Inter- and intra-observer variation in soft-tissue sarcoma target definition.

PURPOSE

To evaluate inter- and intra-observer variability in gross tumor volume definition for adult limb/trunk soft tissue sarcomas.

PATIENTS AND METHODS

Imaging studies of 15 patients previously treated with preoperative radiation were used in this study. Five physicians (radiation oncologists, orthopedic surgeons and a musculoskeletal radiologist) were asked to contour each of the 15 tumors on T1-weighted, gadolinium-enhanced magnetic resonance images. These contours were drawn twice by each physician. The volume and center of mass coordinates for each gross tumor volume were extracted and a Boolean analysis was performed to measure the degree of volume overlap.

RESULTS

The median standard deviation in gross tumor volumes across observers was 6.1% of the average volume (range: 1.8%-24.9%). There was remarkably little variation in the 3D position of the gross tumor volume center of mass. For the 15 patients, the standard deviation of the 3D distance between centers of mass ranged from 0.06 mm to 1.7 mm (median 0.1mm). Boolean analysis demonstrated that 53% to 90% of the gross tumor volume was common to all observers (median overlap: 79%). The standard deviation in gross tumor volumes on repeat contouring was 4.8% (range: 0.1-14.4%) with a standard deviation change in the position of the center of mass of 0.4mm (range: 0mm-2.6mm) and a median overlap of 93% (range: 73%-98%).

CONCLUSION

Although significant inter-observer differences were seen in gross tumor volume definition of adult soft-tissue sarcoma, the center of mass of these volumes was remarkably consistent. Variations in volume definition did not correlate with tumor size. Radiation oncologists should not hesitate to review their contours with a colleague (surgeon, radiologist or fellow radiation oncologist) to ensure that they are not outliers in sarcoma gross tumor volume definition. Protocols should take into account variations in volume definition when considering tighter clinical target volumes.