Data comprised raw data from two previously published 18F-FDGal PET studies: eight patients with cirrhosis [1] and seven subjects with no parenchymal liver disease (healthy subjects) [2]. In addition, unpublished data from nine healthy subjects and eight patients with cirrhosis were included. In total, 16 healthy subjects and 16 patients with cirrhosis were thus included in the study. The study was approved by The Central Denmark Region Committees on Health Research Ethics and conducted in accordance with the Helsinki Declaration.
PET studies
All subjects were studied in supine position using the same PET/CT camera, a 64-slice Siemens Biograph Truepoint PET/CT camera (Siemens AG, Erlangen, Germany). Before the PET recording, a topogram of the upper abdomen was performed for optimal positioning of the liver within the 21.6-cm transaxial field-of-view of the PET camera followed by a low-dose CT scan (50 effective mAs with CARE Dose4D (Siemens Healthcare, Erlangen, Germany), 120 kV, pitch 0.8, slice thickness 5 mm) for definition of anatomical structures and attenuation correction of PET data. A bolus of 100 MBq of 18F-FDGal in 10 ml saline was administered intravenously during the initial 15 to 20 s of a dynamic PET recording (list-mode). In PET recordings exceeding 20 min, only measurements up to 20 min were used in the present study since this is now the standard protocol [1-3]. 18F-FDGal was produced in our own radiochemistry laboratory (radiochemical purity ≥ 97%) [6].
All subjects had an Artflon catheter (Artflon, Ohmeda, Swindon, UK) placed percutaneously in a radial artery. During the PET recording, arterial blood samples (0.5 ml) were manually collected for determination of 18F-FDGal blood concentrations (kBq/ml) using a well counter (Packard Instruments, Meriden, CT, USA) and corrected for radioactivity decay back to start of the scan.
Reconstruction of PET data
Data were reconstructed using two different methods: without resolution modelling (336 matrix, voxel size 2 × 2 × 2 mm3, 6 iterations, 21 subsets, 2-mm Gaussian filter, separate prompts/randoms) and with resolution modelling (336 matrix, voxel size 2 × 2 × 2 mm3, 4 iterations, 21 subsets, 2-mm Gaussian filter, separate prompts/randoms). Measurements were corrected for radioactivity decay back to start of the scan. Time-frame structure was 18 × 5, 15 × 10, 4 × 30, 4 × 60, and 2 × 300 s, total 20 min.
Image analysis
Liver-VOI
Using fused PET/CT images, regions of interest (ROIs) were drawn in successive image planes in the parenchyma of the right liver lobe avoiding large blood vessels and summed into a volume of interest (Liver-VOI). The same VOI was used to generate a liver time-activity curve (Liver-TAC; liver activity concentration in kBq/ml liver tissue vs. time in minutes) for both reconstruction methods.
Aorta-VOI
Four different VOIs in the abdominal aorta were tested; for all four VOIs, ROIs were drawn in adjacent transaxial planes starting 3 planes below the diaphragm (6 mm) and continued for 40 planes in the caudal direction (total length, 8 cm). The ROIs were summed into an Aorta-VOI, and an Aorta-TAC (blood concentration in kBq/ml blood vs. time in minutes) was generated for both reconstruction methods. The four Aorta-VOIs were defined as (Figure 1):
-
Aorta-VOI-1: A semicircle drawn in the posterior half of the aorta with the straight line (length one third of the total diameter of the aorta) going through the centre of the vessel. Average volume was 2.95 ml (range, 2.03 to 4.21 ml).
-
Aorta-VOI-2: A circular ROI drawn in the centre of the aorta with a diameter of one third of the diameter of the aorta. Average volume was 4.75 ml (range, 3.04 to 6.23 ml).
-
Aorta-VOI-3: A semicircle drawn in the posterior half of the aorta with a diameter equal to that of the aorta and with the straight line going through the centre of the vessel. Average volume was 10.23 ml (range, 6.17 to 15.59 ml).
-
Aorta-VOI-4: A circular ROI including the entire aorta in each plane included in the VOI. Average volume was 32.1 ml (range, 21.62 to 45.78 ml).
Ventricle-VOI
For the VOI in the left ventricle of the heart (Ventricle-VOI), circular ROIs were drawn in four to six adjacent transaxial planes (Figure 1). Average volume was 1.88 ml (range, 0.89 to 5.65 ml).
Data analysis
Hepatic systemic clearance of 18F-FDGal (ml blood/ml liver tissue/min) was calculated according to the Gjedde-Patlak representation of data [7,8] as validated in previous studies [1-3]. In short, the hepatic systemic clearance is defined as the slope of the asymptote fitted to the linear part of the data representation 6 to 20 min after tracer administration [1-3,7,8]. The Liver-TAC was used as output function, and the hepatic systemic clearance of 18F-FDGal was calculated for each non-invasive input function (K*) and compared to the value calculated using the Artery-TAC (K) which was used as reference value.
Statistics
Correlation between K* and K was tested by the Pearson product–moment correlation coefficient, r, and a p value < 0.05 was interpreted to indicate a statistically significant correlation. Mean relative deviations of K* from K, i.e., (K* − K)/K, were tested by a one-sample t-test, and deviations with a p value < 0.05 were considered statistically significant.