Impact of delineation uncertainties on dose to organs at risk in CT-guided intracavitary brachytherapy

Purpose: This study quantifies the inter- and intraobserver variations in contouring the organs at risk (OARs) in CT-guided brachytherapy (BT) for the treatment of cervical carcinoma. The dosimetric consequences are reported in accordance with the current Gynecological Groupe Européen de Curiethérapie/European Society for Therapeutic Radiology and Oncology guidelines.

Methods and materials: A CT planning study of 8 consecutive patients undergoing image-guided BT was conducted. The bladder, rectum, and sigmoid were contoured by five blinded observers on two identical anonymized scans of each patient. This provided 80 data sets for analysis. Dosimetric parameters analyzed were D0.1 cc, D1 cc, and D2 cc. The mean volume of each OAR was calculated. These endpoints were compared between and within the observers. The CT image sets from all patients were evaluated qualitatively.

Results: The interobserver coefficient of variation for reported D2 cc was 13.2% for the bladder, 9% for the rectum, and 19.9% for the sigmoid colon. Unlike the variation seen in bladder and rectal contours, which differed largely in localization of the organ walls on individual slices, sigmoid colon contours demonstrated large differences in anatomic interpretation.

Conclusions: Variation in recorded D2 cc to the bladder and rectum is comparable with the previous published results. Inter- and intraphysician variations in reported D2 cc is high for the sigmoid colon, reflecting varying interpretation of sigmoid colon anatomy. Variation in delineation of the OARs may influence treatment optimization and is a potential source of uncertainty in the image-guided BT planning and delivery process.