P03.08.A A PILOT EXPERIENCE OF APPLICATION OF RANO 2D AND 3D IN PATIENTS WITH NEWLY-DIAGNOSED ASTROCYTOMA IDH MUTANT CNS WHO GRADE 4

Abstract BACKGROUND The WHO 2021 classification of central nervous system (CNS) tumors introduced astrocytoma IDH mutant WHO grade 4 (AG4) as a novel and distinct entity with better outcome compared to glioblastoma IDH wild type (GBM). Volumetric analysis (RANO 3D) did not significantly improve correlation with overall survival (OS) over RANO 2D in GBM, while ensured more stable and reliable measures of tumor growth rate in IDH mutant lower-grade gliomas. Specific studies exploring performance of RANO 3D in AG4 are currently lacking. MATERIAL AND METHODS We retrospectively reviewed clinical and radiological data of 44 patients diagnosed with AG4 at our Institution, selecting 14 new diagnosis with available imaging from post-surgery to first progressive disease (PD) to chemo-radiation. Imaging series were assessed by a neuro-radiologist panel for RANO 2D and by a deep learning neural network-based algorithm (3D U-Net) for RANO 3D. 3D U-Net screened four areas (contrast enhancement, non-enhancing tumor core, FLAIR hyperintensity and surgical cavity) and selected all appropriate signal alterations for volumetric analysis, with a PD threshold set at ≥40% increase of enhancing and/or non-enhancing components. Kaplan-Meyer method were used to test progression-free survival (PFS) and overall survival (OS) and Pearson test was used to correlate PFS and OS, with threshold for significance at p≤0.05. RESULTS A total of 12/14 patients experienced PD to first-line chemo-radiation and 11/14 patients were deceased at time of analysis. Median age was 39.5 years and most received a gross-total resection (11/14), had a Karnosky Performance status ≥80% (13/14), were MGMT methylated (6/14) and had non-R132H IDH mutation (6/14). All patients (100%) received post-operative chemo-radiation followed by maintenance temozolomide. With a median follow-up of 36.4 months (range: 15.6-84.6 months), median OS was 36.2 months. RANO 3D identified PD earlier than RANO 2D in 6/12 patients, with median PFS of 9.7 and 13 months, respectively. Of these 6 cases, 4 reached a threshold for PD in the FLAIR hyperintensity and 2 in the contrast enhancement. Correlation between PFS and OS appeared higher for RANO 2D (Pearson 0.80, p 0.0005) compared to RANO 3D (Pearson 0.69, p 0.0056) CONCLUSION In our pilot experience (1) median PFS was shorter with RANO 3D compared to RANO 2D, especially for PD of FLAIR hyperintensity; (2) correlation with OS was higher for RANO 2D compared to RANO 3D. Given the stronger association with prognosis and lower complexity, RANO 2D should remain the standard criteria for clinical trials of AG4. Limits of the study included small sample size and no definitive criteria to clearly distinguish edema/necrosis from tumor in the FLAIR area, particularly for RANO 3D. A prospective, observational, multi-center trial across selected hospitals from Italy and Spain is now opening to recruitment.