STEM-10. SOX2 POSITIVE, PRESUMED TUMOR INVASION MEASURED WELL BEYOND CONTRAST ENHANCEMENT AND FLAIR HYPERINTENSITY IN BOTH TREATED AND UNTREATED GLIOBLASTOMA PATIENTS ASSESSED AT AUTOPSY

Abstract

Abstract In the treatment of glioblastoma (GBM), MRI is used to determine location and extent of cancer. It is poorly understood how far tumor invasion may exist outside of contrast enhancement (CE) and FLAIR hyperintensity (FL). This study identifies presumed non-enhancing tumor invasion using SOX2 staining on tissue collected at autopsy, which may highlight invading glioma cells. Three patients with primary GBM were included for this study. Patient 1 (P1) with an overall survival (OS) of 148 days received no treatment following sub-total resection. Patient 2 (P2) (OS = 1601) and patient 3 (P3) (OS = 629 days) underwent the clinical standard of care treatment including chemo-radiation and bevacizumab at recurrence. At autopsy, brains were sliced axially to match patients’ last MRI scan. Samples were obtained including regions suspicious of tumor and regions appearing normal. SOX2 positive cells were identified visually on digitized autopsy histology. Samples were visually matched to the patient’s last T1C and FLAIR MRI acquisition prior to death. Approximate distance from CE and FL from SOX2 positive cells was measured digitally, limited to the same MRI slice (2D assessment). For P1 who received no treatment, SOX2 positive cells were identified 3.9cm from CE and 4.1cm from FL within the same hemisphere as the primary tumor, and 2.1cm and 2.3cm within the opposite hemisphere, respectively. For P2 and (P3), SOX2 positive cells were identified approximately 4.4cm (5.2 cm) from CE and 4.0cm (2.0cm) from FL within the same hemisphere as primary tumor, and 4.7cm (4.4cm) from CE and 4.1 cm (4.7cm) from FL within the opposite hemisphere. This study uses SOX2 to identify presumed GBM cells well beyond clinically identified regions of tumor, on T1C and FLAIR imaging. Additional research is necessary to determine SOX2 sensitivity and specificity for identifying GBM cells outside regions with readily identifiable GBM characteristics.