Purpose:
This study aims to characterize clinical and proteomic differences between HR⁻/HER2-low and HR⁻/HER2-zero breast cancers, identify protein biomarkers to enhance HER2 status assessment and inform personalized therapy.
Experimental Design:
41 HR⁻ breast cancer specimens were stratified by HER2 status into HER2-low and HER2-zero groups. Clinical characteristics were compared between groups. Proteomic profiling of tumor samples by LC-MS/MS was followed by different expression, pathway enrichment, and immune microenvironment analyses. Key proteins were identified using LASSO-logistic regression, XGBoost, and SVM, and a multivariate classifier was constructed. Independent validation was performed in 20 additional samples by IHC.
Results:
Patients with HR⁻/HER2-low tumors exhibited less advanced nodal involvement than those with HR⁻/HER2-zero tumors (p = 0.009). Proteomic profiling revealed 94 differentially expressed proteins in HER2-low tumors, which were enriched for inflammatory response, detoxification, and oxidative-stress pathways. Machine learning by integrating LASSO-logistic regression, XGBoost, and SVM methods identified UBA7 and ARHGAP17 as the best discriminators with AUCs of 0.844 (internal) and 0.867 (external). Compared to HR⁻/HER2-zero tumor, UBA7 was higher whereas ARHGAP17 was lower in HR⁻/HER2-low tumors. IHC validation of 20 samples confirmed these expression patterns. Immune analysis revealed that ARHGAP17 expression was positively correlated with ESTIMATE immune scores, while both UBA7 and ARHGAP17 showed significant positive associations with immune checkpoint molecules. In addition, high expression of both markers predicted improved survival.
Conclusion:
HR⁻/HER2-low breast cancers display unique clinical and proteomic signatures. UBA7 and ARHGAP17 serve as robust biomarkers, and our predictive model offers promise for refining HER2 status assessment and guiding personalized therapy.