WCLC 2025 - Posters & ePosters-EP.07.18 Optimizing Intraoperative Lymph Node Dissection in Stage I NSCLC: A Risk-Stratified Approach Based on CNN Model Predictions

EP.07.18 Optimizing Intraoperative Lymph Node Dissection in Stage I NSCLC: A Risk-Stratified Approach Based on CNN Model Predictions

Back to course

Pdf Summary

This study by Fu and Hou from Shanghai Chest Hospital addresses the lack of consensus on intraoperative lymph node (LN) dissection standards for stage I non-small cell lung cancer (NSCLC). Various guidelines differ in recommended LN dissection extent, ranging from specific lymph node stations to a minimum number of nodes. To optimize surgical treatment, the authors developed a 3D convolutional neural network (CNN) model integrating chest CT images, clinical features (e.g., age, smoking history, tumor markers), and pathological data to predict occult lymphatic metastasis (OLM) risk in clinical stage I NSCLC patients. The model showed strong predictive performance with area under the curve (AUC) values exceeding 0.9 in internal and prospective test cohorts, demonstrating high accuracy, sensitivity, and specificity.<br /><br />Using the 85th percentile risk score as a threshold, patients were stratified into high- and low-risk groups. The study found that in high-risk clinical T2a patients, resection of more than six lymph nodes improved disease-free survival (DFS) and cancer-specific survival (CSS). Conversely, low-risk patients benefited from less extensive LN dissection, specifically fewer than six nodes or removal limited to 3 N1 and 3 N2 stations. No survival differences were observed in other subgroups based on LN dissection extent.<br /><br />The results suggest a tailored approach to intraoperative LN dissection in stage I NSCLC: limited dissection is sufficient and beneficial for low-risk patients, while more extensive LN removal is warranted for high-risk individuals to improve outcomes. This risk-stratified strategy, guided by the CNN model’s OLM prediction, offers a personalized surgical treatment framework that may improve prognosis and avoid overtreatment in stage I NSCLC patients.

Asset Subtitle

Yuanyuan Fu

Meta Tag

Speaker Yuanyuan Fu

Topic Early-Stage Non-small Cell Lung Cancer

Keywords

Fu and Hou study

Shanghai Chest Hospital

intraoperative lymph node dissection

stage I non-small cell lung cancer

3D convolutional neural network

chest CT images

occult lymphatic metastasis prediction

risk stratification

disease-free survival

personalized surgical treatment