2023 World Conference on Lung Cancer (Posters)-EP08.01. Subtyping EGFR-mutated NSCLC Patients with Adjuvant TKIs or Chemotherapy Using Deep Radiomic Signatures

EP08.01. Subtyping EGFR-mutated NSCLC Patients with Adjuvant TKIs or Chemotherapy Using Deep Radiomic Signatures - PDF(Slides)

Back to course

Pdf Summary

Researchers in China have developed a deep learning model that uses radiomic signatures on computed tomography (CT) scans to subtype patients with epidermal growth factor receptor (EGFR)-mutated non-small cell lung cancer (NSCLC) and predict their response to adjuvant EGFR-tyrosine kinase inhibitors (TKIs) or chemotherapy. This study aimed to identify patient subgroups with different TKI or chemotherapy outcomes to improve NSCLC treatment.<br /><br />The researchers collected data from 194 EGFR-mutated NSCLC patients who underwent complete resection at Guangdong Provincial People's Hospital between 2008 and 2021. They designed a deep learning model to predict disease-free survival (DFS) for adjuvant TKI and chemotherapy. The model provided deep hazard (dh) indexes, dhT and dhC, which were used to classify patients into four clusters: cluster 1 (poor DFS for both therapies), cluster 2 (good DFS for both therapies), cluster 3 (better DFS with TKIs than chemotherapy), and cluster 4 (better DFS with chemotherapy than TKIs). They compared the subtypes of EGFR mutations, clinicopathological features, and DFS between these clusters.<br /><br />The results showed that 92.27% of patients had 19del or L858R EGFR mutations, and 7.73% had other types of mutations. Of the patients, 88 received adjuvant TKIs and 106 received chemotherapy. The median follow-up was 1540 days. The analysis found that the four clusters varied in tumor diameter, T stage, TNM stage, and pathological grade. Cluster 1 had the largest tumor diameter and the worst clinicopathological features, while cluster 4 had more unclassical EGFR mutations. The DFS varied among the clusters, with cluster 3 patients benefiting more from TKIs and cluster 4 patients benefiting more from chemotherapy.<br /><br />Based on these findings, the researchers concluded that CT-based deep radiomic signatures may help determine whether patients with stage IB-III EGFR-mutated NSCLC should receive adjuvant chemotherapy or TKIs. This study highlights the potential of using deep learning models and radiomic analysis to personalize treatment decisions for NSCLC patients.

Asset Subtitle

Yi-Fan Qi

Meta Tag

Speaker Yi-Fan Qi

Topic Local-Regional NSCLC: Biomarkers

Keywords

China

deep learning model

radiomic signatures

computed tomography

CT scans

epidermal growth factor receptor

EGFR-mutated non-small cell lung cancer

NSCLC

adjuvant EGFR-tyrosine kinase inhibitors

chemotherapy