2023 World Conference on Lung Cancer (Posters)-P1.21. Radiomic Signature of Identifies Outcome and Prognosis to Immune Checkpoint Inhibitors (ICI) in PD-L1 Low Non-small Cell Lung Cancer (NSCLC)

P1.21. Radiomic Signature of Identifies Outcome and Prognosis to Immune Checkpoint Inhibitors (ICI) in PD-L1 Low Non-small Cell Lung Cancer (NSCLC) - PDF(Slides)

Back to course

Pdf Summary

This document presents a retrospective study that aimed to develop an AI-powered tool to identify patients with PD-L1 negative non-small cell lung cancer (NSCLC) who are likely to respond and benefit from immune checkpoint inhibitors (ICI). The study used a training set of 144 patients from a single institution and a testing set of 87 patients from two institutions. Radiomic features were extracted from target lesions using a deep learning segmentation model. A radiomics-based Cox proportional hazards model was trained to predict post-ICI progression-free survival (PFS).<br /><br />The results showed that the radiomic risk score was significantly associated with PFS and overall survival (OS), and had an area under the curve (AUC) of 0.65 in predicting the best overall response (BOR). The radiomic risk groups effectively separated PD-L1 negative patients by both PFS and OS. The low radiomic risk group had a higher response rate among PD-L1 negative patients, while the high radiomic risk group had a higher progression rate and did not respond to ICIs.<br /><br />The authors concluded that the AI-powered radiomic signature could stratify patients by immunotherapy outcomes in the PD-L1 negative patient population, which currently lacks validated biomarkers. This signature has the potential to better target immunotherapy and reduce overtreatment among PD-L1 negative NSCLC patients. The study highlights the promise of non-invasive machine learning tools in predicting ICI benefit in NSCLC and addresses critical gaps in existing biomarkers.<br /><br />Further validation in larger cohorts is recommended. The study references clinical trial data to support the need for biomarkers in the PD-L1 negative subset of patients. Overall, this research contributes to the development of personalized medicine approaches for NSCLC patients receiving ICIs.

Asset Subtitle

Nathaniel Braman

Meta Tag

Speaker Nathaniel Braman

Topic Pathology & Biomarkers: Biomarkers for Immuno-oncology

Keywords

retrospective study

AI-powered tool

PD-L1 negative NSCLC

immune checkpoint inhibitors

radiomic features

Cox proportional hazards model

progression-free survival

radiomic risk score

validated biomarkers

personalized medicine approaches