Date of Award

9-20-2010

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Science

First Advisor

Steven Jennings

Abstract

Malignant astrocytic gliomas, like Glioblastoma Multiforme and Anaplastic Astrocytoma, are among the deadliest forms of brain tumors with varying periods of survival. Currently, the risk of a glioma patient is mostly measured by prognostic models with clinical prognostic markers and has limited success. In this study a Bioinformatics & Computational Biology approach to identify candidate prognostic markers from a combination of clinical, genetic and gene expression variables (related to multiple significant pathways targeted by classical genetic alterations) involved in malignant glioma progression, was developed to explain variation in survival. The initial selection of molecular variables for prognosis modeling was done by querying multiple databases using data-mining tools. Gene expression profiling and mutational analyses were done using reliable experimental quantification methodologies. Differential gene expression and correlations, direct and indirect effects of molecular and clinical variables, prognosis modeling, and their effect on the overall survival of patients were examined. The study identified genes such as PTEN, VEGF, CDK4, BCCIP, BIRC5 and mutation of the TP53 gene as candidate prognostic markers, and was able to explain about 41 % of the variation in survival. The pathway-centric approach with experimental methodologies and multiple statistical analyses for modeling prognosis was able to identify positive and negative candidate prognostic markers. The results from the study indicate a strategy to develop a prognostic system to identify prognostic markers not only for malignant astrocytic gliomas but for other forms of cancer as well.

Download

Share

COinS