Early prediction of immune checkpoint inhibitor-related pneumonitis in advanced non-small cell lung cancer based on primary tumor Delta-radiomics features - Report - MDSpire
Advertisement
Early prediction of immune checkpoint inhibitor-related pneumonitis in advanced non-small cell lung cancer based on primary tumor Delta-radiomics features
Clinical Report: Prognostic Assessment of Immune Checkpoint Inhibitor-Induced Pneumonitis
Overview
This study evaluates the predictive capability of Delta radiomics features from CT imaging for immune checkpoint inhibitor-related pneumonitis (ICIP) in advanced non-small cell lung cancer (NSCLC). The Delta radiomics model demonstrated strong performance, particularly when integrated with clinical features, suggesting its potential utility in early risk stratification for ICIP.
Background
Immune checkpoint inhibitors (ICIs) have become a cornerstone in the treatment of advanced NSCLC, but they are associated with significant toxicities, including ICIP. Accurate risk stratification and early recognition of ICIP are crucial for optimizing patient management and improving outcomes. This study explores the use of advanced imaging techniques to enhance predictive capabilities for this serious adverse effect.
Data Highlights
Model
AUC (Training Set)
AUC (Validation Set)
Delta Radiomics (LR)
0.92 (95% CI: 0.88-0.97)
0.85 (95% CI: 0.78-0.92)
Combined Model
0.94 (95% CI: 0.90-0.98)
0.86 (95% CI: 0.79–0.93)
Key Findings
46 out of 131 patients (35.1%) developed ICIP, with 8 patients (17.4%) experiencing grade 3-5 ICIP.
From 2153 initial features, 22 key Delta radiomics features were identified for model construction.
The Delta radiomics model using Logistic Regression showed the highest performance in predicting ICIP.
The combined model integrating Delta radiomics and clinical features improved predictive performance.
Calibration curves indicated good calibration and favorable clinical utility for the models.
Clinical Implications
The findings suggest that Delta radiomics can serve as a valuable imaging biomarker for early identification of patients at risk for ICIP. Clinicians may consider integrating these predictive models into routine assessments to enhance patient management strategies and mitigate the risks associated with ICIs.
Conclusion
This study highlights the potential of Delta radiomics in predicting ICIP in advanced NSCLC, warranting further validation in independent cohorts before clinical implementation.
