We aimed to develop a more accurate model for predicting severe radiation pneumonitis (RP) after concurrent chemoradiotherapy for non–small-cell lung cancer.
We retrospectively analyzed data from 122 patients with locally advanced non–small-cell lung cancer treated with concurrent chemoradiotherapy. Several dose-volume histogram metrics including absolute lung volume spared from a 5 Gy dose (VS5) were analyzed for an association with RP above NCI-CTC grade 3 (RP ≥ G3). Clinical factors including pulmonary fibrosis score (PFS) and pulmonary emphysema score on baseline chest computed tomography (CT) were also analyzed.
Fourteen patients (11.4%) developed RP greater than or equal to G3. On univariate analysis, all dose-volume histogram metrics, sex, and PFS on baseline CT were significantly (p < 0.05) associated with occurrence of RP greater than or equal to G3. Multivariate analysis revealed that V20 greater than or equal to 26%, VS5 less than 1500 cc, age greater than or equal to 68 years, and PFS on baseline CT greater than or equal to 2 were significant risk factors. Thus, we defined a new predictive risk score (PRS) that combines these factors. The cumulative incidence of RP greater than or equal to G3 at 12 months were 0%, 7.8%, 26.6%, and 71.4% when the PRS was 0, 3–5, 6–8, and 9–14, respectively (p < 0.001). This PRS was superior at predicting RP than both V20 and VS5 combined, or V20 alone by receiver operating characteristic analysis (area under the curve, 0.888 versus 0.779 versus 0.678).
V20, VS5, age, and PFS on baseline CT are independent and significant risk factors for occurrence of severe RP. Combining these factors may improve the predictability of severe RP.