ABSTRACT
PURPOSE: With the results of the INT0116 study, adjuvant radiochemotherapy has become the standard treatment after complete resection of gastric adenocarcinoma. However, the implementation of radiotherapy (RT) remains a concern. In response, consensus guidelines on RT technique have been published. Our objective was to measure the inter- and intraclinician variability in RT field delineation using conventional two- (2D) and three-dimensional (3D) techniques. METHODS AND MATERIALS: Between 1999 and 2003, five radiation oncologists (ROs) treated 45 patients with completely resected, gastric adenocarcinoma using postoperative radiochemotherapy (INT0116). Two cases were included in this study (Patient 1 had cardia and Patient 2 had antral disease). Standardized vignettes (with surgical and pathologic findings) and preoperative and postoperative imaging for each case were developed. Each RO designed AP-PA fields for each patient (2D planning) on two separate occasions. This was repeated using a 3D planning technique. RESULTS: Patient 1 had a mean field area of 250.2 cm(2) (SD 12.0) and 227.9 cm(2) (SD 26.5) using 2D and 3D planning, respectively (p = 0.03). The mean clinical target volume (CTV) volume was 468.3 cm(3) (SD 65.9). Patient 1 had a significantly greater inter- than intra-RO variation for the field area designed with 3D planning; however, no difference occurred with 2D planning or CTV contouring. Patient 2 had a mean field area of 234.8 cm(2) (SD 33.1) and 226.8 cm(2) (SD 19.3) using 2D and 3D planning, respectively (p = 0.5). The mean CTV was 729.4 cm(3) (SD 67.3). For Patient 2, the inter-RO variability was significantly greater than the intra-RO variability for the field area using both 2D and 3D planning, and no difference was seen for the CTV. Composite beam's-eye-view plots revealed that the superior, inferior, and right lateral borders proved to be most contentious. CONCLUSION: Despite published guidelines and a departmental protocol, significant variations in the RT field areas were seen among ROs for both 2D and 3D planning. However, in general, CTV contouring was reproducible. Because 3D-RT hinges on accurate target identification, caution should be exercised before migrating to 3D planning for postoperative gastric cancer.
