ID #1110 Evolutionary history of dordaviprone-treated H3K27-altered diffuse midline glioma
Bavani Subramaniam, Francesco Feher, Truman Knowles, Sridevi Yadavilli, Ji Sunjoing, Carl Koschmann, Sabine Mueller, Javad Nazarian, Sebastian WaszakAbstract
Introduction
H3K27-altered diffuse midline glioma (DMG) is a high-grade glioma that primarily affects children and young adults and associates with poor clinical outcomes. We have shown that dordaviprone (ONC201), a ClpP agonist, disrupts metabolic pathways and associates with improved clinical outcomes. Molecular data from our early phase studies suggested that somatic driver mutations in EGFR associated with poor radiographic response to dordaviprone, yet the molecular mechanism of this potential resistance remained elusive. Here, we aimed to map mechanism of resistance to dordaviprone using molecular profiling of autopsied H3K27M-mutant DMGs and cell lines treated with ONC201 and the EGFR inhibitor osimertinib.
Methods
Multi-focal tissue samples (pons, thalamus) were collected at autopsy (n = 25; 13 treated with dordaviprone and 12 with other therapies) and processed for WES, lcWGS, mRNA-seq. Structural variants, somatic copy number alteration, and somatic driver gene mutations were predicted with DELLY2/Strelka2 and custom pipelines. Gene expression levels were quantified with STAR and tumour microenvironments were predicted with multiple cell type deconvolution algorithms. Two H3K27M DMG cell lines were treated with dordaviprone, osimertinib, and dordaviprone/osimertinib combination at multiple time points and DESeq2 was used for differential gene expression analysis.
Results
Analysis of DMG genomes revealed characteristic somatic mutations (H3K27M) in histone genes and oncogenic partner mutations in p53, PI3K/AKT, RTK, and MAPK signaling genes. No significant association between dordaviprone treatment and selection of driver mutations was identified. Notably, somatic EGFR mutations and amplifications were detected in 23% (3/13) of dordaviprone-treated DMGs yet absent in DMG patients treated with other therapies. Chromosomal instability, a hallmark DMG event, was significantly associated with somatic TP53 driver mutations, yet not with dordaviprone therapy. This data suggests that DMGs exhibit resistance to dordaviprone via non-genetic mechanisms. Transcriptomes Data demonstrated a global change in gene expression profiles. Cell type deconvolution demonstrated that changes were driven by changes in the tumor microenvironment with reduced rates of myeloid cell populations in dordaviprone-treated patients and validated by canonical myeloid cell marker genes. We further assessed the therapeutic potential of dordaviprone plus osimertinib (EGFRi) in in vitro models as a potential combination therapy. We observed activation of the mitochondrial unfolded protein response in dordaviprone-treated cells as well as rapid and adaptive responses to osimertinib treatment.
Conclusions
Our results suggest that dordaviprone is not driving genetic selection in DMGs yet associates with potential changes in the tumor microenvironment. These results warrant validation and may inform biomarker development and rational combination strategies.