Validation and optimization of a predictive model for radiation pneumonitis in patients with lung cancer.

The aim of the current retrospective study was to validate a predictive model for radiation pneumonitis (STRIPE) in an independent dataset and to investigate whether the addition of other potential risk factors could strengthen the accuracy of the model. Consecutive patients with non-small cell lung carcinoma (NSCLC; n=71) treated with definitive concurrent chemotherapy and radiotherapy were retrospectively assessed for radiation pneumonitis (RP). The results identified that 16 (23%) patients developed grade ≥2 RP. Furthermore, STRIPE score (intermediate vs. low risk) was independently associated with the development of RP [odds ratio (OR), 3.72; 95% confidence interval (CI), 1.00-13.89], whereas current smoking status was found to be protective against RP (OR, 0.09; 95% CI, 0.01-0.78). Similar discriminatory power of the STRIPE score was observed as in the original study. The addition of smoking status strengthened the model's discriminatory ability to predict RP. Thus, the addition of smoking status as a risk factor may strengthen the accuracy of the model for predicting RP in patients with NSCLC.