84P - T cell receptor repertoire profiles of tumor -infiltrating lymphocytes improves neoantigen prioritization for personalized cancer immunotherapy

Background

Tumor-specific neoantigens arising from non-synonymous mutations have become increasingly attractive targets for immunotherapies. The accurate identification and prioritization of neoantigens is crucial to design personalized therapies for each cancer patient. Putative immunogenic neoantigens are currently selected by using several peptide-HLA-I binding predictors. While HLA-I binding affinity is essential to tumor cell presentation of neoantigens, it is insufficient to determine neoantigen immunogenicity. As such, the vast majority of predicted tumor neoantigens (98%) are not immunogenic and ineffective in activating antitumor responses. To address this challenge, we propose integrating T-cell receptor (TCR) repertoire profiles of tumor infiltrating lymphocytes (TILs), a key determinant of immunogenicity, in the current neoantigen prediction workflow.

Methods

We sequenced the CDR3 regions of TCR beta chain (TCRB) repertoire in 27 colorectal cancer (CRC) patients and predicted neopeptide-TCR binding by using our previously published predictor, epiTCR and another existing tool pMTNet. We constructed a machine learning model which considers both peptide-HLA-I and peptide-TCR binding features to prioritize high quality immunogenic neoantigen candidates. We further performed ELISpot assay on autologous PBMCs from four CRC patients to experimentally validate the effectiveness of our model. Finally, we selected the highest immunogenic candidate to perform 10X single-cell sequencing to identify the potential neoantigen-specific TCRs for cell-based vaccine development.

Results

We showed that 5.5% of neopeptide candidates despite having weak HLA-I binding affinity are capable of binding to TCR at predicted rank < 2%. The integrated model achieved higher accuracy for ranking immunogenic neopeptides than the conventional NetMHCpan tool solely relying on peptide-HLA-I binding features.

Conclusions

Our study presents an efficient process to select and prioritize neoantigens exhibiting both HLA presentation and immunogenicity for personalized cancer immunotherapy.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Medical Genetics Institute.

Funding

Nexcalibur Therapeutics.

Disclosure

All authors have declared no conflicts of interest.