Characterization of the immune microenvironment in gliomas and its potential clinical value

Diffuse gliomas constitute the most frequent malignant primary brain tumors in adults. Every year ~600 adultsare diagnosed with diffuse glioma in Denmark. Diffuse gliomas are graded into World Health Organization(WHO) malignancy grades II-IV based on histomolecular features. Glioblastomas, grade IV, account for ~60%of all gliomas and remain one of the deadliest cancer types. Despite standard-of-care treatment with surgery,radiation, and chemotherapy, only 5-10% of patients are alive five years after diagnosis. Especiallyglioblastomas are very heterogeneous and plastic in nature, presenting a cellular and molecular diversity bothamong tumors and within the same tumor, and some glioblastoma subtypes and cellular subpopulations aredescribed as exceptionally aggressive and treatment resistant. Until recently most therapeutic strategies havefocused on the tumor cells. However, over the past decade accumulating data suggest that the tumor immunemicroenvironment plays a crucial role in tumor progression. This has led to the introduction of immunotherapy,such as checkpoint inhibitors, as a strategy to treat and even cure cancer.The predominant immune cell population in glioma is the tumor-associated microglia/macrophages(TAMs). Traditionally, TAMs are categorized into two phenotypes: the M1 phenotype which promotesimmunity and inhibits tumor growth, and the M2 phenotype which is immunosuppressive and promotes tumorprogression. The pattern recognition/scavenger receptor cluster of differentiation 204 (CD204) is a relativelynovel marker of M2-like TAMs. CD204 has been reported to negatively influence patient outcome in manycancer types including glioma, making it a possible target for anti-cancer therapy, however; little knowledgeexists on its independent prognostic value. To date, studies of glioblastomas have unveiled a high level ofresistance to immunotherapy, most likely caused by a lymphocyte-depleted and an immunosuppressivemicroenvironment with deficient immune recognition of tumor cells due to low neoantigen burden. The mostcommon neoantigen in diffuse glioma is mutant isocitrate dehydrogenase (IDH), and increasing evidencesuggests that D-2-hydroxyglutarate produced by neomorphic IDH mutations may suppress the immunemicroenvironment. The overall aim of this PhD thesis was to characterize the immune landscape in gliomasfocusing on 1) the possible clinical value of CD204+ TAMs in glioblastoma and 2) the association betweenIDH mutation and different components of the immune system in diffuse glioma.In manuscript I, the prognostic influence of TAMs and CD204+ TAMs in gliomas was investigated. Tissuesamples from 240 patients with primary glioma were stained using an automated quantitative doubleimmunofluorescence techniques with antibodies against ionized calcium-binding adaptor molecule-1 (IBA1)and CD204 to detect all TAMs and M2-like TAMs, respectively. A pixel-based algorithm was developed toquantify individual expression signals of the makers and their co-expression pattern. We found that the amountof IBA1 and CD204 increased with malignancy grade. In grade III–IV gliomas, high CD204 levelsindependently predicted shorter survival when adjusting for clinico-pathological parameters, while highstaining intensity of IBA1 was associated with better outcome. TAMs were heterogeneously dispersed ingliomas at both an intra- and intertumoral level. The density of TAMs was highest in the tumor core, butCD204+ TAMs were also present at the invasive front, indicating a potential involvement in promotion oftumor migration/invasion. TAMs also accumulated in perivascular and perinecrotic areas where glioblastoma cells with stem-like cell properties are known to reside, and necrosis was more prominent in glioblastomaswith high CD204 expression. We therefore investigated the possible crosstalk between these two cellpopulations by performing double immunohistochemistry using IBA1, CD204 and panel of stem cell-relatedmarkers, e.g. podoplanin and CD133. We found that especially podoplanin+ glioblastoma cells and CD204+TAMs were in close proximity to each other in perinecrotic regions. Additionally, we subtyped a cohort ofglioma using immunohistochemistry and observed that particularly CD204+ TAMs were more abundant intumors with the aggressive mesenchymal phenotype. Immunofluorescent phenotyping of CD204+ TAMsshowed conjunctional expression of proteins related to the M1 and M2 polarization, and CD204+ TAMs alsoco-expressed markers related to tumor aggressiveness. Collectively, these results indicate that CD204+ TAMsmay favor tumor progression by stimulating formation of more aggressive and invasive tumors.In manuscript II, we explored the possible biological role of CD204 more comprehensively using myeloidtranscriptome profiling. Profiling showed that CD204-enriched glioblastomas had a distinct signature withupregulation of genes related to hypoxia, angiogenesis and invasion, including interleukin-6, interleukin-8,programmed death-ligand 1, podoplanin, and CD44. The gene profile correlated with poor prognosis, but wasnot a stronger prognosticator of survival than CD204. Cluster and network analyses of the upregulated genesrevealed that interleukin-6 formed an epicenter with the highest number of gene-gene interactions, andpathway analyses demonstrated an overrepresentation of terms related to e.g. immune activation andextracellular matrix organization, including the tumor necrosis factor (TNF) and interferon signaling pathways.In manuscript III, we investigated the potential role of IDH mutation and its oncometabolite D-2-hydroxyglutarate as an intercellular regulator of the innate and adaptive immune system in glioma. Weevaluated the level of complement deposition and T cell infiltration using immunohistochemistry and a patientcohort with 72 astrocytic gliomas, grade III-IV, stratified according to IDH mutation status. We found thatIDH-mutant tumors had reduced complement activation and decreased numbers of CD4+ T helper cells, CD8+cytotoxic T cells, and forkhead box P3+ regulatory T cells compared to IDH-wildtype gliomas. Ex vivoexperiments demonstrated that D-2-hydroxyglutarate impaired complement-mediated destruction of botherythrocytes and a glioblastoma cell line by inhibiting effective cell lysis and by reducing the level ofopsonization-mediated phagocytosis. In contrast, D-2-hydroxyglutarate neither affected the differentiation ofdendritic cells nor their ability to present antigens. The proliferation, cytokine secretion, and migration of Thelper cells as well as the migration of cytotoxic T cells were inhibited by D-2-hydroxyglutarate. Notably, thedifferentiation of regulatory T cells was stimulated by exposure to D-2-hydroxyglutarate, while theirproliferation and secretion of interleukin-10 was inhibited.In manuscript IV, we explored the association between IDH mutation and presence of cells expressing theco-inhibitory immune checkpoint proteins galectin-9 and/or T cell immunoglobulin and mucin-domaincontaining-3 (TIM3). We established a chromogenic 3-plex immunohistochemistry protocol with antibodiesagainst galectin-9, TIM3, and the tumor marker oligodendrocyte transcription factor 2 (OLIG2) and stainedtumor sections from the patient cohort of 72 astrocytic gliomas described above. Quantification was done usingstereological-based cell counting. We found that the number of TIM3+ cells was significantly lower in IDHmutant compared to IDH-wildtype tumors. In contrast, the number of galectin-9+ cells did not significantly differ. Using double immunofluorescence, most TIM3+ and galectin-9+ cells were found to be IBA1+ TAMs.OLIG2+ tumor cells accounted for ~45-50% of the cells in both IDH-mutant and IDH-wildtype tumors, andOLIG2+ cells rarely co-expressed galectin-9 and TIM3. The presence of TIM3+ galectin-9- cells wereinfrequent, and these cells morphologically resembled TAMs and T cells. Double immunofluorescence showedthat only ~10% of the CD3+ T cell population expressed TIM3. IDH-mutant tumors had lower frequency ofTIM3+ T cells than wildtype tumors, and only 50% of these T cells interacted with galectin-9+ TAMs comparedto the ~85-100% interaction rate observed in IDH-wildtype tumors. In contrast, almost no crosstalk occurredbetween TIM3+ T cells and galectin-9+ OLIG2+ tumors cells. We employed an in silico dataset from to screenfor transcriptional changes comparing the glioblastomas with the highest and lowest TIM3 mRNA levels. Atotal of 75 genes were differentially upregulated in glioblastomas with the highest TIM3 mRNA expression,while no genes were differentially downregulated. Many of the upregulated genes were associated withleukocyte regulation, and chemotaxis. Among the upregulated genes were CD204 and interleukin-6 whoseexpression levels were significantly reduced in IDH-mutant astrocytic gliomas. The upregulated genes wereenriched in various signaling pathways including TNF, complement, and the phagosome pathways.In conclusion, this thesis demonstrates that the prognostic value of TAMs in gliomas does not depend onthe total amount of TAMs, but on the acquired phenotype, as only high levels of CD204+ TAMs predictedshorter survival in patients with grade III-IV glioma. Profiling CD204+ TAMs and CD204-enrichedglioblastomas revealed that CD204 is associated with an inflamed and immunosuppressive milieu with highexpression of several pro-tumorigenic factors especially interleukin-6. These findings suggest that CD204and/or interleukin-6 could serve as targets for reprogramming of TAMs as a means of potentiating the antiglioma response of current and future treatment strategies including other immunotherapies. The results of thisthesis support the hypothesis that the neomorphic activity of IDH mutations conveys a global gliomaassociated immunosuppression affecting innate and adaptive immune responses as well as immune checkpointsignaling. This aspect underlines that IDH mutation status could serve as a useful tool for selecting patientsmore likely to respond to immunotherapy, but also indicates that mutant IDH may be an immunotherapeutictarget, and combining immunotherapy with mutation-specific IDH inhibitors/vaccines could be clinicallyrelevant.