Predicting Survival Rates in Brain Metastases Patients from Non-Small Cell Lung Cancer Using Radiomic Signatures Associated with Tumor Immune Heterogeneity.

Abstract

Non-small cell lung cancer (NSCLC) frequently metastasizes to the brain, significantly worsened prognoses. This study aimed to develop an interpretable model for predicting survival in NSCLC patients with brain metastases (BM) integrating radiomic features and RNA sequencing data. 292 samples are collected and analyzed utilizing T1/T2 MRIs. Bidirectional stepwise logistic regression is employed to identify significant variables, facilitating the construction of a prognostic model, which is benchmarked against four machine learning algorithms. BM tissue samples are processed for RNA extraction and sequencing. The optimal model achieved an AUC of 0.96 and a C-index of 0.89 in the train set and an AUC of 0.84 with a C-index of 0.78 in the test set, indicating strong predictive performance and generalizability. Patients from Xiangya Hospital are stratified into high-risk (n = 11) and low-risk (n = 30) groups. RNA sequencing revealed an enrichment of immune-related pathways, particularly the interferon (IFN) pathway in the low-risk group. Immune cell infiltration analysis identified a significant presence of CD8⁺-T cells, IFNγ-6/-18 in the low-risk group, suggesting an immunologically favorable tumor microenvironment. These findings highlight the potential of combining radiomic and RNA sequencing data for improved survival predictions and personalized treatment strategies in BM patients from NSCLC.