• DNA hypomethylation;
  • head and neck squamous cell carcinoma;
  • pyrosequencing;
  • epigenetics;
  • smoking exposure;
  • tobacco


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Epigenetic changes have been implicated in the pathogenesis of solid tumors, including head and neck squamous cell carcinoma (HNSCC). Prior efforts have primarily examined regional promoter hypermethylation as a silencer of tumor suppressor gene expression. To analyze the global state of methylation in the HNSCC genome, we utilize pyrosequencing of repetitive elements (LINEs) to compare the state of global methylation in HNSCC to normal aerodigestive mucosa. 137 samples (119 HNSCC tumors and 18 normal mucosal tissues) were digested to extract DNA and subjected to bisulfite treatment. Treated DNA was amplified using PCR primers for the repetitive LINEs sequence and produced a heterogeneous sample of products, from many genomic loci. These products were pyrosequenced to quantitatively evaluate their global genomic methylation status. HNSCC specimens showed global hypomethylation, with a mean level of genomic methylation of 46.8% methylated with a standard deviation of 9.0. Conversely, the normal upper airway mucosa had a global methylation level of 54.0 and a standard deviation of 4.6 (Mann–Whitney p value < 0.001). The tumor specimens also showed an increasing degree of hypomethylation associated with advanced tumor stage (ANOVA p-value of 0.003). About 67% of HNSCC's are globally hypomethylated when evaluated against the minimum level of methylation in the normal mucosal specimens. Degree of global hypomethylation was associated with smoking history, alcohol use and stage in univariate analysis (p-value 0.02), however, only HNSCC diagnosis remained significant on multivariate analysis. Despite the presence of regional promoter hypermethylation, HNSCC demonstrates global genomic hypomethylation. The effects of stage, alcohol use and smoking on global hypomethylation were not independently significant. © 2007 Wiley-Liss, Inc.

Head and neck squamous cell carcinomas (HNSCCs) account for 3% of all cancers in the United States and 40,000 new cases each year.1 Although significant progress has been made in the areas of early detection, diagnosis and treatment, the 5-year survival rate for patients with HNSCC has shown only modest improvement in the past 40 years.2 Comprehensive analysis of clinical and treatment factors has shown tumor-site specific improvements in 5-year survival for cancers of the nasopharynx, oropharynx and hypopharynx, and late-stage laryngeal cancer.3

The molecular mechanisms of HNSCC carcinogenesis are undergoing intensive investigation. Despite significant genetic/epigenetic alterations found in these cancers, few known alterations correlate with the clinical outcomes of the disease. Epigenetic changes have been characterized in the pathogenesis of HNSCC. By far, the most studied epigenetic mechanism has been promoter hypermethylation of tumor suppressor genes including: Cyclin A1, MGMT, DCC and p16.4 Methylation of cytosine-guanine dinucleotides by the enzyme class of DNA methyltransferases transfers a methyl group from S-adenosyl-methionine and is associated with gene silencing. Promoter hypermethylation has shown marked promise for identification of biomarkers and functional tumor suppressor genes implicated in the carcinogenesis of HNSCC.

Reports have noted a paradox between regional DNA hypermethylation in the promoters of tumor suppressor genes and a state of global genomic hypomethylation in solid tumors over the last few years.5, 6, 7 In fact, there are several reasons to believe that hypomethylation may be a significant factor driving oncogenesis. Mice with disruption of DNA methyltransferase 1 (DNMT1) function demonstrate significant genomic hypomethylation in all tissues and develop aggressive T-cell lymphomas with chromosomal instability.8 Murine embryonic cells with homozygous deletion of DNMT1 exhibit significantly elevated rates of genetic deletions.9 DNMT mutation can lead to chromosomal instability in humans as well. Mutation of DNMT3b leads to numerous chromosome aberrations.10 Meta-analysis of DNA global hypomethylation across various cancer types suggest a correlation between global hypomethylation and tumor stage.7 Additionally, extensive demethylation of centromeric sequences is common in human tumors and may play a role in aneuploidy.11

To date, the global methylation status of head and neck squamous cell cancer has not been adequately studied in patients. In a limited cohort of only 6 tumor samples, Chalitchagorn et al.12 showed a promising trend toward hypomethylation in HNSCC versus normal mucosal tissues using combined bisulfite restriction analysis. In contrast, Piyathilake et al. found a higher percentage of cells positive for 5-methyl-cytosine immunostaining in 39 HNSCC samples versus normal oral epithelial cells. This evaluation was looking at the percentage of cells positive for staining and staining score, and does not necessarily correlate to the degree of DNA methylation per cell. To quantitate this finding globally, our group utilized the technique of pyrosequencing of LINE sequences to examine the degree of global genomic hypomethylation in head and neck cancer tumor specimens compared to normal control mucosal tissue in an effort to determine if global DNA hypomethylation exists in head and neck cancer.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References


All samples were analyzed by the Pathology Department at Johns Hopkins Hospital. Tissues were obtained via Johns Hopkins Institutional Review Board approved protocols. Normal samples were microdissected and DNA prepared from the mucosa. Normal upper aerodigestive mucosa was obtained during upper airway procedures for benign disease, primarily Uvulopalatopharyngoplasty. Samples were confirmed to be upper aerodigestive mucosa by pathology and originated primarily from the soft palate. Tumor and normal samples were chosen randomly from available tissue banks comprised of patients treated at the Johns Hopkins Hospital from 1998 to 2006. Tumor samples were confirmed to be head and neck squamous and subsequently microdissected to separate tumor from stromal elements to yield at least 80% tumor cells. Microdissection was initially performed by making hematoxylin and eosin stained slides of tissue blocks. After reviewing the slides with a pathologist, areas of tumor were outlined and the tissue blocks were then cut with areas of interest dissected by scalpel. Tissue DNA was extracted as described later.

DNA extraction

Samples were centrifuged and digested in a solution of detergent (sodium dodecylsulfate) and proteinase K, for removal of proteins bound to the DNA. Samples were first purified and desalted with phenol/chloroform extraction. Digested sample was subjected twice to ethanol precipitation, and subsequently resuspended in 500 μl of LoTE (EDTA, 2.5 mmol/l and Tris–HCl, 10 mmol/l) and stored at −80°C.

Bisulfite treatment

DNA from salivary rinses was subjected to bisulfite treatment, as described previously.13 In short, 2 μg of genomic DNA was denatured in 0.2 M of NaOH for 30 min at 50°C. This denatured DNA was then diluted into 500 μl of a solution of 10-mmol/l hydroquinone and 3-M sodium bisulfite. This was incubated for 3 hr at 70°C. After, the DNA sample was purified with a sepharose column (Wizard DNA Clean-Up System; Promega, Madison, WI). Eluted DNA was treated with 0.3 M of NaOH for 10 min at room temperature, and precipitated with ethanol. This bisulfite-modified DNA was subsequently resuspended in 120 μl of LoTE (EDTA, 2.5 mmol/l and Tris–HCl, 10 mmol/l) and stored at –80°C.

LINES amplification

The LINES element PCR was designed for pyrosequencing-based methylation analysis. A 50-μl PCR was carried out in 60 mM Tris–HCl pH 8.5, 15 mM ammonium sulfate, 2 mM MgCl2, 10% DMSO, 1 mM dNTP mix, 1 unit of Taq polymerase, 50 pmol of the forward primer (5′-TTTTGAGTTAGGTGTGG GATATA-3′), 50 pmol of biotinylated reverse primer (5′-AAA ATCAAAAAATTCCCTTTC-3′) and 50 ng of bisulfite-treated genomic DNA. The reverse primer was biotin-labeled, so the final PCR product could be purified using Sepharose beads. PCR cycling conditions were 95°C for 30 s, 50°C for 30 s and 72°C for 45 s for 45 cycles.


Many methods exist to evaluate global methylation status.7 We used methods described by Yang et al.14 utilizing pyrosequencing to quantitate methylation of the repetitive LINEs sequences. The biotinylated PCR product was purified and made single-stranded to act as a template in a pyrosequencing reaction as recommended by the manufacturer using the Pyrosequencing Vacuum Prep Tool (Pyrosequencing, Westborough, MA). In brief, the PCR product was bound to streptavidin sepharose beads (Amersham Biosciences, Uppsala, Sweden). These sepharose beads containing the boundPCR product were purified, washed, denatured using a0.2-M NaOH solution and washed again. Finally, 0.3-μM pyrosequencing primer (5′-GGGTGGGAGTGAT-3′) was annealed to the purified single-stranded PCR product and pyrosequencing was performed using the PSQ HS 96 Pyrosequencing System (Pyrosequencing). Methylation quantification was performed using the provided software.

Statistical analysis

Statistical analyses were performed using STATA/SE 9.1 (StataCorp LP, College Station, TX) values were calculated using Mann–Whitney U test, analysis of variance (ANOVA) or χ2 as detailed.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

Study population

In Table I, our 2 sample populations are compared. The ages of the patients from which the normal mucosal samples were obtained are slightly lower than the population of head and neck cancer patients, mean ages 48 years (range, 28–76) and 58 years (range, 31–88), respectively. Both sample groups have a similar male and Caucasian predominance. Tumor patients were composed of those without tobacco history (23%), current smokers (45%) and those who had quit smoking (32%). Tumor samples (119) were obtained from patients with Stage I (15%), Stage II (11%), Stage III (15%) and Stage IV (59%) lesions at the time of presentation. These were from primary tumors of the oral cavity (41), oropharynx (33), nasopharynx (1), hypopharynx (6), larynx (18), nose/sinus (2) and other/neck (18).

Table I. Demographics of Study Patients
 Head and neck cancer (N = 119)Normal patients (N = 18)
 Mean (range)58.2 ± 11.4 (31–88)47.9 ± 14.9 (28–76)
 M92 (77%)14 (78%)
 F27 (23%)4 (22%)
 Caucasian95 (80%)17 (94%)
 Black18 (15%)1 (6%)
 Other6 (5%)0 (0%)
Smoking status
 Never27 (23%)11 (61%)
 Former38 (32%)3 (17%)
 Current54 (45%)4 (22%)
 I18 (15%)
 II13 (11%)
 III18 (15%)
 IV70 (59%)
 Nose/Sinus2 (2%)
 Oral Cavity41 (34%)
 Nasopharynx1 (1%)
 Oropharynx33 (27%)
 Hypopharynx6 (5%)
 Larynx18 (15%)
 Neck/other18 (15%)

Global hypomethylation assay

Table II summarizes the results of our global methylation assay. Overall the differences were found comparing tumor to normal (p = 0.0001), smoking status (p = 0.02) and stage (p = 0.003). Race, sex, tumor site and lymph nodes showed no statistically significant differences between groups.

Table II. Results of Global Methylation Assay
 AverageSD95% CIp-value
  1. SD, standard deviation; CI, 95% confidence interval. p values calculated by Mann–Whitney U comparing the two groups (tissue, sex, race, nodes), smoking p-values calculated by Mann–Whitney U test comparing former to never smokers and current to never smokers, stage p value from ANOVA, individual stages and sites calculated by Mann–Whitney U comparing site to all others.

Tissue   0.0001
Sex   0.22
Race   0.80
Smoking status    
Stage   0.003
 Nose/sinus61.217.5 0.13
 Oral cavity46.39.2(43.4–49.2)0.92
 Nasopharynx43.7  0.58
Nodes   0.36

Head and neck squamous cancer specimens showed a marked trend toward hypomethylation compared to control samples (Fig. 1). The normal upper airway mucosal samples demonstrated a mean global methylation level of 54.0% (standard deviation 4.6). The tumor specimens showed a mean level of global methylation of 46.8% with a standard deviation of 9.0. This degree of hypomethylation in the tumor cohort was statistically significant (p < 0.001, Mann–Whitney). About 67% of head and neck squamous cell cancers are hypomethylated when compared to the most hypomethylated normal sample.

thumbnail image

Figure 1. Global hypomethylation in head and neck squamous cell cancer compared to upper aerodigestive mucosal controls. Tumors show increased variability and global hypomethylation assessed by LINEs-specific pyrosequencing (46.8%, standard deviation of 9.0) versus normal upper airway mucosa (54.0%, standard deviation of 4.6), Mann–Whitney p value < 0.001.

Download figure to PowerPoint

An increasing degree of hypomethylation is associated with increased stage of the tumors (see Fig. 2). The ANOVA revealed a p-value of 0.003 confirming the relationship of increasing tumor stage with increasing hypomethylation. Comparisons of the individual stages to each other utilizing a Mann–Whitney U test did not achieve statistical significance, but when considered separately, each individual stage was significantly different than the normal population: Stage 1 (p = 0.04), Stage 2 (p = 0.002) Stage 3 (p = 0.006) and Stage 4 (p < 0.001).

thumbnail image

Figure 2. Global DNA hypomethylation compared to increasing tumor stage. All stages show higher proportion of hypomethylated samples compared to control tissues. Increasing stage shows increased trend of global hypomethylation. Normal mucosa is 54.0% methylated, Stage I, II, III and IV lesions have a mean methylation percent of 48.4, 47.0, 48.8 and 45.7, respectively.

Download figure to PowerPoint

Degree of global hypomethylation did relate to smoking status (Fig. 3). Cancer patients who had no tobacco exposure had a mean level of global methylation of 49.7 (standard deviation 9.2). However, smokers showed global hypomethylation. Patients who had quit smoking had a mean level of 46.7 with an 8.2 standard deviation, while current smokers had the lowest mean levels of 45.6 and 9.4 standard deviation. Current smokers (p = 0.02) showed significantly hypomethylated lesions compared to those who have never smoked and smokers who quit tended toward hypomethylation also (p = 0.07). By Mann–Whitney U test within the HNSCC cohort, current smokers were not distinguishable from smokers who quit (p = 0.64), with respect to global methylation. Taken as a group altogether, smokers (current and former) showed global hypomethylation compared to those who had never smoked (p = 0.02). To evaluate to effect of alcohol use, we performed a subset analysis of 76 HNSCC patients with characterized drinking habits. Of the 27 nondrinking patients, the mean level of global methylation was 51.1 (SD 9.4), and in the 49 described as drinkers the mean level was 46.7 (SD 6.3), associated with a statistical significance of p < 0.015 (Mann–Whitney U test). We then conducted a multivariate analysis (see Table III) of the influence of smoking, tumor presence, stage and alcohol use within the cohort of tumor patients and normal subjects. In multivariate analysis controlling for stage and alcohol use, tobacco use did not reach statistical significance (p = 0.11), but tumor diagnosis remained a significant difference (p < 0.001).

thumbnail image

Figure 3. Smoking status compared to global DNA methylation in patients with HNSCC. Patients who had no history of smoking had a mean global level of hypomethylation of 49.7 with a standard deviation of 9.2. Smokers who had quit had a level of 46.7 with 8.2 standard deviation, while current smokers had the lowest levels of 45.6 and 9.4 standard deviation. Mann–Whitney showed current smokers (p = 0.02) differed from never smokers and smokers who quit has a degree of hypomethylation (p = 0.07). As a group smokers (current and quit) were different from never smokers (p = 0.02), while current smokers could not be differentiated from smokers who quit (p = 0.64).

Download figure to PowerPoint

Table III. Results of Multivariate Analysis
VariableUnivariate resultMultivariate result
  1. A multivariate model was constructed to define the relationship between degree of global hypomethylation and the independent variables: tissue type, stage, alcohol use and smoking status. p values from the regression model are shown. A regression model that did not include tissue type was also conducted with p values for stage, alcohol and smoking status of (0.71, 0.22, 0.11, respectively). Smoking and alcohol use among patients were often correlated.

Tissue typep < 0.001p < 0.001
Stagep = 0.003p = 0.71
Alcohol usep < 0.015p = 0.22
Smoking statusp = 0.02p = 0.11


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References

DNA hypermethylation studies continue to elucidate epigenetic silencing of novel candidate tumor suppressor genes. However, evidence exists showing a decrease in the global genomic methylation state in solid tumors. Using LINE elements, we examined the global level of DNA methylation in primary HNSCC and compared it with normal aerodigestive mucosa. LINE elements have been used as a surrogate for global methylation level because of their marked presence in the genome, and this reproducible assay has been shown to correlate with methylation-specific restriction digests.14 There were differences in patient ages reflecting the fact that HNSCC has a median presentation in the late 50s versus the normal samples from a population that presented for benign disease. About 67% of head and neck squamous tumors in our study showed a degree of global hypomethylation that exceeded the measure of any normal mucosal specimen. Although there is a prominent degree of hypomethylation in the tumor samples, a few samples are actually hypermethylated. Our results suggest that, despite promoter hypermethylation of individual tumor suppressor genes, head and neck cancers are globally hypomethylated. These findings are actually consistent with previous reports of other solid tumors that have shown global DNA hypomethylation, and regional areas of promoter hypermethylation.7 These findings suggest an opposite conclusion to the report by Piyathilake et al.15 showing a higher percentage of cells with antibody staining to a5-methyl-CpG antibody in cancers versus normal mucosa. Although these results were not accounted for by increased DNA content of the cancer, it would require further efforts to see if this was a result of reduced chromosomal compaction, as well as the known presence of regional hypermethylation in cancers, or variation from qualitative assessment. Also, a small subset of our tumors did indeed have hypermethylation, indicating what may be an overall dysregulation of methylation in the genome of cancer cells.

To further elucidate the potential role of global hypomethylation in carcinogenesis, we assessed the relationship between increasing tumor stage and the degree of hypomethylation. Global methylation mean levels were reduced in the Stage IV lesions 45.7 compared to the Stage I/II/III (48.4/47.0/48.8), suggesting this epigenetic change worsens as tumorigenesis progresses. Although many molecular alterations continue to increase in frequency as stage increases, this does not necessarily imply causation, but remains an association.

The association between smoking status and global state of methylation in tumors suggest tobacco exposure may be causing genome-wide damage apparent in this epigenetic assay. In our study, patients who had quit smoking (p = 0.07), and current smokers (p = 0.02), with HNSCC, had lower levels of global methylation. This association was present on univariate analysis. Smoking has previously been linked to promoter methylation of genes such as SFRP in HNSCC16 and TSLC1/IGSF4, P16, MGMT in nonsmall cell lung cancer.17, 18, 19 Smoking has not been previously shown to cause global genomic hypomethylation. It has been reported that smoking is associated with reduced levels of vitamins including B12, which is required for synthesis of S-adenyl-methionine.20 This might be one factor causing the association of hypomethylation and smoking, but will require further study. Subset analysis within the tumor group also showed an association of alcohol use with hypomethylation, although it is difficult to separate tobacco and alcohol effects as they tend to be confounding characteristics in this patient population. Alcohol ingestion can also affect the B12 pathway. Dietary differences were not collected here, which can also affect methylation, especially when considered globally. Subsequent studies are planned to look at the causality of the 2 main head and neck cancer risk factors: tobacco use and ethanol in contributing to global hypomethylation.

To analyze the possible functional role of DNA hypomethylation, several theories exist supporting its contribution to molecular pathogenesis of cancer.21 Prominent theories include the direct reactivation of transposable repetitive elements, such as LINEs and Alus which have been shown to be activated by hypomethylation,22 and which commonly have deleterious effects such as functional repression through insertion into transcribed sequences. Hypomethylation may also represent a loss of imprinting in the genome that causes activation of a wide spectrum of genes that convey various growth advantages or even possible oncogenes.21 Alternatively, global hypomethylation may be an epiphenomenon associated with genomic alterations common to most solid tumors. Subsequent work will be necessary to determine the causation of global hypomethylation, and factors that influence it (e.g., tobacco exposure).


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. References