OP37THREE-DIMENSIONAL MRI ASSESSMENT OF DIFFUSION AND PERFUSION STRATIFIES OVERALL SURVIVAL IN PATIENTS WITH GLIOBLASTOMA

Abstract

INTRODUCTION: Imaging assessment of glioblastoma (GBM) currently involves 2-dimensional analysis of conventional T1- and T2-weighted images. This study used advanced imaging to assess 3-dimentional (3-D) patterns of diffusion and perfusion spread in GBM. METHOD: 25 patients with retrospectively-confirmed GBM underwent pre-operative MRI. Maps of apparent diffusion coefficient (ADC) and relative cerebral blood volume (rCBV) were processed and regions of interest (ROIs) were developed using thresholds of low ADC with overlapping high rCBV. ROIs were visually assessed using interactive 3-D images. Patterns of spread within the T1- and T2-weighted abnormality (entire lesion) were qualitatively categorised and the presence or absence of ‘ROI clusters' within different tumour components was also observed. Overall survival (OS) was assessed using Kaplan-Meier statistics with significance accepted at p = 0.05 with statistical power > 95%. RESULTS: Four 3-D phenotypes were identified based on ROI pattern of spread: localised (4/25 patients, 16%), showing minimal and confined ROI spread; diffuse (19/25, 76%), shapeless ROI spread throughout the ‘entire lesion'; massive (2/25, 8%), ROI composing most of the ‘entire lesion'; and encapsulated (5/25, 20%), ROI diffusely spreading around a spherical component. The presence of ‘ROI clusters' was observed in CE (12/25, 48%) and necrotic (6/25, 24%) regions. Kaplan-Meier statistics found that an encapsulated ROI shape was associated with poor OS (p < 0.001). The presence of ‘ROI clusters' within CE was associated with improved survival patterns (p < 0.001). CONCLUSION: Qualitatively assessing 3-D patterns of ADC and rCBV spread stratifies overall survival in patients with GBM.