Graduation Year

2024

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Medical Sciences

Major Professor

Alexander Anderson, Ph.D.

Committee Member

Thomas Yankeelov, Ph.D.

Committee Member

David Basanta, Ph.D.

Committee Member

Amer Beg, Ph.D.

Committee Member

Robert Gatenby, Ph.D.

Committee Member

Heiko Enderling, Ph.D.

Keywords

bridging-scales, host-tumor interaction, mathematical oncology, population-based modeling, volumetric imaging

Abstract

To understand the dynamics of cancer, mathematical oncologists have developed models of tumor growth and treatment response. Some models are mechanistic and approach tumor growth at the cell-scale, focusing on the evolution of cancerous cells within the ecology of normal tissue, and are often simulated with agent-based modeling. Other models are more clinically motivated and model tumor growth operating at the organ-scale, using patient data to predict treatment response, and are often simulated with partial differential equations. We developed the Hybrid Automata Library which includes both agent-based modeling and partial differential equations for modeling at either of these scales. We further aimed to bridge these scales by developing the population-based modeling method. This novel method applies probabilistic rules to discrete cell counts with cubic millimeter spatial resolution, thus linking mechanistic cell-scale dynamics to clinically observable organ-scale tumor growth.

We have applied this method to early stage non-small cell lung cancer to explore the effect of normal tissue density on tumor growth. We found that high density tissue causes the tumor cells to become more glycolytic, while low density tissue causes the tumor cells to become more angiogenic. We then fit the model to pre-cancerous lesions from the National Lung Screening Trial, and supported that lesions in low-density tissue appear angiogenic. We extended the model to simulate adaptive therapy and studied the impact of monitoring the tumor using imaging compared to using a blood biomarker. We found that blood biomarker guided adaptive therapy extends time to progression by providing better control of the sensitive population, but it does not control the spread of the tumor cells spatially as much as imaging guided adaptive therapy. We also studied how tissue properties measured via MRI sequences impact glioblastoma by implementing another novel method using convolution to propagate the segmented tumor region. We found that the FLAIR sequence is highly correlated with both the tumor growth trajectory and where it will survive immunotherapy. The implications of these models present new insights into improving patient-specific therapy and are supported by/verifiable with imaging data.

Download

Share

COinS