Tumor neoantigens – novel peptide sequences formed by somatic mutations arising within a tumor – are promising therapeutic targets insofar as they are able to mediate highly-specific discrimination between tumor and self. T cell reactivity to neoantigens can act at different stages of the tumor life-cycle: naturally-arising immunity shapes the mutational landscape of emergent tumors through immunoediting, and subsequently-induced immunity can be potently therapeutic against established tumors. An ability to view the neoantigen landscapes of individual tumors therefore offers information about both the tumor’s previous interactions with the immune system, and its potential to be treated by future immunotherapy. Such landscapes may also highlight specific targets for therapeutic intervention and monitoring.
Despite the fact that it has now become routine to catalog the somatic tumor variants in individual patients, methods for identifying which variants are immunogenic and for tracking responses against them remain underdeveloped. This is largely because neoantigen identification is a highly-personalized inquiry that depends upon: (i) the set of particular mutant sequences in the tumor, (ii) the expression of proteins containing those sequences, (iii) the ability of mutant sequences to be presented by autologous MHC proteins, and (iv) the activation and expansion of p:MHC-reactive T cells. Using a combination of novel multiplexed binding assay technology and in silico methods, we are working to represent these events and thereby generate personalized neoantigen landscapes, both in patients studied prospectively and in large retrospective tumor genome datasets. We hypothesize that these analyses will reveal the impact of the immune system on tumor genomes, and help predict the efficacy – and guide the design – of immunotherapies.