Date of Award
Doctor of Philosophy (PhD)
Tumors cannot survive, progress and metastasize without recruiting new blood vessels. Vascular properties, including perfusion and permeability, provide valuable information for characterizing cancers and assessing therapeutic outcomes. Dynamic contrast-enhanced (DCE) MRI is a non-invasive imaging technique that affords quantitative parameters describing the underlying vascular structure of tissue. To date, the clinical application of DCE-MRI has been hampered by the lack of standardized and validated quantitative modeling approaches for data analysis.
From a therapeutic perspective, radiation therapy is a central component of the standard treatment for patients with cancer. Besides killing cancer cells, radiation also induces parenchymal and stromal changes in normal tissue, limiting radiation dose and complicating treatment response evaluation. Further, emerging evidence suggest that the radiation-modulated tumor microenvironment may also contribute to the enhanced tumor regrowth and resistance to therapy.
Given these clinical problems, the objectives of this dissertation were to: i) improve the DCE MRI-based measurements of vascular properties; and ii) assess the radiation treatment effects on normal tissue (parenchyma) and the interaction between radiation-modulated parenchyma and tumor growth. For the first goal, Bayesian probability theory-based model selection was employed to evaluate four commonly employed DCE-MRI tracer kinetic models against both in silico DCE-MRI data and high-quality clinical data collected from patients with advanced-staged cervical cancer. Further, a constrained local arterial input function (cL-AIF) modeling approach was developed to improve the pharmacokinetic analysis of DCE-MRI data. For the second goal, a novel mouse model of radiation-mediated effects on normal brain was developed. The efficacy of anti-vascular endothelial growth factor (VEGF) antibody treatment of delayed, radiation-induced necrosis (RN) was evaluated. Also, the effects of radiation-modulated brain parenchyma on glioblastoma cell growth were studied.
It was found that 1) complex DCE-MRI signal models are more sensitive to noise than simpler models with respect to parameter estimation accuracy and precision. Caution is thus advised when considering application of complex DCE-MRI kinetic models. It follows that data-driven model selection is an important prerequisite to DCE-MRI data analysis; 2) the proposed cL-AIF method, which estimates an unique local-AIF amplitude and arrival time for each voxel within the tissue of interest, provides better measurements vascular properties than the conventional approach employing a single, remotely measured AIF; 3) anti-VEGF antibody decreased MR-derived RN lesion volumes, while large areas of focal calcification formed and the expression of VEGF remained high post-treatment. More effective therapeutic strategies for RN are still needed; 4) the radiation-modulated brain parenchyma promotes aggressive, infiltrative glioma growth. The histologic features of such tumors are consistent with those commonly observed in recurrent high-grade tumors in patients. These findings afford new insights into the highly aggressive tumor regrowth patterns observed following radiotherapy.
Chair and Committee
Joseph J. Ackerman
Joel R. Garbow, Alexander B. Barnes, Dewey Holten, Christina I. Tsien,
Duan, Chong, "MRI in Cancer: Improving Methodology for Measuring Vascular Properties and Assessing Radiation Treatment Effects in Brain" (2017). Arts & Sciences Electronic Theses and Dissertations. 1237.