In Vivo Assessment of the Spatial Distribution of Liver Injury Following Stereotactic Body Radiation Therapy for Hepatocellular Carcinoma

Abstract

To improve understanding of the relationship between delivered radiation dose and functional response of the liver using Gd-EOB-DTPA (EOB) enhanced MRI. EOB is a hepatobiliary-specific gadolinium based intravenous contrast agent for MR imaging with approximately 50% of the injected dose taken up by functional hepatocytes allowing for the distinction of functional regions of hepatocytes from non-functional zones. We hypothesized that decreased EOB uptake in post-treatment imaging will correspond to a specific iso-dose distribution after liver stereotactic body radiation therapy (SBRT). We reviewed pre/post liver function tests (LFT) for patients treated at our institution with SBRT for hepatocellular carcinoma (HCC) between 2010 and 2015. Patients with pre- and 3-month post-treatment MRI with EOB were retained for the analysis. The 3D-dosimetric data from CT planning scan were deformed to MRI with quality of deformation assessed by DICE coefficient. Images were visually reviewed for the regions of liver hypointensity (LHI) on T1-w images by independent reviewers that were blinded to 3D-dosimetric data. Based on the volumetric data, a dose-volume-histogram was calculated. We estimated the threshold function loss as the D90 (i.e., the dose achieved in 90% of the LHI volume). The volume of LHI were also recorded and correlated to the volume corresponding to various iso-dose using Pearson’s correlation as confirmation. Associations between LFT and liver enhancement normalized to the spleen were estimated with linear regression. Thirty patients met inclusion criteria. All patients were treated with five fractions of SBRT, with doses ranging from 31.5 to 54 Gy based on mean liver dose (MLD) and tumors’ proximity to organs at risk. MLD was 8.8 ± 3.8 Gy and the median PTV volume was 70.9 cc (IQR = 44.7-126.9). The median MELD score at baseline and 3 months after treatment were 9 (IQR = 7-11) and 9 (IQR = 7-12) respectively which were not significantly different (P = 0.70). Higher direct bilirubin was associated with decrease hepatobiliary enhancement (-0.25 signal intensity ± 0.1, P < 0.05). The 20 Gy volume corresponded most closely to the LHI volume with a ratio of LHI to iso-dose of 1.1 ± 0.16 and r = 0.58 (P < 0.01). The mean DICE coefficient for deformation of 3D-dosimetric data to MRI was 0.82 ± 0.05. The D90 was 16.5 ± 7.9 Gy for the whole cohort and 19.4 ± 5.5 Gy for patients with DICE coefficient > 0.8. The mean liver volume was 1590 ± 483 cc and mean liver – hypointense volume was 1357 ± 480 cc. We found that SBRT to the liver resulted in decreased spatial function on MRI that corresponded to a threshold dose. However, direct bilirubin is inversely correlated to hepatobiliary enhancement and may decrease the sensitivity of this method in patients with elevated direct bilirubin. This spatial function relationship can potentially be used to image and monitor regions of radiation liver injury and can be used for better dose allocation in patients needing liver re-irradiation.