Date of Award

8-27-2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Bioinformatics

First Advisor

Rick Edmondson

Abstract

Multiple myeloma is a cancer of antibody-producing plasma B cells with patients presenting heterogeneous progression and survival outcome. Myeloma is currently incurable and the goal of current treatment options, including chemotherapy and stem cell transplantations, is to manage the disease so as to increase overall survival (OS) and/or event-free survival (EFS). Currently, patient risk is assessed using gene expression profiling based on a seventy gene risk signature (GEP70). However, the GEP70 high-risk group has poor prognosis and survival outcome and the low-risk patients with poor survival do exist. Therefore, in order to improve risk stratification, we undertook a mass spectrometry (MS)-based proteomics study to complement the present genomics studies of myeloma patient samples with the goal of identifying biomarkers indicative of overall survival. IPI2Affy, an ID-converter tool, was built to facilitate interoperation between the proteomics and genomics datasets. IPI2Affy converts 88% of the human International Protein Index IDs to their corresponding Affymetrix gene probe set ID(s). Both MS-based proteomics and microarray gene expression profiles were obtained using 85 baseline patient samples. IPI2Affy was then used to generate the genomics counterpart of the proteomics dataset. The 85 samples were divided into two survival-based groups (OS less than or greater than 3 years). Both proteomics and genomics datasets were analyzed by two independent Significant Analysis of Microarray (SAM) analyses to find differential peptides and gene probe sets respectively. Nine probe sets mapping to seven genes (CACYBP, CORO1A, ENO1, FABP5, IQGAP2, PAICS, and STMN1) were identified as significantly differentially expressed in both genomics and proteomics SAM analyses indicating their association with short OS. Using these nine gene probes as candidate biomarkers, the survival difference between patient groups was examined statistically by the logrank test and graphically by Kaplan-Meier plots. Furthermore, to find which genes have independent prognostic value, a multivariate Cox regression analysis (MVA) was performed on genomics data on two large test sets of two patient cohorts respectively. Based on the MVA results, CACYBP was selected to be tested in combination with the GEP70 model and was found to determine an intermediate-risk myeloma subgroup. Thus, seven statistically-valid potential OS-indicative prognostic biomarkers were identified from this combined proteomics-genomics study.

Download

Share

COinS