Meeting Banner
Abstract #2419

Comparison of Perfusion MRI-Based Methods to Estimate Histologic Tumor Fraction & Predict Survival in Recurrent GBM

Leland S. Hu1,2, Jennifer M. Eschbacher3, Amylou C. Dueck4, Seban Liu5, Kris A. Smith6, Kasuen Kotagama5, Stephen W. Coons3, Joseph E. Heiserman7, John P. Karis7, Todd Jensen8, William Shapiro9, Josef Debbins5, Peter Nakaji6, Burt G. Feuerstein9, Leslie C. Baxter5

1Radiology, Mayo Clinic Arizona, Phoenix, AZ, United States; 2Radiology, Barrow Neurological Institute, Phoenix, AZ, United States; 3Neuropathology, Barrow Neurological Institute; 4Biostatistics, Mayo Clinic Arizona; 5Keller Center for Imaging Innovation, Barrow Neurological Institute; 6Neurosurgery, Barrow Neurological Institute; 7Neuroradiology, Barrow Neurological Institute; 8Imaging Biometrics, LLC; 9Neurology, Barrow Neurological Institute

Perfusion-MRI (pMRI) measures or relative cerebral blood volume (relCBV) can distinguish subregions of tumor from radiation-injury within non-specific Contrast-Enhanced MRI lesions in recurrent GBM. As histologic tumor fraction impacts prognosis and management, we study three different pMRI-based methods of estimating tumor fraction and compare their correlations with outcome. Specifically, we report a new voxel-based relCBV thresholding method called Fractional Tumor Burden (pMRI-FTB), compared with previously published histogram-based Peak Height Position (PHP) and mean relCBV methods. pMRI-FTB showed the highest correlation with Histologic tumor fraction (r=0.82,p<0.0001) and was the only method to correlate with Overall Survival (p<0.006), suggesting its clinical utility.