Advisory Committee Chair
Hassan Fathallah
Advisory Committee Members
Ian Knowles
S S Ravindran
Hemant Tiwari
Shan Zhao
Document Type
Dissertation
Date of Award
2017
Degree Name by School
Doctor of Philosophy (PhD) College of Arts and Sciences
Abstract
Glioblastoma multiforme (GBM) is a uniformly fatal form of brain cancer that can exhibit unpredictable responses to dierent rate-reducing therapies. Here, we construct a mathematical model of brain tumor growth and utilize in silico clinical trials to investigate the unusual behavior of GBM. We also use the model to study the progression of grade 2 and grade 3 tumors to higher grades like GBM. Simulations in Chapter 1 lead to the identication of a previously unknown pattern of progression under anti-angiogenic (AA) therapy, which we call \Expanding FLAIR + Necrosis." Model analysis predicts that clinical features of this growth would include expanding regions of brain necrosis and low-concentration invasive tumor cells, which we nd in the MRIs of 11 out of 70 patients treated with AA. In Chapter 2, model simulations suggests key hypotheses on the pathogenesis of all GBM progression patterns under AA therapy. Namely, we identify tumor motility as an important biomarker because it appears to exert a signicant impact on patient prognosis, tumor progression, the ecacy of dierent therapies, and overall survival times. The model used in Chapters 1, 2, and 4 includes two types of cancer cells: proliferative and invasive. This dichotomy is based on the Go-or-Grow (GoG) phenotype{the idea that cancer cell motility and proliferation are mutually exclusive. In an attempt to simplify the model and prepare it for future clinical applications, we introduce a single glioma cell model in Chapter 3. The ability of the new model to replicate results from Chapter 2 leads to the hypothesis that the GoG phenotype may not be necessary for the formation and progression of GBM. In studying the development of glioma grades in Chapter 4, we also unearth a new concept in cancer genetics{the idea of natural tumor evolution vs. transformation. Both model simulations and clinical data support the conclusion that the natural evolution from grade 3 gliomas to GBM could occur, in the absence of rate-changing mutations, in a period of a few months. We also identify a subgroup of patients whose tendency to recur at the same grade is associated with favorable genetic variations including mutations in the isocitrate dehydrogenase (IDH1-R132H) and ATRX genes and/or 1p/19q deletion. We hypothesize that the transformation of some gliomas to higher grades necessitates the accrual of molecular changes aecting the key biological rates of replication, angiogenesis, and dispersion. All of these findings serve to advance the basic medical sciences, particularly the eld of oncogenesis. They also herald the future use of in silico clinical trials in drug discovery by showing how computer models can safely and eectively evaluate therapeutic choices. Finally, our research promotes a multidisciplinary approach to cancer research by demonstrating the capability of mathematical modeling in testing hypotheses, validating phenotypes, uncovering novel clinical phenotypes, and predicting tumor response to treatment as well as patient survival times.
Recommended Citation
Scribner, Elizabeth Yates, "Mathematical Modeling of Brain Tumors Advances Patient Care, Oncogenesis, and the Use of In Silico Clinical Trials" (2017). All ETDs from UAB. 2932.
https://digitalcommons.library.uab.edu/etd-collection/2932
