1069 - Characterization of Glioblastoma Infiltration and Survival by Brain Regions
Presenter(s)
W. J. Yeo1, M. Liu2, H. Feng1, D. Ruan3, W. Yang1, and K. Sheng4; 1University of California, San Francisco, San Francisco, CA, 2University of California San Francisco, San Francisco, CA, United States, 3Department of Radiation Oncology, University of California, Los Angeles, Los Angeles, CA, 4Department of Radiation Oncology, University of California, San Francisco, San Francisco, CA
Purpose/Objective(s): It has been shown that glioblastoma (GBM) growth may be promoted by specific neuronal signals. We aim to investigate spatial and connectivity patterns of GBM infiltration by brain regions, as well as their associated risks for mortality. We also assess associations between demographic and clinical characteristics with the brain regions.
Materials/Methods: The study population was 434 patients from Río Hortega University Hospital and University of California, San Francisco with incident GBM (mean age 60.4 years, 38.9% women). Their T1-weighted magnetic resonance (MR) images were registered to the Neuromorphometrics atlas (136 brain regions), and GBM was defined with the necrotic, enhancing, and non-enhancing regions segmented from T1, T1ce, and T2-FLAIR MR images, respectively. The prevalences of GBM occurring in each brain region was first assessed. Infiltration dependencies between brain regions was assessed with the residual of a linear regression between pairwise Kendall correlations of GBM occurrence by brain regions and their centroids’ Euclidean distance. Logistic regressions adjusted for age, sex, MGMT methylation, IDH mutation, and tumor volume were used to assess associations between each brain region and mortality. Associations between these covariates and GBM brain regions were assessed via logistic regressions as well. Significance in associations were defined with the Bonferroni-corrected threshold p < 0.05/136.
Results: The most prevalent regions for GBM occurrence were the cerebral white matter (WM) (left: 73.5%, right: 63.4%) and lateral ventricles (left: 58.8%, right: 52.3%), complementing previous studies suggesting high GBM infiltration along WM, as well as the subventricular zone being a neural stem cell niche. GBM was more prevalent in the left hemisphere than the right, an observation consistent with our connectivity analysis that showed strong negative correlations between contralateral brain regions even after confounding effects of pairwise region distances were removed. The strongest negative correlation was between the left and right cerebral WM (-0.64), suggesting low inter-hemispherical GBM infiltration. The strongest positive correlations were between the fronto-occipital lobes (>0.50), potentially highlighting the fronto-occipital fascicle as a tract of interest. No specific brain region was significantly associated with mortality, suggesting the low survival rates of GBM may not be specific to brain regions. Of note, IDH mutation was a strong risk factor for GBM occurring in the left subcallosal area (hazard ratio 9.4, 95% confidence interval 3.1 – 27.9).
Conclusion: GBM infiltration exhibits spatial and connectivity patterns by brain regions that can be discerned through statistical methods. IDH mutation was found to be a strong risk factor for GBM growth in the left subcallosal region. Our results contribute to understanding the pathophysiology of GBM and may have clinical implications towards improving GBM treatment.