Author of

• 11481

  +

  P2.02 - Poster Session 2 - Novel Cancer Genes and Pathways (ID 148)

  • Event: WCLC 2013
  • Type: Poster Session
  • Track: Biology
  • Presentations: 1
  • Moderators:
  • Coordinates: 10/29/2013, 09:30 - 16:30, Exhibit Hall, Ground Level

  • 11484

    +

    P2.02-003 - DNA methylation and expression patterns in adenocarcinomas of the lung of never-smokers with discordant methylation and expression patterns of homeobox related genes (ID 856)

    10:04 - 10:21  |  Author(s): C. Kolbert
    • Abstract

    Background
    Lung cancer occurs in a significant number of never-smokers. Epigenetic changes in lung cancer potentially represent important diagnostic, prognostic and therapeutic targets. Accordingly, we sought to determine if there are differences in DNA methylation between lung adenocarcinomas and adjacent non-malignant lung tissue in never-smokers.

    Methods
    Using the Illumina Infinium HumanMethylation27 BeadChip we compared the DNA methylation profiles of 28 adenocarcinomas of the lung of never-smokers with their paired adjacent non-malignant lung tissue. The β value represents each methylation data point as the ratio of fluorescent signals between the methylated sites and the sum of methylated and unmethylated sites, which ranges continuously from 0 (unmethylated) to 1 (fully methylated). We used the Mann Whitney U test to compare β values between groups using JMP (SAS Institute Inc. Cary, NC USA). Then, using the Illumina Human WG DASL beadchip, we correlated differential methylation changes with gene expression changes from the same 28 samples. We validated our findings with 24 samples of lung adenocarcinomas and paired non-malignant tissue from The Cancer Genome Atlas (TCGA). Database for Annotation, Visualization, and Integrated Discovery was used to determine gene enrichment.

    Results
    We observed a distinct separation in methylation profiles between tumor and adjacent nonmalignant lung tissue using principal component analysis. Tumors were generally hypomethylated (β=0.15) when compared to adjacent non-malignant tissue (β=0.17; p=0.02). There were 1906 differentially methylated (Bonferroni corrected p value <0.05) CpG sites between tumor and adjacent non-malignant lung tissue. Of these sites, 1198 were within classically defined CpG islands where tumors (median β=0.38) were hypermethylated compared to adjacent non-malignant tissue (median β=0.22; p<0.0001), and 708 sites were outside of CpG islands where tumors (β=0.49) were hypomethylated compared to adjacent non-malignant tissue (β=0.56; p<0.0001). When compared to a dataset of 24 lung adenocarcinomas and paired non-malignant tissue from TCGA, 1841 of these 1906 (96.6%) sites were also differentially methylated with the same direction of change between tumor and non-tumor lung tissue. We matched 1483 genes with the differentially methylated CpG sites and found that these genes were enriched with the terms glycoprotein, signal, plasma membrane part, homeobox in addition to a few other terms. There were significant differences in expression of 376 genes (Bonferroni corrected p<0.05) of the 1483 (25.4%) differentially methylated CpG sites. There was an inverse correlation between methylation and gene expression in 80% of these genes. Genes that were not significantly differentially expressed and were hypermethylated within CpG sites were enriched for homeobox genes (false discovery rate = 1.2E-27).

    Conclusion
    The methylation profiles of lung adenocarcinomas of never-smokers and their adjacent non-malignant lung tissue are significantly different. Differential methylation of a CpG site was associated with altered gene expression in approximately 25% of cases. Despite the differential methylation of homeobox genes, no significant changes in expression of these genes were detected.