Spatial phenotypic profile of CD163⁺ cells in tumor-rich and tumor-sparse regions. (A) Tissue biopsies from 100 patients were evaluated with the GeoMx DSP technology and stained for CD20 (red), CD3 (light blue), CD163 (yellow), and Syto13 (dark blue) to identify MCL cells, T cells, and macrophages, respectively, and to measure the expression of 63 target proteins. Segmented and selected cells for which the protein quantification was measured are indicated as follows: CD20⁺ cells are marked in orange, CD3⁺ cells are marked in turquoise, and CD163⁺ cells are marked in pink. The biopsies from patients with MCL were then categorized into 2 groups, namely CD163-rich regions (n = 56) in which CD163⁺ cells could be sampled (at least 20 cells) and CD163-sparse region (n = 44) in which low or no CD163 infiltration was detected. The molecular comparisons of CD20⁺ cells and CD3⁺ cells using these groups are referred to as CD163-sparse vs CD163-rich TIMEs. Patients with CD163-rich MCL TIMEs were subsequently stratified based on the sampling of CD163⁺ cells in tumor-sparse regions (n = 17) or in tumor-rich regions (n = 24). The molecular comparison of CD163⁺ cells in these 2 tissue regions is referred to as tumor-sparse vs tumor-rich. (B) Significantly deregulated proteins among the 63 targeted proteins are visualized by plotting the LMM coefficient (x-axis) vs the false discovery rate (FDR) values (y-axis) in which proteins (marked in red) in the right part of the plot are higher in CD163⁺ segments collected in tumor-rich regions, and proteins (marked in pink) higher in CD163⁺ cells in tumor-sparse regions are shown to the left. To account for different numbers of segments collected per patient, a LMM with patient ID and tissue included as random effects were used to assess both the significance and relevance of each marker. Significantly differentially expressed proteins (FDR < 0.1) are indicated by name. (C) Heat map representation of the significant differential protein expression between CD163⁺ cells in tumor-rich and tumor-sparse ROIs. Data are normalized by column and protein intensities are displayed as colors ranging from red to blue as shown in the key. Rows are clustered based on spatial localization and columns are clustered using Ward error sum of squares method. Coefficients and FDR values are listed in supplemental Table 4.