Background: A novel prediction algorithm is needed for the identification of effective tumor associated mutated neoantigens. Only those with no homology to self wild type antigens are true predicted neoantigens (TPNAs) and can elicit an antitumor T cell response, not attenuated by central tolerance. To this aim, the mutational landscape was evaluated in HCV-associated hepatocellular carcinoma.
Methods: Liver tumor biopsies and adjacent non-tumor liver tissues were obtained from 9 HCV-chronically infected subjects and subjected to RNA-Seq analysis. Mutant peptides were derived from single nucleotide variations and TPNAs were predicted using two prediction servers (e.g. NetTepi and NetMHCstabpan) by comparison with corresponding wild-type sequences, non-related self and pathogen-related antigens. Immunological confirmation was obtained in preclinical as well as clinical setting.
Results: The development of such an improved algorithm resulted in a handful of TPNAs despite the large number of predicted neoantigens. Furthermore, TPNAs may share homology to pathogen's antigens and be targeted by a pre-existing T cell immunity. Cross-reactivity between such antigens was confirmed in an experimental pre-clinical setting. Finally, TPNAs homologous to pathogen's antigens were found in the only HCC long-term survival patient, suggesting a correlation between the pre-existing T cell immunity specific for these TPNAs and the favourable clinical outcome.
Conclusions: The new algorithm allowed the identification of the very few TPNAs in cancer cells, and those targeted by a pre-existing immunity strongly correlated with long-term survival. Only such TPNAs represent the optimal candidates for immunotherapy strategies.
Keywords: Cancer vaccine; Immunotherapy; Liver cancer; Neoantigens; Personalized treatment.