2022 World Conference on Lung Cancer (Posters)-P2.12-03. External Validation of a Novel CT-Based Prognostic Radiomic Signature in Patients With Metastatic NSCLC in SWOG S0819 Phase III Randomized Trial

P2.12-03. External Validation of a Novel CT-Based Prognostic Radiomic Signature in Patients With Metastatic NSCLC in SWOG S0819 Phase III Randomized Trial

Back to course

Pdf Summary

This study aimed to externally validate a novel radiomic signature (NivoSig) derived from computed tomography (CT) scans to predict overall survival (OS) in patients with metastatic non-small cell lung cancer (NSCLC) enrolled in the SWOG S0819 phase III randomized trial. The NivoSig radiomic signature, based on imaging biomarkers extracted from CT scans, was previously associated with OS in patients treated with immunotherapy. In this study, the NivoSig was validated in a separate cohort of patients receiving different systemic therapy regimens. The study included 1,229 patients who were randomized to receive either chemotherapy alone or the same chemotherapy regimen in combination with cetuximab. The NivoSig radiomic signature ranged from 0 (most favorable estimated OS) to 1 (least favorable estimated OS) and classified patients as high-risk or low-risk. The analysis showed that the NivoSig was a significant predictor of OS, with low-risk NivoSig patients having longer median OS compared to high-risk NivoSig patients. Additionally, the radiomic features included in the NivoSig, such as tumor texture, tumor volume, and number of target lesions, were found to be significantly different between high-risk and low-risk patients. The study concluded that the NivoSig radiomic signature could be leveraged to prognosticate cancer patients and guide clinical decision-making. The use of AI-based radiomics signatures can provide objective and reproducible predictions at an early timepoint, which is clinically relevant given the subjective nature of radiologist interpretation of medical images. Further research will focus on incorporating these radiomic features to better understand mechanisms of resistance and improve treatment strategies.

Asset Subtitle

Laurent Dercle, United States

Meta Tag

Speaker Laurent Dercle, United States

Topic Tumour Biology and Biomarkers - Immune Biology & Immunotherapy

Keywords

radiomic signature

computed tomography scans

overall survival

metastatic non-small cell lung cancer

chemotherapy regimen

cetuximab

high-risk patients

low-risk patients

tumor texture

tumor volume