Abstract Understanding the spatial distribution of gliomas in the brain and their molecular subtypes can aid in diagnosis and development of targeted therapies. This study aims to create probabilistic radiologic maps of glioma locations using large MRI datasets and the most recent consensus brain tumor classification. Neuroimaging data from multiple databases were analyzed. Patients included had MRI T1 images and validated tumor segmentations. Probabilistic tumor maps were generated whereby binary tumor masks were aligned to a standard brain template and aggregated into compute voxel-wise frequency maps of glioma occurrence detailing glioma volume, molecular subtype, age, sex, and overall survival with tumor location. The study included 2164 patients with gliomas. Key findings include distinct spatial patterns associated with glioma size and molecular subtype: smaller tumors favored the left temporal region, medium-sized tumors the medial frontoparietal and bilateral temporal regions, and larger tumors the frontotemporoparietal regions, predominantly on the right. Isocitrate dehydrogenase (IDH) IDH wild-type tumors were more common in medial parietotemporal regions, while IDH-mutant tumors were preferentially found in frontotemporal regions. Younger patients had more frontal tumors, while older patients had higher parieto-occipital tumor burdens. Tumors in medial structures and parietal lobes were linked to lower survival whereas right temporal tumors were associated with higher rates of survival. These findings likely correlate with isocitrate dehydrogenase mutation status. Leveraging 8 glioma databases, probabilistic tumor maps revealed significant relationships between brain regions molecular subtypes and clinical outcomes. These findings could be used in clinical decision-making and offer insights into glioma pathogenesis and treatment of patients impacted by this disease.
Samuel et al. (Tue,) studied this question.