UCSF

Background and purpose

Super-resolution track density imaging generates anatomic images with submillimeter voxel resolution by using high-angular-resolution diffusion imaging and fiber-tractography. TDI within the diseased human brain has not been previously described. The purpose of this study was to correlate TDI with histopathologic features of GBM.

Materials and methods

A total of 43 tumor specimens (24 contrast-enhancing, 12 NE, and 7 centrally necrotic regions) were collected from 18 patients with treatment-naïve GBM by use of MR imaging-guided neurosurgical techniques. Immunohistochemical stains were used to evaluate the following histopathologic features: hypoxia, architectural disruption, microvascular hyperplasia, and cellular proliferation. We reconstructed track density maps at a 0.25-mm isotropic spatial resolution by using probabilistic streamline tractography combined with constrained spheric deconvolution (model order, 8; 0.1-mm step size; 1 million seed points). Track density values were obtained from each tissue site. A P value of .05 was considered significant and was adjusted for multiple comparisons by use of the false discovery rate method.

Results

Track density was not significantly different between contrast-enhancing and NE regions but was more likely to be elevated within regions demonstrating aggressive histopathologic features (P < .05). Significant correlation between relative track density and hypoxia (odds ratio, 3.52; P = .01), architectural disruption (odds ratio, 3.49; P = .03), and cellular proliferation (odds ratio, 1.70; P = .05) was observed irrespective of the presence or absence of contrast enhancement.

Conclusions

Numeric values of track density correlate with GBM biologic features and may be clinically useful for identification of regions of tumor infiltration within both enhancing and NE components of GBM.