DIPG-20. PRE-RANDOMISATION CENTRAL REVIEW AND REAL-TIME BIOMARKERS SCREENING IN THE MULTICENTRE BIOLOGICAL MEDICINE FOR DIPG ERADICATION (BIOMEDE) TRIAL: LESSONS LEARNT FROM THE FIRST 120 BIOPSIES

• Published in: Neuro-oncology (Charlottesville, Va.) Va.), 2018-06, Vol.20 (suppl_2), p.i52-i53
• Main Authors: Varlet, Pascale, Debily, Marie-Anne, Teuff, Gwenael Le, Tauziede-Espariat, Arnault, Pages, Mélanie, Andreiuolo, Felipe, Lechapt-Zalcman, Emanuele, Barret, Emilie, Picot, Stephanie, Nysom, Karsten, Pravong, Sophie, Vassal, Gilles, Castel, David, Puget, Stephanie, Grill, Jacques
• Format: Article
• Language: English
• Subjects: Abstracts
• ISSN: 1522-8517; 1523-5866
• DOI: 10.1093/neuonc/noy059.113

Abstract INTRODUCTION BIOMEDE is a multicentre, phase-II trial comparing efficacy of erlotinib, everolimus, and dasatinib in combination with radiotherapy, in newly diagnosed diffuse intrinsic pontine glioma (DIPG) (NCT02233049). Randomisation is based on biomarkers determined in upfront stereotactic biopsies: mTOR pathway activation detected by the loss of PTEN expression and EGFR overexpression. METHODS A central pathology review is mandatory in BIOMEDE for DIPG diagnosis confirmation, assessment of H3K27M immunostatus, PDGFRA amplification by FISH and biomarkers evaluation for treatment allocation (no erlotinib if no EGFR overexpression; no everolimus if no PTEN loss, no specific marker for dasatinib allocation). We evaluated the process feasibility and the agreement between H3.3 immunoprofil and genotyping on the first 120 patients. RESULTS DIPG was centrally confirmed in 94% of cases (median time=1 day; range 0–6). The differential diagnoses were HGNET-MYCN (n=1) and angiocentric glioma MYB-QKI (n=1). No tumoral infiltration was found in 3 cases and extensive necrosis in 1 case. Overall, 38/118 evaluable cases showed EGFR overexpression (32%) and 90/118 had PTEN loss (76%). 15/105 evaluable cases had PDGFRA amplification (14%). Trimethylation loss at H3K27 was detected in all samples but 4 corresponding to HGNET-MYCN, angiocentric glioma and to 2 H3 wild-type gliomas. Agreement between genotyping (H3.3 and/or H3.1) and H3K27M immunostatus (loss of trimethylation and/or H3K27M) is substantial (Kappa coefficient=0.79, 95%-confidence interval, 0.41 to 1). CONCLUSIONS These data document the feasibility of real-time molecular pathology and biomarkers screening in international multicentre trials.