The ability to differentiate solid tumor from peritumoral edema is of clinical importance (e.g., for surgical planning and therapeutic assessment). When there is progression or recurrence, elevated vascularity is usually observed, particularly in contrast-enhanced tumor, and has been assessed with cerebral blood volume (CBV) as measured by dynamic susceptibility-contrast (DSC) MRI ¹. Intravoxel incoherent motion (IVIM) MRI is a contrast-material-free alternative for CBV measurement ², but few studies had examined its performance in brain tumor ^3,4. Here we performed IVIM MRI in a group of glioma patients and compared the intravascular fraction f (a parameter directly proportional to CBV) with DSC-MRI-derived CBV (CBV_DSC).

The institutional review board approved this study. Twenty-five patients with histologically proven brain gliomas were included. Tumors that had been treated were included if they showed definitive signs of residual or recurrent tumor at conventional MR imaging. Written informed consent was obtained from each participant beforehand.

MR imaging was performed on a 3T clinical system (Tim Trio, Siemens). Diffusion-weighted imaging was based on a single-shot twice-refocused spin-echo echo-planar readout: TR = 3.8 s, TE = 94 ms, field-of-view = 20 cm, in-plane matrix = 98x98, GRAPPA acceleration factor = 3, 18 slices, slice thickness = 4 mm, b-value = 0, 400, 600, 850, 1200, 1700 s/mm², 8 repetitions after 1 dummy scan. Diffusion encoding was applied along three orthogonal directions in separate series. DSC imaging was performed after intravenous injection of 0.1 mmol/kg bodyweight Gd-DTPA followed by 15-ml saline flush (TR = 1 s, TE = 25 ms, 120 measurements).

Echo-planar images were corrected for head motion. On a voxel-wise basis, f was estimated by fitting the multi-b-value data with a previously described model ⁵ in which pseudo-diffusion coefficient was dropped considering its low precision. The contrast leakage effect on CBV_DSC was remedied with a method described by Boxerman et al ⁶. Contrast-enhanced tumor, peritumoral edema, and normal-appearing white matter were defined by two raters independently on conventional images, and overlapped areas were used as the final regions of interest. The ability of f in identifying contrast-enhanced tumor was assessed by repeated-measures analysis of variance and the area under the curve (AUC) derived from receiver operating characteristic analysis. A p-value < 0.05 was considered significant.

Figure 1 demonstrates the typical observation that unaccounted contrast extravasation causes CBV_DSC miscalculation. In a lot of the voxels enhanced in the post-contrast T1-weighted image, the concentration-time curve undershoots the baseline after the passage of contrast agent, leading to underestimation in CBV (sometimes even negative CBV). By contrast, f is able to measure blood volume regardless of the integrity of blood-brain-barrier. Figure 2 shows the group comparison between f and CBV_DSC in glioma patients. Negative CBV_DSC is found in several contrast-enhanced tumors as well as some peritumoral areas when contrast leakage is not accounted for. After correction, negative CBV_DSC cases notably decrease and the correlation between f and CBV_DSC increases (Pearson’s r = 0.61 as opposed to 0.40 with uncorrected CBV_DSC). Repeated-measures analysis of variance revealed significant f difference among areas (peritumoral edema, contrast-enhanced tumor, and normal-appearing white matter): Greenhouse-Geisser F(1.654, 39.688) = 17.666, p < 10^-3. Post hoc analysis further indicated that f is lowest in peritumoral edema but there is no statistical difference between normal-appearing white matter and contrast-enhanced tumor. The AUC for differentiating contrast-enhanced tumor from peritumoral edema is 0.766 (p < 10^-3) with f and 0.804 (p < 10^-3) with corrected CBV_DSC, but there is no statistical difference between the two parameters (p = 0.62).

Correlation between f and CBV_DSC has been previously reported in healthy volunteers ^7,8. In this study, we further demonstrate that f correlates with CBV_DSC in brain glioma. However, the two measures have inherent differences that should be noted. First, f measures blood volume without contrast delivery and is thus less susceptible to alterations in capillary permeability. CBV_DSC originally works on a basis of intravascular tracer and demands correction when contrast leakage is present. The correction method ⁶ we adopted is computationally robust but its underlying assumptions (small leakage and homogeneous mean transit time) may not be always valid. Second, f is defined as the intravascular volume fraction of the protons that are moving or flowing in a random pattern and detectable by MR imaging. Thus, f is normally greater than CBV_DSC, particularly when partial volume is present with large vessels that are not randomly organized.

In conclusion, IVIM-derived f is able to differentiate contrast-enhanced glioma from peritumoral edema by detecting increased vascularity.