ID #222 Antigen Discovery in Diffuse Midline Glioma: What We Know, What Limits Us, and Where We Go Next

Abstract

Diffuse midline glioma (DMG) is characterised by a low mutational burden, raising uncertainty regarding the feasibility of antigen-directed immunotherapy and limiting the predictive value of genomics-based target discovery. To directly define the antigens naturally presented by DMG cells, we performed systematic immunopeptidomic profiling of H3K27M-mutant DMG models across multiple HLA backgrounds, integrating baseline and therapy-perturbed conditions.

Using high-resolution mass spectrometry coupled with functional immunogenicity assays, we identified naturally processed and presented neoantigens across four distinct classical HLA class I allotypes. Despite limited mutation load, these neoantigens elicited robust antigen-specific T cell responses, demonstrating that actionable neoantigen presentation is detectable in DMG. Beyond canonical neoantigens, we uncovered dozens of immunogenic peptides derived from the dark proteome, including non-canonical translation products and atypical processing events not predicted by standard genomic approaches, substantially expanding the DMG antigenic target space.

Comparative analyses revealed that baseline antigen presentation in DMG is constrained and heterogeneous across models. Notably, the H3K27M mutation itself reshaped the immunopeptidome by inducing the presentation of a distinct subset of dark antigens that were absent in H3-wild-type comparators, establishing a direct mechanistic link between oncogenic epigenetic dysregulation and tumour antigen emergence. These findings indicate that tumour-intrinsic epigenetic state is a critical determinant of antigen visibility in DMG.

We further demonstrate that antigen presentation in DMG is dynamic and therapeutically modifiable. Radiation treatment induced the presentation of additional neoantigens and enhanced peptide presentation by non-classical HLA molecules, including HLA-E and HLA-G, revealing a previously underappreciated source of shared, therapy-induced targets.

Together, these data define a multi-layered and dynamic antigen landscape in paediatric DMG, comprising classical neoantigens, H3K27M-induced dark antigens, and therapy-induced targets presented by both classical and non-classical HLA molecules. This work provides a rational framework for integrating immunopeptidomics into therapy-aware antigen selection strategies to enable both personalised and shared immunotherapies for DMG.