Systematic evaluation of neoepitope predictions challenges clinically observed T-cell responses and their impact on immune evasion.

Peptide presentation on human leukocyte antigens (HLAs) is essential for initiating T-cell responses and all consequences of this presentation including anticancer immunity or immune escape. Many studies have relied on in silico prediction tools rather than biological measurement of HLA presentation to study these effects. To better assess the frequency and consequences of neoantigen presentation, we overexpressed 125 combinations of full-length neoantigens and one HLA class I allele to experimentally validate presentation of mutated and non-mutated HLA ligands through HLA ligand isolation followed by tandem mass spectrometry. A successful presentation was observed only in 22% of predicted cases with strong implications on previously described downstream effects. For example, the association of HLA loss of heterozygosity with predicted neoepitopes was challenged for 58% (73/125) of combinations. Furthermore, when testing 51 sequences used for personalized messenger RNA neoepitope vaccines, we observed that clinical responses were independent of the presentation status of the neoepitopes. Even a presumably neoepitope-specific and strongly expanded T cell receptor clone from a neoantigen vaccination study could not be linked to a successfully presented neoepitope. Overall, these data highlight the importance of validating the presentation of neoepitopes to fully understand our interpretation of clinical mutation-specific responses and their related effects, including immune evasion.