Expanding TNM for lung cancer through machine learning

Area of Science:

• Oncology
• Machine Learning
• Biostatistics

Background:

• The tumor, lymph node, metastasis (TNM) staging system is crucial for cancer patient stratification and survival prediction.
• Incorporating additional prognostic and predictive factors can enhance the TNM system's accuracy.
• Lung cancer staging requires refinement for improved patient outcomes.

Purpose of the Study:

• To apply machine learning for integrating additional prognostic factors into the conventional TNM staging system for lung cancer.
• To improve patient stratification and survival prediction accuracy in lung cancer.
• To evaluate the efficacy of the Ensemble Algorithm for Clustering Cancer Data (EACCD) in refining cancer staging.

Main Methods:

• Utilized data from the Surveillance, Epidemiology, and End Results (SEER) database, analyzing 77,953 lung cancer patients.
• Employed the Ensemble Algorithm for Clustering Cancer Data (EACCD) to generate prognostic groups and expand the TNM staging system.
• Incorporated factors including primary tumor (T), regional lymph node (N), distant metastasis (M), age, and histology type.

Main Results:

• The EACCD model stratified patients into 11 groups, achieving significantly higher survival prediction accuracy (C-index = 0.7346) than the 10 AJCC stages (C-index = 0.7247).
• Integrating age and histological tumor type with TNM further improved prediction accuracy, creating 12 prognostic groups (C-index = 0.7468).
• A strong association was observed between EACCD groupings and AJCC staging, indicating robust model performance.

Conclusions:

• The Ensemble Algorithm for Clustering Cancer Data (EACCD) effectively integrates additional prognostic factors with the TNM staging system for lung cancer.
• Machine learning offers a powerful approach to enhance cancer staging and improve survival prediction.
• Refined staging systems lead to better patient stratification and personalized treatment strategies.