Dosimetric consequences of interobserver variability in delineating the organs at risk in gynecologic interstitial brachytherapy.

Abstract

To investigate the dosimetric variability associated with interobserver organ-at-risk delineation differences on computed tomography in patients undergoing gynecologic interstitial brachytherapy. The rectum, bladder, and sigmoid of 14 patients treated with gynecologic interstitial brachytherapy were retrospectively contoured by 13 physicians. Geometric variability was calculated using κ statistics, conformity index (CIgen), and coefficient of variation (CV) of volumes contoured across physicians. Dosimetric variability of the single-fraction D0.1cc and D2cc was assessed through CV across physicians, and the standard deviation of the total EQD2 (equivalent dose in 2 Gy per fraction) brachytherapy dose (SD(TOT)) was calculated. The population mean ± 1 standard deviation of κ, CIgen, and volume CV were, respectively: 0.77 ± 0.06, 0.70 ± 0.08, and 20% ± 6% for bladder; 0.74 ± 06, 0.67 ± 0.08, and 20% ± 5% for rectum; and 0.33 ± 0.20, 0.26 ± 0.17, and 82% ± 42% for sigmoid. Dosimetric variability was as follows: for bladder, CV = 31% ± 19% (SD(TOT) = 72 ± 64 Gy) for D0.1cc and CV = 16% ± 10% (SD(TOT) = 9 ± 6 Gy) for D2cc; for rectum, CV = 11% ± 5% (SD(TOT) = 16 ± 17 Gy) for D0.1cc and CV = 7% ± 2% (SD(TOT) = 4 ± 3 Gy) for D2cc; for sigmoid, CV = 39% ± 28% (SD(TOT) = 12 ± 18 Gy) for D0.1cc and CV = 34% ± 19% (SD(TOT) = 4 ± 4 Gy) for D2cc. Delineation of bladder and rectum by 13 physicians demonstrated substantial geometric agreement and resulted in good dosimetric agreement for all dose-volume histogram parameters except bladder D0.1cc. Small delineation differences in high-dose regions by the posterior bladder wall may explain these results. The delineation of sigmoid showed fair geometric agreement. The higher dosimetric variability for sigmoid compared with rectum and bladder did not correlate with higher variability in the total brachytherapy dose but rather may be due to the sigmoid being positioned in low-dose regions in the cases analyzed in this study.