Gene-Signatures Predicting Survival

Authors:

Augusto Villanueva, MD, PhD (Spain)

Recent genomics technology has enabled simultaneous, comprehensive analysis of wide variety of biomolecules, including coding- and non-coding-RNA, genomic DNA, and protein for their abundance, structural alterations, and biochemical properties such as methylation, phosphorylation, and acetylation. This has been expected to facilitate and greatly increase the likelihood of successful identification of traditional, single molecular biomarkers for each specific clinical problem. Application of these technologies in hepatocellular carcinoma (HCC) has enabled tumour classification based on molecular features. HCC subclasses correlate with signalling pathway de-regulation, specific gene mutations and some suggest progenitor cell origin. A number of prognostic gene signatures have been generated from tumour and adjacent non-tumoral tissue. Presumably, incorporation of additional layers of molecular information such as DNA methylation or non-coding RNA will improve prediction power of current mRNAbased clusters. However, there is a remarkable imbalance in the amount of signatures reported and the ones that reach full clinical implementation. Small studies are prone to overestimate or underestimate the actual association, and the lack of appropriate validation has also jeopardized study interpretation when considering cancer outcome with microarrays. Additional issues related to tumour heterogeneity could become critical, particularly in advanced tumour stages.