Techniques for analysis and integration of histology data and agent-based modelling of T cell-tumour interactions

Thesis

Cytotoxic (CD8+) T cells form an important part of the adaptive immune response. They kill infected and abnormal cells, including cancerous cells, via a process known as immuno-surveillance. When CD8+ T cells infiltrate tumour tissue, they identify antigens on the surface of tumour cells, and begin tumour cell lysis. Immunotherapies are increasingly being developed to target aspects of the immune system, such as CD8+ T cells, in order to boost its anti-tumour response. In clinical trials of such therapies, patients are monitored in several ways, often including core-needle biopsies in which a small piece of the tumour is removed, stained for specific cell markers, and imaged.

The increasing availability of such spatially-resolved imaging data motivates the development of mathematical models which capture the spatial dynamics of individual CD8+ T cells and tumour cells. Agent-based models (ABMs) allow the simulation of complex biological systems with detailed temporal and spatial resolution. Spatial statistics, for example the pair correlation function (PCF), can quantify the spatial heterogeneity inherent in both biological and synthetic data, and generate insight into the underlying biological system.

The main aims of this DPhil thesis are to explore how mathematical modelling and image analysis techniques can be used in oncology clinical trials to gain insight into the mechanism of action of immunotherapeutics. In the first part of the thesis, we adapt an existing on-lattice ABM of tumour-CD8+ T cell interactions in order to elucidate the key mechanisms driving oscillatory dynamics caused by tumour-T cell interactions, and investigate how specific modelling assumptions influence the spatial dynamics. We then undertake an exploratory analysis of histology data from a Phase I clinical trial of Simlukafusp (FAP-IL2v), a cancer immunotherapy developed by Roche. Here we aim to ascertain whether using the PCF to analyse the imaging data can uncover further insights into the efficacy of the drug, as opposed to the current clinical practice of recording cell counts and visual inspection of CD8+ T cell infiltration in the tumour. In the final part of the thesis we develop a new on-lattice ABM which builds on the findings of our first model, and use it to identify conditions under which spatial statistics enable us to make earlier and more accurate predictions of tumour responses to immunotherapy.

Authors: Stephens, R

• DOI: 10.5287/ora-2rmraeaee
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: sensitivity analysis; agent-based model; histology data; CD8+ T cell infiltration; spatial statistics
• Subjects: Mathematical Oncology