Machine learning algorithms can effectively predict the development of brain metastases (BM) in patients with non-small cell lung cancer (NSCLC) who are at a high risk of BM and could benefit from surveillance brain MRI. NSCLC, which accounts for about 85% of lung cancers, has a BM incidence of 20-40% among patients. Early detection of BM is crucial for better patient outcomes.

Researchers conducted a study using classifier models to identify NSCLC patients at high risk of developing BM. The study included a clinical cohort of 395 patients with an initial NSCLC diagnosis between January 2011 and April 2019. The models were trained and validated using clinical and 3D radiomics features of the primary tumor from staging chest CT scans.

The study found that classification models using only clinical features as input had the best performance, achieving a sensitivity of approximately 70% and a specificity just above 60%. Radiomics features did not provide sufficient information due to the heterogeneity of imaging data. Certain factors, such as adenocarcinoma histology, lymph node invasion, and histological tumor grade at diagnosis, were positively correlated with the prediction of BM. On the other hand, age and squamous cell carcinoma histology at diagnosis were negatively correlated with BM prediction.

Subpopulation analyses revealed that female patients with adenocarcinoma histology and patients with adenocarcinoma without other distant metastases were at a higher risk of BM. The study also highlighted the importance of the oxidative phosphorylation pathway in BM risk.

Although some known predictors, such as the presence of EGFR mutations, did not show significant results in this study, the researchers acknowledged the limitations of the small dataset size and high fraction of missing values. Prospective application of these prediction models could help identify high-risk patients who would benefit from early detection and treatment of asymptomatic BM through surveillance brain MRI.

The post Machine Learning Helps Predict Brain Metastasis in Non-Small Cell Lung Cancer Patients appeared first on TS2 SPACE.