Distribution and prognostic impact of microglia/macrophage subpopulations in gliomas

Abstract

While the central nervous system is considered an immunoprivileged site and brain tumors display immunosuppressive features, both innate and adaptive immune responses affect glioblastoma (GBM) growth and treatment resistance. However, the impact of the major immune cell population in gliomas, represented by glioma-associated microglia/macrophages (GAMs), on patients' clinical course is still unclear. Thus, we aimed at assessing the immunohistochemical expression of selected microglia and macrophage markers in 344 gliomas (including gliomas from WHO grade I-IV). Furthermore, we analyzed a cohort of 241 IDH1R132H-non-mutant GBM patients for association of GAM subtypes and patient overall survival. Phenotypical properties of GAMs, isolated from high-grade astrocytomas by CD11b-based magnetic cell sorting, were analyzed by immunocytochemistry, mRNA microarray, qRT-PCR and bioinformatic analyses. A higher amount of CD68-, CD163- and CD206-positive GAMs in the vital tumor core was associated with beneficial patient survival. The mRNA expression profile of GAMs displayed an upregulation of factors that are considered as pro-inflammatory M1 (eg, CCL2, CCL3L3, CCL4, PTGS2) and anti-inflammatory M2 polarization markers (eg, MRC1, LGMN, CD163, IL10, MSR1), the latter rather being associated with phagocytic functions in the GBM microenvironment. In summary, we present evidence that human GBMs contain mixed M1/M2-like polarized GAMs and that the levels of different GAM subpopulations in the tumor core are positively associated with overall survival of patients with IDH1R132H-non-mutant GBMs.

Keywords: glioma; glioma-associated microglia and macrophages; immune polarization; tumor microenvironment.

Figure 1

GAMs display a mixed M1/M2 immune phenotype. A. Heatmap showing the expression profiles of 979 genes that were differentially expressed between GAMs from 6 GBMs (n = 5 IDH1R132H‐non‐mutant, n = 1 IDH1R132H‐mutant) and reference microglia (from three pooled samples from normal WM) based on an absolute Log2 Fold‐change >= 2. The expression pattern was similar among the different GAM samples and clearly different from the reference. Hierarchical clustering was performed using the Ward’s minimum variance method. Red indicates a high raw intensity in the microarray, gradually decreasing, indicated in green. B. The most significantly regulated molecules of immune polarization in GAMs vs. WM microglia are depicted (FC: fold‐change GAM/WM). C. The mRNA expression profile of selected M/M polarization markers differentially expressed in the microarray (CCL2, TLR2, IL1B, MRC1, CD14, CD163, PTGS2, MSR1, CD163L1) was assessed via qRT‐PCR in GAMs of eight GBM patients (blue, n = 7 IDH1R132H‐non‐mutant, n = 1 IDH1R132H‐mutant) and one patient with IDH1R132H‐mutant anaplastic astrocytoma WHO grade III (gray) in comparison to the median expression of the respective genes in WM microglia (n = 3, yellow). D. In the Volcano plot, the pfp (−log10) vs. gene expression ratio (log2) of all features present on the microarray is shown. Blue and red correspond to significantly (pfp < 0.01) down or upregulated genes. Upregulated M1 or M2 genes are highlighted in green or purple triangles.

Figure 2

Different M/M markers show a heterogeneous distribution in astrocytomas. A. Representative immunohistochemistry stainings for Iba1, CD68, CD163 and CD206 in normal appearing white matter (NAWM) and in patients with astrocytomas of WHO grade I–IV (original magnification: 20×, scale bars = 100 µm). B. Iba1, CD68, CD163 and CD206 expression (see also Table 1) was statistically assessed in IDH1R132H‐mutant and ‐non‐mutant astrocytomas WHO grade I–IV using the nonparametric Wilcoxon’s test. Box and Whisker plots for positive cells (in %) are depicted. P‐values were indicated (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001). For adjustment of the P‐values because of multiple testing, we used the method of Bonferroni–Holm. Only significantly different expression levels between the different entities were highlighted. Statistical analysis was performed using JMP 14.0 software (SAS).

Figure 3

High levels of CD68‐, CD206‐ and CD163‐positive GAMs in IDH1R132H‐non‐mutant GBM patients are associated with better survival. A. Representative immunohistochemistry stainings for Iba1, CD68, CD163 and CD206 as well as MHCII, CD45, CD15 and MPO in either the vital tumor core or the perinecrotic tumor area (necrosis = red asterisk) of selected GBM patients (original magnification: 4×, scale bars = 100 µm; inserts = 40×). Red arrows indicate a predominant perivascular localization of CD206‐positive GAMs. B. Kaplan–Meier survival curves of IDH1R132H‐non‐mutant GBM patients were obtained performing median split for the different subpopulations of Iba1‐, CD68‐, CD163‐ and CD206‐positive GAMs (respected median splits in % depicted). Curves were compared by log–rank and Wilcoxon’s test (P‐values depicted).

Figure 4

High levels of Iba1‐positive GAMs in the vital tumor core of GBM PDOX models are associated with better survival. A. Representative immunohistochemistry stainings for Iba1 in GBM patient‐derived orthotopic xenografts (PDOX) showing distinct GAM infiltration levels (original magnification: 20×, scale bars = 100 µm). B. Parametric survival analysis of tumor‐bearing mice (n = 26, corresponding to 12 PDOX models, 1–3 mice per model) shows a significant association of higher levels of Iba1‐positive GAMs with longer survival (P = 0.0008). C. Kaplan–Meier survival curves were obtained performing median split for Iba1 levels (high expression >12% Iba1‐positive GAM, low expression ≤12% Iba1‐positive GAM). Curves were compared by log–rank and Wilcoxon’s test (P‐values depicted). D. PDOXs displayed heterogeneous GAM distribution patterns (original magnification: 10×, scale bars = 50 µm). Kaplan–Meier survival curves of Iba1‐positive GAMs in E. infiltration zone (IZ) and in F. normal appearing brain tissue (NAB) of PDOXs were obtained performing median split [in (E/F): high expression >14.5%/10% Iba1‐positive GAMs; low expression ≤14.5%/10% Iba1‐positive GAMs] levels. Curves were compared by log–rank and Wilcoxon’s test (P‐values depicted).

Figure 5

Ingenuity pathway analysis (IPA) reveals molecular pathways and processes involved in the immunological polarization profile of GAMs. A. The top five ranked IPA canonical pathways and B the top ten ranked diseases and disorders detected by Ingenuity in the dataset of the Affymetrix GeneChip Human gene 2.0 ST array (GAMs compared to normal WM microglia) are depicted. The threshold (dashed line) corresponds to a P‐value of 0.01.

Figure 6

Distribution of GAM subpopulations. The figure illustrates the heterogeneous distribution of different GAM subtypes in the microenvironment of IDH1R132H‐non‐mutant GBMs. Iba1‐, CD68‐, CD206‐ and CD163‐positive GAMs were quantified by IHC in the vital tumor center, the infiltration zone and the normal appearing brain tissue (see Figure S5 for statistical analyses).