SU-E-J-124: FDG PET Metrics Analysis in the Context of An Adaptive PET Protocol for Node Positive Gynecologic Cancer Patients.

PURPOSE: To compare PET extracted metrics and investigate the role of a gradient-based PET segmentation tool, PET Edge (MIM Software Inc., Cleveland, OH), in the context of an adaptive PET protocol for node positive gynecologic cancer patients. METHODS: An IRB approved protocol enrolled women with gynecological, PET visible malignancies. A PET-CT was obtained for treatment planning prescribed to 45-50.4Gy with a 55- 70Gy boost to the PET positive nodes. An intra-treatment PET-CT was obtained between 30-36Gy, and all volumes re-contoured. Standard uptake values (SUVmax, SUVmean, SUVmedian) and GTV volumes were extracted from the clinician contoured GTVs on the pre- and intra-treament PET-CT for primaries and nodes and compared with a two tailed Wilcoxon signed-rank test. The differences between primary and node GTV volumes contoured in the treatment planning system and those volumes generated using PET Edge were also investigated. Bland-Altman plots were used to describe significant differences between the two contouring methods. RESULTS: Thirteen women were enrolled in this study. The median baseline/intra-treatment primary (SUVmax, mean, median) were (30.5, 9.09, 7.83)/(16.6, 4.35, 3.74), and nodes were (20.1, 4.64, 3.93)/(6.78, 3.13, 3.26). The p values were all < 0.001. The clinical contours were all larger than the PET Edge generated ones, with mean difference of +20.6 ml for primary, and +23.5 ml for nodes. The Bland-Altman revealed changes between clinician/PET Edge contours to be mostly within the margins of the coefficient of variability. However, there was a proportional trend, i.e. the larger the GTV, the larger the clinical contours as compared to PET Edge contours. CONCLUSION: Primary and node SUV values taken from the intratreament PET-CT can be used to assess the disease response and to design an adaptive plan. The PET Edge tool can streamline the contouring process and lead to smaller, less user-dependent contours.

Duke Scholars

Author Oana Craciunescu Radiation Oncology