Imaging patterns of necrosis after treatment with chemoradiation are prognostic for survival in glioblastoma

Abstract Objectives Tumor necrosis is a hallmark of glioblastoma. While recent data suggest that necrotic tissue plays an active role in glioblastoma growth, the prognostic relevance of necrosis imaging remains understudied. This study evaluated necrotic tissue on MRI to identify a prognostic morphological signature and characterize its underlying biological characteristics. Materials and methods 82 patients with glioblastoma imaged after chemoradiation were retrospectively analyzed. A combination of two shape features (volume density and surface-to-volume ratio) was used to define prognostic necrosis phenotypes (low- vs. high-risk). The association between necrosis phenotypes and survival was tested with log-rank and multivariable Cox models adjusting for other prognostic variables. Diffusion, perfusion, and pH-weighted MRI were used to biologically characterize the two groups. Results The low-risk necrosis phenotype was more compact (not spatially dispersed), with smoother margins and fewer fragmentations. Patients with this phenotype had an almost three-fold lower hazard of death (HR: 0.35, p = 0.001), which was confirmed after adjusting for tumor volume, MGMT status, age, and Karnofsky performance score (KPS) (HR: 0.29, p = 0.001). Low-risk necrosis was characterized by greater necrosis volume and necrosis proportion, fewer isolated clusters, lower perfusion values, and higher oxygen extraction values. Conclusions The necrosis morphology bears prognostic information in post-chemoradiation glioblastoma and exhibits distinct biological features. Key Points Question Are quantitative MRI-derived necrosis phenotypes associated with survival in post-chemoradiation glioblastoma? Findings The low-risk necrosis phenotype, derived from two quantitative necrosis features, is independently associated with patient survival and reflects distinct perfusion characteristics . Clinical relevance Quantitative assessment of necrosis morphology from routine MRI may complement established prognostic factors and enable improved risk stratification of patients with glioblastoma after chemoradiation .