The Validation of Predictive Risk Scoring Model of Radiation Pneumonitis (RP) in Lung Cancer Patients

Abstract

<h3>Purpose/Objective(s)</h3> In our previous research, pulmonary emphysema (PE) was reported as a risk factor of radiation pneumonitis (RP) in patients with non-small cell lung cancer (NSCLC). Subclinical interstitial lung disease (SILD) was associated with severe RP in small cell lung cancer patients. A predictive risk scoring model of RP was developed in stage IIIA or Stage IIIB NSCLC patients receiving radical radiotherapy. The risk score has not been validated using an independent dataset. The aim of this study was to validate the predictive risk score which established in our previous study and evaluate the predictive value of the model. <h3>Materials/Methods</h3> From January 2019 to October 2021,129 lung cancer patients were recruited in study which included in validation group. Those patients were received thoracic radiotherapy with ≥50Gy total radiotherapy dose. The risk scoring model in our previous research as follow: PRS=5 (if subclinical ILD grade≥1) + 3 (if age >68 years) + 3 (if PE grade >1). According to the risk score, the cases in validation group were scored, predictive value of this model were evaluated by receiver operator characteristic curve (ROC) using statistical software. The decision curve analysis was constructed using Stata 17.0 software. <h3>Results</h3> In 129 eligible patients, 13(10.08%, 13/129) developed grade 1 RP, 15(11.62%, 15/129) developed grade 2 RP, grade 3 and grade 4 RP were not observed. 23 (16.60%, 23/129) had PE grade >1. Risk scores of all patients were calculated according to the risk score, 37 had 3 points which incidence of RP ≥2 grade was 13.50%, 2 had 5 points which incidence of RP ≥2 grade was 100.0%, 3 had 6 points which incidence of RP ≥2 grade was 33.33%. ROC curve was plotted according to the risk score of all patients, the area of ROC curve (AUC) was 0.695 (95%CI: 0.541-0.849, P < 0.05). The decision curve indicated that the net benefit of the model was higher in the high-risk threshold of 0.2-0.3. <h3>Conclusion</h3> The previous predictive risk scoring model indicated good discrimination and clinical application value, which can be used in clinical treatment.