Comparison of Tumor Volumes Derived from Glucose Metabolic Rate Maps and SUV Maps in Dynamic 18F-FDG PET

Abstract

Tumor delineation using noninvasive medical imaging modalities is important to determine the target volume in radiation treatment planning and to evaluate treatment response. It is expected that combined use of CT and functional information from 18F-FDG PET will improve tumor delineation. However, until now, tumor delineation using PET has been based on static images of 18F-FDG standardized uptake values (SUVs). 18F-FDG uptake depends not only on tumor physiology but also on blood supply, distribution volume, and competitive uptake processes in other tissues. Moreover, 18F-FDG uptake in tumor tissue and in surrounding healthy tissue depends on the time after injection. Therefore, it is expected that the glucose metabolic rate (MRglu) derived from dynamic PET scans gives a better representation of the tumor activity than does SUV. The aim of this study was to determine tumor volumes in MRglu maps and to compare them with the values from SUV maps. Twenty-nine lesions in 16 dynamic 18F-FDG PET scans in 13 patients with non-small cell lung carcinoma were analyzed. MRglu values were calculated on a voxel-by-voxel basis using the standard 2-compartment 18F-FDG model with trapping in the linear approximation (Patlak analysis). The blood input function was obtained by arterial sampling. Tumor volumes were determined in SUV maps of the last time frame and in MRglu maps using 3-dimensional isocontours at 50% of the maximum SUV and the maximum MRglu, respectively. Tumor volumes based on SUV contouring ranged from 1.31 to 52.16 cm3, with a median of 8.57 cm3. Volumes based on MRglu ranged from 0.95 to 37.29 cm3, with a median of 3.14 cm3. For all lesions, the MRglu volumes were significantly smaller than the SUV volumes. The percentage differences (defined as 100% x (V MRglu - V SUV)/V SUV, where V is volume) ranged from -12.8% to -84.8%, with a median of -32.8%. Tumor volumes from MRglu maps were significantly smaller than SUV-based volumes. These findings can be of importance for PET-based radiotherapy planning and therapy response monitoring.