A cancer neuroscience-related gene model for prognosis prediction and immunotherapy response evaluation in breast cancer

Abstract Background Breast cancer prognosis remains challenging, and the emerging field of cancer neuroscience suggests that the nervous system plays a crucial yet underexplored role in tumor progression. This study aimed to construct and validate a novel prognostic signature for breast cancer based on genes involved in neural-tumor interactions. Methods Differential expression analysis and univariate Cox regression were performed on the TCGA-BRCA dataset to identify genes associated with both cancer neuroscience and patient prognosis. A prognostic model was constructed using LASSO and multivariate Cox regression analyses. Its predictive performance was validated in external datasets. The immune microenvironment, tumor mutation burden, and immunotherapy response were compared between the high- and low-risk groups. Drug sensitivity was predicted using the oncoPredict algorithm. Results A 12-gene prognostic model was developed. Patients stratified into high- and low-risk groups showed significant survival differences in all cohorts. The signature demonstrated reliable predictive accuracy, with AUCs of 0.700, 0.744, and 0.759 for 1-, 3-, and 5-year survival in the TCGA dataset. The low-risk group exhibited a more immunologically active tumor microenvironment,suggesting a potentially better response to immunotherapy. Drug sensitivity analysis identified three compounds with lower predicted IC50 values in the high-risk group. Conclusion This study establishes and validates a novel 12-gene cancer neuroscience-related prognostic model for breast cancer. This model not only effectively stratifies patient risk but also reveals distinct immune landscapes and predicts differential responses to immunotherapy and potential therapeutic agents. These findings offer new insights for prognostication and personalized treatment strategies in breast cancer.