Multiplexed methylation profiles of tumor suppressor genes and clinical outcome in lung cancer

Six NSCLC cell lines consisting of three adenocarcinoma cell lines (A549, H522 and H358), two large carcinoma cell lines (H460, H661), and one squamous cell carcinoma cell line (H226), as well as one small cell lung cancer cell line (SCLC) (H841) were obtained from the American Type Culture Collection (Rockville, MD, US), grown in RPMI-1640 medium (Sigma) supplemented with 10% fetal bovine serum, and collected under standard tissue culture protocols. The lung cancer cell line, H460, was included in all sample runs in order to test the reproducibility of the MS-MLPA test.

The study cohort consisted of a series of archived paraffin-embedded blocks from 54 NSCLC patients. Patients with local disease (stage I to resectable stage III) were treated surgically and those with advanced disease (stage III and IV) received systemic and/or local treatment. Primary lung tumors were collected after institutional review board approval and handled anonymously following ethical and legal protection guidelines of human subjects. The observation period ranged from 2 to 79 months, with a median follow-up of 20,5 months. Inclusion criteria of newly diagnosed lung cancer patients were based on the histopathologic information, covering from early to advanced stages. It was also required to have tissue material available for obtaining high-quality DNA for methylation analyses. Of the 54 NSCLC patients, 22 had TNM Stage I-II, 18 had Stage III, and 14 stage IV defined under standard criteria [29]. The tumors were histologically classified as adenocarcinomas (n = 32), squamous cell carcinomas (n = 21) and large cell carcinoma (n = 1) according to the histological typing of lung tumors of the World Health Organization [29]. Demographic and clinicopathologic information of the lung cases analyzed is described in Table 1.

Genomic DNA from cell lines and tissue was extracted using standard methods. Paraffin-embedded tissues were macro-dissected based on hematoxylin-eosin evaluations to ensure a minimum of 75% of tumor cells [30]. Corresponding slides were digested using proteinase K (Roche Diagnostics GmbH, Mannheim, Germany) overnight before DNA extraction. Concentration and purity of DNA samples were determined with a ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). DNA quality was evaluated based on 260/280 ratios of absorbances and the integrity was also checked by gel electrophoresis analysis using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto CA).

The present study used the MS-MLPA probe set ME003 (MRC-Holland, Amsterdam, The Netherlands) which can simultaneously check for aberrant methylation at one or two CpG dinucleotides of the following proven or suspected 18 tumor suppressor genes (Table 2). Probe sequences, gene loci and chromosome locations can be found at http://​www.​mlpa.​com (date of accession: 25-May-2010). Several genes were evaluated by two probes, which recognized different Hha1 restriction sites in their promoter regions. The experimental procedure was carried out and results analyzed according to the manufacturer's instructions, with minor modifications. In short, DNA (200 ng) was dissolved up to 5 μl TE-buffer (10 mM Tris pH 8.2, 1 mM EDTA pH 8.0), denatured and subsequently cooled down to 25°C. After adding the probe mix, the probes were allowed to hybridize (16 h at 60°C). Subsequently, the samples were divided in two: one half of the samples were ligated, whereas for the other part of the samples, ligation was combined with the HhaI digestion enzyme. This digestion resulted in ligation of only the methylated sequences. PCR was performed on both parts of the samples in a volume of 50 μl containing 10 μl of the ligation reaction mixture using a thermal cycler (MJ Research Inc., Waltham, MA, USA), with 35 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 30 s and extension at 72°C for 1 min with a final extension of 20 min at 72°C. Aliquots of 2 μl of the PCR reaction were combined with 0.12 μl LIZ-labeled internal size standard (Applied Biosystems, Foster City, CA, USA) and 9.0 μl deionized formamide. After denaturation, fragments were separated and quantified by electrophoresis on an ABI 3700 capillary sequencer and the Peak Scanner v1.0 analysis software (both Applied Biosystems). Peak identification and values corresponding to peak size in base pairs (bp), and peak areas were used for further data processing. Automated fragment and data analysis was performed exporting the peak areas to an excel-based analysis program (Coffalyser V8, MRC-Holland). For hypermethylation analysis the 'relative peak value' or the so-called 'probe fraction' of the ligation-digestion sample is divided by the 'relative peak value' of the corresponding ligation (undigested) sample, resulting in a so-called 'methylation-ratio' (M-ratio). Aberrant methylation was scored when the calculated M-ratio was ≥0.30, corresponding to 30% of methylated DNA. The methylated ratios were interpreted as absence of hypermethylation (0.00-0.29), mild hypermethylation (0.30-0.49), moderate hypermethylation (0.50-0.69), and extensive hypermethylation (>0.70). In genes with more than one probe, their ratios were calculated independently for methylation analysis.

Coefficients of variation for each probe were estimated based on the ratio of the standard deviation and the respective mean of four replicates of the H460 cell line. Associations among MS-MLPA methylation and tumor stage and grade were evaluated using non-parametric Wilcoxon-Mann-Whitney and Kruskall-Wallis tests using Bonferroni adjustment for multiple testing. Associations between methylation candidates were analyzed using Kendall's tau ß test, considering only two-sided p-values 0.05 to be statistically significant. For each probe of the assay, methylation was scored when the calculated M-ratio was ≥0.30. Associations of methylation of each gene probe with overall survival were also evaluated using the log-rank test in those cases for which follow-up information were available. Overall survival time was defined as the months elapsed between surgery and death as a result of disease (or the last follow-up date). Patients who were alive at the last follow-up or lost to follow-up were censored. Survival curves were plotted using the standard Kaplan-Meier methodology [31]. Statistical analyses were performed using the SPSS statistical package (SPSS 17.0.1 for Windows 2009, Chicago, IL, USA).

In order to test the reproducibility of the assay, a lung cancer cell line, H460, was included as control samples each assay run. The methylation ratios of these replicated experiments of the cell line analyzed and their coefficients of variation are shown in additional file 1, Table S1. Using the thresholds defined above for methylation detection suggested high reproducibility of the methylation profiles. In conclusion, these initial analyses revealed reproducible results allowing methylation assessment using the selected panel of candidate genes.

The methylation profiles of the 18 genes under study were initially tested in seven cell lines derived from lung tumors of different histopathologic variants. Additional file 2, Table S2 provides an overview of the methylation patterns of these cell lines grouped based on the histopathology of the tumors from which these lung cancer cell lines were derived from. The percentual methylation for each gene is provided as well. Several genes: SCGB3A1, ID4, SFRP5, CCND2, and CACNA1A, were found methylated in at least 4 out of the 7 cell lines analyzed, covering various histopathologic types. These initial analyses suggested that the panel of candidate genes selected could be appropriate to detect aberrant methylation profiles in human lung tumors.

In the next step, we tested whether MS-MLPA could be applied to lung tumors (Figure 1, Table 3). Overall, the most frequent hypermethylated genes found by MS-MLPA were DLC1 (50%), SCGB3A1 (50.0%), CCND2 (48.1%), ID4 (46.3%), BNIP3 (44.4%), RUNX3 (42.5%), and PRDM2 (40.7%). Notably, genes methylated in lung tumor specimens frequently overlapped with those found to be methylated in the lung cancer cell lines as shown above. Promoter hypermethylation of genes previously reported methylated in lung cancer included PRDM2, RUNX3, RARB, SCGB3A1, TWIST1, SFRP4, DLC1, SFRP5, CCND2, BCL2 and TIMP3 (Reviewed in additional file 3, Table S3). Methylation was newly identified for HLTF, ID4, BNIP3, H2AFX, CACNA1G, CACNA1A, and TGIF. The percentual methylation rates of each gene depending on the different clinicopathologic variables are shown in Table 3. The genes more frequently methylated in adenocarcinomas were: RARB, TWIST1, and CACNA1A, while the most commonly methylated genes in squamous tumors were SCGB3A1, ID4, SFRP4, SFRP5, DCL1, BNIP3, H2AFX, CACNA1G, TGIF, TIMP3 and BCL2. Statistically significantly different methylation rates were observed for ID4-2 (p = 0.011), DCL1 (p = 0.019), BNIP3 (p = 0.003), H2AFX (p = 0.001), H2AFX-2 (p = 0.005), CACNA1G (p = 0.007) and TIMP3 (p = 0.021) when comparing squamous versus adenocarcinoma cases. When comparing methylation rates in localized tumors versus metastatic disease, the methylation of RUNX3 (p = 0.013), SCGB3A1-2 (p = 0.008), SFRP4-2 (p = 0.022), and DLC1 (p = 0.016) was significantly associated with the presence of metastatic disease. The methylation of the same genes was also associated with tumor stage RUNX3 (p = 0.040), SCGB3A1 (p = 0.032), SFRP4 (p = 0.033), and DLC1 (p = 0.035). RARB methylation (p = 0.028) was associated with the Karnofsky status. Methylation of several genes was simultaneously present in the lung tumors analyzed, as revealed by Kendall's tau correlations shown in additional file 4, Table S4. We did not find any significant association between methylation of the genes under study and age, gender, smoking history (data not shown). Methylation rates regarding histology lung subtypes, differentiation grade and tumor stage (comparing localized versus advanced disease), are provided in Table 3. In conclusion, this set of analyses suggested that the panel of candidate genes selected could be of clinical relevance for the clinicopathologic staging of human lung tumors.

In the next step, we tested whether MS-MLPA could be applied to differentiate patients with different clinical outcome, using overall survival as the clinical endpoint. We observed that patients with tumors methylated for the HTLF gene showed an overall survival significantly shorter as compared to patients unmethylated for HTLF (log rank, p = 0.035; Figure 2A). In contrast, survival was significantly longer in patients with methylation for SFRP5 (probe 2) (log rank, p = 0.021; Figure 2B); and TIMP3 (log rank, p = 0.030; Figure 2C), as compared to patients with no aberrant methylation of these genes. Importantly, this set of analyses indicated that the methylation of three genes was significantly associated with overall survival, suggesting that the panel of candidate genes under analyses could be of clinical relevance as prognosticators of the clinical outcome of patients affected with lung tumors.