Low perfusion compartments in glioblastoma quantified by advanced magnetic resonance imaging and correlated with patient survival

Background and purpose: Glioblastoma exhibits profound intratumoral heterogeneity in perfusion. Particularly, low perfusion may induce treatment resistance. Thus, imaging approaches that define low perfusion compartments are crucial for clinical management. Materials and methods: A total of 112 newly diagnosed glioblastoma patients were prospectively recruited for maximal safe resection. The apparent diffusion coefficient (ADC) and relative cerebral blood volume (rCBV) were calculated from diffusion and perfusion imaging, respectively. Based on the overlapping regions of lowest rCBV quartile (rCBV(L)) with the lowest ADC quartile (ADC(L)) and highest ADC quartile (ADC(H)) in each tumor, two low perfusion compartments (ADC(H)-rCBV(L) and ADC(L)-rCBV(L)) were identified for volumetric analysis. Lactate and macromolecule/lipid levels were determined from multivoxel MR spectroscopic imaging. Progression-free survival (PFS) and overall survival (OS) were analyzed using Kaplan-Meier's and multivariate Cox regression analyses, to evaluate the effects of compartment volume and lactate level on survival. Results: Two compartments displayed higher lactate and macromolecule/lipid levels compared to contralateral normal-appearing white matter (each P < 0.001). The proportion of the ADC(L)-rCBV(L) compartment in the contrast-enhancing tumor was associated with a larger infiltration on FLAIR (P < 0.001, rho = 0.42). The minimally invasive phenotype displayed a lower proportion of the ADC(L)-rCBV(L) compartment than the localized (P = 0.031) and diffuse phenotypes (not significant). Multivariate Cox regression showed higher lactate level in the ADC(L)-rCBV(L) compartment was associated with worsened survival (PFS: HR 2.995, P = 0.047; OS: HR 4.974, P = 0.005). Conclusions: Our results suggest that the ADC(L)-rCBV(L) compartment may potentially indicate a clinically measurable resistant compartment. (C) 2019 The Authors. Published by Elsevier B.V.