Background and aims:
Development of primary liver cancer is a multi-stage process. Pre-neoplastic dysplastic
lesions emerge on the basis of chronic liver damage and evolve into early hepatocellular
carcinoma (eHCC) and, subsequently, progressed HCC (pHCC). Detailed molecular characterization
and prediction of pre-neoplastic lesions at high risk for malignant transformation
would significantly advance our diagnostic and therapeutic approaches. We here utilized
integrative molecular analyses to characterize the sequential evolution of liver cancer
and aimed to define key epigenetic drivers and biomarkers of HCC development and progression.
Methods:
Methylation 450k-beadchip analyses were performed on cirrhotic liver (n = 7), low-
(n = 4) and high-grade (n = 9) dysplastic lesions, eHCC (n = 5) and pHCC (n = 3) from
8 HCC patients with chronic hepatitis B infection. Differentially methylated gene
regions (DMGR) were identified in comparison to non-cirrhotic and non-infected liver
(n = 9). Potential epi-drivers and biomarkers were identified by integrative analyses
of transcriptomic changes and validated in an independent cohort from the TCGA database.
Results:
The proportion of hypermethylated DMGR progressively increased from cirrhosis over
dysplastic- to HCC and peaked in eHCC lesions. Early epigenetic alterations involved
signaling pathways related to cell death, apoptosis and immune regulation, while late
changes centered on cell survival, growth and migration. A common regulation of stem
cell-associated pathways including Wnt/b-catenin signaling was revealed in dysplastic
as well as eHCC potentially predisposing tumor progression. Moreover, we identified
101 genes with significant methylom changes in dysplastic and cancerous lesions with
concomitant progressive gene expression alterations in cancer tissue. We further defined
an epi-panel of early epigenetic marks in dysplastic lesions including selected CpG-sites
with confirmed differential methylation in cancer tissue and consequential transcriptional
alterations of the target genes using an independent cohort of 362 HCC and 49 surrounding
liver samples. Unsupervised hierarchical clustering confirmed a robust classification
in malignant and non-malignant lesions.
Conclusion:
Our results confirm that epigenetic changes occur early during hepatocarcinogenesis.
Epigenetic modifications, therefore, might be of high diagnostic/predictive utility
for the identification of dysplastic lesions at risk for cancer progression. The identified
epi-panel of oncogenic epigenetic marks might be useful to complement phenotypic classifications
and facilitate selection of lesions amenable to early therapeutic interventions.
