Variation of clinical target volume definition in three-dimensional conformal radiation therapy for prostate cancer.
PURPOSE: Currently, three-dimensional conformal radiation therapy (3D-CRT) planning relies on the interpretation of computed tomography (CT) axial images for defining the clinical target volume (CTV). This study investigates the variation among multiple observers to define the CTV used in 3D-CRT for prostate cancer. METHODS AND MATERIALS: Seven observers independently delineated the CTVs (prostate +/- seminal vesicles [SV]) from the CT simulation data of 10 prostate cancer patients undergoing 3D-CRT. Six patients underwent CT simulation without the use of contrast material and serve as a control group. The other 4 had urethral and bladder opacification with contrast medium. To determine interobserver variation, we evaluated the derived volume, the maximum dimensions, and the isocenter for each examination of CTV. We assessed the reliability in the CTVs among the observers by correlating the variation for each class of measurements. This was estimated by intraclass correlation coefficient (ICC), with 1.00 defining absolute correlation. RESULTS: For the prostate volumes, the ICC was 0.80 (95% confidence interval [CI]: 0.56-0.96). This changed to 0.92 (95% CI: 0.75-0.99) with the use of contrast material. Similarly, the maximal prostatic dimensions were reliable and improved. There was poor agreement in defining the SV. For this structure, the ICC never exceeded 0.28. The reliability of the isocenter was excellent, with the ICC exceeding 0.83 and 0.90 for the prostate +/- SV, respectively. CONCLUSIONS: In 3D-CRT for prostate cancer, there was excellent agreement among multiple observers to define the prostate target volume but poor agreement to define the SV. The use of urethral and bladder contrast improved the reliability of localizing the prostate. For all CTVs, the isocenter was very reliable and should be used to compare the variation in 3D dosimetry among multiple observers.