Adjustment of vascular 2-deoxy-2-[18F]fluoro-D-glucose uptake values over time through a modeling approach.

Abstract

To develop and test a model predicting 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) standardized uptake value (SUV) changes over time in the aorta and the superior vena cava (SVC). Maximum aortic SUV and mean SVC SUV were determined at two time points (T1 and T2) in the ascending (ASC), descending(DSC), abdominal (ABD) aorta, aortic arch (ARC) and SVC of patients who have undergone [18F]FDG PET/CT for clinical purposes. For SUV prediction at T2, linear and non-linear models of SUV difference for a given time change were developed in a derivation group. The results were tested in an independent validation group, whilst model reproducibility was tested in patients of the validation group who have undergone a second clinically indicated scan. Applying the linear model in the derivation group, there were no statistically significant differences in measurements obtained in the examined segments: mean differences ranged from 0 ± 0.10 in SVC to 0.01 ± 0.13 in ARC between measured and predicted SUV. In contrast, in the non-linear model, there were statistically significant differences in measurements, except in ARC, with mean differences ranging from 0.04 ± 0.14 in ARC to 0.28 ± 0.13 in ABD. In the validation group using the linear model, there were no statistically significant differences, with mean differences ranging from - 0.01 ± 0.08 in ASC to - 0.03 ± 0.11 in ABD. Regarding reproducibility, mean differences were no statistically significant, ranging from 0.004 ± 0.06 in ASC to - 0.02 ± 0.16 in ABD. We have developed a linear model allowing accurate and reproducible prediction of SUV changes over time in the aorta and SVC.