Lung Inflammation Predictors in Combined Immune Checkpoint-Inhibitor and Radiation Therapy-Proof-of-Concept Animal...

Area of Science:

• Oncology
• Radiology
• Immunology

Background:

• Combined radiotherapy (RT) and immune checkpoint-inhibitor (ICI) therapy offer synergistic tumor response.
• Checkpoint-inhibitor-induced (CIP) pneumonitis is a severe, unpredictable side effect of ICI therapy.
• Predicting and managing CIP pneumonitis is crucial for patient safety.

Purpose of the Study:

• To develop predictive models for lung inflammation associated with combined RT/ICI therapy.
• To investigate the utility of routinely collected data, including imaging, blood counts, and cytokines, for predicting lung inflammation.
• To test the hypothesis that quantitative imaging, blood counts, and cytokine levels can predict RT/ICI-induced lung inflammation.

Main Methods:

• A preclinical study using a Lewis lung carcinoma murine model.
• Mice received combined RT and PD-1 inhibitor therapy.
• Quantitative radiomics features were extracted from CT and MRI scans.
• Pre-treatment blood counts (neutrophil-to-lymphocyte ratio) and cytokine levels (GM-CSF) were analyzed.
• Logistic regression models were built incorporating radiomics, blood counts, and cytokines to predict lung inflammation (CD45 infiltration).

Main Results:

• Four pre-treatment radiomics features, neutrophil-to-lymphocyte ratio (NLR), and GM-CSF levels correlated with lung inflammation.
• Predictive models combining radiomics with NLR and GM-CSF were developed.
• Internal cross-validation showed excellent predictive performance: AUC of 0.834 for MRI-based model and 0.787 for CT-based model.

Conclusions:

• Quantitative imaging, blood counts, and cytokine data can be used to build effective models for predicting lung inflammation.
• The developed models demonstrate high predictive accuracy (AUC > 0.78) for RT/ICI-induced lung inflammation.
• These findings support the hypothesis and offer a potential strategy for early detection and management of CIP pneumonitis.