This retrospective study consecutively enrolled 10 patients (5 females and 5 males, age: 53.7±13.9 year, weight: 63.3±8.6 kg) who underwent a total-body 18F-FDG dPET imaging at Shanghai Jiao Tong University Renji Hospital between 28th Dec 2020 and 27th Nov 2021. In this study, the clinical purpose of dPET imaging included cancer staging/restaging (hepatic carcinoma, pancreatic cancer, breast cancer, lung cancer, gallbladder cancer, and cervical cancer) and tumour screening. Blood glucose level of all patients was not higher than 7 mmol/L prior to the scan. This study was approved by the Institutional Review Board of Shanghai Jiao Tong University Renji Hospital, and written informed consents were obtained from all patients in this study.
Dynamic acquisition protocol
All patients fasted for at least 6 h before 18F-FDG administration. Total-body dynamic 18F-FDG PET/CT scan was performed with a weight-based injection of 4.4 MBq/kg (247-319 MBq). The patients were instructed to be immobilized and fastened with a belt during the acquisition and positioned in supine position with feet pointing towards the scanner (feet first supine) with arms down. A CT scan was performed before PET acquisition for attenuation correction and anatomical localization using a fixed tube voltage of 120 kV with an auto-mAs technique for dose modulation. All patients underwent a 60-min dPET scan, and one of them had an additional delayed PET scan at 180-min post-injection for qualified display of the pelvic lesion. The workflow is schematically depicted in Fig. 1.
PET raw data were reconstructed into 92 frames (24×5s, 20×30s and 48×60s). The standard order subset expectation maximization (OSEM) algorithm was used as well as time-of-flight (TOF) and point-spread function (PSF) modelling with the following parameters: 3 iterations, 20 subsets, 256×256 matrix, 600-mm FOV, 2.89-mm slice thickness, and a Gaussian post filter with a full width at half maximum (FWHM) of 3 mm. In addition, all PET reconstructions included standard corrections like decay, scatter, random, dead time, attenuation, and normalization.
The Patlak graphic analysis (linear fitting) and the irreversible two tissues compartment model (i2TCM) analysis (non-linear fitting) were both used in this study. For Patlak graphic analysis, the selected frames include the first 10 to 60 min, 10 to 50 min, 10 to 40 min, 10 to 30 min and 10 to 20 min (referred to as G60, G50, G40, G30 and G20, respectively). The G60 was considered as a reference in this study. For the i2TCM analysis, the TACs from dPET were truncated into different groups: G60 (0 to 60 min), G50 (0 to 50 min), G40 (0 to 40 min), G30 (0 to 30 min) and G20 (0 to 20 min).
Moreover, a hybrid approach was proposed which combines the initial 20-min dPET data and the static PET scan at 55-60 min post-injection, allowing for providing both the kinetic and the static parameters, referred to as GHybrid20. The i2TCM analysis was used to obtain the kinetic parameters in this proposed approach.
To determine the shortest dPET acquisition time, we compared the quantitative parameters derived from the two different analysis methods – Patlak graphic analysis and i2TCM. For both methods, the input function was derived from the descending aorta. For the analysis of healthy organs/tissues, the regions-of-interest (ROIs) were manually delineated on the homogenous area as large as possible at gray matter, lung, muscle, spleen, pancreas and liver. ROIs were drawn on lesions at the transverse slice with the largest cross-section area.
Patlak graphic analysis
The glucose metabolism rate (Ki) of ROIs can be quantitatively measured by linear fitting method. The pixel-wise parametric Ki images were calculated with a commercialized software (uKinetics, United Imaging Healthcare, China) on the dedicated workstation (uWS-MI, United Imaging Healthcare, China).
Irreversible two tissues compartment model (i2TCM) analysis
Utilizing a research tool (PMOD Technologies Ltd., Zuerich, Switzerland), the kinetic parameters (K1, k2 and k3) were calculated based on the i2TCM model of FDG. K1, k2, k3 represent the influx of the tissues, the venous clearance out of the tissues, and the phosphorylation rate inside the tissues, respectively . Notably, while Ki is referred to as the net flux of FDG transported from plasma to the tissue and metabolized, k3 represents the phosphorylation rate which is much higher in cancer cells for its aggressive proliferation [24-26].
The statistical analyses were performed using R project (version 4.0.4). Wilcoxon signed rank test was applied to evaluate the consistency in the organs/tissues in the following groups: G20 and G60, G30 and G60, G40 and G60, as well as GHybrid20 and G60. To test the correlations of kinetic parameters, linear regression (last-square approach) was performed on the K1, k2, and k3 between GHybrid20 and G60, where the confidence interval (CI) was set as 95%. Pearson’s coefficient (r2) was calculated. Statistical significance was considered for a p-value less than 0.05.