Combining Perfusion and High B-value Diffusion MRI to Inform Prognosis and Predict Failure Patterns in Glioblastoma.

Abstract

Advanced imaging modalities such as high b-value diffusion and dynamic contrast enhancement magnetic resonance imaging have the potential to improve the clinical management of glioblastoma by informing prognosis, predicting sites of progression, and guiding dose-escalated radiation to maximize tumor control and minimize toxicity. Fifty-two patients with de novo glioblastoma underwent magnetic resonance imaging before chemoradiation therapy. Enhanced tumor volumes (TVs), excluding the surgical cavity, hypercellularity (TV_HCV) and increased cerebral blood volume (TV_CBV) were defined using conventional gadolinium-enhanced T1-weighted images, high b-value (3000 s/mm²) diffusion-weighted images, and cerebral blood volume maps from T1-weighted dynamic contrast enhancement images, respectively. The image-phenotype TVs were analyzed for prediction of progression-free survival (Cox proportional hazard models), and sites of progression (pattern of failure tumor volume). The median progression-free survival (PFS) of the cohort was 13 months. The TV_CBV and TV_HCV were spatially distinct, with a mean overlap of only 21%. Univariate analysis showed that increasing age, decreasing radiation dose, larger TV_HCV, and larger overlap of TV_HCV and TV_CBV were significantly associated with inferior PFS. Multivariate analysis identified that TV_HCV was the most adversely prognostic imaging-defined variable. Enhanced TVs, excluding the surgical cavity, and the union of TV_HCV and TV_CBV showed a high likelihood of containing the pattern of failure tumor volume, and the volume composed of the intersection of TV_HCV and TV_CBV had an especially high likelihood of progression. TV_HCV and the overlap of TV_HCV and TV_CBV are prognostic for PFS. Combinations of gadolinium-enhanced TVs, TV_CBV, and TV_HCV could predict tumor progression locations better than could individual subvolumes. Radiation dose escalation to these subvolumes could be a promising therapeutic strategy.