Altered immune signatures in breast cancer lymph nodes with metastases revealed by spatial proteome analyses

Abstract

Background: Metastasis to lymph nodes is strongly associated with reduced survival in breast cancer patients. To increase the understanding on how lymph node metastasis impairs the local immune response in affected lymph nodes, we here studied spatial proteomic changes of critical lymph node immune populations in uninvolved lymph nodes (UnLN) and paired lymph nodes with metastases (LNM) from five breast cancer patients.

Methods: The proteome was analyzed for cortical lymphocyte compartments, subcapsular sinus (SCS) and medullary sinus (MS) CD169⁺ macrophages, using the Digital Spatial Profiling (DSP) platform from NanoString.

Results: Our results identified a stable proteome of SCS CD169⁺ macrophages in LNM, with the exception for downregulation of the anti-apoptotic protein Bcl-xL and FAPα, but a clear reduction in numbers of SCS CD169⁺ macrophages in LNM. In contrast, the proteome of MS CD169⁺ macrophages, B-cell compartments and interfollicular T-cells showed altered immune signatures in LNM, indicating that the decline in SCS CD169⁺ macrophages coincide with a malfunction in the local, anti-tumor immune responses.

Conclusions: The findings from our study support the notion that metastasis to lymph nodes in breast cancer patients modifies local immune responses. These changes may contribute to explain unsuccessful therapeutic responses, and thereby worsened prognosis, for breast cancer patients with LNM.

Keywords: B-cell follicles; Breast cancer; CD169+ macrophages; Germinal centers; Interfollicular T-cells; Lymph node metastasis.

Fig. 1

(A-B) Representative of breast cancer lymph node tissue stained with H&E and IHC (CD169) (A) SCS CD169⁺ macrophages and (B) MS CD169⁺ macrophages of whole lymph node sections from LNM together with zoomed visualizations of areas (DAB and segmentation in QuPath) (C) Quantification of bulk CD169⁺ cells (D) SCS CD169⁺ macrophages (E) MS CD169⁺ macrophages in UnLN compared to LNM using QuPath Software. Error bars indicate SEM, Statistics were performed with paired students T-test, * P < 0.01, ** P < 0.05, *** P < 0.001. N = 5

Fig. 2

Visualization of ROI strategy applied to all samples. Four regions of cells were investigated with GeoMX DSP: Subcapsular CD169⁺ macrophages (ROI #1), Medullary sinus CD169⁺ macrophages (ROI # 2–5), CD45⁺ lymph node follicle regions (ROI #6–10), and the metastasis (ROI #11). ROIs from each lymph node are represented in the table below the figure

Fig. 3

(A) General heat map with supervised clustering performed based on ROI. The color scale represents the log2 change from the geometric mean off all probes in the analysis, 4 outliers were excluded from CD169 ROI. Supervised clustering was performed based on ROI. (B) PCA plot representing all samples ROI illustrated with the first three principal components. Annotations: CD169 (macrophages), CD45 (lymphoid follicles), Metastasis (malignant cells metastasis), Outliers. (C-D) Volcano plots showing the statistical significance versus the magnitude of change in protein expression between five paired LNM (left) versus UnLN (right). Linear mixed model (LMM) adjusted for matched donors or paired T-test adjusted with Benjamini & Hochberg method was applied as statistical tests. Proteins with significant P-value < 0.05 are shown in blue, with proteins showing a log2 fold change above 0.5 plotted in red or green depending on the direction. (C) SCS CD169⁺ macrophage associated proteins with significant difference in expression between paired LNM and UnLN. (D) MS CD169⁺ macrophage associated proteins with significant difference in expression between paired LNM and UnLN

Fig. 4

(A) PCA plot representing all CD45 ROIs illustrated with the first three principal components. Three clusters are shown. (B) General heat map with unsupervised clustering performed based on CD45 ROIs. The color scale represents the log2 change from the geometric mean off all probes in the analysis. Annotations: Neg (UnLN); Pos (LNM); B-cell follicles with GC (Cluster 1); B-cell follicles without GC (Cluster 2); Interfollicular T-cell rich region (Cluster 3). (C) Quantification of bulk CD20⁺ B-cells in UnLN compared to LNM using QuPath software. Statistics were performed with paired students T-test. (D-F) Volcano plots showing the statistical significance versus the magnitude of change in protein expression between five paired LNM (left) versus UnLN (right). Linear mixed model (LMM) adjusted for matched donors was applied as statistical tests. Proteins with significant P-value < 0.05 are shown in blue, with proteins showing a log2 fold change above 0.5 plotted in red or green depending on the direction (D) B-cell follicles with GC (Cluster 1) associated proteins with significant difference in expression between paired LNM and UnLN. (E) B-cell follicles associated proteins (Cluster 1 and 2) with significant difference in expression between paired LNM and UnLN. (F) Interfollicular T-cell rich region (Cluster 3) associated proteins with significant difference in expression between paired LNM and UnLN