To evaluate clinical pulmonary and radiographic bronchial toxicity after stereotactic ablative radiation therapy and hypofractionated radiation therapy for central lung tumors, and perform normal tissue complication probability modeling and multivariable analyses to identify predictors for toxicity.

A pooled analysis was performed of patients with a central lung tumor treated using ≤12 fractions at 2 centers between 2006 and 2015. Airways were manually contoured on planning computed tomography scans, and doses were recalculated to an equivalent dose of 2 Gy per fraction with an α/β ratio of 3. Grade ≥3 (≥G3) clinical pulmonary toxicity was evaluated by 2 or more physicians. Radiographic toxicity was defined as a stenosis or an occlusion with or without atelectasis using follow-up computed tomography scans. Logistic regression analyses were used for statistical analyses.

A total of 585 bronchial structures were studied in 195 patients who were mainly treated using 5 or 8 fractions (60%). Median patient survival was 27.9 months (95% confidence interval 22.3-33.6 months). Clinical ≥G3 toxicity was observed in 24 patients (12%) and radiographic bronchial toxicity in 55 patients (28%), both mainly manifesting ≤12 months after treatment. All analyzed dosimetric parameters correlated with clinical and lobar bronchial radiographic toxicity, with V having the highest odds ratio. Normal tissue complication probability modeling showed a volume dependency for the development of both clinical and radiographic toxicity. On multivariate analyses, significant predictors for ≥G3 toxicity were a planning target volume overlapping the trachea or main stem bronchus (P = .005), chronic obstructive pulmonary disease (P = .034), and the total V (P = .012). Radiographic bronchial toxicity did not significantly correlate with clinical toxicity (P = .663).

We identified patient and dosimetric factors associated with clinical and radiographic toxicity after high-dose radiation therapy for central lung tumors. Additional data from prospective studies are needed to validate these findings.