Evaluating the prognostic potential of telomerase signature in breast cancer through advanced machine learning model.

Background: Breast cancer prognosis remains a significant challenge due to the disease's molecular heterogeneity and complexity. Accurate predictive models are critical for improving patient outcomes and tailoring personalized therapies. Methods: We developed a Machine Learning-assisted Telomerase Signature (MLTS) by integrating multi-omics data from nine independent breast cancer datasets. Using multiple machine learning algorithms, we identified six telomerase-related genes significantly associated with patient survival. The predictive performance of MLTS was evaluated against 66 existing breast cancer prognostic models across diverse cohorts. Results: The MLTS demonstrated superior predictive accuracy, stability, and reliability compared to other models. Patients with high MLTS scores exhibited increased tumor mutational burden, chromosomal instability, and poor survival outcomes. Single-cell RNA sequencing analysis further revealed higher MLTS scores in aneuploid tumor cells, suggesting a role in cancer progression. Immune profiling indicated distinct tumor microenvironment characteristics associated with MLTS scores, potentially guiding therapeutic decisions. Conclusions: Our findings highlight the utility of MLTS as a robust prognostic biomarker for breast cancer. The ability of MLTS to predict patient outcomes and its association with key genomic and cellular features underscore its potential as a target for personalized therapy. Future research may focus on integrating MLTS with additional molecular signatures to enhance its clinical application in precision oncology.