Quantification of perineural satellitosis in pretreatment glioblastoma with structural MRI and a diffusion tensor imaging template

Abstract

Background Survival outcomes for glioblastoma (GBM) patients remain unfavorable, and tumor recurrence is often observed. Understanding the radiological growth patterns of GBM could aid in improving outcome. This study aimed to examine the relationship between contrast-enhancing tumor growth direction and white matter, using an image registration and deformation strategy. Methods In GBM patients two pretreatment scans (diagnostic and neuronavigation) were gathered retrospectively, co-registered to a template and DTI atlas. The GBM lesions were segmented and co-registered to the same space. Growth vectors were derived and divided into vector populations parallel (Φ = 0-20 degrees) and perpendicular (Φ = 70-90 degrees) to white matter. To test for statistical significance between parallel and perpendicular groups, a paired samples Students T-test was performed. O6-methylguanine-DNA methyltransferase (MGMT) methylation status and its correlation to growth rate was also tested using a one-way ANOVA test. Results For 78 GBM patients (mean age 61 years ± 13 SD, 32 men), the included GBM lesions showed a predominant preference of perineural satellitosis (p<0.001), with a mean percentile growth of 30.8% (95CI 29.6% – 32.0%) parallel (0° < |Φ| <20°) to white matter. Perpendicular tumor growth with respect to white matter microstructure (70° < |Φ| <90°) showed to be 22.7% (95CI 21.3% – 24.1%) of total tumor growth direction. Conclusions The presented strategy showed that tumor growth direction in pretreatment GBM patients correlated with white matter architecture. Future studies with patient-specific DTI data are required to verify the accuracy of this method prospectively to identify its usefulness as a clinical metric in pre- and posttreatment setting.