HG-85. INTER-OBSERVER AGREEMENT IN NEUROPATHOLOGICAL HGG DIAGNOSIS : EXPERIENCE OF THE PRE-RANDOMISATION CENTRAL REVIEW IN THE HERBY TRIAL Pascale Varlet1, Marie-Cecile Le Deley2, Felice Giangaspero3, Christine Haberler4, Thomas S. Jacques5, Dominique Figarella-Branger6, Torsten Pietsch7, Felipe Andreiuolo1, Darren Hargrave5, Maura Massimino8, Tim Jaspan9, Chris Jones10, Amedeo A. Azizi11, Adela Canete12, Eric Bouffet13, Helen Smith14, Josep Garcia14, Gilles Vassal2, and Jacques Grill2; Sainte-Anne Hospital, Paris, France; Institut Gustave Roussy, Villejuif, France; Sapienza UniversityofRome,Roma, Italy; InstituteofNeurology,Wien,Austria; Great Ormond Street Hospital, London, UK; Hopital de la Timone, Marseille, France; University of Bonn, Bonn, Germany; Fondazione IRCCS Istituto Nazionale Tumori, Milan, Italy; University Hospital, Nottingham, UK; InstituteofCancerResearch,Sutton,UK;UniversitatsKlinik furKinderund Jugendheilkunde, Wien, Austria; H.U. LA FE, Valencia, Spain; Hospital for Sick Children, Toronto, ON, Canada; F. Hoffmann-La Roche Ltd, Basel, Switzerland In previous large paediatric multicentre trials for High-Grade Glioma (HGG), highproportions of diagnosticdisagreementacrossneuropathologists were reported. The HERBY trial is a Phase-II open-label, randomised, multicentre trial evaluating bevacizumab in paediatric patients with newly diagnosed supratentorial HGG. METHODS: Confirmation of HGG diagnosis by reference pathology was mandatory before randomisation, followed by a review by five independent expert neuropathologists. Inter-observer agreement was assessed using Cohen’s Kappa coefficients. RESULTS: Slides from 169/187 patients screened for eligibility were centrally reviewed, within 3 days in 69% of cases. HGG diagnosis was centrally confirmed in 149 cases (88%), whereas diagnosis was low-grade glioma in 5% (n 1⁄4 8), anaplastic ganglioglioma/PXA (n 1⁄4 8), miscellaneous diagnoses (n 1⁄4 4). 124 patients entered the randomised trial. Three cases were not confirmed by the pathology panel (reclassified in grade II oligodendroglioma or anaplastic gangliogliomas). The Kappa coefficient between local and reference pathologistwasmoderate for the subtype and substantial for the WHO grade (Kappa 1⁄4 0.45 and 0.68, respectively). Agreement varied according to the initial diagnosis with the confirmation of subtype and grade for 23% cases of oligodendroglioma or mixed oligo-astrocytoma, 76% of WHO grade III astrocytoma and 95% of glioblastoma. The review panel registered some degree of discordance for subtype or grade in 52% cases, with dispersed conclusions from the experts (. 3 out of 5) in 10% cases. However, Kappa coefficient reached 85% for the grading between the reference pathologist and the consensus. These data highlight the interest and feasibility of a pre-randomised real-time central neuropathological review in multicentre trials. Neuro-Oncology 18:iii48–iii77, 2016. doi:10.1093/neuonc/now073.81 #The Author(s) 2016. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.