Lessons learned from the multicentre QA program in SBRT for perihilar cholangiocarcinoma: benchmark-case results from the STRONG II trial.

Abstract

SBRT for unresectable perihilar cholangiocarcinoma (pCCA) is technically challenging, yet delineation and planning guidelines remain limited. Within the STRONG II trial, we implemented a QA program to evaluate protocol compliance and harmonize treatment practices. Six centres participated in this study. One radiation oncologist per centre delineated the target and organs at risk (OAR) on a single CT scan, supported by MR imaging and endoscopic reports. Agreement with reference contours was evaluated using the Dice Similarity Coefficient (DSC), Mean Distance to Agreement (MDA), Hausdorff Distance 95% (HD95), and contour volume. Each centre then generated a treatment plan using reference contours. Dose distributions underwent quantitative assessment for major and minor protocol violations. Outcomes were discussed in multicentre meetings. Most OAR delineations showed good consensus (median DSC > 0.8). Initially, the target, central biliary tract, and cystic duct demonstrated poor agreement (median DSC: 0.57, 0.63, 0.00). Following two additional delineation rounds using updated radiologist-informed guidelines, redelineation of these three structures improved DSC values, though not all significant (median DSC, [range]: 0.74, [0.53-0.88], p = 0.08; 0.75, [0.51-0.88], p = 0.35; 0.42, [0.28-0.67], p = 0.04). Corresponding MDA and HD95 values were reduced in most cases. Treatment plans showed no major protocol violations. All plans except one had a minor PTV-underdosage violation required to remain within OAR constraints. Implementing consensus sessions and updated guidelines enhanced delineation consistency; nevertheless, delineation and dose distribution variability remained. This QA program and its resulting guidelines are an important step toward SBRT standardization for pCCA, which is essential for reliable evaluation of its therapeutic value.