2023 World Conference on Lung Cancer (Posters)-P2.07. Checkpoint Inhibitor Pneumonitis Model for NSCLC based onPreexistingLung Ground Glass Opacities: Development and Validation

P2.07. Checkpoint Inhibitor Pneumonitis Model for NSCLC based onPreexistingLung Ground Glass Opacities: Development and Validation - PDF(Abstract)

Back to course

Pdf Summary

This study aimed to identify non-small cell lung cancer (NSCLC) patients at high risk of checkpoint inhibitor pneumonitis (CIP) using quantitative imaging and computational analysis. The study included a training cohort of 206 patients and a validation cohort of 111 patients. A deep learning algorithm labeled interstitial lesions and computed their volume. Two prediction models were developed, one to predict grade 2 CIP and another to predict grade 3 CIP. Model 1 used age, histology, and preexisting ground glass opacity (GGO) percentage in the whole lung, while Model 2 used histology and GGO percentage in the right lower lung. Both models showed good discriminatory ability and clinical utility in guiding treatment decisions. The validation cohort had similar accuracy to the training cohort.<br /><br />The study also found that GGOs involving more than 1 lobe on pretreatment CT were a risk factor for progression-free survival (PFS). Assessing the volume and distribution of GGOs on pretreatment CT may help monitor and manage the risk of CIP in NSCLC patients receiving immune checkpoint inhibitor (ICI) treatment.<br /><br />The study highlights the potential of using deep learning algorithms and quantitative imaging analysis to identify NSCLC patients at high risk of CIP. These models could assist clinicians in making treatment decisions and managing the risk of CIP in patients receiving ICI therapy. Additionally, the study emphasizes the importance of assessing GGOs on pretreatment CT scans to monitor and manage the risk of CIP and improve patient outcomes.

Asset Subtitle

Xinyue Wang

Meta Tag

Speaker Xinyue Wang

Topic Metastatic NSCLC: Immunotherapy - Retrospective

Keywords

non-small cell lung cancer

NSCLC

checkpoint inhibitor pneumonitis

CIP

quantitative imaging

deep learning algorithm

ground glass opacity

GGO

treatment decisions

patient outcomes