ID #1121 Actionable alterations in diffuse midline glioma: the yield of extended NGS in real-world practice

Background Diffuse midline glioma (DMG) H3 K27-altered remains a lethal brain tumor. While integrated histo-molecular diagnosis is increasingly recommended, real-world feasibility and the yield of extended molecular profiling in terms of actionable alterations are still incompletely described. Methods We retrospectively reviewed DMG patients treated from 2009–2025 (n = 77; age 2–31 years), collecting histopathological, immunohistochemical and molecular data whenever available. Somatic extended NGS profiling was performed in a subset of patients, based on tissue availability; variants were interpreted according to ACMG/AMP criteria. Molecular features were compared between pontine and non-pontine tumors. Exploratory univariate overall survival (OS) comparisons (log-rank test) were performed within the extended-NGS subset for select co-mutations. Results Histological confirmation was achieved in 62/77 patients (80.5%), with no permanent morbidity/mortality associated with stereotactic biopsy. Molecular testing was available in 36/77 (46.8%) and extended NGS in 25/77 (32.5%). In the extended-NGS subset (n = 25), recurrent co-alterations involved TP53 (52.0%), ATRX (32.0%), PTEN (24.0%), NF1 (24.0%), PIK3CA (20.0%), PDGFRA (16.0%), FGFR1 (16.0%), and ATM (16.0%). TP53 alterations were more frequent in non-pontine DMG compared with pontine tumors (76.9% vs 25.0%; p = 0.017). Alterations clustered in potentially actionable signaling pathways (RTK/RAS and PI3K/AKT/mTOR) and in genes involved in chromatin remodeling/DNA damage response. In exploratory univariate analyses, no co-mutation or increased CNA burden was significantly associated with OS (TP53 p = 0.5845; >1 CNA p = 0.3624), despite a numerically shorter median OS with >1 CNA (11.31 vs 21.63 months). Conclusions In a real-world setting, integrated histo-molecular characterization is feasible for most DMG patients and extended NGS reveals recurrent co-alterations with site-specific patterns that cluster in potentially actionable pathways. These data support systematic tissue acquisition and molecular tumor board discussions to optimize biology-driven stratification and targeted therapy opportunities. Larger, prospective and multicenter cohorts are required to better clarify the prognostic and predictive roles of individual co-mutations and CNA burden.