Prediction of Lung Metastasis in Breast Cancer Patients Using Machine Learning Classifiers

Area of Science:

• Oncology
• Genomics
• Biomarker Discovery

Background:

• Breast cancer is the most common cancer in women, with lung metastasis indicating a poor prognosis.
• Accurate biomarkers are crucial for early detection and improved clinical decision-making in breast cancer patients.
• Identifying patients at high risk for lung metastasis is essential for timely intervention.

Purpose of the Study:

• To develop and validate a gene expression-based risk signature for predicting lung metastasis in breast cancer.
• To identify a panel of genes that can reliably predict the risk of lung metastasis.

Main Methods:

• Microarray data from primary breast cancer patients with follow-up for lung metastasis outcomes were analyzed.
• High-throughput screening, Venn diagram analysis, and least absolute shrinkage and selection operator (LASSO) regression were employed.
• A 10-gene risk signature was constructed using logistic regression and validated in independent datasets.

Main Results:

• A 10-gene risk signature (including CDK19, GLUD1, GTPBP4, HLCS, HYI, KCND3, MAP2K1, NMUR1, PRKD3, SLC16A3) demonstrated strong predictive performance (AUC >0.87 in training/validation).
• The signature generalized to an independent dataset (METABRIC, AUC = 0.706) and retained predictive value in early-stage disease.
• High-risk scores correlated with shorter lung metastasis-free survival, recurrence-free survival, and overall survival.

Conclusions:

• The developed 10-gene risk signature accurately identifies breast cancer patients at risk of lung metastasis.
• This biomarker enables better risk assessment and facilitates tailored treatment strategies for improved patient management.
• The signature provides independent predictive information beyond traditional clinical variables.