Dose-Volume Predictors of Fatal Bronchopulmonary Bleeding After Stereotactic Body Radiation Therapy of Centrally Located Lung Tumors: An Analysis of the Expanded HILUS Cohort

Purpose The expanded HILUS study showed that bronchial dose and bronchial tumor compression are risk factors for bronchopulmonary bleedings after stereotactic body radiation therapy of centrally located lung tumors. In the current analysis of the same cohort, the aim was to identify the dose-volume histogram (DVH) parameter that best predicts fatal bronchopulmonary bleeding and to develop a predictive model for this endpoint. Methods and Materials The HILUS cohort included 230 patients with 238 central targets treated with stereotactic body radiation therapy of 7 Gy × 8 to the periphery of the planning target volume, where 21 patients developed grade-5 bronchopulmonary bleeding. Cox regression-based normal-tissue complication probability models were developed, accounting for the dose to the main and intermediate bronchi and bronchial tumor compression. Three alternative DVH parameters were explored: the dose to a certain volume (D v ), the volume receiving a certain dose (V d ), and the equivalent uniform dose. Internal validation was performed with the bootstrap method. Results The best fits of the bivariable normal-tissue complication probability models included bronchial tumor compression in combination with the DVH parameters of D 0.31 cm3 , V 82Gy,EQD2 (equivalent dose in 2-Gy fractions) , and equivalent uniform dose with n = 0.024, respectively. This indicates that a high dose to a small volume may lead to grade-5 bronchopulmonary bleeding. The probability of fatal bleeding at 2 years without bronchial tumor compression was 10% for a D 0.31 cm3 of 107 Gy 3 EQD 2 , and 20% at 165 Gy 3 EQD 2 , whereas the probability with bronchial tumor compression was 10% at 0 Gy 3 and 20% at 58 Gy 3 EQD 2 . The model showed good discrimination and calibration. Conclusions Bronchial tumor compression is a strong predictor for grade-5 bleeding, and the most relevant dose parameter appears to be the dose to a small volume of the main and intermediate bronchi.