Presenter(s)
C. D. S. Smith1, V. L. Doss1, E. Hales1, D. Liu1, B. M. Anderson2, H. K. Kooner1, K. Yu3, C. Hu1,4, H. Li1, C. T. Lin5, A. N. Viswanathan1, X. Jia1, R. K. Hales1, K. R. Voong1, T. R. McNutt1, and R. Ger1; 1Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, 2University of California San Diego, Department of Radiation Medicine and Applied Sciences, La Jolla, CA, 3Johns Hopkins University Department of Biomedical Engineering, Baltimore, MD, 4Division of Biostatistics and Bioinformatics, Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD, 5Department of Radiology and Radiological Science, Johns Hopkins School of Medicine, Baltimore, MD
Purpose/Objective(s):
Thoracic reirradiation (reRT) seeks local control but risks severe toxicity, with G5 rates of 2–20%. Despite proposed cumulative constraints, substantial uncertainty remains. We evaluated cumulative EQD2 esophageal doses and their association with toxicity in high-risk reRT.Materials/Methods:
We retrospectively reviewed 139 patients (2012 - 2025) with two thoracic RT courses with any 95% prescription isodose overlap. Esophageal toxicity (CTCAE v5.0 and attribution) after reRT start was evaluated. Grade =4 events defined the primary endpoint; grade 5 deaths were reported descriptively. Prior plans were deformably registered to the reRT CT; voxel-wise EQD2 (a/ß=3) was computed to derive cumulative doses. Continuous variables were compared by Mann–Whitney U test. Firth penalized logistic regression (scaled per 10 Gy) assessed associations between dose and toxicity. Given low event counts, adjusted analyses were limited to concurrent chemotherapy. Cumulative EQD2 esophageal Dmax was compared to the American Radium Society’s 100 Gy recommendation. Normal Tissue Complication Probability (NTCP) was estimated from the fitted model with bootstrap resampling.Results:
Median age was 65 years (range 19-91). Median RT interval was 16.6 months (IQR 9.4–30.4), and follow-up 11.0 months. Overall, 101 patients (72.7%) had no esophageal toxicity. Patient-reported dysphagia was none in 59.3%. Grades 1-5 events occurred in 5 (3.6%), 19 (13.7%), 9 (6.5%), 1 (0.7%), and 4 (2.9%) patients. Median cumulative esophageal Dmean was 18.3 Gy (range 0.3-73.2) and Dmax 66.7 Gy (range 0.7-163). Patients with =G4 events had higher Dmean (34.3 vs 17.2 Gy; p=0.01) and Dmax (81.5 vs 65.9 Gy; p=0.014). The lowest cumulative doses with =G4 toxicity were Dmax 79.8 Gy and Dmean 25.3 Gy. Increasing cumulative dose correlated with =G4 toxicity (Dmean OR 1.73, 95% CI 1.08–2.85, p=0.023; Dmax OR 1.39, 95% CI 1.02–1.92, p=0.035). Results were consistent after leave-one-out sensitivity analyses and adjustment. Of 10 (7%) cases with Dmax > 100 Gy, there was one G4 and one G5 event. Three G5 events occurred at Dmax < 100 Gy. Model-estimated NTCP in this cohort (=G4) at 100 Gy was 9.1% (95% CI 3.4–20.8%). All four G5 events involved fistula or perforation with multifactorial contributions: all had recurrence involving the esophageal wall at event, and two had mechanical trauma from esophageal stent placement <2 months prior to death.Conclusion:
Higher cumulative esophageal dose was associated with increased toxicity risk, yet defining a single safe constraint remains challenging. The risk–benefit of reRT is complex as untreated disease can be catastrophic. Severe events occurred across wide dose ranges with other competing contributors. Estimates are imprecise given low event counts, and Dmax may be more sensitive to uncertainty than Dmean. These findings support incorporating dose, interval, and recovery into individualized reRT decisions in prospective studies.