Molecular features derived from tumor samples (e.g., somatic mutations, gene expression, or DNA methylation) can be very useful biomarkers for epidemiology studies. Recent success of immunotherapy demonstrated that tumor immune microenvironment plays a crucial role for tumor growth and inhibition. Therefore, biomarkers derived from tumor immune microenvironment are great additions to many large epidemiology studies that have access to tumor samples. In this project, we propose to develop a set of statistical methods and computational tools to study biomarkers in tumor immune microenvironment, and as a demonstration, apply them to analyze the omic data from The Cancer Genome Atlas (TCGA). Specifically, we will estimate immune cell composition in the TCGA samples using gene expression and/or DNA methylation data, which can be collected from either fresh frozen or formalin-fixed paraffin-embedded (FFPE) samples. Next we will use immune cell composition to construct prognostic signatures of patient survival time. Our methods and software packages will provide important resources that will enable new epidemiology studies, such as association of immune features with environmental/genetic factors, or cancer risk prediction for cancer subtypes defined/refined by immune biomarkers.
We propose to develop statistical methods and software packages to study tumor immune microenvironment, and associate immune cell composition with survival time. Our project will break new ground for epidemiology studies, for example, to enable stratified association analysis for patients with particular tumor immune microenvironment.