High-Temporal Resolution Kinetic Modeling on Total-Body PET Differentiates the Hepatic Interstitial Space in Nonalcoholic Fatty Liver Disease and Healthy Subjects

Introduction: Interstitial space is the fluid space surrounding tissue cells. Transport and uptake properties of the radiotracer ¹⁸F fluorodeoxyglucose (FDG) in this space may be distinct in health and disease, hypothetically due to differently programmed cell death processes in disease. Dynamic PET using a standard temporal resolution (e.g., 10-20s/frame) commonly assumes that the radiotracer concentrations in the interstitial space and intracellular space reach equilibrium rapidly and hence does not permit separate characterization of the interstitial space. In this paper, we demonstrate the use of high-temporal resolution (2s/frame) dynamic PET imaging and kinetic modeling enabled on the EXPLORER total-body PET system to explicitly characterize the interstitial space of the liver in both healthy human subjects and nonalcoholic fatty liver disease (NAFLD) patients.

Methods: Fourteen healthy subjects and ten NAFLD patients were included in this study and scanned on the EXPLORER system for a one-hour dynamic acquisition. Prior IRB approval and informed consent were obtained. The data were reconstructed into a total of 66-time frames: 30x2s, 12x10s, 6x30s, 12x120s and 6x300s. Regions of interest (ROI) were placed in the descending aorta to extract an image-derived blood input function, and in the liver to extract hepatic tissue time activity curve (TAC). Frames from the 60 seconds, with a 2-second temporal resolution (TR), were also rebinned to generate three additional temporal sampling schemes with 4-second, 6-second and 10-second TR. Both conventional two-tissue (2T) model, which combines the interstitial space and intracellular space into a free-state space, and the proposed three-tissue (3T) model, which separately models the interstitial space (Fig. 1), are used to fit the liver TACs. To account for the dual blood supply in the liver, an optimization-derived dual-blood input function (DBIF) approach with time delay correction was used with the 2T and 3T models. The 2T-DBIF and 3T-DBIF models were compared for statistical TAC fit quality which was evaluated using the Akaike information criteria (AIC). Different kinetic parameters, including K1 (the rate of FDG transport from plasma to the interstitial space) and V_x (the distribution volume of FDG in the interstitial space) are compared between the healthy subjects and NAFLD patients using statistical T-tests for characterizing biopsy-determined liver steatosis (score 0-3), liver inflammation (score 0-5) and total NAFLD activity score (NAS: score 0-8). Nonalcoholic steatohepatitis (NASH) was defined as NAS>4.

Results: The 3T-DBIF model provided a much better fit quality than the 2T-DBIF model for fitting liver TACs with high temporal resolution (Fig. 2A and 2B). Modeling of the interstitial space is increasingly favored as the temporal resolution becomes better from 10s/frame to 2s/frame according to the AIC comparison (Fig. 2C and 2D). Both the transport rate FDG-K1&nbsp;(Fig. 3A) and the interstitial distribution volume Vx&nbsp;(Fig. 3B) estimated with 3T-DBIF were statistically lower in the NAFLD patients with high inflammation (score: 3-5) than in the healthy subjects and NAFLD patients with low inflammation (score 0-2), according to the T-test statistical criteria. FDG Vx&nbsp;was higher in the group of healthy subjects and patients with low steatosis (score 0-1) than in the NAFLD patients with high steatosis (score 2-3) (Fig. 3C). Other kinetic parameters did not correlate with liver inflammation or steatosis. FDG Vx&nbsp;alone can also differentiate NASH (NAS>4) from non-NASH (NAS<=4) (Fig. 3D).

Conclusions: This study explored explicit kinetic modeling of interstitial space in the liver by using high-temporal resolution dynamic FDG-PET imaging. The modeling is better suited as the temporal resolution improves. The results indicate that the FDG kinetic characterization of the interstitial space has the strong potential to derive multiparametric PET imaging biomarkers to assess both liver inflammation and liver steatosis in NAFLD.

• Download figure
• Open in new tab
• Download powerpoint