The Impact of Fluorodeoxyglucose Positron Emission Tomography (FDG-PET) Variability On the Assessment of Treatment Response in Head-And-Neck Cancer Patients Undergoing Chemoradiotherapy
S Das1*, J Hoang2, K Roy Choudhury2, D Yoo1, D Brizel1, (1) Dept. of Radiation Oncology, Duke Univ Medical Center, Durham, NC (2) Dept. of Radiology, Duke Univ Medical Center, Durham, NCWE-A-217A-2 Wednesday 8:00:00 AM - 9:55:00 AM Room: 217A
Purpose: To compare intrinsic, temporal variability of tumor glucose metabolism versus early treatment-induced change. This assessment is necessary to use changes in imaging parameters during therapy to properly segregate metabolic responders from non-responders
Methods: As part of an IRB-approved study, 12 patients underwent 2 baseline PET scans (mean separation: 11 days) prior to chemoradiotherapy (70 Gy, 2 Gy/fraction). The intra-RT PET scan was acquired at 10-20 Gy (mean 14 Gy) into therapy. For each patient, the two baseline scans were analyzed using Bland-Altman analysis to determine the PET repeatability coefficient as a function of average baseline voxel SUV (repeatability coefficient for any voxel is the treatment-induced SUV change from the average baseline value required for the change to be deemed significantly higher than baseline variation). Using the repeatability coefficient, the intra-treatment PET nodal/GTV volume was divided into 3 groups of voxels: significantly reduced SUV, no significant difference, significantly increased SUV. A volume difference metric, defined as the difference in the percentages of node/GTV voxels with significantly decreased and increased SUV, was matched against each patient's post-RT assessment of residual/no-residual disease.
Results: The PET repeatability coefficient decreased from 49%-17% as the baseline voxel SUV increased from 2-12. The repeatability coefficient identified large volumes of the nodes (mean:52;std:35%) and GTV (mean:49;std:26%) as not having significant change. One patient had clinically-assessed residual disease. The node/GTV volume metric for this patient was positive (22%/46%), compared to an average node/GTV volume metric of -36%/-40% for patients without residual disease. Among the patients with no residual disease, only one had a positive nodal volume metric and one other had a positive GTV volume metric.
Conclusions: The repeatability coefficient can be used in future head-and-neck studies with only one baseline PET. A positive volume metric on an early intra-treatment scan may identify patients with residual disease post-RT.