Predicting Survival Rates in Brain Metastases Patients from Non-Small Cell Lung Cancer Using Radiomic Signatures...

Area of Science:

• Oncology
• Radiology
• Genomics

Background:

• Non-small cell lung cancer (NSCLC) brain metastases (BM) significantly worsen patient prognosis.
• Accurate survival prediction is crucial for personalized treatment strategies.

Purpose of the Study:

• To develop an interpretable prognostic model for NSCLC patients with BM.
• Integrate radiomic features from MRI and RNA sequencing data for enhanced prediction.
• Identify biological pathways associated with survival outcomes in NSCLC with BM.

Main Methods:

• Collected and analyzed 292 NSCLC patient samples with BM using T1/T2 MRIs.
• Employed bidirectional stepwise logistic regression to build a prognostic model.
• Performed RNA sequencing on BM tissue and analyzed immune cell infiltration.

Main Results:

• The developed model demonstrated high predictive accuracy (AUC 0.96, C-index 0.89 train; AUC 0.84, C-index 0.78 test).
• Low-risk patients exhibited enrichment in immune-related pathways, including the interferon (IFN) pathway.
• Increased CD8+ T-cell and IFNγ presence correlated with a favorable tumor microenvironment in low-risk NSCLC BM patients.

Conclusions:

• Combining radiomic and RNA sequencing data offers a powerful approach for predicting survival in NSCLC patients with BM.
• The findings suggest an immunologically driven basis for survival differences.
• This integrated approach can inform personalized treatment strategies for improved patient outcomes.