Validation of RTOG 0813 Normal Tissue Constraints for Pulmonary Toxicity for Stereotactic Body Radiation Therapy in Central Non-Small Cell Lung Cancer

Abstract

Stereotactic body radiation therapy (SBRT) yields excellent local control rates for medically inoperable early stage “centrally” located non-small cell lung cancer. Normal tissue constraints provided in RTOG 0813, a phase I/II trial testing safety and efficacy of lung SBRT for central tumors, were largely based on expert estimates. Validation of these protocol constraints with clinical data is limited. We sought to identify the sensitivity and specificity of the current RTOG constraints for predicting pulmonary toxicity in a large institutional data set. We identified 136 lesions located within 2 cm of the proximal bronchial tree (PBT), treated with definitive lung SBRT from 2005 to 2014 from a prospective registry of 1,462 patients. Dose was 50 Gy/5 fx to all but 2 lesions, which received 60 Gy/5 fx. Pulmonary toxicity was categorized as pneumonitis or non-pneumonitis (fistula, bronchial stenosis or necrosis, atelectasis/collapse, hemoptysis, or clinically significant pleural effusion). A series of dose endpoints for the PBT was generated based on dose volume histograms, where dose levels ranged from 0 Gy to 80 Gy in increments of 10 cGy, and volumes ranged from 0.05 cc to 50 cc in increments of 0.05 cc. A total of 801,000 dosimetric endpoints was analyzed. The sensitivity and specificity of each of these endpoints in predicting pulmonary toxicity was calculated. The optimal dosimetric endpoint was chosen by identifying the highest F-score. There were 9 Grade 2 pneumonitis and 10 Grade ≥2 non-pneumonitis toxicities, of which 3 were Grade 5 (broncho-pleural fistula, left mainstem bronchus necrosis, and bronchial stenosis causing collapse). Of the 801,000 endpoints analyzed, we found that the optimal point dose to avoid Grade ≥ 2 pulmonary toxicity was D0.05 cc <49.5 Gy to the PBT (defined as dose to 0.05 cc less than 49.5 Gy), which had a sensitivity of 59% and specificity of 77%. The optimal point dose to avoid Grade 3-5 non-pneumonitis toxicity was D0.33 cc <46.5 Gy, which had a sensitivity of 100% and specificity of 85%. Applying the PBT RTOG constraints to our data set achieved a sensitivity of 18% and specificity of 91% for D4 cc <18 Gy and a sensitivity of 29% and specificity of 93% for D0.03 cc <52.5 Gy. Clinical results from this large institutional data set validate current RTOG constraints for PBT as predictive for pulmonary toxicity. The results also suggest that the RTOG constraints of D4 cc <18 Gy and D0.03 cc <52.5 Gy to the PBT have moderate sensitivity but excellent specificity for pulmonary toxicity. We identified D0.05 cc <49.5 Gy to the PBT as having the highest sensitivity and specificity for toxicity prediction, and this value parallels the current RTOG constraint of D0.03 cc <52.5 Gy. This analysis suggests that an additional constraint of D0.33 cc <46.5 Gy be considered for avoidance of Grade 3-5 non-pneumonitis pulmonary toxicity.