Date of Award

2019

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Chemical and Biomolecular Engineering

First Advisor

Ravi . Radhakrishnan

Abstract

Systems models of key signaling pathways in cancer have been extensively used to under-

stand and explore the mechanisms of action of drugs and growth factors on cancer cell

signaling. In general, such models predict the effect of environmental stimuli (both chemical such as for e.g., growth factor and drugs as well as mechanical such as matrix stiffness)

in terms of activities of proteins such as ERK or AKT which are important regulators of cell

fate decisions. Although such models have helped uncover important emergent properties

of signaling networks such as ultrasensitivity, bistability, and oscillations, they miss many

key features that would make them useful in a clinical setting. 1) The predictions of activity

of proteins such as ERK or AKT cannot be directly translated into a clinically useful parameter such as cell kill rate. 2) They don’t work as well when there are multiple biological

processes operating under different time and length scales such as receptor-based signaling

(4-6 hours) and cell cycle (24-48 hours). 3) The parameter space of such models often exhibits sloppy/stiff character which affects the accuracy of predictions and the robustness of

these models. Apart from single-cell systems models of signaling, pharmacokinetic and cell

population-based pharmacodynamic models are also extensively used to predict the efficacy

of a particular therapy in a clinical setting. However, there are no direct or consistent ways

of incorporating patient-specific gene/protein expression data in these models. This thesis

describes the development and applications of a multiscale and multiparadigm framework

for signaling and pharmacodynamic models that helps us address some of the above short-

comings. First two single scale systems models are described which introduces methods of

exploration of parameter space and their effect on model predictions. Then the multiscale

framework is described and it is applied to two different cancers - Prostate Adenocarcinoma

and Nephroblastoma (Wilm’s Tumor). Special mathematical techniques were used to de-

velop algorithms that can integrate models of disparate time scales and time resolutions

(continuous vs. discrete-time). Such multiscale modeling frameworks have great potential

in the field of personalized medicine and in understanding the physics of cancer taking into

account the biology of the cells.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS