Variation in gross tumor volume delineation using CT, MRI, and FDG-PET in planning radiotherapy of anal cancer.

Abstract

652 Background: Radiotherapy (RT) outcome depends on accurate delineation of gross tumor volume (GTV) regarding tumor size and position. CT and MRI are standard image modalities as they provide the necessary structural information. Positron emission tomography with flourodeoxyglucose (FDG-PET) visualizes glucose metabolism in active tumor tissue. The main objective was to explore the intra-observer and inter-modality variation of the GTV delineated on CT, MRI and PET images. Methods: Between November 2006 and July 2009, 22 patients with biopsy-proven anal carcinoma stage T1-T4/N0-3/M0-1 scheduled for curative RT were included in the study. All patients underwent FDG-PET/contrast enhanced CT and MRI. The delineations of GTV were done twice for each imaging modality with a minimum of 3 months in between. Delineations on the CT and MRI were done by routine methods. For the PET part three different cut-off values were used: SUV 2.5, 40% and 50% of maximum SUV. Non-overlap was calculated by subtracting the intersection from the union of two sets both regarding intra-observer variation and inter-modality variation. The GTV were compared for size, intra-observer variation and non-overlap using Wilcoxon rank sum test, where p < 0.05 was considered significant. Results: Volume size: the median GTV-CT was 25.6 cm3 which was significantly larger compared to GTV-MRI (7.2 cm3) and the GTV from the different PET cut offs (6.2 cm3, 10.6 cm3 and 7.0 cm3). Intra-observer variation: the median GTV delineated on CT, MRI and PET did not vary significant between delineation number 1 and 2. Non-overlap: the non-overlapping regions in GTV between delineation 1 and 2 using CT and MRI (20.5% and 22.5%) were significantly different from the PET cut offs (5.5% , 6.4% and 5.8%). Conclusions: Delineation of anal tumors is difficult; CT gives rise to the largest GTV with considerable intra-observer variation. MRI has the same degree of inaccuracy, but smaller GTV. PET-GTVs are the smallest with significant smaller intra-observer uncertainty.