| dc.description.abstract | Intra-tumour heterogeneity in the tumour microenvironment (TME) of CRCs may lead to multiple immune escape mechanisms coexisting within CRCs, affecting sensitivity to immunotherapy. This thesis describes the spatial intra-tumour heterogeneity of the immune tumour microenvironment and immune escape alterations as well as spatially resolved neoantigen burden in superficial and invasive cancer and lymph node metastases from 11 patients with stage III microsatellite-stable colorectal cancer. In chapter 4, I present evidence that substantial within-tumour TME differences are not seen in most lymphocyte, neutrophils or macrophage subsets. Significant within- tumour differences in spatial organisation of epithelial cells and CTLs are also not detected using a range of spatial statistics (such as the Morisita-Horn index, the nearest neighbour index and Ripley’s H function). This may lead to a lack of substantial differences in immune selection pressure experienced by tumour cells in superficial, invasive and node deposits of CRC. In chapter 5, I present evidence that LOH HLA and HLA mutations are not specifically enriched in superficial, invasive or node ROIs of CRC. I show that fraction of non-epithelial PDL1+ cells is increased in invasive ROIs compared to superficial cancer. I also present evidence that the fraction of cells assigned to cellular neighbourhood 3, which is highly enriched for PDL1 and PDL1+ epithelial cells, is significantly higher in invasive and node compared to superficial ROIs of CRC. It is unclear whether the increase in fraction of PDL1+ cells in invasive regions (compared to superficial cancer) is likely to drive immune escape here. In the next chapter, I investigate whether regions showing an increased PDL1 fraction or any HLA alterations show an increase in neoantigen burden, due to accrual of neoantigens following establishment of functional immune escape. In chapter 6, I investigate whether there are substantial differences in neoantigen burden within individual tumours. I use the normalized metric of proportional neoantigen burden to compare differences across cancer region type. Proportional neoantigen burden is defined as proportion of subclonal protein-changing SNVs that are missense neoantigen mutations giving rise to subclonal novel strong-binding neoantigen peptides OR proportion of subclonal protein-changing frameshift mutations that are indel neoantigen mutations giving rise to subclonal novel strong- binding neoantigen peptides. No significant differences in proportional neoantigen burden are seen across superficial, invasive or node deposits of CRC. I then use the normalised metric of immune dN/dS to identify whether non- synonymous mutations are depleted relative to synonymous mutations in regions of the genome likely to generate neoantigens in any tumour regions. Immune dN/dS did not vary by cancer region type. I show that neoantigens are similarly immunogenic between cancer region types and are just as likely to be shared across multiple tumour regions as all other protein- changing mutations, suggesting that there is no substantial negative selection that preferentially acts on neoantigens. 6 Finally, I present evidence that regions showing an increased PDL1 fraction or HLA alterations do not show an increase in proportional neoantigen burden, suggesting that immune-specific (ongoing) subclonal selection for functional immune escape mechanisms (with an associated gain of neoantigens) is weak. This thesis shows that there is no clear enrichment for immune escape mechanisms (with associated increase in neoantigen burden or immune dN/dS) in superficial, invasive or lymph node metastatic tumour regions. These results together suggested a lack of differential immune editing/ escape between superficial, invasive and node deposits in advanced CRC. | en_US |
