2022 World Conference on Lung Cancer (Posters)-P2.09-03. A Radiomics Approach Using Baseline CT Can Predict Response to 1st-Line Pembrolizumab in Advanced NSCLC with High PD-L1

P2.09-03. A Radiomics Approach Using Baseline CT Can Predict Response to 1st-Line Pembrolizumab in Advanced NSCLC with High PD-L1

Back to course

Pdf Summary

A study was conducted to determine if a Machine Learning (ML) model using radiomic features derived from lung tumor CT images at baseline can predict the response to pembrolizumab in advanced non-small cell lung cancer (NSCLC) patients with high PD-L1 expression. The study included 97 patients with advanced NSCLC who received first-line pembrolizumab. Treatment response was assessed using RECIST criteria. Radiomic features were extracted from the lung tumors using baseline CT scans and combined with clinical factors to train a ML model (Linear Discriminate Analysis, LDA) to classify treatment response.<br /><br />The results showed that the ML model incorporating radiomic features and clinical factors had the best classification performance compared to models that included only clinical or radiomic features. This model achieved an area under the curve (AUC) of 0.92, indicating its ability to predict treatment response.<br /><br />The study concluded that the ML model can be a valuable decision-making tool for guiding therapy in advanced NSCLC patients with high PD-L1 expression. It has the potential to identify patients who are likely to respond to pembrolizumab and those who may need alternative treatments such as triplet chemotherapy and immune checkpoint inhibitors.<br /><br />These findings highlight the potential of radiomics and ML in predicting treatment response and personalizing therapy for NSCLC patients. By using non-invasive imaging data, the ML model offers advantages over tissue-based biomarkers and could be implemented into standard clinical practice.<br /><br />Overall, the study suggests that the ML model trained on radiomic features from baseline CT scans, combined with clinical factors, can predict the response to pembrolizumab in advanced NSCLC patients with high PD-L1 expression. This model could help guide treatment decisions and improve outcomes for this patient population.

Asset Subtitle

Ren Yuan, Canada

Meta Tag

Speaker Ren Yuan, Canada

Topic Pulmonology, Radiology, and Staging

Keywords

Machine Learning

radiomic features

lung tumor

CT images

pembrolizumab

non-small cell lung cancer

NSCLC

PD-L1 expression

treatment response

Linear Discriminate Analysis