In the present study, 54 pairs of lung cancer and adjacent NT samples were obtained from 54 patients who underwent surgical resection from January 2014 to June 2014 at the Department of Pathology, Shanghai Zhongshan Hospital (Shanghai, China). Of these patients, 12 were diagnosed with squamous cell carcinoma and 42 were diagnosed with adenocarcinoma. All these samples were obtained with informed consent and were stored at −80°C until later total DNA extraction. This research was approved by the Institutional Review Board of Shanghai Zhongshan Hospital (Shanghai, China).

No more than 30 mg of tissue was obtained using sterilized operating scissors. Tissue was then ground in tissue lyser (Jingxin, Shanghai, China) for 80 sec at 65 Hz. Whole-genome DNA extraction was then performed using ALL Prep DNA/RNA Mini kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Sodium bisulfite was used to convert the extracted DNA using an EZ DNA Methylation-Gold kit (Zymo Research Corp., Irvine, CA, USA), in which ~500 ng of DNA was used as the proper addition, according to specification. The final elution volume was 10 µl.

Converted DNA was amplified using a LightCycler 480 (Roche Diagnostics GmbH, Mannheim, Germany). The whole reaction volume was 10 µl, which consisted of, using the MGMT gene as an example: 1 ng/µl DNA template, 1X Master Mix (Roche Diagnostics GmbH), 0.5 µM primers and 4 mM magnesium ions. Polymerase chain reaction (PCR)-grade water was used to bring the final reaction volume to 10 µl. The detailed amplification thermocycling protocol was set to the following: Preheat for 10 min at 95°C before starting a 50-cycle process involving 10 sec at 95°C, 20 sec at a temperature between 61 to 55°C for 20 sec (temperature dropped 2.2°C per sec), and 20 sec at 72°C. The following MS-HRM melting protocol was used: Heating at 95°C for 1 min, followed by 40°C for 1 min (29), 65°C for 1 sec, and continuous heating to 95°C at a ramp rate of 0.02°C per second. To ensure veracity and repeatability, each reaction was conducted in duplicate. 5-Aza-dc-treated Jurkat Genomic DNA and CpG Methylated Jurkat Genomic DNA (New England Biolabs, Ipswich, MA, USA) that had been subjected to bisulfite conversion were used as fully methylated and unmethylated control DNA samples, respectively. The former was added into the latter in gradient proportions (0, 5, 10, 25, 50, 75, 90 and 100%) and used as artificial DNA standards of different methylation levels. These DNA standards were used in each assay to establish gradient standard curves. These curves were then used to evaluate the methylation levels of the tumor and normal samples.

In the present study, pyrosequencing was used for 3 of the 6 genes (PCDHGB6, HOXA9 and miR-126), to validate the accuracy of the present results. Pyrosequencing was performed at Shanghai Medical College, Fudan University (Shanghai, China). Table I shows the basic primer information for both MS-HRM and pyrosequencing. Primers were designed by MethPrimer (http://www.urogene.org/cgi-bin/methprimer/methprimer.cgi), which is used for designing methylation PCR primers. Subsequent to bisulfite conversion, cytosine was converted to thymine (except for CpG islands) and the newfound sequence was used as the template for primer design.

SPSS v17.0 (SPSS, Inc., Chicago, IL, USA) was used for all statistical analysis. The dependence between the HRM-assessed methylation status and temperature difference was analyzed using simple linear regression. To verify the sensitivity and specificity of high and low methylation status, the receiver-operating characteristic (ROC) curve was generated and the AUC was calculated to obtain the highest specificity and sensitivity. Differences between tumor and surrounding tissue were evaluated by t-test and P<0.05 was considered to indicate a statistically significant difference. Additionally, Pearson's correlation coefficient was evaluated to examine the consistency between HRM assay and pyrosequencing of PCDHGB6, HOXA9 and miR-126.

In the present MS-HRM assay, melting temperature was obtained as a T_m value in the T_m calling analysis. This refers to the temperature at which half of the double-stranded DNA melts into single strands, thus undergoing a sharp decline in fluorescence intensity (8,9). Fig. 1 shows the melting curves of control samples of the six genes in the T_m calling analysis. A gradient of diluted methylated DNA with unmethylated DNA (0–100%) subsequent to bisulfite conversion were used for each gene as controls. Melting profiles of all these gene promoters showed higher T_m in methylated controls and lower T_m in controls without methylation. Among them, T_m values of SOCS3 gene exhibited the largest span, from 77 to 84°C (Fig. 1D), whereas small temperature differences were found in HOXA9 and miR-126 as less than 2°C (Fig. 1C and F). Based on the difference in melting temperatures between the amplicons, single nucleotides can be distinguished (30).

To further explore this association, a correlational analysis was conducted between temperature difference and methylation status using a simple linear regression (Fig. 2). Every standard curve yielded a corresponding temperature difference value. As shown in Fig. 2, there was a negative correlation between the methylation status of DNA standards in a series of dilutions and temperatures. HOXA9, PCDHGB6, MGMT and miR-126 exhibited good liner association with significant R² values of 0.994, 0.982, 0.949 and 0.998, respectively. From these data, accurate methylation degrees could be calculated.

According to the generated standard curves, 54 pairs of tumor and NT tissues were investigated. Methylation levels of each gene are shown in Fig. 3. From the experimental results, PCDHGB6 methylation was found in 35 of the 54 tumors (64.8%) and HOXA9 methylation was found in 23 of the 54 tumors (42.6%). miR-126 also had a high methylation frequency (68.5%), but methylation of miR-126 was found to be nonspecific in normal tissue, with a frequency of 46.3%. Notably, no promoter methylation of either SOCS3 or NORE1A was identified in the tumor or normal tissue. As for MGMT, which has been affected in several malignancies that include colorectal and lung cancer, did not show any prominent impact (14 positive samples out of the 54 tumor samples) in the early diagnosis of NSCLC.

The precise methylation values obtained from our linear regression analysis were then compared with pyrosequencing. In the present study, methylation values were derived from an average of all the CpG sites. Ten pairs of malignant and control tissue from each selected DNA sequence were chosen to detect methylation degrees of PCDHGB6, HOXA9 and miR-126 by pyrosequencing with 10, 6 and 5 possible mutations of CpG sites, respectively. Fig. 4 illustrates the pyrosequencing results of PCDHGB6 from one tumor tissue sample. The blue sections indicate successful detection of CpG sites. Fig. 5 shows comparisons of methylation degrees, as tested by MS-HRM and pyrosequencing. Every two samples joined by a line represent a pair of tumor and para-carcinoma tissue excised from the same patient. As shown, PCDHGB6 had the most closely associated values of the three genes, while HOXA9 came second. The sensitivity of assessing miR-126 methylation by MS-HRM was generally lower than by pyrosequencing. To examine statistical significance, Pearson's correlation coefficient was evaluated for PCDHGB6, HOXA9 and miR-126, deriving values of P<0.001, P=0.001 and P=0.03, respectively. Collectively, these results demonstrated a statistically significant coincident tendency between MS-HRM and pyrosequencing (P<0.05).

The association between clinical manifestations of NSCLC and the MS-HRM analysis for selected genes are shown in Table II (31). Based on the present study, the methylation frequency of PCDHGB6 and HOXA9 increased with disease progression, while the methylation frequency of PCDHGB6 was higher than that of HOXA9 in every disease stage. For PCDHGB6, the frequency was at 53.8% at stage I, with the ratio reaching 83.3% at later stages. In the case of HOXA9 gene, the ratio increased between 38.5% at stage I to 66.7% at stages III–IV. When compared with the PCDHGB6 and HOXA9 genes, MGMT showed no evident association with tumor-node-metastasis (TNM) stage and possessed low detection rates in both normal and tumor tissue. For the miR-126 gene, methylation frequency was significantly increased in early stages, with 73.1% methylation frequency in stage I. Additionally, results from all selected genes showed that males appeared to be more susceptible than females. In the case of histopathological classification, these four genes appeared to have increased sensitivity in squamous cell carcinoma, with detection rates of ≥75%. PCDHGB6 and miR-126 were both reliable biomarkers for the detection of adenocarcinoma, with positive rates of 54.8 and 61.9%, respectively.

A multi-gene analysis was then conducted to evaluate the sensitivity and specificity of methylation. Results were analyzed by ROC curve. As shown in Table III, values were calculated using SPSS software, with various combinations used to acquire the best result. PCDHGB6 had the highest sensitivity (66.7%) out of the four genes, with 75.9% of the patients having at least two methylated genes. Finally, the combination of four genes resulted in the highest sensitivity and specificity (85.2 and 81.5%, respectively), with the AUC being 0.891 (Fig. 6).