Utility of Intravoxel Incoherent Motion MR Imaging for Distinguishing Recurrent Metastatic Tumor from Treatment Effect following Gamma Knife Radiosurgery: Initial Experience

BACKGROUND AND PURPOSE:Intravoxel incoherent motion MR imaging can simultaneously measure the diffusion and perfusion characteristics of brain tumors. Our aim was to determine the utility of intravoxel incoherent motion–derived perfusion and diffusion parameters for assessing the treatment response of metastatic brain tumor following gamma knife radiosurgery.MATERIALS AND METHODS:Ninety-one consecutive patients with metastatic brain tumor treated with gamma knife radiosurgery were assessed by using intravoxel incoherent motion imaging. Two readers independently calculated the 90th percentile and the 10th percentile histogram cutoffs for perfusion, normalized CBV, diffusion, and ADC. Areas under the receiver operating characteristic curve and interreader agreement were assessed.RESULTS:With the combination of the 90th percentile histogram cutoff for perfusion and the 10th percentile histogram cutoff for diffusion, the sensitivity and specificity for differentiating recurrent tumor and treatment were 79.5% and 92.3% for reader 1 and 84.6% and 94.2% for reader 2, respectively. With the combination of the 90th percentile histogram cutoff for normalized CBV and the 10th percentile histogram cutoff for ADC, the sensitivity and specificity for differentiating recurrent tumor and treatment were 69.2% and 100.0% for reader 1 and 74.3% and 100.0% for reader 2, respectively. Compared with the combination of 90th percentile histogram cutoff for normalized CBV and the 10th percentile histogram cutoff for ADC, adding intravoxel incoherent motion to 90th percentile histogram cutoff for normalized CBV substantially improved the diagnostic accuracy for differentiating recurrent tumor and treatment from 86.8% to 92.3% for reader 1 and from 89.0% to 93.4% for reader 2, respectively. The intraclass correlation coefficients between readers were higher for perfusion parameters (intraclass correlation coefficient range, 0.84–0.89) than for diffusion parameters (intraclass correlation coefficient range, 0.68–0.79).CONCLUSIONS:Following gamma knife radiosurgery, intravoxel incoherent motion MR imaging can be used as a noninvasive imaging biomarker for differentiating recurrent tumor from treatment effect in patients with metastatic brain tumor.

Perfusion MR imaging techniques have significantly advanced and can now provide information regarding tumor physiology. There are several reports suggesting the usefulness of dynamic susceptibility contrast-enhanced perfusion MR imaging for differentiating recurrent metastatic brain tumor from stereotactic radiosurgery–induced radiation necrosis.^1–3 However, quantitative brain perfusion measurement remains a challenge for currently available MR perfusion methods. DSC and dynamic contrast-enhanced MR imaging are inhibited by their signal nonlinearity, and arterial spin-labeling exhibits, in addition to a low signal-to-noise ratio, a strong dependence on the transit time.Le Bihan et al⁴ defined intravoxel incoherent motion (IVIM) as the microscopic translational motion occurring in each image voxel in MR imaging. In biologic tissue, this incoherent motion includes molecular diffusion of water and microcirculation of blood in the capillary network, referred to as “perfusion.” These 2 phenomena account for the biexponential decay of the signal intensity on DWI when different diffusion b-values are applied. With the IVIM theory, both true molecular diffusion and water molecule motion in the capillary network can be estimated by using a single diffusion imaging-acquisition technique. As opposed to DSC, dynamic contrast-enhanced imaging, and arterial spin-labeling, IVIM has a unique capillary dependence that is not sensitive to the coherent laminar flow of arteries and veins. The measurement of IVIM is intrinsically local (ie, the encoding and readout are performed at the same location).⁵In our clinical experience, the major advantage of IVIM MR imaging is that because it allows the simultaneous acquisition of diffusion and perfusion parameters, it can provide both measures within corresponding solid lesions without the requirement for a further coregistration processing step. In the current study, we attempted to validate the IVIM-derived perfusion and diffusion parameters by using the clinicoradiologic correlation in patients with post-gamma knife radiosurgery (GKRS) metastatic brain tumor. We also assessed the diagnostic accuracy and added value of the IVIM method for differentiating recurrent tumor from treatment effect, compared with the combination of DSC perfusion MR imaging and DWI, which has commonly been used as a parameter for brain tumor imaging.Our hypothesis was that the difference in vascularity between recurrent tumor and the treatment effect can be assessed by using an IVIM-derived perfusion fraction (f); and the combination of f and the true diffusion parameter (D) would show diagnostic performance comparable with the combination of normalized CBV (nCBV) and the ADC. The purpose of this study was to determine the utility of IVIM-derived perfusion and diffusion parameters for assessing the treatment response of metastatic brain tumor following GKRS.