Abstract A25: Array-based Comparative Genomic Hybridization of Hepatocellular Carcinoma Reveals a Unique Genomic Aberration Pattern in Tumors with Poor Prognosis

Stephanie Roessler,Lori Long,Anuradha Budhu,Yidong Chen,Xuelian Zhao,Robert Walker,Hu-Liang Jia,Qing-Hai Ye,Lun-Xiu Qin,Zhao-You Tang,Kent Hunter,Snorri Thorgeirsson,Paul Meltzer,Xin Wang

DOI: https://doi.org/10.1158/0008-5472.fbcr09-a25

IF: 11.2

2009-01-01

Cancer Research

Abstract:Abstract Hepatocellular carcinoma (HCC) is the third leading cause of cancer death worldwide. Like most solid tumors, HCC is very heterogenous in its clinical presentation and genetic makeup, due to genomic instability. The poor survival of HCC patients is partly due to inefficient treatment regimens and the lack of accurate prognostic subgroup stratifications. The discovery of genomic changes in other cancer types has led to the development of novel specific therapies, underscoring the importance of understanding molecular mechanism and tumor biology to provide the best therapy for each cancer patient. To this end, we performed high-resolution array-based comparative genomic hybridization of 76 HCC clinical specimens (cohort1) to search for genetically disrupted genes associated with outcome. In agreement with previous publications, recurrent gains and losses were associated with 1q, 6p, 8q and 4q, 8p, 13q, 16, 17p, respectively. However, only deletions of a few distinct loci on 1p,4q, 8p and 9p were associated with survival (FDR<0.05). Of the 419 genes which mapped to regions of loss, analysis of cohort1 revealed that gene expression and DNA copy number of 134 genes (31.98%) were significantly correlated (Pearson coefficient r>0.3, p<0.001). Next, we built a 10-gene-survival signature based on the expression of these 134 genes utilizing cohort1 as a training and testing set with validation in an independent cohort2 (N=180). Kaplan-Meier survival analysis revealed that the 10-gene signature was able to predict survival outcome in tumor tissue (log-rank p=0.008) but not in paired non-tumor samples of cohort2 (log-rank p=0.421). Cox proportional hazards regression analysis controlling for potentially confounding clinical covariates showed that the 10-gene signature was significantly associated with a 2.1-fold increased risk of death. In addition, the 10-gene signature was able to predict survival in three publicly available breast cancer cohorts. In this study, we found a 10-gene signature that allows for the stratification of patients with poor prognosis in HCC and breast cancer. Ongoing functional analysis of the genes in this signature may lead to the discovery of HCC subtype-specific drugs that might be used for personalized therapy. Citation Information: Cancer Res 2009;69(23 Suppl):A25.