UVM Theses and Dissertations
Format:
Online
Author:
Lescault, Pamela Jean
Dept./Program:
Microbiology and Molecular Genetics
Year:
2013
Degree:
PhD
Abstract:
The goal of my work was to develop a process that uses cancer genome-wide expression profiles to inform clinical decisions. My overall hypothesis was that multivariate analysis of expression profiles would improve the development of this process and I tested this hypothesis in two ways. First, division of neuroblastoma solid tumor biopsies into sections allowed me to use formal statistical methods to study differences between patients in the context of variation within patients. My analytical process focused on three distinct, sequential representations of genome-wide expression data (1) expression profiles, (2) comparative expression profiles (compared to a standard reference) and (3) predicted drug lists (derived from comparative expression). Each patient biopsy (n = 5) was dissected into three sections. I quantified the extent of intra-tumor biopsy expression profile variation, identified genes associated with intra-tumor biopsy expression profile variation, described the nature of this variation in terms of biological processes and evaluated how different types of intra-biopsy variation might impact drug prediction.
Additionally, variation within and between biopsies was calculated using comparative expression (the input for drug prediction) based on a collection of whole body reference expression profiles. I show that intra-biopsy variation, in terms of expression, comparative expression, and drug lists is small in the context of variation among patient biopsies in the same contexts (p < 0.001). I identified biological processes attributed to patient-specific intra-biopsy expression profile variation and show that intra-biopsy variation in neuroblastoma has little impact on the reproducibility of identifying personalized therapeutics.
Second, bone marrow aspirates (from neuroblastoma patients with bone marrow involvement) potentially provide an important source of information. For example, when it is important to capture variation over time (e.g., response to therapy) or between primary tumor and metastatic disease (e.g., cancer cells in different microenvironments may exhibit different therapeutic targets). Unlike solid tumors, however, neuroblastoma cells may be represented in the bone marrow by less than 1% of cells so isolation is essential. I show that genome-wide expression profiles can be obtained from neuroblastoma cells disseminated to the bone marrow based on a paired design, where neuroblastomacells isolated by each of two methods were paired by patient. I show that patient-specific expression profiles can be obtained from bone marrow aspirates using two isolation methods, conditioned cell culture and fluorescence-activated cell sorting (FACS) (p < 0.002).
Patient-specific variation was correlated between methods (R² = 0.75). Among the most significant biological pathways associated with variation between patients were, both previously reported (ie., those known to be associated with neuroblastoma) and novel (ie., those yet to be reported associated with neuroblastoma). These results show, for the first time, that FACS can be used to obtain patient-specific expression profiles from neuroblastoma cells disseminated in bone marrow. Overall, my work paves the way for a formal and comprehensive study of the variation in gene expression profiles within and between neuroblastoma patients as well its impact on personalized medicine.
Additionally, variation within and between biopsies was calculated using comparative expression (the input for drug prediction) based on a collection of whole body reference expression profiles. I show that intra-biopsy variation, in terms of expression, comparative expression, and drug lists is small in the context of variation among patient biopsies in the same contexts (p < 0.001). I identified biological processes attributed to patient-specific intra-biopsy expression profile variation and show that intra-biopsy variation in neuroblastoma has little impact on the reproducibility of identifying personalized therapeutics.
Second, bone marrow aspirates (from neuroblastoma patients with bone marrow involvement) potentially provide an important source of information. For example, when it is important to capture variation over time (e.g., response to therapy) or between primary tumor and metastatic disease (e.g., cancer cells in different microenvironments may exhibit different therapeutic targets). Unlike solid tumors, however, neuroblastoma cells may be represented in the bone marrow by less than 1% of cells so isolation is essential. I show that genome-wide expression profiles can be obtained from neuroblastoma cells disseminated to the bone marrow based on a paired design, where neuroblastomacells isolated by each of two methods were paired by patient. I show that patient-specific expression profiles can be obtained from bone marrow aspirates using two isolation methods, conditioned cell culture and fluorescence-activated cell sorting (FACS) (p < 0.002).
Patient-specific variation was correlated between methods (R² = 0.75). Among the most significant biological pathways associated with variation between patients were, both previously reported (ie., those known to be associated with neuroblastoma) and novel (ie., those yet to be reported associated with neuroblastoma). These results show, for the first time, that FACS can be used to obtain patient-specific expression profiles from neuroblastoma cells disseminated in bone marrow. Overall, my work paves the way for a formal and comprehensive study of the variation in gene expression profiles within and between neuroblastoma patients as well its impact on personalized medicine.