Mapping cell density and hypoxia in glioblastoma using time-dependent diffusion MRI: improved cell density assessment...

Objective.Tumor heterogeneity, including differences in cell density and hypoxic fraction (HF), can impact the efficacy of radiation therapy (RT). Quantitative imaging biomarkers (QIBs) are required to assess spatial tumor heterogeneity to allow personalization of RT with biologically guided adaptive RT. Time-dependent diffusion MRI (TDD-MRI) derived metrics are promising QIBs, as they are potentially able to characterize different aspects of tissue microstructure better than conventional diffusion-weighted MRI (DW-MRI). Time-dependent diffusion contrast (TDDC), a recently proposed TDD-MRI method, has shown promise in characterizing tissue microstructure by subtracting images acquired with different timing parameters. The current study aimed to biologically validate the TDDC in a small cohort of glioblastoma patients and determine the added value of TDDC compared to conventional diffusion metrics.Approach.12 patients with glioblastoma underwent MRI scanning before surgery. A total of 27 image-guided biopsy specimens were collected during surgery to obtain the histological cell density and HF. TDDC and conventional diffusion metrics (apparent diffusion coefficient (ADC) from a monoexponential model, mean diffusivity (MD) and mean kurtosis (MK) from the diffusion kurtosis imaging model) were obtained from the TDD-MRI acquisition in locations corresponding to the biopsies to determine the associations with histology.Main results.TDDC correlated with cell density (ρ= 0.71,p= 0.003), whereas no significant correlation was found between cell density and ADC (ρ= -0.12,p= 0.6), MD (ρ=-0.11,p= 0.7), or MK (ρ=0.03,p= 0.9). Moreover, no significant correlation was found between HF and any of the diffusion metrics (p> 0.1).Significance.For the first time, we related a clinically implementable TDD-MRI method to histology in glioblastoma and found that TDDC was a better cell density biomarker compared to conventional diffusion metrics. This positions TDDC as a potential QIB for biologically guided adaptive RT.