The purpose of this study is to evaluate changes in tumor size and ADC measured volumetrically before and after 90-yttrium (Y90) radioembolization as definitive treatment for patients with metastatic liver cancer.

Eighteen patients with metastatic liver cancer treated with 90-yttrium radioembolization between January 2013 and August 2015 at the City of Hope National Medical Center were identified in a retrospective study. Of the 18 patients, 5 had primary diagnosis of cholangiocarcinoma, 5 colorectal, 4 hepatocellular carcinoma and the remaining 4 were of mixed primary diagnosis. MRIs were obtained prior to therapy (median: 12 days) and an average of 3 months (median: 86 days) following administration of Y90. Post gadolinium T1-weighted (T1 PG) and apparent diffusion coefficient (ADC) images created from diffusion-weighted images (DWI) with b fields of 50 and 800 were analyzed. Three dimensional tumor volumes were collected on the T1 (PG) and ADC images by a trained expert (S.R.) and were reviewed by a board certified radiologist (J.M.P.). ADC voxel data was extracted from the greatest cross-sectional area of the lesions in the axial plane to interrogate intra-lesion heterogeneity. Mean ADC increased for the cohort following Y90, with a median increase of 222.5 mm2/s (p < 0.001, Wilcoxon). The increase in ADC suggests a decrease in the cellularity in the tumor. The change in tumor volume measured on T1 (PG) was not statistically significant (p = 0.32, Wilcoxon). Twelve patients showed a decrease in both T1(PG) volume and mean volumetric ADC. Four patients showed response on ADC but did not show a decrease in T1 PG tumor volume. Hepatocellular carcinoma patients (N = 4) showed the most dramatic decrease in T1(PG) volume, with an average 54% reduction with only 6% reduction in mean ADC. Cholangiocarcinoma patients showed the least response on T1 (PG), with an 88% average increase in tumor volume and a 31% increase in mean ADC. Volumetric analysis is an increasingly important means of quantifying imaging findings. Volumetric measures have been shown to have more stronger and superior criteria than 2D or 1D measures. This is due primarily to the subjectivity of region of interest placement on single two-dimensional views of spatially heterogeneous lesions. To determine if the mean ADC was a good representation of the lesion as a whole, we analyzed the kurtosis of the ADC values in a region of interest defined to be the largest cross-sectional area of the lesion in the axial plane. Prior to therapy, kurtosis ranged from -0.91 to 30.91 with a mean of 2.14. In the post treatment images kurtosis ranged from -.05 to 2.48 with a mean of 0.33, indicating peaked distributions. This suggests that the use of the distribution mean for the volume is not a misleading measure of ADC in this cohort. We interpret our findings to indicate that changes to tissue treated with Y90 radioembolization may be realized on ADC before T1 PG MRI. The stark changes observed on ADC compared to the modest changes on T1 PG MRI suggest that changes in mean ADC using the entire tumor volume may provide an early indicator of response to localized therapies such as Y90. Additional MRIs following therapy along with clinical outcomes are needed to evaluate the value of early changes in ADC in predicting response to therapy.