Dummy run for a phase II multi-institute trial of chemoradiotherapy for unresectable pancreatic cancer: inter-observer variance in contour delineation.

UNLABELLED: The aim of this study was to examine the interobserver variance in delineating the contour of unresectable pancreatic cancer for chemoradiotherapy. PATIENTS AND METHODS: CT images of two cases of unresectable pancreatic tumors (head and body cancer) were sent to eight radiation therapy facilities in a CD-ROM. Gross tumor volume (GTV) and planning target volume (PTV) were delineated using the radiotherapy treatment planning system (RTP) of the respective facilities. The mean and variance of the GTV and PTV of 11 plans by the eight facilities were analyzed.
RESULTS: The respective mean volumes of the GTV of pancreatic head and body cancer cases were 34.8 cm3 (SD, 30.4; median, 31.8; range, 13.5-122 cm3) and 73.4 cm3 (SD, 28.1; median, 67.9; range, 46.3-152 cm3). The ratios of the largest to the smallest contoured GTV were 9 and 3, respectively. The corresponding average volumes of PTV were 148 cm3 (SD, 84.3; median, 129; range, 69.6-363 cm3) and 240 cm3 (SD, 79.8; median, 227; range, 148-420 cm3). The ratios of the largest to the smallest contoured volume were 5 and 2.8 for PTV delineation, respectively.
CONCLUSION: Dummy run using CD-ROM is possible on a multi-institute scale but also disclosed interobserver variance. Unified protocol interpretation to reduce inter-observer variance is therefore essential for successful multi-institute clinical trials.