2212 - Neck Volume as a Novel Surrogate to Trigger Adaptation for Head and Neck Radiotherapy

Y. C. Hu¹, P. Zhang¹, E. Aliotta¹, P. Lichtenwalner¹, S. Abraham², P. Zhang¹, L. I. Cervino¹, N. Y. Lee³, and M. Aristophanous²; ¹Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, ²Memorial Sloan Kettering Cancer Center, New York, NY, ³Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center, New York, NY

Purpose/Objective(s): This study investigates the use of neck volumes auto-segmented on cone beam CT (CBCT) as a novel surrogate for monitoring tumor shrinkage to select candidates for adaptation in head and neck (HN) radiotherapy. Materials/

Methods: We hypothesize that the auto-segmented neck volume from CBCT is an effective surrogate for monitoring tumor shrinkage without needing to contour gross tumor, which is often indistinguishable on CBCT. Twenty-six HN patients who underwent adaptive radiotherapy (re-simulation ~3 weeks into treatment) in 2023 at our institution were selected for this study. Patient selection criteria included patients who had weekly or daily CBCTs, a re-simulation CT for re-planning, gross tumor volume (GTV) >5cc, and the neck contours encompassing at least 10% of the GTV. Our in-house AWARE (Automatic Watchdog in Adaptive Radiotherapy Environment) system automatically extracted weekly neck volumes on CBCTs by auto-segmenting the patient external and selecting a fixed longitudinal distance beginning from the bottom of the brainstem to 5cm inferior. We correlated (Pearson correlation coefficient, PCC) the ratio of the week-3 neck volume to the week-0 (first CBCT) neck volume with the ratio of the re-planned planning target volume (PTV) volume to the original PTV. We used receiver operating characteristic curve (ROC) analysis to measure the predictive performance of identifying large PTV changes.
Results: The mean shrinkage of neck volumes was 5.2% (±2.9%.) There was a significant correlation between the change in neck volumes and the change in PTV volumes (PCC=0.39, p-value=0.047.) When predicting PTV shrinkage of more than 15%, the area under the curve (AUC) was 0.74, the true positive rate was 94%, and the false positive rate was 33% if the neck volume shrank more than 5%.
Conclusion: The results indicated that easy-to-segment neck volumes can be an effective surrogate for target volume changes, which may help select candidates for adaptation in HN radiotherapy. Future work will include testing with additional cases and other predictive endpoints.