MRI-defined patterns of infiltration and outcome in patients with glioblastoma

Abstract Background Glioblastoma is a rare primary tumor of the brain. Infiltration of glioblastoma into the brain parenchyma may influence prognosis. We therefore aimed to investigate possible influences of distinct patterns of MRI-defined infiltration on the prognosis. Methods We performed a retrospective analysis of sequential patients with glioblastoma between April 2005 and December 2017. Patient data were collected from the hospital data management system, MRI images from the hospital PACS and from cooperative radiology units. Patients were divided into subgroups based on the tumor growth pattern (frame-like, palisade-like, infilling). The impact of various factors on overall survival and progression-free survival was then calculated and compared between the groups. Results 259 patients were included. Of the 258 evaluable patients, 117 showed a palisade-like infiltration, 98 were classified as non-infiltrating frame-like, and 43 as infilling dense-solid. Standard prognostic factors aligned to published data. In multivariate analysis, no significant influence of palisade-like growth on overall survival and progression-free survival could be detected. In Cox regression analyses, we found a significant effect for overall survival in palisade-like tumors in the univariate analysis (OR 1.354, 95% CI 1.032–1.776, P = .029). Conclusion We show here a possible correlation of MRI-based infiltration patterns and survival in patients with glioblastoma. Our results correspond well to published literature that shows that certain subtypes of glioblastoma exhibit an enhanced invasion pattern and decreased survival. Our results should be verified in a prospective setting in a large patient cohort and by using automated methods for the classification of infiltration patterns in glioblastoma.