Development of a novel MHC/epitope prediction tool PrDx trained on data from the Neoscreen platform, and presentation of a first case where NeoScreen and PrDx where applied for the selection of 40 neo-epitopes amongst 20 000 mutations in a colorectal cancer patient. (October 2022)

• Record Type: Journal Article
• Authors: Prachař, Marek; Bagger, Frederik Otzen; Justesen, Sune; Steen-Jensen, Daniel
• Is Part Of: Molecular immunology. Volume 150(2022)
• Page Start: 26
• Publication Date: 2022-10
• Subjects: Immunochemistry -- Periodicals; Molecular biology -- Periodicals; Immunochemistry -- Periodicals; Allergy and Immunology -- Periodicals; Molecular Biology -- Periodicals; Immunochimie -- Périodiques; Biologie moléculaire -- Périodiques; Immunochemistry; Molecular biology; Periodicals; Electronic journals; 571.96
• DOI: 10.1016/j.molimm.2022.05.089
• Languages: English
• ISSNs: 0161-5890

Abstract

Abstract : Mutations in cancer can cells may lead to the formation of neo-epitopes potentially presented by both major histocompatibility complex (MHC) class I or II. These neo epitopes may be recognized by CD8+ or CD4+ T-cells, and trigger an immune response. Only a small fraction of the neo-epitopes will be displayed by the MHC class I or II. One of the challenges of cancer immuno-therapy is therefore to predict which neo-epitopes are susceptible to elicit a T cell response. Software tools such as netMHC, Mhc Flury and many others are over-predictive as the bulk part of the data used to train these methods are based on affinity assays. Several publications have indicated that stability assays may be better to train prediction tools for epitope presentation by MHC, therefore paving the way to effective cancer vaccine design. We performed stability assay measurements through our Neoscreen platform for 10 MHC class I and 10 MHC class II alleles using a peptide scan library approach. The data is used to train a prediction tool, PrDx, that relies on a combination of different machine learning methods (random forest, feed forward neural networks and recurrent neural networks), of which the outputs are gathered in an assemble model, able to predict T-cell epitopes, based on both stability and affinity data. The models are then further trained with peptides predicted to have strong stability, until satisfactory performances are attained. Our method is able to further filter potential peptides for cancer vaccine design, compared to state-of-the-art epitope prediction methods. The PrDx and NeoScreen platforms have been applied in a first patient case where 20, 000 mutations were identified, with 40 of the most likely neoepitopes being selected. The patient is expected to be treated in April/May 2019.