Better Tools Needed for Neoantigen Prediction

Immunitrack Report from the NeoAg Summit, 14-16 November 2018, Boston

As next generation sequencing (NGS) becomes increasingly important in cancer therapy, clinicians are faced with an overwhelming flood of mutanome data. NGS identifies many potential neo-epitopes from somatic mutations in tumors, but the question remains: which neo-epitopes are immunogenic?

Recently, Immunitrack attended the 2-day NeoAg Summit in Cambridge, Boston. The summit covered plenty of ground with twin tracks in ‘Prediction and Discovery’ and ‘Clinical Development & Manufacturing.’ The NeoAg summit highlighted significant progress - but also outstanding issues - in the neo-antigen field. One issue of high interest to us at Immunitrack is improving the selection of appropriate neoantigen candidates. Given the use of peptides in cancer vaccines, this selection is crucial to a vaccine’s therapeutic value.

If personalised cancer therapeutics is to be realized, there is an urgent need to identify immune system-targeting neo-antigens or neo-epitopes. A typical first screen for immunogenic epitopes relies upon bioinformatics tools for prediction of binding to the different classes of major histocompatibility complex, MHC. One speaker at the summit described the current epitope-binding prediction tools as “not fit for purpose.”

Our CSO, Sune Justesen, was honoured to speak at the NeoAg summit. Sune’s presentation highlighted Immunitrack’s work in running best-in-class peptide-MHC (pMHC) stability screens. Our own view is that the stability of the pMHC complex is a better indicator of immunogenicity than traditional affinity-based assays. By coincidence, a paper from Blaha et al., published the day before Sune’s presentation, fully supports this view (Cancer Immunol Res. 2018 Nov 13). Furthermore, Sune introduced Immunitrack’s software tool for the prediction of immunogenic epitopes, PrDx. PrDx is trained upon Immunitrack’s pMHC stability assay data, generated from our NeoScreen™ platform.

PrDx generates significantly less false positives (up to 80% fewer) than the industry standard (NetMHC). In this way, PrDx generates larger subsets of truly immunogenic T-cell epitopes.

In other highlights from the NeoAg Summit, the theme of improving MHC presentation prediction featured in other talks. Laronna Colbert from the FDA presented a regulatory perspective on neoantigen-based cancer therapies.

We also really appreciated the shout out from Drew Pardoll, from Johns Hopkins University School of Medicine. Professor Pardoll’s own presentation also featured the hot topic of neoantigen prediction, with better predictors being described as a “holy grail” in this field.

Finally, we’d like to thank all those attendees who spoke with Sune and Robin from Immunitrack at the NeoAg summit, who asked interesting questions and overall made our stay in Boston a very enjoyable experience.