Similar to other cancers, MB presents a high heterogeneity, both in terms of histology and molecular characteristics. Advances in tumor biology and genetics have led to a classification system that improves both diagnosis and treatment strategies. Originally, MB was categorized by histological properties into three main subtypes: classic (C), desmoplastic/nodular (DN), and large cell/anaplastic (LCA) depending on cellular phenotype (53–57).

The classic MB group represents the most common subtype, accounting for 66–72% of cases. Tumoral cells typically do not grow significantly in volume and rarely show structural alterations such as desmoplasia or nodules exhibiting a high nucleus-to-cytoplasm ratio, rounded nuclei, and significant mitotic and apoptotic activity. The DN group is much less frequent than the classic subtype, representing about 15% of all the cases. It is often associated with a favorable prognosis and a distinct molecular subtype (SHH), facilitating a quicker and more precise diagnosis. MB cells have the ability to deposit collagen in the pericellular space. Cytologically, these cells appear small and round, with characteristic arborizations and pericellular reticulin deposition. This subtype of MB tumors presents a specific variant - MB with Extensive Nodularity (EN) – that primarily occurs in newborns (50, 55). The LCA subtype accounts for approximately 15% of the cases that have more aggressive behavior. It is characterized by nuclear pleomorphism, increased cell volume, and a tendency for cells to cluster. Within this subgroup, the large cell variant is particularly rare, occurring in only 2–4% of cases. Cells show high proliferative and apoptotic activity and are associated with a significantly poorer prognosis compared to other subtypes (54–57).

The 2021 fifth edition of the WHO Classification of CNS Tumors categorizes MBs into four distinct molecular subgroups: WNT-activated, SHH-activated with TP53 wild type, SHH-activated with TP53 mutation, and non-WNT/non-SHH (formerly classified as groups 3 and 4 (50, 58–60). ( Figure 1 ). These subcategories are further segmented into more specific subtypes using advanced methylation profiling, which has identified multiple WNT-, SHH- and non-WNT/non-SHH-associated subtype. This molecular stratification has significantly enhanced the understanding of the biological diversity and clinical variability of MB, which enables more precise classification, prognosis and therapy (57–70). Information about epidemiological, clinical and molecular features of MB molecular subcategories is summarized in Table 1 . It is to be noted that the identification of distinct methylation patterns and histological features continues to drive research into targeted therapies and personalized medicine treatment protocols, aiming to improve outcomes for patients with MB (61, 62). Furthermore, besides the sporadic medulloblastoma, this tumor can occur in association with cancer predisposition syndromes such as colon polyps in Turcot syndrome or basal-cell carcinomas in Gorlin syndrome [reviewed in (46, 47, 73, 74)]. This event should be considered in relation to each molecular subgroup and this knowledge can guide oncologist to perfom the cancer surveillance to diagnose and treat early the medulloblastoma in collaboration with other specialists. This topic is of great relevance for MB as germline mutations are 5-6% and these genetic alteration affect specific molecular pathways leading to tumor development (75).

Like in other cancers, despite advancements in multimodal therapy, MB still remains a challenge due to its high heterogeneity and therapy resistance. To better understand biology and develop targeted therapies, researchers rely on well-characterized cell lines as essential tools for studying tumor pathogenesis, investigating signaling pathways, and screening therapeutic compounds ( Figure 1 ). Several cell lines have been established, each representing different molecular subtypes and biological characteristics. Herein, we present a brief description of the most used cells, which include DAOY, D283-Med and D341-Med, UW228, ONS-76, HD-MB03 and D425-Med and D458-Med (76, 77) ( Figure 1 , Table 2 ).

Several others cell lines have been selected and described in the literature reviewed partly elsewhere (78, and listed in https://www.cellosaurus.org/search?query=medulloblastoma+cell+line). WNT and group 4 MB subtype cell lines are underrepresented compared to the frequency of these MB (78). For instance, the CHLA-01-MED and CHLA-01R-MED have been used to investigate features of primary and metastatic cells of group-4 MB (79) or the MED6 and MED5R of WNT group and MED1 of group 4 to evaluate the function of the MDR1 (ABCB1 molecule) to study drug-resistance mechanisms (80).

The most used and represented in literature is the DAOY cell line. This cell line has been derived from a desmoplastic tumor in a 4-year-old male and categorized within the SHH subgroup (81). It harbors a heterozygous PTCH1 mutation (c.1312G>T; p.G438X), leading to constitutive SMO activation (54, 55, 76, 77). Genomically, it exhibits 9q loss (PTCH1 locus) and 10q gain (MYCN locus), consistent with SHH-MB. DAOY retains wild-type TP53, differentiating it from TP53-mutant SHH MB. These adherent cells form neurospheres under serum-free conditions, expressing CSC markers such as CD133, nestin, and SOX2 (77). DAOY xenografts show desmoplastic histology and it expresses SHH targets (GLI1, MYCN) responding to SMO inhibitors like vismodegib (82). However, resistance via SUFU mutations or G88LI2 amplification can emerge (70, 82, 83). DAOY also displays HDAC1 overexpression, and HDAC inhibitors reduce cell viability (84). Limitations include non-metastatic behavior in xenografts and genomic drift over time (78, 85).The D283-Med was established from a metastatic tumor obtained postmortem from pediatric patient while the D341-Med were from the primitive medulloblastoma tumor at craniectomy (71, 86). These cells present MYC amplification and aggressive behavior (71, 78). Due to their non-adherent, suspension-like growth, these models are widely used to study metastasis and therapy resistance. Both exhibit MYC proto-oncogene amplification, a hallmark of Group 3 MB, correlating with their aggressive phenotype and poor prognosis (71, 87).

D283-Med presents isochromosome 17q (i17q), characteristic of Group 4, while D341-Med often has chromosome 8 gains. Their morphology mimics anchorage-independent growth, facilitating metastasis studies (71, 86). D341-Med xenografts show leptomeningeal spread; D283-Med is less metastatic but grows aggressively. Both express CSC markers (CD133, SOX2, nestin) and show intrinsic resistance to cisplatin and etoposide due to upregulated ABC transporters (e.g., ABCG2) and anti-apoptotic proteins (e.g., BCL2) (78, 86, 88–91). Notably, D341-Med responds to BET inhibitors that suppress MYC transcription (91). However, long-term culture can lead to genomic divergence, necessitating periodic validation. Nevertheless, these cell lines present some indicators of genomic drifts upon prolonged in vitro culture time, which can lead to unwanted clonal selection and genomic divergence from the initial primary tumor patients. This important fact raises the need to perform periodic molecular validations.

The UW228 cell line was derived from a MB recurrence (92). Unlike D283-Med and D341-Med, UW228 cells display moderate differentiation and exhibit a more adherent, epithelial-like morphology with moderate differentiation (78, 92). It carries inactivating PTCH1 mutations and wild-type TP53. Genomic alterations include 9q loss, aligning it with SHH-MB. UW228 expresses intermediate neuronal markers (β3-tubulin, MAP2), distinguishing it from less differentiated models. Cultured in monolayers, it supports studies on differentiation and SHH signaling (93, 94). UW228 shows partial sensitivity to SMO inhibitors and transient suppression of GLI1, with resistance mechanisms involving SUFU downregulation or GLI2 amplification. Resistance to cisplatin and etoposide correlates with elevated ABCB1 expression. It is useful for modeling therapy-naïve SHH MB and differentiation therapy, though limited by low metastatic potential and genomic instability in long-term cultures. This cell line displays a partial sensitivity to SMO inhibitors, with reduced GLI1 expression and transient growth suppression. Nevertheless, some resistance arises via SUFU downregulation of GLI2 amplification (78, 95, 96). Studies have shown that this cell line also shows medium resistance to cisplatin and etoposide. UW228 cell line presents low metastatic potential in vivo, which limits its utility for invasion and tumor dissemination studies. Like other cell lines, it presents some genomic instability over in vitro passages, which fosters low-passage genetic validation.

The ONS-76 cell line was established in 1989 from a 2-year-old Japanese patient with nodular/desmoplastic MB (98). ONS-76 cells display neuronal differentiation when exposed to retinoic acid, making this cell line a suitable model for studying differentiation therapy and SHH inhibitors (98). Integrated in the SHH molecular subgroup, ONS-76 has PTCH1 mutations or deletions, disrupting its inhibitory function on SMO and promoting GLI-mediated transcription (98). These cells retain wild-type TP53 and harbor PTCH1 mutations, supporting SHH pathway activation. ONS-76 expresses neuronal markers (β3-tubulin, GFAP) and undergoes differentiation upon retinoic acid exposure, marked by neurite outgrowth and NeuroD1 upregulation (98, 99). It grows as a 2D monolayer and can form desmoplastic xenografts. ONS-76 is sensitive to SMO inhibitors, and unlike DAOY, rarely develops GLI2-driven resistance, likely due to functional TP53. Retinoic acid reduces proliferation via HDAC inhibition (100). This cell line also shows modest cisplatin resistance. ONS-76 is ideal for SHH-targeted differentiation therapies, though differentiation capacity may decline with extended culturing.

The HD-MB03 cell line, established in 2014 from a MYC-amplified, represents a critical preclinical model for studying high-risk, metastatic MB biology (101). It exhibits MYC amplification, 1q/7 gains, and loss of 10q (PTEN) and 16q. These alterations enhance proliferation and apoptotic resistance (56, 57). HD-MB03 forms floating spheroids and shows robust CSC marker expression (CD133, SOX2, nestin) (102). In vivo, it replicates leptomeningeal and spinal metastases. The cell line is intrinsically resistant to cisplatin and vincristine due to high ABC transporter and BCL2 expression. HDAC inhibitors such as panobinostat disrupt MYC-driven transcription and reduce tumor growth (91, 103). HD-MB03 serves as a strong platform for testing MYC and epigenetic-targeted therapies, though periodic molecular verification is necessary due to potential genomic evolution.

D425-Med and D458-Med were established in the late 1990s from high-risk, recurrent cases, both showing MYC amplification (104). Both have high proliferative and metastatic capacities. D425-Med shows 7q gain and 10q loss (PTEN), driving PI3K/AKT pathway activation, while D458-Med presents i17q and chromosome 8 amplification. Both grow as suspension spheroids and demonstrate strong in vivo metastatic capabilities. D425-Med shows leptomeningeal dissemination in orthotopic models; D458-Med metastasizes to lung and liver via systemic injection (104, 105). D425-Med may harbor MYCN amplification in subclones, enhancing proliferation. These models are vital for studying hematogenous dissemination and high-risk MB mechanisms.

Like in other human tumors, traditional cultures of cell monolayer remain widely used in research, providing a controlled environment for rapid drug screening and genetic manipulation. Still, in vitro culture conditions can influence both phenotype and cell signaling and drug sensitivity to a large extent. Simplistic conventional, also inappropriately called 2D, monolayer cultures with a single layer of cells attached to treated polystyrene have unnaturally low cell densities, lack cell-cell interaction or cell-peri and extracellular matrix interactions, and lack the exhibition of the brain tumor nutrient gradients or physiological levels of oxygen (112). For instance, it has been shown that some MB cell lines including DAOY (SHH), ONS-76 (SHH), D458 (Group-3), HD-MB03 (Group-3), CHLA-01-MED (Group-4) and CHLA-01R-MED (Group-4) showed the growth and metastatization features of MB tumors of each subgroup they belong only when they were cultured in a 3D hyaluronic acid hydrogels but not when cultured in conventional conditions as adherent cells (113). This would suggest that 3D cultures may be used to study MB behavior and drug sensitivity, mimicking better the in vivo physical conditions (113).

Ivanov et al. highlighted the importance of using different 3D culture systems with the relevant tissue architecture and phenotype as well as normal tissues and how the establishment of a collaborative online database linked to distinct cell banks would catalyze preclinical MB research (78, 112–114). The use of 3D spheroid models from cell lines and patient-derived xenografts (PDX) demonstrates having more representative resistance to conventional chemotherapies, e.g. etoposide and cisplatin, in comparison to conventional cultures, and the same 3D experiments were key to identifying hypoxia-induced genes that drive resistance. Brabetz and his co-authors generated PDX-derived 3D organoids retaining genetic and transcriptional heterogeneity of primary tumors that were strategic to demonstrate that group 3 organoids with MYC amplification exhibit invasive growth patterns depending on the 3D extracellular matrix (ECM) (such as collagen) and that venetoclax (a BCL-2 inhibitor) synergizes with chemotherapy, overcoming apoptosis resistance (115, 116). The use of 3D culture assays has become central in research due to their ability to recapitulate key aspects of tumor biology, such as CSC niche enrichment, intra-tumoral heterogeneity preservation, and therapy resistance modeling. Using 3D models, Vinci et al. have identified and demonstrated MYC-dependent metabolic vulnerabilities in Group 3, including sensitivity to glutaminase inhibitors (117). In the past decade, many 3D bio-printed brain tumor models were developed for glioblastoma aiming for the recapitulation of TME and developing better drug screening platforms (118, 119). Still, to the best of our knowledge, there are no studies that developed 3D bio-printed constructs for other brain tumors such as MB. ( Figure 1 ).

Similar to other brain malignancies, the study of MB frequently relies on established cell lines, from murine and/or human origin, and PDX models. Models have provided critical insights into tumor biology and therapeutic response. The heterogeneity of MB has been addressed in vivo through the development of genetically engineered mouse models (GEMMs), through the orthotopic implantation of murine cerebellar progenitor cells and by PDX murine models. This section aims to provide a concise and short overview of the MBs in vivo models employed in research. It is not intended to serve as a comprehensive account of all available models, but rather to offer a general perspective on the principal systems currently utilized in the field. For more exhaustive information on in vivo models applied for MB research, readers are referred to the detailed reviews (120, 121).

GEMMS involve targeted modifications of the murine genome, such as gene knockout (deletion), knock-in (mutagenesis), or transgenic overexpression, enabling the study of tumor biology within an intact immune system and native tissue microenvironment. Unlike transplantation-based systems, GEMMs recapitulate spontaneous tumorigenesis, providing insights into the multistep progression from initiation to malignancy (120, 121). In research, GEMMs have been pivotal in confirming genetic drivers and elucidating the cellular origins of molecular subgroups, particularly SHH and WNT. Despite their high cost, technical complexity, and time-intensive nature, GEMMs remain among the most informative and widely utilized systems in cancer biology (120, 121). This model has been instrumental in validating oncogenic drivers and elucidating the cellular origins of distinct molecular subgroups, particularly SHH and WNT MB. For example, key studies on SHH MB utilized PTCH1 heterozygous mice (PTCH1^+/−), which develop tumors at low penetrance (~20%) following loss of the wild-type PTCH1 allele (122). Subsequent conditional or global knockout models have identified cooperative oncogenic alterations and solidified the granule cell progenitor (GCP) lineage as the cell of origin, with GCP identity being essential for SHH-driven tumorigenesis (123).

In contrast, WNT MB is proposed to arise from dorsal brainstem progenitors derived from the lower rhombic lip, rather than cerebellar compartments. Conditional knock-in models expressing a constitutively active β-catenin variant demonstrated that aberrant WNT pathway activation induces pathological cell accumulation in the brainstem but not the cerebellum, which can be aligned with putative extra-cerebellar origins (124). Additionally, transgenic mouse models overexpressing NMYC in cerebellar tissue have further revealed oncogenic versatility by inducing resembling MB in multiple subgroups, such as groups 3, 4, and SHH. This fact highlights the fact that the context-dependent effects of oncogene activation on pathogenesis.

PDX models have gained prominence in oncology research due to their capacity to closely mirror the biological and histopathological characteristics of the primary tumors from which they originate. In the context of MB, PDXs are typically generated by engrafting freshly resected tumor tissue either subcutaneously or within the cerebellar parenchyma-orthotopic engraftment into immunodeficient mice (125). The genetic diversity of recipient mice—covering inbred, outbred, or hybrid strains—does not preclude successful xenotransplantation. Nevertheless, the fact that these animals are immunosuppressed, a condition needed to allow tumor engraftment excluding the rejection in adult mice, represents a restraint in their use as immunotherapies testing (126). PDX models have been successfully established from all major molecular MB subgroups and have demonstrated stability across multiple passages, although subclonal populations may undergo selection during serial transplantation. We highlight that most of the PDX models have been derived from high-risk cases, suggesting that tumors with more aggressive phenotypes are more amenable to engraft. Also, orthotopic implantation into the cerebellum has been associated with improved engraftment efficiency, particularly for tumors with lower proliferative potential, compared to subcutaneous approaches (127, 128). These types of in vivo models have been instrumental in validating key molecular drivers of pathogenesis and serve as versatile platforms for evaluating a range of therapeutic strategies for preclinical assessment of pharmacologic agents, dosing regimens, delivery routes, and combination treatments, but not so key for cell-based and immunotherapies due to animal immunodeficient environments. ( Figure 1 ).

More recently, both MB cell lines and patient-derived cells (from SHH MB, SHH PDX, the Group 3 MB PDX cell line MB-LU-181) have been transplanted into blastula stage of zebrafish embryos leading to orthotopic MB growth (129) The localization to the hindbrain region of transplanted cells was increased by culturing MB cells in neural stem cell-like medium. This model could be used to test the efficacy of SMO inhibitor sonidegib and an active metabolite of cyclophosphamide (129). The model of zebrafish has been also used for group-3 MB cells (130) to identify the subgroup MB cellular origin (131), to mimic the SHH-MB with specific mutations (132), to generate transcription activator-like effector nucleases TALEN-mediated somatic gene inactivation of CDKN2A/B or RB1 tumor suppressor genes (133). Altogether, zebrafish models can well resemble the growth and the aggressiveness of MB in humans to analyze the effects on MB biology due to the genome editing and the presence of specific mutation and/or activation of oncogenes (130–134) ( Figure 1 ).

It has been reported that NK cells can recognize and kill the MB cells ( Figure 2 ) (160–164). The established cell lines DAOY and D283-Med and the primary cell line 1603-Med from an anaplastic MB can express different levels of ligands for the NK cell activating receptor NKG2D including MICA/B and ULBP3; further, they express nectin-2 and PVR, ligands for the DNAM1 activating receptor, low levels of LFA3/CD58 (a ligand of CD2 antigen express by most lymphocytes) and intercellular adhesion molecule (ICAM) 1, 2 and 3 ligands for the lymphocyte function associated antigen (LFA) 1. More relevantly, the use of specific mAb against NK cell activating receptors such as NKG2D, DNAM1, NKp30 and NKp46 could inhibit the NK cell-mediated cytolysis of these cell lines (161). The expression of ligands for NKG2D has been further confirmed both in immunohistochemistry of MB specimens and in cell lines, together with the relevance of NKG2D and HLA-class I molecules in NK cell-mediated recognition of the MB cell line DAOY (162). This finding indicates that several counter-ligands of NK cell-triggering molecules may be considered suitable targets for the elimination of MB cells in patients (162). The adoptive transfer of IL15-activated NK cells can induce a delay in the growth of the MB DAOY cell line in a subcutaneous xenograft mouse model (160), leading to an increase in the OS rate. The infiltration of these xenografts was characterized by NK cells showing expression of several activating receptors and bearing several markers of cytotoxic cells, including perforin, granzyme, tumor necrosis factor (TNF)α, and interferon (IFN)γ. It is well known that NK cells can recognize tumor cell targets independently by the recognition of HLA-I differently from T lymphocytes (141, 142). In this context, it is essential to be better able to define the role of HLA-I expressed on MB cells, as typically NK cells express potent inhibitory receptors for self-HLA-I, leading to blocking of autologous cell killing (141, 142). It has been reported in a series of 10 MB primary tumors the absence of reactivity with anti-HLA-class I antibodies in immunohistochemistry assays (165). This finding has been confirmed on a large series (n=106) of MB, but in addition, it has been shown that high levels of HLA-I are strongly expressed only in MB with evident anaplastic features in association with MYC expression (166). A role of HLA-I peptide complexes and the endoplasmic reticulum aminopeptidase (ERAP)1 in the regulation of MB cell killing has been reported (167). However, the increase of cell killing upon blocking of HLA-I recognition by NK inhibitory receptors was faint (161, 167). Furthermore, the degree of killing of the polyclonal NK cell populations against the same target (DAOY cell line) used in these reports is markedly different (161, 167). This variability would indicate a variable expression of the activating receptors expressed on NK cells. Thus, the killing of MB cells is the result among positive and negative signals transduced by activating or inhibiting receptors upon binding with the corresponding ligands on MB cells ( Figure 2 ). It is of note that an evident cytotoxic effect was also detected by injecting ¹⁹F-labeled NK cells intratumorally or contralaterally to an orthotopic MB tumor in immunodeficient mice (168). Further, it is of note that NK cells are more infiltrated in patients with a better prognosis (169). Altogether, these findings further support the possibility of using adoptive NK cell transfer as a useful tool to eliminate MB cells and control tumor expansion.

The relevance of some unconventional T cells, such as invariant NK-like T cells, Vδ2 (γδ TCR) T cells, CD1-restricted T cells, mucosal associated invariant T cells (MAIT) and other NK-like T cells, has not been studied in detail with MB cell lines and MB preclinical mouse models (65, 170–175). It has been reported that MB do not express CD1d, besides HLA-I, suggesting a molecular mechanism involved in MB escape from recognition by unconventional T cells (173). Furthermore, the molecular MB SHH subgroup can express elevated levels of CD1d mRNA compared to other MB subgroups (173). Importantly, CD1d-positive MB cells (DAOY and MED8A cell lines) presented glycolipid antigens (α-galactosylceramide) to NK T cells, inducing the production of cytokines such as IFNγ and IL4. Along this line, NKT cells induced remission of orthotopic injected MB xenografts of DAOY MB cells. In addition, the NKT cells in MB patients were present and functional, suggesting the possibility of using these NKT cells to kill MB cells at least in a subset of patients (173). More recently, it has been reported that γδT cells can infiltrate the MB, and in particular the group 4 MB, but there was not a significant correlation between the infiltration of γδT cells and patients OS. Furthermore, EphA2-expressing MB cells trigger Vγ9Vδ2T cell activation while amino bisphosphonates sensitize MB cells but not healthy neuronal cells to Vγ9Vδ2T cell lysis (175). This would indicate that γδT cells can be a useful tool to target MB cells. In this context, we reported that antibodies to tumor antigens conjugated with the aminobisphosphonate zoledronic acid can kill efficiently tumor and stromal cells (176). Thus, we can hypothesize that the use of antibodies to MB antigens conjugated with aminobisphosphonates can be effective in eliminating MB cells, as shown for colorectal carcinoma (176). Finally, it has been reported that six MB cell lines (DAOY, ONS-76, UW228, D341, D425 and D283) can be a good target for protein tyrosine kinase (PTK)7-targeted CAR γδ T cells against MB (177). Altogether, these findings support the idea that besides conventional T cells and NK cells, other effector lymphocytes can be an immunotherapeutic tool to eliminate MB cells ( Figure 3 ).

The antigen-specific immune response could be considered one of the cellular-mediated mechanisms by which MB cells can be eliminated. Briefly, an adaptive immune response should be elicited against neoantigens and/or tumor-associated antigens (TAA) (178). The presence of neoantigens and TAA is also necessary for awaking exhausted tumor-infiltrating lymphocytes (TIL) upon treatment with immune checkpoint inhibitors (ICI) such as anti-CTLA4 or anti-PD1/PDL1 antibodies (179–183). Importantly, the possibility of evoking an adaptive immune response is one of the requisites to plan ICI therapy, and the identification of neoantigens or TAA together with antigen-specific TIL justify the cost of the treatment and predict, at least in principle, the response (183). Conceivably, more are the neoantigens and TAA present together with their immunogenic potency; more probable is that the immune system will control the tumor growth. As reported above, the patients can be subdivided on the basis of subgroup-specific genetic alterations, and more recently it has been shown that it is possible to identify potential tumor antigens, possibly suggesting the development of antigen-directed cellular therapies for MB (184). In detail, it has been developed an algorithm predicting antigens able to trigger an immune response in the context of the patients’ HLA-class I and class II. This point is relevant as it is well-known that the immune response needs both the help mediated by the recognition of CD4+ T cells of MHC-II-restricted peptides and the lytic activity of CD8+ T cells restricted to MHC-I antigens to eliminate tumor cells (185) ( Figure 4 ). The antigens considered include neoantigens, TAA and fusion antigens (184). This pipeline has been named Open Reading frame Antigen aNalysis (ORAN) and it uses the gene expression data for identifying different classes of antigens (184). Importantly, the algorithm can predict putative peptide antigens that indeed trigger an efficient antitumor immune response after vaccination in preclinical glioblastoma models (186). It is of note that ORAN identified that only a subset of the genes mutated in MB could be immunogenic, and this subset was present in about 80% of patients; also, 44% of patients expressed three or more neoantigens. This would indicate that not all the MB patients may benefit from neoantigen-based immunotherapy. About 90% of MB expressed at least one TAA, and a quite high proportion of these patients expressed three TAA. Noteworthy, the TAA prediction showed a strong and better concordance with respect to neoantigens with proteomic data. The overall survival (OS) and the progression-free survival (PFS) of patients in the Group 3 of MB well correlate with the presence of MHC-I and MHC-II peptides of TAA. This work analyzed by RNA-seq 170 cases of MB of which 18 WNT, 46 SHH, 41 Group 3, and 65 Group 4, and it used a data set of microarray technology of 763 MB (64, 184). Moreover, the immune landscape and the pathways for antigen processing and presentation in tumor cells have been studied using up-to-date deconvolution computational methods. This analysis has given some insights on the possibility of identifying the so-called “recurring antigen”. It is conceivable that to design an MB vaccine potentially functional in any patient independently of the molecular subgroup and stage of development, a good immunogenic antigen should be identified. It is of note that MB expressed several private and immunogenic antigens. Except for the SHH-MB subgroup, several TAA were usually expressed. Furthermore, some cancer testis and neurodevelopmental antigens were expressed throughout all the subgroups. In detail, neoantigens from oncogenic driver mutations including CTNNB1, DDX3X, and SMARCA4 and TAA such as NEUROG1 and PIK3R3 could be considered as potential therapeutic targets for immunotherapy. None of these identified potential targets have been validated in experiments showing that it is possible to trigger an immune response upon the use of a personalized vaccine. However, the same ORAN pipeline applied to the murine glioma GL261 cell line identified, after appropriate selection, 192 putative neoantigens and 37 TAA with a predictive immunogenic effect (186). Importantly, it has been developed a gene enrichment platform for the production of tumor open reading frames that are unique (TOFU) specific for tumor antigens. This platform allowed the generation of mRNA by in vitro transcription for immunogenic tumor antigens without the need of providing large tissue samples from patients to obtain these mRNAs, overcoming this bottleneck to produce vaccines. Noteworthy, the TOFU mRNA vaccine was efficient to evoke an antitumor response in a murine model of glioma when loaded into dendritic cells in combination with immune-checkpoint inhibitors (ICI) and/or adoptive cell therapies (72, 187).

It is clear that these novel approaches in identifying the MB neoantigens and TAA support the idea that immunotherapy might be feasible and efficient. Previous observations in MB on the immunogenic property of the fusion proteins composed by the enhancer of polycomb homolog 2 (EPC2) and GULP PTB domain containing engulfment adaptor 1 (GULP1) have shown that this protein can trigger the release of IFNγ by CD8⁺ T cells (188). Also, the finding that a CD8-specific T-cell response can be elicited in neoepitopes derived from Histidine Ammonia-Lyase (HAL), Neuraminidase 2 (NEU2), Proprotein Convertase Subtilisin (PCSK9), Programmed Cell Death 10 (PDCD10), Supervillin (SVIL) and tRNA Splicing Endonuclease Subunit 54 (TSEN54) variants is in line with the notion that specific T cell-mediated immunotherapy can be applied for MB (189). Finally, the proteogenomic approach allowed the identification of neoantigens from MB tumors with a low mutational burden and a limited amount of tissue (189). It is of note that T cells with different T cell antigen receptors (TCR) and producing several antitumor cytokines can be obtained using these neoantigens. This finding further supports the more recent publications (184, 186, 189) that it is possible to discover immunogenic-specific peptides that are the basis for generating appropriate anti-MB vaccines.

Beside targeting tumor-specific antigens, immunotherapy can target receptors present on MB cells but widely expressed on other cell types (144, 190–207). The first point to consider for an efficient therapeutic effect without too many side effects is the expression of appropriate target molecules at the cell surface of MB cells (196–198). The ideal target should be expressed mainly, if not exclusively, on MB cells but not on healthy cells. The second critical point is the degree of the immune response; it should be enough strong to control and eliminate tumor cells, but sufficiently milder not to damage large amounts of healthy cells. This second point is much more important, more relevant: it is the function of healthy cells as a potential target of therapy. Proved target receptors on MB cells include HER2, B7H3 (CD276), Epha2 and GD2 (144, 196, 197, 199–203). These antigens can be targeted mainly by two therapeutic tools: monoclonal antibodies (mAb) ( Figure 5 ) and CAR cells ( Figure 6 ).

Antibodies represent a key therapeutic tool for several kinds of cancers, including tumors of the nervous system (204, 205) and CAR cells could be considered one of the more recent tools developed from classical mAb (206–209). As for the native mAb, CAR shows a component of the engineered chimeric receptor that recognizes the target molecule at the cell surface of MB cells (202–206). The antigen recognition domain of the CAR molecule is typically derived from the variable regions of a mAb as a single-chain variable fragment (scFv) recognizing a tumor expressed antigen. The main difference consists in the molecular and cellular mechanisms of the therapeutic effects (210, 211). Antibodies can affect receptor-ligand interactions and signaling, as well as trigger complement- and/or antibody-dependent cellular cytotoxicity (CDC and ADCC) and cytokine release (212). On the other hand, CAR cells, upon the interaction of the CAR molecule, deliver an activating signal that leads to the killing of the target cells (69). Also, CAR cells produce and/or release cytokines typical for the type of cell in which the CAR has been transduced (69). CAR molecules can be transduced into classical CD4⁺ or CD8⁺αβ T cells, γδ T cells, innate cells including NK cells, monocyte/macrophages (Mo/MΦ) and in vitro assays show that both mAb and CAR cells are efficient tools to eliminate MB cells (213–215). The main matter with both antibodies and CAR cells is the tumor localization of the antibody and the effector cells (216, 217). This could be considered the key factor that distinguishes the strong efficacy reported for CAR cells in hematological malignancies, compared to the disappointing results found for solid tumors (217–220).

The features of the tumor microenvironment (TME) are relevant for the growth and diffusion of several cancers, including MB (255). The knowledge of the features of MB during the interaction with the components of the TME could be considered as a therapeutic target in order to enhance the immune response to MB together with the impairment of the growth and spreading of the MB itself (255). This targeting can either reduce, or even convert, the immunosuppressive TME to immunostimulatory. The characteristics of the MB TME have been described in detail in several reviews elsewhere (255). Herein, we will point out in evidence the relevance of some components of the TME, such as the angiogenic factors, the role of podoplanin (PDPN) in MB spreading, “do not eat me” signals, and the interaction between the astrocytes and MB cells ( Figure 5 ). This is to show some of the key cellular and molecular players against which it is conceivable to use old and new immunotherapeutic tools as antibodies or CAR-T cells to further improve the outcome of MB patients.