Predictors and Models for Stereotactic Radiosurgery Outcomes in NSCLC With Brain Metastases
Posted: Monday, March 9, 2026
Immune-inflammatory and radiologic-clinical factors independently predicted stereotactic radiosurgery efficacy and overall survival in patients with non–small cell lung cancer (NSCLC) and brain metastases, according to a retrospective cohort study published in Frontiers in Immunology, in which investigators also developed decision tree and nomogram models incorporating the systemic immune-inflammation profile for prognostication. Aihong Cao, MD, of The Second Affiliated Hospital of Xuzhou Medical University, China, and colleagues wrote that the models “provide user-friendly tools to assist clinicians in optimizing personalized management for this patient population.”
The investigators focused on 464 patients with NSCLC and brain metastases treated with stereotactic radiosurgery. In a 7:3 ratio, the cohort was randomly divided into training and validation sets. A C5.0 algorithm was used to develop a decision tree model for treatment response. Prognostic factors for overall survival were identified using univariate and multivariate Cox regression and incorporated into graphical and online nomograms. Model performance was evaluated with calibration curves and the Concordance Index (C-index).
In the entire cohort, the median duration of overall survival was 15.8 months. The decision tree analysis identified neutrophil-to-lymphocyte ratio as a key predictor of treatment response, in addition to brain metastasis volume, Score Index for Radiosurgery (SIR), edema index (EI), and maximum diameter. Based on multivariate Cox analysis, age, volume of brain metastases, EI, and SIR were identified as independent prognostic factors for overall survival.
These factors were used to construct graphical and dynamic nomograms. The investigators reported that calibration curves demonstrated good concordance between predicted and actual survival, while the C-index indicated moderate discrimination.
Disclosure: The study authors reported no conflicts of interest.
