Tumor location and seizure risk in glioblastoma population

Introduction: Epilepsy is common among glioblastoma multiforme (GBM) patients. Antiepileptic drug prophylaxis is widely extended in clinical practice despite it is not recommended by lack of efficacy and increase risk of adverse events (class I evidence). Here we present the first study using a voxel-based lesion symptom mapping (VLSM) to identify brain areas with a higher risk of presenting seizures in GBM population Material And Methods: One hundred and forty four GBM patients diagnosed between 2007-2013 were retrospectively reviewed from Hospital Universitari de Bellvitge-ICO L'Hospitalet databases. Patients were split into 2 cohorts: seizures at onset (n=37) and no seizures at onset (n=107). The last group (no seizures at onset) were also divided into 2 subgroups: patients who had seizures during the course of the disease (8 days from surgery) (n=24) and patients who never had seizures (n=71). Patients who suffered from seizures during surgery or 7 days post-surgery were excluded (n=12). A VLSM analysis using Gadolinium-enhanced T1-weighted Magnetic Resonance Imaging (MRI) was performed to determine whether tumor location was associated with an increased seizure risk. Additionally, external validation sets of 44 cases and 54 cases retrospectively collected from other two centers, between 2007 and 2014 was performed Results: GBM patients with a tumor location involving frontal regions (rolandic operculum, postcentral and precentral frontal area, middle frontal gyrus, insula and superior longitudinal fasciculus) were significantly more susceptible to suffer from seizures at onset. Among GBM patients who had no seizures at onset, an increased risk of presenting seizures during the course of the disease were identified if the tumor was located in subcortical frontal areas (corona radiata, corpus callosum, caudate, putamen and cingulum) and in superior occipital lobe, inferior temporal and fusiform gyrus and cuneus. Conversely, those GBM located in subcortical temporal areas (retrolenticular part of internal capsule, external capsule, posterior thalamic radiation, hippocampus, putamen and globus pallidus) and superior temporal lobe had a significantly lower risk of suffer from seizures during the course. These brain regions locations were consistently observed in both the discovery and the validation datasets Conclusions: This study suggests that tumor location is a useful marker to identify which GBM patients are at a high risk of suffering from seizures. These results may help both to lay a solid basis for proper stratification of patients in clinical trials focused on antiepileptic prophylaxis and to support the use of antiepileptic prophylaxis in a selected GBM population.