Abstract Background: Inflammatory breast cancer (IBC) is a rare, aggressive subtype of breast cancer with unfavorable prognosis. Despite its distinct clinical presentation and molecular features, the immune landscape of IBC and its role in driving the aggressive phenotype remain poorly understood. This study aimed to characterize the spatial immune landscape of IBC using digital image analysis (DIA), compare it with subtype-matched non-inflammatory breast cancer (nIBC), and evaluate prognostic implications of immune cell composition and localization. Methods: We examined pretreatment tumor samples from 161 IBC and 115 subtype-matched nIBC patients using an AI-assisted DIA pipeline. Immunostainings (H-DAB) were conducted using validated protocols for CD8 (cytotoxic T cells), FOXP3 (Tregs), CD79α (activated B cells), PDL1 (SP124), and CD163 (tumor-associated macrophages, TAMs). Slides were digitized and assessed using VISIOPHARM® software, enabling virtual multiplexing. We quantified DAB+ immune cells and their locations using XY coordinates to calculate density. Spatial colocalization was analyzed using ecological statistics, based on point pattern or quadrant analysis. In point pattern analysis, we counted immune cells within a 30 µm radius around CD8+ cells or tumor cells. For quadrant analysis, we applied the Morisita-Horn Index (MHI) after square tessellation (250×250 µm tiles). The MHI compares cell composition between tile pairs, ranging from 0 to 1, with higher values indicating greater dissimilarity. Associations with clinicopathological features and outcome were assessed using multivariate regression models. Transcriptomic validation was performed using Affymetrix gene expression data and consensusTME deconvolution. Results: IBC showed higher CD163+ tumor-associated macrophage (TAM) infiltration than nIBC. Gene expression data confirmed the IHC findings, and pathway analysis linked high TAM density with inflammatory and proliferative pathways. The spatial distribution of immune cells was prognostically relevant, with high CD8+ T-cell (OR: 0.41, 95% CI: 0.22-0.76, P = 0.004) and low CD79α+ B-cell infiltration (OR: 3.19, 95% CI: 1.68-6.03, P 0.001) correlating with improved overall survival in IBC. The ratio of CD8+ T cells to FOXP3+ regulatory T cells within tumor area was a significant prognostic indicator (OR: 0.34, 95% CI: 0.14-0.83, P = 0.018), whereas absolute densities of CD8+ or FOXP3+ cells alone did not associate with outcome. Beyond density alone, spatial colocalization of B cells with tumor cells emerged as a key prognostic factor. A high tumor colocalization index for CD79α+ B cells was associated with significantly worse survival (OR: 3.59, 95% CI: 1.31-9.84, P = 0.013). Similarly, greater spatial similarity between CD79α+ B cells and tumor cells, assessed via the MHI, was linked to poor outcome (OR: 2.61, 95% CI: 1.19-5.74, P = 0.017). Incorporating spatial B-cell features improved the prognostic model (P = 0.0005). Conclusions: This comprehensive spatial immune profiling effort in a large IBC cohort revealed the immunosuppressive nature of the IBC microenvironment and underscored the central role of TAMs in promoting its aggressive behavior. Spatial context, not immune cell abundance, was most predictive of outcome. High colocalization of CD79α+ B cells with tumor cells associated with poor survival, suggesting a tumor-promoting role. These findings demonstrate the value of spatial immunophenotyping beyond conventional quantification and support exploring TAM- and B cell-targeted therapies in IBC. Citation Format: C. Van Berckelaer, S. Van Laere, C. Vermeulen, M. Kockx, Y. Waumans, K. Mariën, F. Berditchevski, F. Bertucci, P. Vermeulen, L. Dirix, C. Colpaert, G. R. Devi, P. Van Dam. Spatial Immune Profiling in Inflammatory Breast Cancer Reveals a Central Role for Tumor-Associated Macrophages and a Spatial Context-Dependent Negative Prognostic Impact of B Cells 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 PS3-12-19.
Berckelaer et al. (2026) studied this question.