External Validation of Pulmonary Toxicity Risk Model for Ultracentral Lung Tumors

Abstract

Pulmonary toxicity attributed to the proximal bronchial tree (PBT) is dose-limiting in stereotactic body radiation therapy (SBRT) for central lung tumors. The toxicity is of particular concern in ultracentral lesions, which are defined as the gross tumor volume abutting the PBT, esophagus, or other mediastinal structures. In this work we tested published pulmonary toxicity models associated with dose to the airway structures after SBRT for ultracentral tumors in an independent institutional cohort of patients.The PubMed database was scrutinized for toxicity models evaluating pulmonary and airway toxicity for SBRT or hypofractionated radiation treatment for central tumors in the lung. The following search term criteria were used: (central) AND (lung tumor) AND (normal tissue complication probability) AND (radiation). Identified models were explored in all patients with ultracentral lung tumors treated between 2008 and 2017 at one large center (N = 88). This cohort included 60% primary lung and 40% metastatic patients that had been treated to 45 Gy in 5 fractions (fx), 50 Gy in 5 fx, 60 Gy in 8 fx, or 60 Gy in 15 fx. 16% of the patients received anti-VEGF therapy, which as previously shown, was associated with higher incidence of grade 5 hemorrhage.A total of six NTCP models were identified from work by Tekatli et al 2018. Of these, three models were for ≥grade 3 and three for grade 5 toxicity and included bronchial max dose, V60, V100, and V130. Clinical pulmonary toxicity consisted of radiation pneumonitis, atelectasis due to main stem bronchus occlusion, and/or hemoptysis. The pulmonary toxicity rate was 15% and 8% for ≥grade 3 toxicity and grade 5 toxicity, respectively, in the original study. Within the independent cohort, the corresponding values were 18% and 10%. Two maximum dose models were shown to be valid in our institutional dataset; other models failed this validation test. The model was made more robust by expanding the max dose region of the bronchial volume by focusing on the max dose to 0.1 cm3, 1 cm3 and 3 cm3. The max dose (0.1 cm3) model had improved performance with increased steepness. Generally, models from literature underestimated our institutional complication rates.Pulmonary toxicity models for central tumors were identified from one study and explored within an independent cohort. The models derived from primarily central lung tumors systematically underestimated the pulmonary toxicity rate in the solely ultracentral tumor cohort, possibly also due to anti-VEGF therapy used in the independent cohort. A modified model from the best-performing max dose model was designed, and this model could be utilized for rationally constructed airways constraints in ultracentral patients treated with SBRT or hypofractionation.I. Chen: None. M. Thor: None. A.J. Wu: Research Grant; CivaTech Oncology, Inc. Honoraria; 1199SEIU. Consultant; AstraZeneca, MORE Health. Advisory Board; Simphotek, Inc. Travel Expenses; AlphaTau Medical. Stock; Simphotek, Inc. A. Iyer: None. A. Apte: None. A. Rimner: Research Grant; Varian Medical Systems, Boehringer Ingelheim, Astra Zeneca, Pfizer, Merck. Consultant; MoreHealth. Advisory Board; Boehringer Ingelheim, Astra Zeneca, Merck; International Thymic Malignancies Interest Group, International Mesothelioma Interest Group. D.R. Gomez: Research Grant; Merck, AstraZeneca. Honoraria; BMS, Varian, Research to Practice. Speaker's Bureau; Merck. Advisory Board; AstraZeneca, Olympus. J.O. Deasy: Research Grant; Varian Corp., NIH, Breast Cancer Research Foundation. Stock; Paige.AI; AAPM.A. Jackson: None.