Central nervous system (CNS) tumors are children’s and teenagers’ second most common malignancies. Medulloblastoma is the most prevalent and deadliest in pediatric patients (1). Medulloblastoma is histologically classified into five groups: classic, desmoplastic/nodular, extensive nodularity, large cell, and anaplastic (2). The latest World Health Organization (WHO) classification further divides medulloblastoma into four molecularly defined subgroups: WNT-activated, SHH-activated TP53 wildtype, SHH-activated TP53 mutated, and the non-WNT/non-SHH, which comprises two different subgroups numerically named “3” (G3) and “4” (G4) (3). These molecular subgroups are associated with particular genetic and clinical features, leading to specific prognostication. (1, 3–5).

The survival of medulloblastoma patients has increased in the last decade, mainly due to improvements in treatment strategies such as maximum safe surgical resection, craniospinal radiation in patients older than three years, and chemotherapy. However, the survival for high-risk patients remains low; many patients do not respond to therapy, and most suffer from severe side effects, mainly because of radiation effects on the brain (1, 6). Therefore, there is an urgent need to develop new, more effective, and less toxic therapies (6).

Cancer immunotherapy using immune checkpoint (IC) blockade has shown promising results in treating solid tumors (7, 8). Currently, PD-1, PD-L1, and CTLA4 are the best-established targets, already having FDA-approved monoclonal antibodies for the treatment of many different cancer types (9; 10). However, applying a single immune checkpoint blocker for these targets in CNS tumors, particularly medulloblastoma, did not attain the expected results in clinical trials, particularly in pediatric patients (11–13). Clinical trials using PD-1 inhibitors such as pembrolizumab and nivolumab have shown minor clinical significance in patients with CNS tumors, including in medulloblastoma (11–13).

Alternatively to the currently actionable checkpoints mentioned, other immune checkpoints could become attractive targets of blockage, such as B7-H3, Tim-3, IDO, CD47, LAG-3, TIGIT, and PVR (14–16). Additionally, the knowledge of the non-canonic checkpoints is still poorly described in medulloblastoma and could represent new promising targets for therapy (9, 17). Hence, the immune checkpoint expression profile characterization in medulloblastomas is of utmost importance and could provide crucial cues for more effective and appropriate immune checkpoint blockade-based immunotherapy (10, 17).

The present study aimed to characterize the immune checkpoint profile in medulloblastomas, assess whether there is a distinct profile among molecular subgroups, and evaluate its prognostic outcome.

Eight of the 88 medulloblastomas evaluated by nCounter were excluded due to low housekeeping counts, leading to a final number of 80 high-quality samples for the gene expression analysis. The cases were molecularly divided into WNT (n=13, 16.2%), SHH (n=39, 48.8%), Group 3 (n=10, 12.5%), and Group 4 (n=18, 22.5%). The major clinicopathological features are summarized in Table 1 .

The MSI status was successfully analyzed in 56 (70%) cases, and all of them were classified as microsatellite stable (MSS).

In this study, we characterized the mRNA levels of 19 immune checkpoints associated genes by nCounter in a molecularly characterized series of 80 Brazilian medulloblastomas. We found the absence or shallow mRNA levels of most ICs evaluated, including the currently clinically actionable targets PDCD1, CD274, and CTLA4. At variance, we found higher mRNA levels of CD24, CD47, CD276, and PVR genes, and when compared with non-tumor tissue, CD24 and CD276 were significantly overexpressed in the medulloblastomas. These results were further validated in silico, showing the association of high CD276 expression with a worse outcome of medulloblastoma patients from the WNT and Group 4 molecular subgroups, and high levels of CD24 associated with worse survival in Group 3.

The response to currently actionable immune checkpoint blockade depends on different factors such as immunogenicity and the target’s expression levels. It has been reported that medulloblastoma has a low mutational burden and, therefore, is less immunogenic than other types of tumors, which makes it challenging to target with conventional immunotherapies (37–39). One mechanism associated with a high mutation burden is the presence of MSI (40). The MSI analysis of a subset of our series showed the absence of MSI. These results align with previous studies that reported the absence or very low frequency of MSI in medulloblastomas (41, 42). These characteristics may modulate the tumor’s poor response to PD-1, PD-L1, and CTLA4 blockade (43).

In the present study, we also found the absence or very low mRNA levels of PDCD1, CD274, and CTLA4 in both Brazilian and Cavalli cohorts. Moreover, following the mRNA levels, immunohistochemistry for PD-L1 showed no protein expression. Our results are in accordance with other studies that evaluated PD-1 and PD-L1-related gene expression in medulloblastoma (44). Consistent with our findings, Hwang and coauthors reported the absence of PD-L1 expression in 28 medulloblastomas analyzed (45). Moreover, Vermeulen et al. associated the low PDCD1 levels in the tumor with low lymphocytic infiltration (46). Overall, these findings lower the expectations of initial immune checkpoint inhibitors (ICI) such as pembrolizumab and nivolumab (11, 12). Additionally, enrolling medulloblastoma patients in clinical trials for the abovementioned ICIs lacked appropriate case selection or population size (9). Nevertheless, clinical trials, such as NCT02359565 and NCT03173950, which target PD-1 with pembrolizumab and nivolumab, are recruiting patients with CNS tumors, including medulloblastoma, with a high mutational burden (e.g., rare constitutional MMR-deficiency syndrome – CMMRD) and recurrent or refractory tumors (47).

The combination of different immunotherapy agents is another possible approach for CNS tumors, such as nivolumab and ipilimumab, in treating high-grade pediatric CNS (The NCT03130959 clinical trial). The trial reported delayed disease progression in patients with combined therapy compared to nivolumab monotherapy submitted patients. However, based on our results, the levels of both targets are very low, suggesting limited results of these approaches for the vast majority of medulloblastomas.

The refractory response to PD-1/PD-L1 and CTLA4 blockade by immunologically “cold” tumors like CNS, pancreatic, or prostate cancer gave urgency to finding alternatives to these targets that are mainly dependent on an active immune response (48). Immune checkpoints like LAG-3, TIM-3, PVR, B7-H3, CD47, and IDO1 have been gaining attention as possible substitutes for the initial targets (15). Therefore, we performed the profile of the 19 most preeminent immune checkpoints in our study. We found high mRNA levels in four immune checkpoints related genes, namely CD24, CD276 (B7-H3), CD47, and PVR, and when compared with non-tumor tissue, we confirmed the selective higher tumor expression of CD24 and CD276. Notably, there are two ongoing clinical trials targeting B7-H3. The NCT04167618 uses an anti-B7-H3 monoclonal antibody (omburtamab) in combination with radiotherapy, and the NCT04743661 uses omburtamab combined with radiotherapy, chemotherapeutic agents like irinotecan or temozolomide and bevacizumab (anti-VEGF). However, none of these clinical trials established the levels of B7-H3 as eligibility criteria nor focused on the molecular subgroup. The function of B7-H3 was first thought to be immuno-stimulatory, but further associated with several immuno-inhibitory and tumor progressing functions, mainly by its effects on lymphocytic cells, invasion, and angiogenesis (49). Our results are similar to studies that measured the levels of CD276 expression in different brain tumors, including in medulloblastoma samples and cell lines (50, 51). The analysis performed on the Cavalli et al. microarray dataset and additional dataset comprising 1350 medulloblastomas and 291 normal brain samples corroborate the high mRNA levels of CD276 across all subgroups and a significantly higher expression than in normal brain tissue. Based on these results, CD276 (B7-H3) is a strong candidate for future blockade approaches in medulloblastoma.

The higher expressed immune checkpoint-related gene identified in the Brazilian and the Cavalli series was the CD24. Similar results were reported by Robson et al. showing high mRNA and protein expression in all medulloblastoma subgroups except for WNT (52). The CD24 expression profile in the Cavalli et al. cohort is similar to our Brazilian series, also displaying lower mRNA levels in the WNT molecular subgroups. We further observed in our cohort and in a large dataset of medulloblastomas and non-tumor brain tissue that CD24 expression is significantly higher in all molecular medulloblastoma subgroups except WNT, which showed similar expression levels. Additionally, high CD24 expression was associated with worse survival in the early years in Group 3. The CD24 primary function is to inhibit macrophage phagocytic capacity by binding Siglec-10 (28). Additionally, CD24 was suggested to be a tumor-initiating cell marker and has been described with additional functions as an adhesion molecule important for invasion and migration, as well as being capable of activating signaling pathways involved in tumor cell proliferation (52–54). Regarding the tumor immune microenvironment, Bockmayr et al. reported that WNT, SHH, and a subset of Group 4 medulloblastomas present higher myeloid cells infiltration in the tumor and identified an immunosuppressive environment in Group 3 and a subset of Group 4 medulloblastomas as mediated by anti-inflammatory cytokines and immune checkpoints such as TGFβ1, PD-L1, and CTLA4, yet, PD-L1 was absent from their sample analysis (55). Given the higher myeloid infiltration in the SHH and in a subset of Group 4 subgroups, the blockade of CD24 may be a promising approach to induce a more robust anti-tumoral response. Regardless, the CD24 role in medulloblastoma should be further investigated as this may be a promising target for the blockade approach in this tumor.

Another overexpressed immune checkpoint was CD47, inhibiting macrophage phagocytic capacity by binding SIRP-α (28, 56). In both Brazilian and in silico datasets, the WNT subgroup had significantly higher mRNA levels of CD47 compared to the other medulloblastoma molecular subgroups. Of note, CD47 was also highly expressed in non-tumor brain tissue. The CD47 biological impact in medulloblastoma was investigated by Gholamin et al., that reported high expression levels of CD47 in medulloblastoma datasets and further showed in mice models that targeting CD47 by the monoclonal antibody magrolimab resulted in enhanced phagocytic capacity and tackled tumor primary and metastatic capacity in a model of Group 3 medulloblastoma (57).

The PVR (poliovirus receptor) is a stimulatory and inhibitory immune checkpoint when binding to the costimulatory receptor DNAM-1 and inhibitory receptors TIGIT and CD96, respectively, in immune cells (58). This features grants PVR with an immune regulatory capacity that, in the context of the tumor microenvironment, is often switched to a more anti-inflammatory and pro-tumoral activity (58). In our study, PVR expression exhibited distinct levels among molecular subgroups and non-tumor brain tissue. Therefore, additional studies are warranted to clarify the PVR role in medulloblastoma’s biology and therapy.

We further assessed the association between immune checkpoint expression levels and patients’ overall survival for the four expressed genes. We could observe statistical differences in the Cavalli dataset, which showed an overall worst survival probability in WNT and Group 4 patients with higher CD276 mRNA levels. Similar findings were reported in other solid cancer types like non-small cell lung or pancreatic cancer (59). Moreover, high CD24 mRNA levels were associated with poor survival in the first years following diagnosis in Group 3 medulloblastoma patients.

Altogether, CD276 and CD24 are the most attractive immune checkpoints to be targeted in medulloblastoma, given their higher expression in tumors than in non-tumor tissue and their association with worse prognosis. Notably, a nCounter approach for simultaneous and accurate analysis of several immune checkpoint levels can constitute a putative companion diagnostic test for patient immunotherapy selection. Nevertheless, further pre-clinical and clinical studies are needed to confirm these targets’ biological significance and therapeutic impact. One future issue to be addressed is the well-known poor permeability across the blood-brain barrier (BBB), which constitutes a major challenge in advancing medulloblastoma systemic therapy, including immunotherapy (60). In the previously mentioned omburtamab trials (NCT04167618 and NCT04743661), the drug was administered intraventricularly, overcoming, in this way, the BBB obstacle.

Ultimately, we successfully evaluated relevant immune checkpoint targets in the routine FFPE medulloblastoma biopsies. Our study corroborated the lower mRNA levels of the current targeted immune checkpoints, PD-1, PD-L1, and CTLA4. Notably, the immune checkpoints genes coding for CD24 and B7-H3 were highly expressed in medulloblastoma, suggesting that they can constitute more suitable targets for an immune-targeted approach.

The all panel expression data is deposited in the “Gene Expression Omnibus” under the accession number GSE223606. It can be found in the following link https://0-www-ncbi-nlm-nih-gov.brum.beds.ac.uk/geo/query/acc.cgi?acc=GSE223606.

RM and DM; Design of study, organized the database, performed the statistical analysis, results discussion, and drafted the manuscript. LS and LL; Design of study, data generation, results generation, and manuscript review. FP; Data Generation, results discussion, and manuscript review. IS, GT, FS, and LN: Pathological review of tumor samples, data generation results discussion, and manuscript review. CJ and BM: Medical reports analysis, results discussion, and manuscript review. RR: Conception, project coordinator, results discussion, manuscript writing, and review. All authors contributed to the article and approved the submitted version.