We analyzed 53 NSCLC patients who underwent surgery from 2000 to 2007 at Konyang University Hospital. This study had local ethics committee approval obtained from the Chungnam National University Hospital's Institutional Review Board. We retrieved these formalin-fixed, paraffin-embedded tissue blocks for analyzing of promoter methylation and immunohistochemical staining. We also reviewed the medical records of the NSCLC patients. The patients were characterized by clinicopathologic parameters (Table 1). Histopathological classification was assessed according to the World Health Organization (WHO) criteria. Tumor-node-metastasis (TNM) staging followed the American Joint Committee on Cancer (AJCC) staging system as revised in 2010 (7th edition). Squamous cell carcinoma (SqCC), adenocarcinoma, undifferentiated large cell carcinoma (LCC), which are major histological types of NSCLC, accounted for 32 (60.4%), 17 (32.1%), and 4 (7.5%), respectively. TNM stage following surgical resection was stage I in 24 (45.3%), stage II in 10 (18.9%), stage III in 14 (26.4%) and stage IV in 5 (9.4%). The median length of follow-up after the operation was 33 months (range, 1.5~80 months). Within follow-up periods, 23 (43.4%) patients died of cancer-related causes.

All 53 samples were obtained by dissection of tumor areas and 13 samples were obtained by dissection of normal areas apart from tumors. Deparaffinization was carried out by adding 1 mL xylene to the microtube containing paraffin-embedded tissue sections and shaking incubation at normal temperature. This step repeated two or three times then spun down at 11,000 rpm for 3 minutes. The supernatant was removed, two identical washes with 100% ethanol for 15 minutes, spinning at 11,000 rpm for 3 minutes, was followed by drying at 37℃ incubator for 15 minutes. DNA extraction was prepared using the QIAamp DNA Micro Kit (Qiagen, Hilden, Germany) as specified by the manufacturer's instructions.

Bisulfite treatment was performed to convert unmethylated cytosine to uracil and leaving methylated cytosine unchanged. Bisulfite treatment of 200 ng of each sample was subjected using the EZ DNA Methylation-Gold kit (Zymo Research, Orange, CA, USA) according to the manufacturer's instructions. The bisulfite converted DNA was eluted in 20 µL of elution buffer (Zymo Research). DNA samples were immediately stored at -20℃ until to use.

Polymerase chain reaction (PCR) assays were designed to amplify a part of the CpG islands in the CDKN2A, RARβ, and RASSF1A genes. Primers were designed by using the PSQ assay design program (Biotage, Charlotte, NC, USA). The primer sequence of CDKN2A, RARβ, and RASSF1A genes was listed in the Table 2.

PCR reaction was carried out in a volume of 50 µL with 20 ng or less converted gDNA, 5 µL of 10× Taq buffer, 5 unit Hot/Start Taq polymerase (Enzynomics, Daejeon, Korea), 4 µL of each 2.5 mM dNTP mixture, 2 µL of 10 pmole/µL PCR primers. The amplification was carried out according to the general guidelines suggested by pyrosequencing: denaturating at 94℃ for 15 minutes, followed by 40 cycles at 94℃ for 30 seconds, at 58℃ for 40 seconds, at 72℃ for 300 seconds and a final extension at 72℃ for 10 minutes. Confirmation of PCR product quality was established on 1.5% agarose gel stained with ethidium bromide. Pyrosequencing was performed using the PSQ96MA System (Biotage) according to manufacturer's instructions including single strand binding protein (PyroGold reagents; Biotage).

All the hematoxylin and eosin stained slides for each case were reviewed, and the areas of interest were marked on each slide. The tissue microarrays (TMAs) were constructed using the 5 mm punch on the Bee-Cher arrayer (UNITMA, Seoul, Korea). The corresponding region was circled on the 'donor' paraffin block using a marker pen. The samples were then arrayed on a 'recipient' block. A total 53 samples were available for the array.

Four µm-thick sections of the paraffin-embedded TMAs were deparaffinized with xylene and rehydrated in a series of graded alcohol solutions and microwave-treated for 25 minutes in a citrate buffer (pH 6.0). The antigen was retrieved with 0.01 M citrate buffer (pH 6.0) by heating the sample in autoclave (CHS-ACCE-860; JW Pharmaceutical, Seoul, Korea) at a controlled final temperature of 121℃ for 15 minutes. Endogenous peroxidase activity was blocked using 0.3% hydrogen peroxide. Nonspecific binding sites were blocked by incubating in 10% normal goat serum diluted with phosphate-buffered saline (PBS). Tissue sections were then incubated with p16^INK4A mouse mAb (1:200, Abcam, Cambridge, MA, USA) for 60 minutes at room temperature. The sections were visualized with avidin-biotin-peroxidase complex and tissue arrays were counterstained with Harris' hematoxylin. Human brain tissue served as positive internal control for p16^INK4A staining. Negative controls were incubated with PBS instead of a primary antibody.

All statistical analyses were performed using SPSS version 18 (SPSS Inc., Chicago, IL, USA) for windows. Group comparisons of the clinical variables were performed using chi-square tests or linear by linear association analysis. Comparisons of means were accomplished by independent sample t-tests or one-way ANOVA. The difference was considered to be statistically significant if the two-sided p-value was <0.05. Survival curves were plotted using the Kaplan and Meier method and statistical significance was determined by the Breslow test.

We examined the relationship between the CDKN2A methylation level and p16^INK4A immunoexpression in 53 NSCLC tumor tissue samples and also examined the clinicopathological parameters of the NSCLCs (Table 1). Loss of p16^INK4A immunoexpression was found in 36 samples (67.9%). Hypermethylation of CDKN2A was significantly associated with loss of p16^INK4A immunoexpression (p=0.045). In this study, we showed a significant correlation between the CDKN2A gene hypermethylation status, which was identified by pyrosequencing, and the loss of p16^INK4A immunoexpression, which was detected by immunohistochemical analysis, in the NSCLC tumor samples.

With regard to the clinicopathological characteristics of NSCLC, certain histopathological subtypes were found to be strongly associated with the loss of p16^INK4A immunoexpression (p=0.016), i.e., SqCC and undifferentiated LCC showed more frequent loss of p16^INK4A immunoexpression. However, no other clinicopathological parameter was associated with the loss of p16^INK4A immunoexpression.

We evaluated the methylation status of the CDKN2A, RARβ, and RASSF1A gene promoters in 13 lung cancer tissue samples and 13 corresponding normal lung tissue samples by using pyrosequencing (Table 3, Figure 1). The mean methylation levels of normal lung tissues were 1.79, 5.07, and 3.15 and of tumor tissues were 12.16, 20.9, and 8.53 for CDKN2A, RARβ, and RASSF1A, respectively. On the basis of these results, we defined promoter hypermethylation in a tumor tissue as a 2-fold increase in normal tissue5, i.e., 3.58, 10.14, and 6.30 in CDKN2A, RARβ, and RASSF1A, respectively. We further analyzed an additional 40 tumor tissues by using pyrosequencing (Table 4).

We evaluated the methylation status of the promoter regions of the CDKN2A, RARβ, and RASSF1A genes in 53 tumor tissue samples, as well as the correlation between methylation status of these promoter regions and the clinicopathological parameters of the samples (Tables 4, 5).

The incidence of hypermethylation at the CDKN2A promoter in the tumors was higher in the samples of undifferentiated LCC than in the samples of other subtypes (p=0.002) (Table 5). However, the number of LCC samples studied was too small to obtain meaningful results. The RARβ or RASSF1A gene promoters were not correlated with any clinicopathological parameters of NSCLC. We analyzed the effect of CDKN2A, RARβ or RASSF1A methylation on patient survival. Kaplan-Meier survival curves analysis showed that methylation level and patient survival were barely related (Figure 2).