Immune Recognition and Sculpting of Neoantigens in Human Tumors

Abstract

Understanding the extent to which the immune system can recognize and sculpt human tumors is critical to designing and implementing new immunotherapeutic approaches to cancer. Immune recognition of tumors is primarily mediated through segments of mutated proteins, called neoantigens, that can be processed and presented on the cell surface to elicit a CD8+ T cell response. T cell-mediated elimination of tumor cells bearing immunogenic neoantigens may result in preferential growth of less immunogenic tumor cells, sculpting the tumor in a process called immunoediting. While evidence consistent with immunoediting has been identified in human cancer, it remains unclear to what extent the observed changes are attributable to the immune response. Further complicating this challenge are limitations in the prediction of immunogenic neoantigens. Therefore, this dissertation addresses 1) the characteristics of neoantigens that elicit an immune response and 2) the extent to which the immune system shapes the neoantigen profile of human cancer. To assess the impact of the immune system on sculpting the neoantigen profile of invasive tumors, the neoantigen profiles were compared between precursor actinic keratoses (AK) and invasive cutaneous squamous cell carcinoma (cSCC). When accounting for the MHC class I binding of the neoantigen compared to the binding of the unmutated peptide, the predicted T cell recognition, and the mRNA expression of the neoantigen, cSCC had a lower predicted maximum immunogenicity compared to AK. This work demonstrated the potential for the immune system to shape the neoantigen profile in invasive tumors, but also highlighted the need for an improved understanding of the characteristics of immunogenic neoantigens. Therefore, we employed a regularized regression modeling approach on existing datasets with validated immunogenic neoantigens to select the neoantigen characteristics most predictive of an immune response. The resulting model, called the NeoScore, is a linear combination of the neoantigen:MHC class I dissociation constant and stability, in addition to the mRNA level expression. The NeoScore demonstrated high sensitivity and specificity in predicting immunogenic neoantigens. Furthermore, the NeoScore of the most immunogenic neoantigen was highly associated with response to immune checkpoint inhibition in metastatic melanoma. Finally, to further illuminate the role of the immune system in shaping the neoantigen profile of a tumor, the neoantigen and immune profile was compared between cSCC from immunocompetent and immunosuppressed patients. Tumors from immunocompetent patients had a decreased clonal mutational burden relative to immunosuppressed patients, and the neoantigens from clonal mutations had a lower proportion of binding neoantigens, demonstrating the effect of the immune system on shaping the clonal neoantigen profile in the immunocompetent setting. Finally, the comparison of the clonal and subclonal neoantigens was used as an unbiased screen for the characteristics of neoantigens associated with an immune response. The difference in the MHC class I binding of the neoantigen compared to the unmutated peptide was identified as a key characteristic of immunogenic neoantigens. Overall, these three analyses demonstrated both the impact of the immune system in shaping the neoantigen profile and the characteristics of immunogenic neoantigens. The results of this work are expected to contribute to the design and implementation of future immunotherapeutic approaches.