FDG PET/CT scans from eleven patients with grade III or IV soft tissue sarcomas where obtained. Nine of the patients where liposarcomas, one was a schwannoma and one was a myxoid liposarcoma. Four of the patients were subjected to radiation therapy and their progress was recorded through three additional scans during and after the therapy. Time activity curves (TACs) revealed that the investigated sarcomas were a heterogeneous group with a wide spread of PET parameters. Mann Whitney statistics were used to estimate the predictive quality of the different parameters. It was found that none of the PET parameters have any predictive qualities when it comes to tumor type or location.
Compared to the direct PET image 45 minutes after injection, the pharmacokinetic parameter MRFDG (metabolic rate of FDG) was found to have superior tumor-to-tissue contrast in one patient. The tumor-to-tissue ratio was found to amplify the vascular phase of the TACs, although no additional diagnostic information could be detected. A direct early PET measurement was, together with the pharmacokinetic parameters K1 and the vascular fraction, found to be good descriptors of the vasculature. Patlak analysis was attempted but represented less contrast and more noise compared to the pharmacokinetic parameters. For the patients undergoing therapy some changes in the vasculature could be detected but all patients were concluded to be non-responders.
The current material indicates that dynamic PET may in some instances provide additional information compared to a static PET. However no capability to differentiate liposarcomas from other sarcomas could be found and it does not appear that PET is useful for monitoring early treatment response patients with soft tissue sarcomas. As no link to clinical data could be established the conclusion is that the extra time in the scanner could not be justified.
An extended study including subgroups of liposarcomas with different prognosis is recommended. In such a study the number of patients should be increased if results of statistical significance are desired. Improvements to the PET protocol are most likely an increased sampling rate for precise bolus detection while improvements to the software are likely to be found in improved plasma fitting and timing.