2911 - Voxel-Based Analysis Unveils the Optimal Esophageal Contouring and Dose Constraint to Predict Radiation Esophagitis in Postmastectomy Regional Nodal Radiotherapy
Presenter(s)
D. Q. Wang1, S. Yuan1, H. Jing1, Y. Zhai2, H. Fang1, G. Sun2, X. Zhao1, S. Qi2, Y. W. Song1, X. Liu1, Y. Liu1, W. Zhang3, Y. Tang4, N. Lu1, B. Chen2, T. Yu1, Y. X. Li5, J. Dai1, and S. Wang1; 1Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, 2State Key Laboratory of Molecular Oncology and Department of Radiation Oncology, National Cancer Center/ National Clinical Research Center for Cancer/ Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, 3State Key Laboratory of Molecular Oncology and Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, 4State Key Laboratory of Molecular Oncology and Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences (CAMS) and Peking Union Medical College (PUMC), Beijing, China, 5Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 100021, Beijing, China., Beijing, China, Beijing, China
Purpose/Objective(s): Radiation esophagitis (RE) is an under-recognized toxicity in breast cancer patients receiving hypofractionated regional nodal irradiation (RNI). The optimal esophageal contouring range and dose constraint remain undefined. We prospectively quantified RE and explored spatial dose–toxicity relationships using voxel-based analysis (VBA).
Materials/Methods: High-risk breast cancer patients treated after breast-conserving surgery or mastectomy with hypofractionated VMAT RNI (43.5 Gy/15 fractions; tumor bed boost 49.5 Gy/15 fractions) were prospectively enrolled. RE was assessed weekly during RT, 1 week and 2 weeks after RT, as well as at 3 months and 6 months, graded using CTCAE v3.0. Grade =2 RE was the endpoint. Planning CT and 3D dose were exported in DICOM format. Dosimetric parameters (Dmean, Dmax, V5–V45 by 5 Gy increments) were extracted. In a systematically sampled subgroup (n=206), three esophageal contouring extents were evaluated: ESO1 (cricoid to aortic-arch disappearance), ESO2 (cricoid to below carina), and ESO3 (full-length). VBA was performed using elastic registration to a common anatomy and a generalized linear model adjusted for risk factors to identify dose-sensitive subregions. Multivariable logistic regression and restricted cubic splines were applied. Slice-wise esophageal circumference receiving dose was computed using Python workflow.
Results: Among 516 patients (255 left-sided), grade =2 RE incidence was 29.7% including 152 patients with grade 2 and 1 with grade 3. RE peaked in the final treatment week, and fully resolved within 1 month without treatment interruption. Laterality and internal mammary nodal irradiation (IMNI) were independent clinical risk factors. VBA consistently localized the RE-sensitive region to the cervical esophagus, with the inferior border not extending below the sternal notch, supporting ESO1 as sufficient for contouring. With ESO1 contoured in all patients, the dominant sensitive segment was C7–T3 with 18 slices in 5.4 cm, predominantly right-sided. Dosimetric analysis showed significantly higher V5–V45 in patients with grade =2 RE and in left-sided tumors. In multivariable models, Dmean, Dmax and V10–V45 of ESO1 predicted RE, and Dmean, Dmax and V5–V45 of VBA-defined sensitive region predicted RE, with V25 and V35 achieving the best discrimination. Spline analyses identified optimal thresholds of 60% (circumference receiving 25 Gy) and 30% (circumference receiving 35 Gy), and optimal longitudinal cutoff of 10 slices (3.0 cm). RE incidence increased from 24.6% to 63.2% for >10 slices exceeding 25 Gy/60% circumference, and from 24.6% to 60.3% for >10 slices exceeding 35 Gy/30%.
Conclusion: As RE risk is strongly region-dependent and concentrated in the cervical esophagus, contouring the ESO1 is optimal for esophagus dose constrains. Spatial circumferential dose and its longitudinal extent provide complementary predictors beyond DVH and enable refined RE risk stratification.