The human CNS is a tightly-regulated, well-evolved system. Previously thought to be tightly sealed away from the rest of the body (the idea of “immune privilege”), as we have learned in the last 30-40 years, the CNS does interact with the systemic immune system (3–6) in its own unique and regulated way, feasibly because immune-mediated inflammation of the brain could be extremely detrimental to an organism (7, 8). The blood-brain barrier (BBB) is a unique feature of the CNS microvasculature anatomy that tightly regulates the influx and efflux of molecules, substances, fluids, and cells, including immune cells (9). The interaction between the overall immune system (10) and the CNS system includes specific structures that allow this interaction (3). Specialized pathways deep in the dura (the so-called glymphatic system) (3–5, 11) direct interstitial fluid from the CNS towards drainage into the deep cervical lymph nodes and allow interaction with the systemic immune network (10, 12, 13). The pediatric BBB and immune environment is unique when compared to the adult. The junctions of the pediatric BBB are tighter and are known to be developed by the time an infant is born (14). Additionally, despite being immunologically naïve, the young infant CNS is entirely capable of mounting a robust immune response (15), In contrast, age-related stresses and inflammation affect the function and structure of the aged CNS. Additionally, older adults have stiffer, more permeable vasculature in general (15, 16). Furthermore, the expression of transmembrane transport proteins varies in pediatric tissue studies when compared to adult (17). This can affect the influx and efflux of substances into and out of the CNS, including proteins, drugs, and other blood-borne substances (14, 16, 17).

To understand how CNS lesions respond to immunotherapies, the CNS tumor microenvironment (TME) must be considered in the context of the known differences of the pediatric BBB and tumor microvasculature (the blood-brain-tumor barrier, BBTB), especially due to its unique qualities. TME includes the collection of cells (particularly immune cells and other blood cells), blood vessels, stromal cells, and extracellular matrix surrounding a tumor (18–20). In consideration of the immune landscape of tumors, there are “hot” and “cold” tumors. “Hot” tumors are more immunologically active, more prone to inflammatory responses, and have a higher number of active tumor-infiltrating lymphocytes (TILs). In contrast, “cold” tumors have TILs that lower in number and have an exhausted phenotype (6, 20–22). The resident macrophages of the brain are microglia. These cells tightly regulate the immune homeostasis of the CNS and are known to have powerful immunosuppressive functions (21–26). Tumor-associated macrophages (TAMs) are a well-known part of the TME. In brain tumors, TAMs are bone-marrow derived (whereas microglia are resident in normal brain) (27, 28). TAMs are thought to “polarize” depending on signaling, stimulus, and pathology into either pro-inflammatory (immunoactive) or anti-inflammatory (immunosuppressive) states (27–30). Brain tumor-associated TAMs are pro-tumorigenic in a number of ways, such as upregulation of endothelial cell secretion of vascular endothelial growth factor (VEGF), secretion of immunosuppressive cytokines such as transforming growth factor beta (TGF-β) and interleukin 10 (IL-10), secretion of arginase (to starve T-cells), secretion of epidermal growth factor (EGF) to promote tumor migration, secretion of pro-tumor chemokines and cytokines, secretion of prostaglandins (to inhibit activation of T-cells), and direct antigen presentation functions (30–33). Additional non-TAM myeloid-derived suppressor cells (MDSCs), are a group of cells of myeloid origin that are pathologically activated and serve an immunosuppressive function against both adaptive and innate anti-tumor immunity (34–36).

While most of the research on the CNS TME is in adult gliomas (37–39), more information is becoming available regarding the pediatric CNS TME, and how it relates to the pediatric microvasculature and BBB permeability (26). In general, pediatric CNS tumors are considered “cold”, due to the presumed lack of genetic mutations (generation of neoantigens) (21, 40). Additionally, pediatric tumors tend to have loss of expression of major histocompatibility complex I (MHC I), which decreases the ability of T-lymphocytes to recognize and become activated (41). In contrast to adult CNS tumors, the pediatric CNS TME has a relatively high concentration of immunosuppressive cells, such as marrow-derived TAMs, resident microglia, and other MDSCs (27, 41, 42). Figure 1 is a review of the pediatric CNS-specific TME. As mentioned before, the TILs that are present in pediatric CNS tumors exhibit an exhausted phenotype. These TILs have decreased effector function, increased expression of inhibitory receptors, and decreased production of immunostimulatory cytokines (42), further decreasing the recruitment of additional immune response. Notably, there is some variation in the TME of different types of pediatric CNS lesions (21, 39–46). Griesinger, et al. (47) used flow cytometry to analyze the phenotype and frequency of immune cells that infiltrate tumors in the most common pediatric brain tumor types (pilocytic astrocytoma, ependymoma, glioblastoma, and medulloblastoma). Pilocytic astrocytoma and ependymoma demonstrated a significantly higher number of infiltrating myeloid and lymphoid cells, more of which were activated compared with glioblastoma, medulloblastoma, or nontumor tissue. Levine, et al. (48) designed a 103-gene immune-oncology transcriptomic panel elucidating immune cell types and tumor inflammatory signatures (TIS, a biomarker for immune response to immune checkpoint inhibitors). Pediatric glioneuronal tumors had substantial upregulation of T-cell markers and regulatory genes; diffuse astrocytomas had a nearly normal profile. Diffuse intrinsic pontine gliomas (DIPGs) showed strong upregulation of macrophage markers. In work done by Wei, et al. (49), patient-derived xenograft (PDX) models of pediatric high-grade gliomas (pHGG) and diffuse midline gliomas (DMG) were used to examine the vascular permeability and molecular transport differences between these two types of lesions. Staining for specific tumor cell markers hVimentin and endothelial marker CD31 revealed minimal change in the vasculature of DMGs, but significant alterations in vasculature (such as an increased number of branch points and overall vasculature density) in pHGG. Additionally, when staining for tight junction marker Claudin5 (Cldn5, a marker for tight junctions) and Glut1 (a BBB-associated glucose transporter, also known as Slca2a1), differences were noted in the expression of these proteins in the vasculature of pHGG PDX models when compared to normal brain and DMG PDX models. pHGG models showed slightly decreased expression of Glut1 and increased and more disorganized expression of Cldn5, while the DMG vasculature was far similar to the expression patterns of these proteins in normal brain vasculature. In a review by Morris, et al. (50), similar differences in expression patterns were noted across different subtypes of medulloblastoma including wingless (WNT), sonic hedgehog (SHH), and group 3. The heterogeneity of the TME, BBB, and BBTB when compared to pediatric versus adult tumors, and the variability across different pediatric tumor types, could help to explain why some CNS tumors respond better than others to various types of immune therapies, both in the pediatric arena and when compared to adult tumors (40–46).

Part of the standard immune regulation and surveillance in humans is by regulation and modulation of the innate immune system. One way this is accomplished is by immune checkpoint regulation. When programmed cell death 1 (PD1, which is found on the surface of T- lymphocytes, among other immune cells) binds with programmed cell death ligand 1 (PDL-1, found on the surface of most mammalian cells) it dampens the immune response and makes the T-cell mediated kill less effective (35). Similarly, T-lymphocyte-associated protein 4 (CTLA4, expressed on the surface of activated T-lymphocytes) binds to CD80/CD86 (found on antigen-presenting cells), which sends an inhibitory signal to the activated T-cell, as another checkpoint mechanism (51–54). Many types of tumor cells have evolved ways to evade immune detection or cell-kill mechanisms (such as over-expression of PDL1). Immune checkpoint inhibitors (ICIs) are monoclonal antibodies that bind to either PDL1, PD1, or CTLA4 to overcome this dampening down of the immune response, leading to better tumor cell kill by increased immune infiltration (54).

Multiple iterations of immune checkpoint inhibitors are approved for many types of cancers, including cancers of the breast, skin, lung, bladder, colon/rectum, esophagus, cervix, kidney, uterus, liver, some sarcomas, and some lymphomas. Unfortunately, the efficacy of checkpoint inhibitors alone in CNS tumors has been lackluster in both adult and pediatric populations (54–61). CheckMate-908, a study using nivolumab+ ipilimumab in pediatric patients with high-grade CNS malignancies showed similar results, with failure to demonstrate clinical benefit (62).

The lack of efficacy of ICIs to treat most CNS malignancies is likely multifactorial. Firstly, as detailed above, many lesions of the CNS are known to be “cold” in terms of immunogenicity. One of the major barriers to effective ICI action in most CNS tumors is the fact that ICIs rely on the activation of lymphocytes to exert their mechanism of action (63–65). If the T-cells are activated in the periphery, and only a minimal amount enters the CNS, the point is moot (65). A second possible reason is the challenge of larger molecules to pass through the BBB (however, there is evidence to suggest that ICI are effective in treating metastatic lesions to the brain, which suggests that the BBB is at least partially permeable to ICI therapy.) (66, 67) The lack of efficacy in pediatric patients could also relate to the relatively less permeable state of the pediatric BBB when compared to the aged one (14–17, 46). As the use of ICIs in pediatric patients as monotherapy has not shown clinical efficacy for most CNS tumors (45, 62, 68), many of the trials currently open using ICIs for pediatric patients also include other treatment modalities, such as chemotherapy, radiation therapy, multiple ICIs, or vaccinations.

Notably, there is one population where monotherapy with ICIs has shown efficacy, and this is patients with constitutional mismatch repair deficiency (CMMRD). This is an autosomal recessive condition characterized by mutations in DNA repair genes, which tends to result in the nearly universal development of malignancies (69). Because these patients have such a high tumor mutational burden, there are multiple case reports and emerging data that these patients in particular could benefit from monotherapy with ICIs (70). Table 1 outlines the current open clinical trials in pediatric patients involving ICIs, +/- other chemotherapies.

As more targeted approaches are always needed, the development of anti-tumor vaccines is an exciting prospect in harnessing the immune system to control CNS tumors in a controlled and specific way. Cancer vaccines have been in development since the 1980s when autologous tumor cells were administered to patients with colorectal cancer (71, 72). Currently, there is renewed interest in utilizing cancer vaccines, especially for pediatric patients. This is in part due to the wide availability of laboratory capabilities to identify appropriate tumor antigens, in addition to our growing knowledge about the behavior of dendritic cells (DCs), which are a vital factor for vaccination efficacy (71–73). DCs are professional antigen-presenting cells (APC)with a powerful ability to stimulate the immune system. DCs can travel to lymph nodes, and process and cross-present antigens via MHC I and MHC II. When naïve CD8+ T-lymphocytes are presented with tumor antigen, this then causes differentiation into tumor-specific cytotoxic T-lymphocytes (CTLs) (74). DCs also interact with CD4+ T-lymphocytes via MHC II, which provides help for CD8+ CTLs via the release of cytokines to boost the immune response (74, 75). Additionally, through crosstalk mechanisms, DCs stimulate and activate natural killer (NK) cells, which are another vital arm of the anti-tumor immune response (75–79). Multiple different types of vaccines have been developed to treat tumor cells: cell-based, virus-based, nucleic acid-based, and peptide-based vaccines (80). Figure 2 is a visual review of the vaccines covered in the next paragraphs, that are the most widely studied in pediatric CNS malignancies. Table 2 outlines current open clinical trials for pediatric patients using vaccines, with and without other chemotherapies or procedures.

There is preclinical data for other types of immunotherapies that perhaps have a future role in the treatment of pediatric CNS malignancies (108, 109). Bispecific T-cell engagers (BiTEs) are molecules that “usher” T-lymphocytes to tumors by having two binding sites (110): one for a T cell and one for a tumor antigen. Bhojnagarwala (111), et al, designed multiple BiTEs targeting IL-13Rα2, which is a glioblastoma (GBM) surface antigen, and tested them in an orthoptic model. One of these molecules, PB01-forward, controlled tumor growth and resulted in longer animal survival. Sun, et al. (112) designed a bispecific antibody (BsAB) with epidermal growth factor receptor variant III (EGFRvIII, an antigen presents on GBM tumor cells) and CD3 as the targets, which demonstrated significant in vivo and in vitro activity in mouse models (113, 114). CD47 is an anti-phagocytic molecule that is expressed by brain cancer cells. Abbas, et al. (115) treated orthoptic xenograft mouse models of medulloblastoma with either craniospinal irradiation (CSI), anti-CD47 antibody, or both. The group that received both modalities showed marked regression of the tumor. Gholamin, et al. (116) tested a humanized anti-CD47 antibody, Hu5F9-G4, on five types of aggressive pediatric brain tumors (group 3 medulloblastoma (primary and metastatic), ATRT, PNET, pediatric glioblastoma, and DIPG) in patient-derived orthotopic xenograft models, and efficacy was demonstrated in both in vivo and in vitro.

It is well-establish that radiation therapy can kill tumors due to directly causing cell death. But it is also known that radiation increases the immunogenicity of a tumor in several ways. Firstly, radiation-induced DNA damage causes DNA to accumulate in the cytoplasm of cells, resulting in danger-associated molecular patterns (DAMPs) (117, 118). After this DNA is recognized as a DAMP by host receptors—in tumors, cyclic guanosine monophosphate (GMP)–adenosine monophosphate (AMP) synthase (cGAS; also known as MB21D1) is thought to be the main mechanism of this process (119). Binding of cGAS to cytoplasmic DNA activates the cGAS, and triggers a cascade in the cell, ultimately resulting in increased expression and release of interferons and cytokines (119, 120). This sterile inflammation acts as a recruitment signal to the immune system, causing infiltration and maturation of dendritic cells (121) with subsequent antigen presentation, thus resulting in activation and tumor infiltration of tumor-specific T-cells (122, 123).

In addition to local immune infiltration, RT causes an anti-tumor effect distant from the original radiated field, known as the abscopal effect. First described by Mole in 1953, this phenomenon is somewhat rare and thought to be mediated by systemic immune activation by RT (124, 125). A review by Pangal, et al, described multiple instances of patients with intracranial or intraspinal metastatic disease who had other distant tumors that had a response, despite no direct radiation therapy (126). Of note, most of these patients were on concurrent ICI therapy. This suggests that radiation could potentially “prime” the immune system, making ICI therapy more effective (127, 128). Another effect RT is thought to have relevant to CNS tumors is disruption of the BBB. Both preclinical and clinical studies have demonstrated radiation influences the integrity of the BBB (129–131). While this could put the CNS at increased risk of exposure to toxins and infection, it could also be advantageous. Disruption of the BBB could allow better penetration of systemic therapies and possibly improved immune infiltration, for better tumor control.

Tumor cells have many ways to evade the immune system, and one is decreased expression of MHC-I (132, 133). MHC-I presents tumor peptides, which then are recognized by the T-cell receptor (TCR) of CD8+ T-cells. This in turn causes activation of the T-cell and induces tumor cell kill. Multiple studies have shown, even at low doses, RT can increase the expression of MHC-I on the surface. In work by Das, et al (134), exposure of human medulloblastoma cell lines to a low dose (1Gy) of ionization radiation increased the expression of MHC-I and MHC-II on the surface of these cells. Sharma’s, et al. (135) work showed similar findings: multiple cancer cell lines and normal human cells were exposed to a single dose of 20Gy radiation and showed increased expression of MHC-I and cancer testis antigen (CTA). This effect is advantageous because, in addition to the mechanisms already described, it is another way to overcome tumor immune evasion (136–138). Radiation has also been shown to increase the expression of other mAb targets. Wattenberg, et al. (139) demonstrated that exposing human cell lines to a single dose of either 5Gy or 10Gy of ionizing radiation significantly increased the expression of multiple targetable surface antigens (HER2, EGFR, and CD20). Similarly, Garnett, et al. (140) exposed human cell lines to low-dose radiation. The results showed an increase in the expression of Fas (CD95), in addition to other molecules that are involved in T-cell mediated cell tumor cell kill (such as intercellular adhesion molecule 1, mucin-1, carcinoembryonic antigen, and MHC I). In addition, some of the cell lines that did upregulate CEA showed enhanced CTL-mediated cell kill. In addition to antibody targets and MHC, radiation can increase immune cells directly. This has been demonstrated in multiple other pre-clinical studies (141–143). For example, Pandey, et al. (143) demonstrated in a preclinical model that a low dose of radiation increases immunogenicity by increasing the activation of CD8+ lymphocytes and increasing the phagocytic activity of macrophages. Therefore, we now know radiation affects both tumor cells and immune cells to further increase immunogenicity and enhance immunotherapies when used in combination.

Due to the myriad effects radiation therapy has on the systemic immune system, there are multiple clinical trials to examine the effects of a combination of radiation therapy and various immune-oncologic treatments including ICIs, mAbs, and vaccines (137). Research has consistently shown that while immune strategies such as vaccinations or ICIs are not as effective on their own, there does seem to be better efficacy when these therapies are used in combination with each other or with other therapies (144–146). For example, ICIs in combination with vaccinations can overcome immune evasion and improve the efficacy of both therapies and increase CTL-mediated tumor cell kill (80, 146). The current landscape of immunotherapy for pediatric CNS tumors is heavily focused on complementary immunotherapies for better tumor control. Table 3 outlines current open clinical trials for pediatric patients using combined immunotherapy modalities.