To construct a risk model of radiation-induced lung injury(RILT)in middle-aged and elderly patients with esophageal cancer during intensity-modulated radiotherapy(IMRT), and to explore its application value.

From January 2018 to January 2023, 194 middle-aged and elderly patients with esophageal cancer who were treated with IMRT were selected from the Affiliated Shuyang Hospital of Xuzhou Medical University, and were divided into model group(n=145)and verification group(n=49)according to the ratio of 3∶1.The incidence of RILT in the two groups was recorded, and the risk factors affecting the occurrence of RILT in the model group were summarized by Logistic multivariate regression analysis, and the Nomogram prediction model was established based on the risk factors.The application efficiency of the model group and verification group was calculated, and Kolmogorov-Smimov goodness-of-fit test and decision curve analysis were used to evaluate the model calibration and net yield.

The incidence of RILT was 14.48%(21/145)in the model group and 16.33%(8/49)in the verification group, and there was no statistical difference between the two groups(χ²=0.098, P=0.754). Logistic multivariate regression model analysis showed that age≥65 years, KPS score<80, total radiation dose≥50 Gy, concurrent chemotherapy, diabetes, and lung V₅≥55%, V₁₀≥38%, V₂₀≥28%, V₃₀≥20%, and average lung dose≥13 Gy were all independent risk factors for RILT during IMRT in middle-aged and elderly patients with esophageal cancer(all P<0.05). The Nomogram model based on risk factor analysis had good efficiency, Kolmogorov-Smimov goodness-of-fit test showed that the model had good calibration(χ²=5.604, P=0.779), and decision curve analysis showed that the model had good discrimination.

The risk of RILT in middle-aged and elderly patients with esophageal cancer during IMRT is high, and it is related to many factors, such as age, mobility, treatment plan, complications, lung dosimetry and so on.The prediction model based on the above factors can provide reliable reference for RILT risk prediction, and risk assessment and intervention can gain good benefits.