Abstract

Abstract Neoantigens were suggested to be crucial for the outcome of immune checkpoint blockade (ICB) therapies and accordingly, the neoantigen load of a tumor is often being used to predict therapy outcome. Current approaches utilize HLA binding predictions to rank peptides, and the neoantigen load of a given tumor is then usually defined as the number of peptides binding above a certain threshold. Using this approach some reports found a strong correlation of neoantigen load and response to ICB while others found a better correlation when mutational load was used instead. However, comparing results between studies is hard because they used different tools and thresholds to identify mutations. We sought to overcome this lack of uniformity in re-analyzing data from several published studies to consistently define the mutational burden and neoantigen load, and correlate them with therapy outcomes. We obtained sequencing data from several ICB studies and run all datasets through our in-house mapping and variant calling pipelines. We applied different scoring schemes to define the neoantigen load which were based on HLA binding predictions with varying thresholds and different metrics used. We found that the exact definition of neoantigen load has a major impact on how well this metric performs in predicting checkpoint blockade therapy outcome. We also implemented a likelihood score which is based on data of known immunogenic neoantigens and describes the likelihood of a predicted neoantigen to be immunogenic. Based on our preliminary analysis to date, we found that the cumulative likelihood score of a patient is a good predictor of ICB outcome, outperforming tumor mutational burden and predicted neoantigen load in several cohorts.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call