ID #666 Deciphering cellular sates associated with response to dordaviprone (ONC201) in diffuse midline glioma patients from the PNOC022 clinical trial

Abstract

Diffuse midline gliomas (DMG) are rare, high-grade gliomas, predominantly affecting children and young adults. The absence of effective therapies renders DMG universally fatal. A major obstacle to therapeutic development is marked intratumoural heterogeneity, characterised by multiple glial-lineage cellular states: oligodendrocyte precursor-like (OPC-like), astrocyte-like (AC-like), oligodendrocyte-like (OC-like) or mesenchymal-like (MES-like) populations, with cellular diversity likely being a key determinant of therapeutic response. Preliminary results from the PNOC DMG-ACT (DMG-Adaptive Combination Trial) PNOC022 suggest improved overall survival in patients treated with the combination of dordaviprone and paxalisib, compared with outcomes reported across prior DMG studies conducted by the Pediatric Neuro-Oncology Consortium. However, inter-patient variability in treatment response remains, without a clear molecular or clinical predictor. This study therefore seeks to investigate the cellular architecture of DMG, and to identify tumour cell-state features associated with response to dordaviprone and paxalisib.

Spatial transcriptomic profiling (Xenium 5K) was performed on fresh-frozen tumour biopsies from patients enrolled in PNOC022, including patients with inferior (n = 3) and superior survival outcomes (n = 3). Analysis of the tumour compartment identified a conserved spectrum of malignant cellular states across patients, spanning stem-like OPC-like and cycling populations, through to more differentiated AC-like, OC-like, or MES-like phenotypes. Importantly, the relative composition of these cellular states at diagnosis was significantly associated with clinical outcome (p < 0.001). OPC-like (z=-0.61), cycling (z=-0.38), and OC-like states (z=-0.75) were enriched in inferior survivors, whereas AC-like (z = 0.47) and MES-like states (z = 0.41) were dominant predictors of superior survival. Further interrogation identified state-specific biomarkers, including GPR17 (OC-like) and GFAP/CD44 (AC-/MES-like), highlighting candidate biomarkers that may predict therapy response and enable patient stratification.

We are currently expanding our analysis with validation cohorts to strengthen these findings and to define a clinically actionable, predictive hierarchy of DMG cellular states that informs patient outcomes and treatment response following combination therapy.