Use of spatial multiomics to identify endosomal trafficking and S100A8/A9, and fibronectin-driven immune remodeling programs with T-DXd resistance in breast cancer brain metastases.

Abstract

e15199 Background: Trastuzumab deruxtecan (T-DXd) has transformed the treatment of breast cancer brain metastases (BCBM). However, approximately half of patients exhibit acquired resistance, and no predictive biomarkers have been clinically established. Since antibody-drug conjugate (ADC) efficacy depends on the unique brain microenvironment and intracellular processing, we applied spatial multiomics to resolve tumor- and niche-specific resistance mechanisms. Methods: Pre-treatment BCBM specimens from T-DXd responders (R) and non-responders (NR) were profiled using Bruker GeoMX for spatial proteomics and whole transcriptome analysis (n = 2 R, 2 NR; 5-10 ROIs per specimen). Additionally, Bruker CosMX single-cell spatial transcriptomics was employed for pathway enrichment and cell crosstalk analyses (n = 3 R, 3 NR; 200 ROIs per specimen). Results: GeoMX analysis identified fibronectin as a potential spatial biomarker—the significantly stronger spatial correlation with T cell markers (CD3, CD4, CD8) in NR (p < 0.05), suggests a role in T cell exclusion. Both GeoMX and CosMX identified S100A8 and S100A9 as highly differentially regulated. Interestingly, their clinical significance was cell type-dependent: · Luminal A cancer cells in NR-enriched clusters exhibited S100A8/A9 upregulation (FDR q < 1 × 10⁻⁶), accompanied by increased phosphorylation of NF-κB pathway components. · In responders, S100A8/A9 expression was localized to monocyte-dominant clusters (FDR q < 1 × 10⁻⁶). Spatial Single Cell Crosstalk (S2C2) modeling predicted S100A8/A9-TLR4 interactions between monocytes and astrocytes, suggesting that innate immune–glial crosstalk may prime the niche for therapeutic sensitivity. Furthermore, responder HER2+ cancer cells were significantly enriched for ER-to-Golgi transport, vesicular trafficking pathways, and macroautophagy programs (NES > 3.4, FDR q < 0.01 for all). These findings are mechanistically consistent with the requirements of efficient intracellular processing and lysosomal cleavage for ADC payload release. Conclusions: To summarize, two plausible mechanisms of T-DXd resistance in BCBM were identified: 1) remodeling of the immune-glial niche associated with fibronectin and S100A8/9, and 2) impaired intracellular trafficking. Hence, the novelty of this research lies in uncovering potential biomarkers beyond the HER2 and topoisomerase I genes. Future work will explore these spatial signatures as high resolution biomarkers to stratify patients and nominate strategies to restore ADC efficacy in the CNS.