Abstract Purpose: An immunosuppressive tumor microenvironment (TME) contributes to therapeutic resistance in breast cancer. Immune checkpoint inhibitors (ICIs) combined with chemotherapy have emerged as a promising strategy to overcome this barrier. Understanding immune regulation within the TME is essential for developing effective biomarkers and therapies. Our previous work suggested an immunomodulatory role for ZAG (AZGP1), with both clinical and in vitro findings consistently indicating its immunosuppressive activity, potentially contributing to therapeutic resistance. This study evaluates the relationship between ZAG expression and response to neoadjuvant chemotherapy (NAC) in triple-negative breast cancer (TNBC). Methods: To determine the cellular source of AZGP1, we reanalyzed a public TNBC single-cell RNA-seq dataset using Seurat v5. 2. 1. UMAP clustering was used to annotate cell types. Feature plots and violin plots were employed to visualize cell type-specific AZGP1 expression. Gene set enrichment analysis (GSEA) was performed using bulk transcriptomic data from four NAC-treated TNBC cohorts to explore pathways associated with pathological complete response (pCR) and AZGP1 expression. ROC analysis was used to assess the predictive value of AZGP1 for pCR and to determine optimal cut-off values. For validation, we analyzed NAC-treated TNBC cases at our institution. Pre-treatment core needle biopsy (CNB) samples were evaluated for tumor-infiltrating lymphocytes (TILs) based on International TILs Working Group criteria. ZAG expression was assessed by immunohistochemistry. The ZAG score was defined as the percentage of tumor cells showing moderate-to-strong staining. ROC analysis was also performed to evaluate the ZAG score’s predictive value for pCR (ypT0/isN0, ypT0N0), overall ypN0, and ypN0 in cN1 cases. Results: Single-cell RNA-seq analysis confirmed that AZGP1 expression was restricted to tumor epithelial cells. GSEA revealed that pCR was associated with enrichment of immune-related gene sets such as ALLOGRAFTREJECTION and INTERFERONGAMMARESPONSE, while AZGP1 expression negatively correlated with these pathways. In four public TNBC cohorts, AZGP1 showed modest predictive value for pCR (AUC 0. 557-0. 599; not significant), but high-AZGP1 groups had significantly lower pCR rates in three cohorts (p 0. 05). In our institutional cohort of 51 cases (94. 1% stage II-III, mean age 57. 2), 86. 3% received anthracycline-taxane regimens. Pathological CR (ypT0/isN0) was achieved in 31. 4% of cases. ZAG scores were significantly higher in tumors with low TILs (p 0. 05). High ZAG expression was associated with poorer response to NAC, particularly in terms of nodal clearance in cN1 cases (AUC = 0. 748, p 0. 05). The ZAG-high group showed significantly lower rates of pCR (ypT0N0), overall ypN0, and ypN0 in cN1 cases (all p 0. 05). Conclusion: ZAG (AZGP1), shown to be tumor-derived via single-cell RNA-seq, acts as a paracrine immune modulator in TNBC, supporting its validity as a tumor-intrinsic marker in bulk transcriptomic analyses. High AZGP1 expression correlated with suppressed immune gene programs and reduced chemotherapy sensitivity. Our data confirmed that elevated ZAG expression predicts poor NAC response and reduced nodal clearance, particularly in cN1 cases. ZAG represents a promising tumor-intrinsic biomarker and potential immunotherapeutic target in TNBC. Citation Format: T. Hanamura, K. Yokoyama, S. Takahashi, H. Kiyohara, S. Nakagawa, R. Ishida, M. Terao, T. Okamura, N. Kumaki, A. Katayama, S. Kurozumi, N. Niikura. Tumor-derived ZAG expression is associated with immunologically cold tumors and poor response to neoadjuvant chemotherapy in TNBC abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32 (4 Suppl): Abstract nr PS4-01-29.
Hanamura et al. (Tue,) studied this question.