Abstract
<title>Abstract</title> <p> Background Heat shock protein 90α (HSP90α) is implicated in tumor progression, yet its specific diagnostic, prognostic, and biological utility in breast cancer (BC) warrants comprehensive elucidation. Methods Plasma HSP90α, CEA, and CA15-3 levels were quantified in 797 BC patients and 764 benign controls. Diagnostic performance and prognostic associations were evaluated using ROC, Kaplan–Meier, and Cox regression analyses. The therapeutic monitoring value was assessed in 85 advanced BC patients undergoing chemotherapy. Complementary bioinformatics analyses investigated HSP90AA1 expression, DNA methylation, and immune infiltration. Functional assays in T-47D cells verified the biological impact of HSP90. Results Plasma HSP90α was significantly elevated in BC patients compared to controls ( <italic>p</italic> < 0.05) and positively correlated with advanced clinical stage, nodal involvement, and distant metastasis. While HSP90α alone exhibited high specificity, its combination with CEA and CA15-3 maximized diagnostic accuracy. Longitudinally, plasma HSP90α levels significantly decreased in patients achieving clinical benefit from chemotherapy, confirming its value in monitoring treatment response. Multivariate analysis identified elevated plasma HSP90α as an independent risk factor for poor Overall Survival. At the tissue level, HSP90AA1 was upregulated and served as an independent prognostic factor. Mechanistically, HSP90AA1 expression negatively correlated with DNA methylation, enriched in cell cycle and RNA transport pathways, and significantly associated with immune cell infiltration. In vitro, HSP90 knockdown suppressed proliferation and migration, whereas overexpression promoted these malignant behaviors. Conclusion Circulating HSP90α serves as a robust biomarker for diagnosis, prognostic stratification, and chemotherapy monitoring in breast cancer. Integrated with its critical role in tumor proliferation and immune modulation, HSP90 represents a promising therapeutic target and clinical indicator. </p>