Molecular Signatures Bioinformatics and Clinical data in Cancer- Toward Personalized Medicine"""

Abstract

Cancer is a complex disease. There are three major levels to describe the underlying biological processes: genomic DNA, RNA, and protein. Changes of elements in above three levels can be measured by RT-PCR-based assays or high throughput technologies like microarray. Yet how to combine the genomic data with clinical outcome data for shedding light on cancer biology is a major challenge.　　　 　　　In this talk, I will describe the recent progress in utilizing data from gene-expression in mRNA and microRNA to predict survival of lung cancer. At the mRNA expression level, we found a five-gene signature which could predict patients’ overall and relapse-free survival in two sample cohorts from Taiwanese population and in Western population. At the microRNAs expression level, we also found a set of five microRNAs that could predict patients’ outcome in the testing sample and is validated by samples from another hospital which is located far from the original hospital collecting the training samples. Moreover, by functional assay, the 5 miroRNAs were shown to have biological functions in tumor invasiveness. 　　　Our bioinformatics approach is powerful in extracting useful information from large scale genomic and clinical data. The obtained molecular signatures could improve the traditional prognosis by cancer stage. They can be used to guide the proper clinical treatment assignment according to the patients’ molecular risk scores. The combined use of bioinformatics, epidemiology and biostatistics could aid the progress in personalized medicine.