Radiotherapy(RT) remains an important and potentially curative treatment for localized and locally advanced non-small cell lung cancer (NSCLC), but radiotherapy-induced lung toxicity (RILT) is one of the dose-limiting toxicities for radiotherapy in patients with non-small cell lung cancer. This study is to analyze the accuracy and effectiveness of the combination of mean lung dose (MLD) and inflammatory cytokines (IL-8 and TGF-β1) in predicting radiation-induced lung toxicity (RILT).

75 patients with stage Ⅰ-Ⅲ NSCLC treated with definitive radiotherapy (RT) were included in this study. Treatments of patients receiving definitive conventionally fractionated radiotherapy on a clinical trial for inoperable stages Ⅰ-Ⅲ lung cancer were prospectively evaluated. Radiotherapy was given by using a three-dimensional conformal technique. Intensity-modulated radiation therapy (IMRT) was used in only a few challenging cases. A median total physical dose of 70 Gy (range, 44~87.9 Gy) was delivered with 2.0~2.9 Gy daily fractions over 4~7 weeks using 6~16 MV photons. Circulating cytokine levels were measured before and at weeks 2 and 4 during RT. All patients were prospectively evaluated weekly during RT, with follow-up evaluation after completion of RT. At each follow-up, patients underwent a history review and physical examination as well as a chest computed tomography scan. The primary endpoint was symptomatic RILT, and higher radiation pneumonitis or symptomatic pulmonary fibrosis. Logistic regression was performed to evaluate the risk factors of RILT. The area under the curve (AUC) for the Receiver Operating Characteristic (ROC) curves were used for model assessment.

75 patients with stage Ⅰ-Ⅲ NSCLC were included in this study, and 86% patients with RILT after receiving definitive conventionally fractionated radiotherapy. 16 of 65 patients (24.6%) developed RILT2. Lower pre-IL-8 and higher TGF-β1 2 W/pre ratio were associated with a higher risk of RILT2. Among the 30 cytokines measured, only IL-8 and TGF-β1 were significantly associated with the risk of RILT2. MLD, pre IL-8 level and TGF-β1 2 W/pre ratio were included in the final predictive model. Receiver operating characteristic (ROC) curves were evaluated for MLD, baseline IL-8 level, TGF-β1 2 W/pre ratio. The AUC was 0.61 (0.45, 0.77), 0.70 (0.56, 0.84) and 0.68 (0.53, 0.83) with MLD, baseline IL-8 level and TGF-β1 2 W/pre ratio, respectively. The AUC increased to 0.73 by combining MLD, pre-IL-8 and TGF-β1 2 W/pre ratio compared with 0.61 by MLD alone to predict RILT.

This study validated the predictive value of IL-8 and TGF-β1 for RILT. Pre IL-8 level and TGF-β1 2 W/pre ratio provided a more accurate model to predict the risk of RILT2 compared to MLD alone.