Abstract 285: Integrative analyses of genome-wide expression of miRNAs and DNA methylation patterns in hepatocellular carcinoma to improve functional biomarker identification

Abstract

Abstract Previous studies, including ours have examined aberrant expression of microRNAs (miRNAs) and altered DNA methylation, and several miRNAs have been identified as regulated by DNA methylation. However, whether this mechanism occurs at a genome-wide level and is related to the profiles of dysregulated miRNA commonly observed in hepatocellular carcinoma (HCC) patients is largely unknown. Using a two-phase study design, we conducted a genome-wide screening for miRNA expression and DNA methylation profiles in the tumor and adjacent non-tumor tissues from US HCC cases. The discovery and validation sets included, respectively, 10 and 56 paired tumor/non-tumor tissues. TaqMan Low Density Arrays (TLDA) covering 733 miRNAs were used to measure expression profiles, and quantitative RT-PCR was used to validate candidate miRNA signatures. Infinium HumanMethylation 450K BeadChip arrays covering 3,439 CpG sites for 727 miRNAs were used to determine DNA methylation patterns. Comparing expression profiles in HCC tumor and adjacent non-tumor tissues, we found that 25 miRNAs were statistically significantly different with over 2-fold expression changes after adjusting for false discovery rate (FDR); a perfect classification of HCC tissues was observed. Six of the aberrant miRNAs were over-expressed in HCC tumor tissue (range from 2.7 to 18-fold), while 19 miRNAs were downregulated in tumor tissue (range from -2.5 to -7.5-fold). Several miRNAs (miR-139, miR-196b, miR-381, miR-486 and miR-1180) were identified for the first time as having a role in hepatocarcinogenesis. After Bonferroni adjustment, a total of 1,256 CpG sites covering 412 miRNAs showed significant differences in DNA methylation levels between tumor and adjacent non-tumor tissues, including 185 hypermethylated and 1,071 hypomethylated CpG sites in tumor tissues. We further conducted integrative analyses of the correlations between miRNAs expression and DNA methylation patterns for 222 miRNAs detected in over 80% of the tissue samples. As expected, inverse correlations were observed for 91.7% (55/60) of up-regulated miRNAs with a DNA hypomethylation change, and for 48.1% (78/162) of down-regulated miRNAs with a DNA hypermethylation change. However, for only 8 miRNAs’ (3.6%) did expression levels achieve statistically significant differences between tumor and non-tumor tissues in accord with DNA methylation alterations. A validation study of six of the candidates in an additional 56 paired HCC tissues confirmed the aberrant miRNAs expression patterns observed in tumor tissue. The inverse correlations between miRNAs’ expression and DNA methylation were concordant for all 27 CpG sites in the 4 miRNAs (miR-18a, miR-125b-1, miR-182 and miR-199a-1). These data suggest that a small proportion of DNA methylation changes can lead to significant dysregulation of miRNAs in HCC tumor tissue. Citation Format: Jing Shen, Shuang Wang, Abby B. Siegel, Helen Remotti, Qiao Wang, Iryna Sirosh, Regina M. Santella. Integrative analyses of genome-wide expression of miRNAs and DNA methylation patterns in hepatocellular carcinoma to improve functional biomarker identification. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 285. doi:10.1158/1538-7445.AM2014-285