four analysed, suggesting that alcohol exposure might affect DNA methylation in a gene-specific manner 40. Other findings have shown no correlation between methylation and cigarette smoking and alcohol intake 30, 40, 49, 50. Contrary and obscure results were shown by Puri et al.: the promoter of the MGMT gene was hypermethylated in patients with a history of alcohol use but this was significantly associated with lack of hypermethylation in another gene studied, p16 12. Because a high level of MGMT caused failure of therapy, it is noteworthy that expression of the MGMT gene imparted drug resistance of cancer cells to very popular classes of chemotherapeutic and chloroethylating agents 47, 48. Moreover, methylation of MGMT not only led to sensitivity to alkylating drugs used in chemotherapy, but also exposed a mutator phenotype 62. Polymorphisms of genes vulnerable to environmental carcinogens were especially common in those coding for enzymes involved in carcinogen metabolism, such as MGMT, as a mechanism for differential cancer susceptibility 78. In a study by Huang et al. alcohol-related head and neck cancer risks tended to vary with MGMT genotypes 79. An MGMT Ile143Val polymorphism may play a role in modulating the risk of cancer in the presence of alcohol 80. Further studies are required to clarify these interactions of genetics and environmental factors.

Aberrant methylation profiles could be caused by various factors including the above-mentioned tobacco and alcohol consumption and also Human Papilloma Virus (HPV) infection 54, 57. HPV-positive cancers have been shown to have elevated levels of methylation in the same regions of the genome 81. TIMP3 was more hypermethylated in HPV-positive than in HPV-negative patients with oropharyngeal squamous cell carcinoma 54. In addition to HPV, especially HPV type 16, as a cause of cancers of the head and neck 82, Helicobacter pylori was detected in the oropharyngeal area, leading to hypotheses about its participation in the development of cancer 83, although this result has not been confirmed. We plan to study HPV infection in our group of patients.

Our results showed aberrant DNA methylation both in tumour and surgical margins, which might be due to the existence of a heterogeneous preneoplastic field that is not detectable by basic histologic analysis and ipso facto revealing the impact of DNA aberrant methylation in tumorigenesis. Several cancers are known to display a "field effect" region outside the tumour border that harbours histological or molecular changes associated with cancer 84. The initial step of a field effect is associated with various molecular lesions. The genetically altered mucosa that remains after therapy can cause local recurrences and second primary tumours after surgery to remove the primary carcinoma 85, 86. This process arises from exposure to harmful environmental factors such as alcohol and tobacco and is related to 75% of all squamous cell carcinomas of the head and neck 1. Interestingly, research by Tan et al. and Hayashi et al. studied gene methylation in negative surgical margins in HNSCC; methylation was associated with decreased survival, and the researchers concluded that these analyses served as predictive markers of postoperative locoregional recurrence 18, 87. Importantly, epigenetic aberration was also found in histologically normal mammary tissues 88. Similar suggestions based on results indicating hypermethylation of tumour suppressor genes in control populations as a consequence of environmental factors were made by Carvalho et al. 16. Increased DNA methylation is also associated with ageing and chronic inflammation 62.

Some authors describe a “molecular surgical margin” (MSM), whose status is estimated not only by histopathologic assessment but also by the presence of molecular markers; this MSM could allow more accurate assessment of cancer recurrence. Moreover, in surgical practice, particularly regarding the head and neck, the problem is to completely remove collateral areas because any remaining cells with molecular abnormalities increase the risk of developing a second primary tumour 89.

Taken together, access to molecular biology methods has allowed recently for the precise analysis of the genetic material, but at the same time the complexity of the physiological and patho-physiological processes often make it difficult to interpret the results. More precise methods of cancer treatment depend on advances in basic research. The diagnosis and treatment of cancer should focus on the field of origin, and not only on the tumour 90.

Conclusions

Based on our observations, aberrant methylation is an important epigenetic event in HNSCC cancer. The hypermethylation of the promoter region of the MGMT and CDH1 genes could be a potential biomarker in HNSCC cancer. Moreover, clinical and habitual factors affect methylation in different manners and cause different patterns of gene promoter methylation. Our observations confirm the presence of epigenetic changes not only in the cancer cells but also in the surrounding mucosa, and represent a basis for the suggestion that appropriate cancer risk assessment based on epigenetic alterations in surgical margins may influence a patient's diagnosis and cure.