Objective:To analyze risk factors for immune checkpoint inhibitor-related pneumonitis (CIP) in non-small cell lung cancer (NSCLC) patients based on clinical and radiological characteristics, and to develop and validate a nomogram model for predicting the risk of CIP.Methods:A retrospective case-controlled study was conducted. The clinical data of 159 patients diagnosed with NSCLC in Shanxi Province Cancer Hospital between January 2020 and December 2023 who received immune checkpoint inhibitor (ICI) therapy were retrospectively analyzed. Based on the development of CIP after immunotherapy, the patients were divided into the CIP group (30 cases) and the control group (129 cases). The clinical data of NSCLC patients, hematological indicators and the data of imaging characteristics before their first ICI treatment were collected. Quantitative assessments were performed on pretreatment chest CT images, including lung total tumor volume, number of involved lung segments, and pulmonary infection index. Logistic regression analysis was used to screen out the factors influencing the development of CIP. R 4.3.0 statistical software was used to construct a nomogram model for predicting CIP based on the statistically significant risk factors identified in the multivariate logistic regression analysis. The predictive performance of the model was evaluated by using receiver operating characteristic (ROC) curves and the area under the curve (AUC). Calibration curves and decision curve analysis (DCA) were employed to assess the model's consistency and clinical benefit.Results:There were statistically significant differences in the proportions of patients with a history of chest radiotherapy and those receiving different immunotherapy regimens between the control group and the CIP group (both P < 0.001). The difference in the lactate dehydrogenase (LDH) [ M ( IQR)] between the both groups was statistically significant [211.00 U/L (57.00 U/L) vs. 276.00 U/L (136.00 U/L), Z = -3.41, P < 0.001]; additionally, the difference in lung status score between the 2 groups was statistically significant ( P < 0.001). Multivariate logistic regression analysis revealed that a history of chest radiotherapy (with vs. without: OR = 4.200, 95% CI: 1.466-12.036), the combination of immunotherapy (monotherapy vs. the combined therapy: OR = 0.106, 95% CI: 0.022-0.509), LDH ≥ 255.5 U/L (< 255.5 U/L vs. ≥ 255.5 U/L: OR = 0.988, 95% CI: 0.981-0.995), and severe lung status score(mild vs. moderate vs. severe: OR = 0.187, 95% CI: 0.059-0.593) were independent risk factors for CIP development in NSCLC patients after immunotherapy (all P < 0.05). A nomogram model for predicting CIP occurrence was constructed based on chest radiotherapy history, immunotherapy regimen, LDH, and lung status score. ROC curve analysis showed the AUC was 0.878 (95% CI: 0.813-0.942). The calibration curve demonstrated the good consistency between the predicted risk probability of CIP and the observed outcomes; DCA indicated that the model had favorable clinical benefits. Conclusions:The constructed nomogram prediction model shows a good predictive performance.

Objective:To analyze risk factors for immune checkpoint inhibitor-related pneumonitis (CIP) in non-small cell lung cancer (NSCLC) patients based on clinical and radiological characteristics, and to develop and validate a nomogram model for predicting the risk of CIP.Methods:A retrospective case-controlled study was conducted. The clinical data of 159 patients diagnosed with NSCLC in Shanxi Province Cancer Hospital between January 2020 and December 2023 who received immune checkpoint inhibitor (ICI) therapy were retrospectively analyzed. Based on the development of CIP after immunotherapy, the patients were divided into the CIP group (30 cases) and the control group (129 cases). The clinical data of NSCLC patients, hematological indicators and the data of imaging characteristics before their first ICI treatment were collected. Quantitative assessments were performed on pretreatment chest CT images, including lung total tumor volume, number of involved lung segments, and pulmonary infection index. Logistic regression analysis was used to screen out the factors influencing the development of CIP. R 4.3.0 statistical software was used to construct a nomogram model for predicting CIP based on the statistically significant risk factors identified in the multivariate logistic regression analysis. The predictive performance of the model was evaluated by using receiver operating characteristic (ROC) curves and the area under the curve (AUC). Calibration curves and decision curve analysis (DCA) were employed to assess the model's consistency and clinical benefit.Results:There were statistically significant differences in the proportions of patients with a history of chest radiotherapy and those receiving different immunotherapy regimens between the control group and the CIP group (both P < 0.001). The difference in the lactate dehydrogenase (LDH) [ M ( IQR)] between the both groups was statistically significant [211.00 U/L (57.00 U/L) vs. 276.00 U/L (136.00 U/L), Z = -3.41, P < 0.001]; additionally, the difference in lung status score between the 2 groups was statistically significant ( P < 0.001). Multivariate logistic regression analysis revealed that a history of chest radiotherapy (with vs. without: OR = 4.200, 95% CI: 1.466-12.036), the combination of immunotherapy (monotherapy vs. the combined therapy: OR = 0.106, 95% CI: 0.022-0.509), LDH ≥ 255.5 U/L (< 255.5 U/L vs. ≥ 255.5 U/L: OR = 0.988, 95% CI: 0.981-0.995), and severe lung status score(mild vs. moderate vs. severe: OR = 0.187, 95% CI: 0.059-0.593) were independent risk factors for CIP development in NSCLC patients after immunotherapy (all P < 0.05). A nomogram model for predicting CIP occurrence was constructed based on chest radiotherapy history, immunotherapy regimen, LDH, and lung status score. ROC curve analysis showed the AUC was 0.878 (95% CI: 0.813-0.942). The calibration curve demonstrated the good consistency between the predicted risk probability of CIP and the observed outcomes; DCA indicated that the model had favorable clinical benefits. Conclusions:The constructed nomogram prediction model shows a good predictive performance.