NIMG-11. AUTOPSY-BASED RADIO-PATHOMIC MAPS OF TISSUE COMPOSITION DELINEATE IDH1 STATS IN GLIOMAS

Abstract

Abstract Isocitrate dehydrogenase 1 (IDH1) mutation status is used as an important prognostic marker for gliomas, where IDH1-wildtype patients see shorter survival than patients with an IDH1 mutation. This study uses radio-pathomic maps of cell, extracellular fluid (ECF), and cytoplasm (Cyt) density developed using conventional MRI and aligned autopsy samples to test the hypothesis that IDH1-mutant tumors differ in contrast-enhancing and non-enhancing tissue composition from IDH1-wildtype tumors. A total of 426 patients from the UCSF-PDGM data set (380 IDH1-wildtype, 46 IDH1-mutant) were included in this study. T1, T1C, FLAIR and ADC images from each patient’s pre-surgical MRI were used to generate whole brain radio-pathomic maps of cell, ECF, and Cyt density using a previously reported algorithm. This tool was trained using aligned autopsy samples as ground truth and is particularly adept at highlighting areas of tumor beyond traditional imaging signatures. The mean value from each feature map was computed for each patient both within non-necrotic contrast enhancement and non-enhancing FLAIR hyperintensity. Higher ECF density (F=15.45, p< 0.001) and lower Cyt density (F=8.51, p=0.004) was observed within contrast-enhancement for IDH1 mutant patients, with no difference observed for cell density within contrast enhancement between mutation statuses (F=0.03, p=0.86). Higher cell density (F=4.16, p< 0.042), Higher ECF density (F=4.65, p=0.032), lower Cyt density (F=4.48, p=0.035) was also found in patients with IDH1 mutations in the non-enhancing FLAIR hyperintense region. These results suggest that the large-scale tissue composition of IDH1-mutant tumors may differ from IDH1-wildtype tumors, suggesting differences in growth patterns and pathological composition between tumor types. Future studies examining how tissue composition post-treatment over time may elucidate how pathological characteristics evolve for each tumor type.