Local and systemic immune interactions in malignant gliomas

Glioblastoma (GBM), the most frequent primary intrinsic brain tumor, is without any doubt one of the most devastating diseases known to mankind. GBM are currently being treated with neurosurgical resection followed by radio- and chemotherapy. However, despite this treatment, prognosis for these patients is grim with a median survival of only 15 months and less than 20% 3-year survival rates. Already at diagnosis, GBM cells are infiltrating beyond the visible tumor margins, making complete resection impossible. A therapy-resistant subpopulation of these remaining cells eventually leads to tumor recurrence.

In the past 2 decades, immunotherapy has gained interest as a possible fourth treatment strategy. Immunotherapy is theoretically appealing, firstly because tumor-infiltrating cytotoxic immune cells could target all invasive cancer cells without damaging surrounding normal tissue. Secondly, when immunological memory develops, responses can be long-lasting without the need of persistent administration of therapy. For GBM, these concepts have been shown in preclinical animal models with several vaccination strategies. However, until now no randomized immunotherapy trials have been able to shown survival benefit in humans.

In this research project we aimed to get more insight in the complex interactions between the immune system and GBM, in particular the differences in immune profiles between the local tumor micro-environment and the systemic / peripheral compartment. We hypothesize that better understanding on how the immune system interacts with a developing GBM is essential to develop effective immunotherapy, or at least to understand why successful preclinical therapies don’t work in the clinical setting. For this research we used tumor tissue and blood samples from glioma patients, as well as the orthotopic murine GL261 malignant glioma model in which malignant brain tumors are induced by stereotactic injection of GL261 cells.

In the first part of the research (Chapter 3 – Research paper 1), we focused on the immune checkpoint molecule PD-1 as a contemporary paradigm target for current immunotherapy in several cancers. PD-1 is a surface protein present on activated T-lymphocytes that, after binding with its ligand PD-L1, leads to a negative feedback signal. This pathway is being used by many cancers to suppress infiltrating lymphocytes, and blockade of PD-1 has been shown to reactivate anti-tumor T cells and prolong survival in advanced melanomas and other cancers. Using flow cytometry, we show a strong upregulation of PD-1 on tumor-infiltrating CD4+, CD8+ and regulatory T cells. This upregulation was not restricted to GBM, but also present in lower grade 2 and 3 gliomas. Moreover, we showed that PD-1 expression on circulating T-lymphocytes is not different between patients and healthy volunteers. In the GL261 model we found a similar upregulation on brain-infiltrating T cells, and markedly prolonged survival with a CD8+ driven T cell response after treatment with a PD-1 blocking antibody. This research provides further evidence for trials with PD-1 blocking strategies to treat GBM.

In the second part of the research (Chapter 4 – Research paper 2), we wondered whether local and systemic immune profiles in human GBM could be related to specific biological subgroups of GBM. We studied a large cohort of patients with histologically proven GBM, which were treated with standard therapy and experimental immunotherapy. Tumor samples were subclassified according to DNA methylation profiles into 4 subgroups and a rest group with unclassifiable tumors. With immunohistochemistry, we quantified tumor-infiltrating CD3+ T cells and found these to be significantly more numerous in tumors belonging to the mesenchymal subgroup. On the other hand, IDH tumors had the lowest CD3+ T cell infiltration. This T cell infiltration contrasted with overall survival, which was significantly longer in IDH tumors than mesenchymal tumors. In a multivariate analysis, DNA methylation-based stratification of these tumors remained an independent variable for survival. We also characterized CD68+ myeloid cell infiltration, but found this not to be related to any methylation subgroups or survival. With flow cytometry, we also assessed T cell populations in the blood before and after radiochemotherapy in these patients. We found an overall reduction in lymphocyte count after radiochemotherapy, but relative increase in CD8+ T and CD8+ PD-1+ T cells in all subgroups except IDH tumors. Together, these finding support the hypothesis that histologically identical GBM can be classified into relevant biological subgroups, related to survival and T cell infiltration. Subclassification seems useful in trials as it is plausible that the efficacy of new therapies could be restricted to one or several subgroups.

In the final part of the research (Chapter 5 – Research paper 3), we investigated the effects of subcutaneous implantations of living syngeneic GL261 cells in the murine GL261 malignant glioma model. We see living tumor cell vaccination as the theoretical most basic type of vaccination strategy, able to provide us with more insight in differences of immunity based on specific tumor location and timing of vaccination. The interest in heterotopic living tumor cell implantations originates from the rarity of extracranial metastasis of GBM, despite its local aggressiveness and frequent systemic dissemination of individual tumor cells. We found that subcutaneously implanted GL261 cells are rejected in > 80% of mice, while intracerebral implantations induce lethal tumors in all mice. Furthermore, 79% of mice that rejected subcutaneous tumors were protected against subsequent intracranial tumor implantation (prophylactic model). Immunologically, the brains of prophylactically vaccinated mice were noticed to have an early CD3+ CD8+ T cell influx and decreased regulatory T cells compared to control mice. Interestingly, unlike in control mice, an initial high PD-1 expression in prophylactically vaccinated mice decreased towards basal PD-1 expression 4 weeks after intracerebral tumor implantation. Despite the high protection rate in the prophylactic model, subcutaneous GL261 implantations did not induce any survival benefit in mice with established cerebral tumors, nor did they induce subcutaneous tumors. These findings confirm that the capability of the immune system to control and reject tumor cells depends on the implantation site, and that location-based prophylactic vaccination is possible with living unmodified tumor cells. However, in the presence of an established brain tumor, peripheral living tumor cell implantation isn’t capable anymore to induce a clinically beneficial immunity, nor does it seem to induce palpable tumors. As there seem to be different immunological answers to subcutaneous GL261 implantations in the prophylactic and established tumor models, the local subcutaneous microenvironment after GL261 implantation in these different models is currently further investigated with pathology and immunohistochemistry.