Visualization of patterns of lymph node metastases in non–small cell lung cancer using network analysisCentral MessagePerspective

Abstract

Objective: We aimed to visualize complicated patterns of lymph node metastases in surgically resected non–small cell lung cancer by applying a data mining technique. Methods: In this retrospective study, 783 patients underwent lobectomy or pneumonectomy with systematic mediastinal lymph node dissection for non–small cell lung cancer between January 2010 and December 2018. Surgically resected lymph nodes were classified according to the International Association for the Study of Lung Cancer lymph node map. Network analysis generated patterns of lymph node metastases from stations 1 to 14, and the degree of connection between 2 lymph node stations was assessed. Results: The median number of lymph nodes examined per patient was 20, and the pathological N category was pN0 in 428 cases, pN1 in 132, pN2 in 221, and pN3 in 2. N1 lymph node stations had strong associations with superior mediastinal lymph node stations for patients with primary tumors in the upper lobes and with station 7 for the lower lobes. There was also a connection from the N1 lymph node stations to superior mediastinal lymph node stations in the lower lobes. In the right middle lobe, an even distribution from station 12m toward stations 2R, 4R, and 7 was noted. We released an interactive web application to visualize these data: http://www.canexapp.com. Conclusions: Lymph node metastasis patterns differed according to the lobe bearing the tumor. Our results support the need for clinical trials to further investigate selective mediastinal lymph node dissection.

Keywords

• network analysis
• data mining
• mediastinal lymph node dissection
• systematic mediastinal lymph node dissection
• selective mediastinal lymph node dissection
• non–small cell lung cancer