Characterization of glioma microcirculation and tissue features using intravoxel incoherent motion magnetic resonance...

Our aim was to investigate the pertinence of diffusion and perfusion magnetic resonance imaging (MRI) parameters obtained at 17.2 T in a 9L glioma rat brain tumor model to evaluate tumor tissue characteristics.

The local animal ethics advisory committee approved this study. 9L glioma cells were injected intracerebrally to 14 Fischer rats. The animals were imaged at 7 or 12 days after implantation on a 17.2-T MRI scanner, using 72 different b values (2-3025 s/mm(2)). The signal attenuation, S/So, was fitted using a kurtosis diffusion model (ADCo and K) and a biexponential diffusion model (fractions ffast and fslow and diffusion coefficients Dfast and Dslow) using b values greater than 300 s/mm(2). To bridge the 2 models, an average diffusion coefficient and a biexponential index were estimated from the biexponential model as ADCo and K equivalents, respectively. Intravoxel incoherent motion perfusion-related parameters were obtained from the residual signal at low b values, after the diffusion component has been removed. Diffusion and perfusion maps were generated for each fitted parameter on a pixel-by-pixel basis, and regions of interest were drawn in the tumor and contralateral side to retrieve diffusion and perfusion parameters. All rats were killed and cellularity and vascularity were quantitatively assessed using histology for comparison with diffusion and perfusion parameters.

Intravoxel incoherent motion maps clearly highlighted tumor areas as generally heterogeneous, as confirmed by histology. For diffusion parameters, ADCo and were not significantly different between the tumor and contralateral side, whereas K in the tumor was significantly higher than in contralateral basal ganglia (P < 0.0001), as well as biexponential index (P < 0.001). ADCo and in the tumor at day 7 were significantly higher than at day 12 (P < 0.01 and P < 0.001, respectively). fIVIM in the tumor from the kurtosis diffusion model was significantly higher than in contralateral basal ganglia (P < 0.001). fIVIM in the tumor at day 7 was significantly higher than in the tumor at day 12 (P < 0.0001). There was no significant difference for D* between the tumor and contralateral side (P = 0.06). A significant negative correlation was found between tumor vascularity and fIVIM (P < 0.05) as well as between tumor cell count and (P < 0.01).

Quantitative non-Gaussian diffusion and perfusion MRI can provide valuable information on microvasculature and tissue structure to improve characterization of brain tumors.