Meta-Analysis of High Throughput Oncology Data

In this chapter, we provide an overview of meta-analysis methods for oncology data with an emphasis on gene expression microarrays. We demonstrate these methods with a case study where we perform a meta-analysis on five publicly available microarray datasets designed to study survival in breast cancer patients. In this case study, we propose a novel metric using gene pathways for stratifying survival in our patient cohorts. We observe that significant genes in the individual studies show little reproducibility across the datasets, while using gene pathways with clinical variables is more reliable across microarray studies. We close with a discussion on extending meta-analyses and the potential for combining meta-analysis with integrative methods.