Targeted therapeutics and immunotherapy strategies are critically dependent on identification of druggable tumor-specific antigens (TSAs) and neoantigens. To achieve maximal clinical efficacy and minimal toxicity, the ideal target should be immunogenic, highly expressed and presented on the surface of the majority of tumor but not normal host cells, and play a role in tumorigenesis. A very few antigens in general (and none of them in pediatric sarcomas) meet such criteria. In contrast to adult cancers, pediatric tumors exhibit very low mutation rates and, consequently, much fewer TSAs and tumor neoantigens ^[1][2][3][22,34,35]. Even osteosarcoma (OS), which has relatively high level of copy-number variations and gene deletions ^[4][70], exhibits on average about 7 neoepitopes per tumor and only 2 of them are predicted to be expressed, while EwS has none ^[5][71]. With regard to cell surface proteins, the best studied targets include B7-H3, GD2, IGF1R (which are shared between OS and EwS) as well as HER2 (for OS) and CD99, endosialin/CD248, TRAIL-R and STEAP-1 (for Ewing sarcoma (EwS)) ^[2][6][7][18,34,45].

One of the potential solutions to the lack of conventional TSAs in pediatric sarcomas may be hidden in noncoding regions of the genome, including introns, alternative splicing variants, gene fusions, endogenous retroelements and other unannotated open reading frames ^[8][75]. According to recent studies, noncoding regions could be the main source of targetable TSAs in human malignancies ^{[9][10][11][12]}[76,77,78,79]. Activation of these regions mainly occurs due to demethylation of the genome and other epigenetic mechanisms that are highly dysregulated in tumor cells, rising a possibility that the resulting TSAs may be widely shared between different tumor types and absent from normal tissues ^[13][14][80,81]. In addition, pediatric sarcomas may also express unique neoantigens from noncoding regions, given that at least a third of them carry recurrent chromosomal translocations and express characteristic fusion proteins which act as transcriptional and epigenetic regulators ^[15][16][17][13,15,82]. 

2. Low Expression of MHC-I and Upregulation of Immune Checkpoints

One of the mechanisms whereby pediatric sarcomas escape T cell-mediated immunosurveillance is impaired expression of MHC class I/Human Leukocyte Antigen (HLA) class I antigens ^[18][39]. About 48–79% of primary EwS and the majority of metastatic lesions, especially pulmonary metastasis, exhibit low-to-absent MHC class I and II expression ^[19][20][84,85]. 

Upregulating MHC-I expression in pediatric sarcomas may thus be a promising strategy to activate CD8⁺ T cell-mediated antitumor responses ^[21][88]. This can be achieved by stimulating proinflammatory pathways, such as TNF-TNF receptor-NFκB, type I IFNs-IFNAR1/2-STAT1/2/3 or type II IFN-IFNGR-STAT1 ^[22][89]. For example, MHC-I expression in EwS cell lines is induced by IFNγ or mediators of dendritic cell maturation, including TNF ^[19][23][24][84,90,91]. In particular, treatment with GM-CSF, IL-4, TNF, IL-6, IL-1β and PGE₂ upregulated MHC-I, ICAM-1 and CD83, and improved recognition of EwS cell lines by TSA-specific TCR transgenic T cells in vitro ^[25][74]. 

Concurrent with reduced MHC-I expression, upregulation of immunosuppressive receptors and checkpoints may contribute to the immune escape in pediatric sarcomas. For instance, HLA-G and HLA-E, the non-classical MHC-I molecules implicated in the protective maternal-fetal barrier in the placenta ^[26][95], are highly upregulated on tumor and myeloid cells in the EwS TME. HLA-G is expressed in ~34% of EwS biopsies, and can be further induced by proinflammatory signaling, including IFNγ and GD2-specific CAR-engineered NK cells ^[24][27][28][56,91,96]. When expressed on tumor cells, HLA-G and HLA-E were shown to interact with inhibitory receptors expressed on T cells and NK cells, negatively affecting cytotoxic functions of both CD8⁺ T cells and NK cells ^[29][97].

On a similar note, therapeutic targeting of PD-L1 and PD-1 immune checkpoints, which are expressed in ~20% of pediatric sarcoma patients in EwS and OS ^{[30][31][32][33][34][35][36]}[46,47,48,49,50,51,52], has not shown clinical efficacy ^[37][38][39][43,53,54]. In line with low expression of immune checkpoints on EwS and OS tumor cells, only 35% of EwS- and OS-infiltrating immune cells express PD-L1 ^[31][47], whereby expression of PD-L1 or PD-1 on T cells is rare for the most part in EwS and OS ^{[34][40][41][42]}[50,55,98,99], and is predominantly observed on macrophages ^[39][54]. Yet, ~10% of post-treatment OS tumors score in the top quartile of immune infiltration, which is comparable to other strongly immune-infiltrated malignancies, including lung cancer and renal clear cell carcinoma ^[43][100]. The respective groups of OS patients may potentially benefit from the immune checkpoint blockade.

3. Immunosuppressive TME in EwS and OS

3.1. Improving CD8

⁺

T Cell Infiltration and Antitumor Activity

Only 12–38% of EwS and ~52–68% OS tumors are infiltrated by cytotoxic CD8⁺ T cells ^[24][32][40][48,55,91]. Poor CD8⁺ T cell infiltration is a negative prognostic marker associated with metastatic progression and worse outcomes ^{[20][42][44][45][46][47]}[85,99,101,102,103,104]. Key chemotactic mechanisms for the recruitment of TILs and activation of antitumor immune responses are the TME-derived C-X-C Motif Chemokine Ligand 9/10 (CXCL9/10) or stromal-derived Chemokine (C-C motif) ligand 5 (CCL5) and their respective receptors C-X-C Motif Chemokine Receptor 3 (CXCR3) and CCR5 ^[48][49][105,106]. In EwS, increased expression of CXCL9/10 and CCL5 correlates with infiltration of CD8⁺ CXCR3⁺/CXCR5⁺ T cells ^[46][103]. In spite of higher proportion of TILs in the OS TME compared to EwS, they exhibit terminally exhausted phenotypes, including expression of co-inhibitory receptors TIGIT, LAG3, PD-1 and TIM3 ^[50][107]. Apart from the PD-1/PD-L1 axis, their blockade in OS may thus enhance TILs cytolytic activity. This may be especially relevant in OS with pulmonary metastases, which show increased T cell infiltration at the interface between the adjacent healthy tissue and tumor stroma ^[43][51][100,112]. These interfaces are enriched with activated exhausted CD8⁺ T cells positive for PD-1, LAG3 and IFNγ and with myeloid cells expressing M-MDSC and DC signatures. The core of pulmonary metastases is devoid of immune infiltrates, suggesting that myeloid cells may exclude TILs ^[51][112].

3.2. Targeting Tumor-Associated Macrophages

The most abundant immune cells in the TME of EwS and OS are tumor-associated macrophages (TAMs) which exhibit immunosuppressive M2 signatures ^{[47][52][53][54]}[104,114,115,116]. Based on recently published transcriptomic analysis, these M2 macrophages may be phenotypically and functionally distinct in EwS and OS ^[50][107]. In line with this, TAM infiltration in EwS was indicative of poorer survival ^[47][53][55][104,115,117], while opposite observations were made in OS, where infiltration with CD14⁺/CD163⁺ myeloid cells and M1/M2 macrophages correlated with improved outcomes ^[42][56][57][67,99,118]. However, infiltration with CD68⁺ macrophages was associated with worse survival in OS ^[33][49], suggesting the existence of different TAM populations with opposite activities. Higher density of CD68⁺ and CD163⁺ macrophages in OS (the CD68⁺ to TIL ratio is 5.9, compared to 2.5 in EwS) may contribute to OS aggressiveness ^[58][119]. Chemotactic signals from the TME recruit monocytes from the bone marrow into the tumor stroma ^[59][120], where they polarize into TAMs and pro-tumorigenic M2 macrophages ^[60][121]. Signaling in the TME promotes sarcoma progression by inducing angiogenesis ^[53][61][62][115,122,123], migration ^[63][124], extravasation ^[64][125] and chemotherapy resistance ^[65][126]. TAMs in pediatric bone sarcomas release pro-inflammatory cytokines ^[53][115], prevent T cells from entering the tumor core ^[51][112] and impede the activation and degranulation of T cells ^[66][127]. 

4. Immunogenicity and Response to Immunotherapy of EwS and OS in the Context of Bone and Soft Tissue Sarcomas

Bone and soft tissue sarcomas (STS) of children and adults comprise a heterogenous group of tumors with distinct biological properties, albeit they all share non-immunogenic properties and non-responsiveness to immunotherapy ^{[1][37][67][68][69]}[22,43,135,136,137]. The most common bone tumors include EwS, OS and chondrosarcoma, while fibrosarcoma, gastrointestinal stromal tumors (GIST), leiomyosarcoma, liposarcoma, rhabdomyosarcoma (RMS), undifferentiated pleomorphic sarcoma and synovial sarcoma are the most frequent STS ^[70][138]. Immune infiltrates are heterogenous between sarcoma entities and age-dependent ^[71][139]. Sarcomas driven by mutations and copy number alteration tend to be T cell-inflamed, while translocation-driven sarcomas are immunologically cold ^[44][72][73][101,140,141]. Remarkably, mutation rates among sarcomas are low compared to other tumor entities ^[74][142], with pediatric sarcomas exhibiting the lowest numbers of mutations per Mb (EwS 0.24; OS 0.38 and RMS 0.33, the most frequent pediatric STS, compared to adult STS 1.06) ^[73][75][141,143]. Expression of immune checkpoints and clinical response to immune checkpoint blockade in sarcomas is dynamic, variable, and dependent on the histologic subtype. PD-L1 expression is sparse on the pediatric sarcomas EwS, OS and RMS, similar to the majority of adult STS (~20% PD-L1 expression) ^[76][77][78][144,145,146].

5. Extracellular Vesicles (EVs) as Means of Immune Escape

Communications between tumor and host cells in local and distant tumor sites are mediated by diffusible molecules such as cytokines, chemokines and lipids as well as to a large extent by extracellular vesicles (EVs) that create permissive environment for tumorigenic progression. EVs do so by transferring nucleic acids, proteins, lipids and various metabolites from tumor to various host cells, and vice versa ^[79][80][81][152,153,154]. As such, the EV cargo reflects the cell of origin and its physiological conditions, representing an important source of cancer-associated biomarkers. Most importantly, the EV cargo is encapsulated into lipid bilayer membrane and thus protected from degradation. When taken up by bystander normal and malignant cells, the EV cargo and is capable of functionally reprogramming the acceptor cells. The EVs are comprised of highly heterogeneous populations of vesicles, whose secretion and composition are influenced by environmental conditions and tissue homeostasis. Most studies are focused on the nanosized vesicles (30–200 nm) originating from endosomal compartments (exosomes) or plasma membrane (ectosomes), which are believed to be important players in extracellular communications in healthy and diseased states ^[81][82][83][154,155,156]. The original hypothesis proposed in early 1980s implicated exosomes (EVs) as garbage bags for removal of unwanted proteins or harmful metabolites from the cells ^[84][85][157,158]. Indeed, secretion of cellular waste in EVs (or in specialized EV subsets) may be important for maintaining cellular homeostasis in normal and cancer cells. Recent findings have indicated that secretion in exosomes is essential for removal of damaged DNA and that blocking exosomal pathways provokes innate immune responses and induces senescence-like phenotype or apoptosis in normal cells due to accumulation of nuclear DNA in the cytosol ^[86][159]. Packaging in EVs is also required for expulsion of chemotherapy drugs and cellular toxins ^[87][88][160,161]. The EV-mediated waste management may be especially important for cancer cells, given their high proliferation and metabolic rates, and deficiencies in DNA repair pathways. Tumor-derived EVs influence all major hallmarks of cancer, including immune evasion, tumor-promoted inflammation, angiogenesis, metabolic and epigenetic reprograming of the recipient cells, extracellular matrix remodeling, cancer metastasis and drug resistance ^[80][89][90][40,153,162]. In bone sarcomas and in cancers that preferentially metastasize to bones (such as prostate and breast carcinomas), EVs secreted by tumor cells are also capable of interfering with osteogenesis to promote tumor-supporting microenvironment inside the bone ^{[91][92][93][94]}[163,164,165,166].  Lack of TSAs and low expression of MHC molecules have been described in previous sections as one of the major impediments for therapeutic targeting of EwS and OS cells. The available evidence suggests that their release in EVs could be one of the mechanisms employed by tumor cells to eliminate their specific antigens and MHCs and to reduce their recognition by cytotoxic T cells. Indeed, presence of tumor-derived MHCs and antigens (including pre-formed functional TSA-MHC complexes) in EVs is a well-documented phenomenon ^[89][95][96][40,41,42], albeit its role in EwS and OS remains to be elucidated. Dissemination of tumor EVs harboring TSAs and their subsequent acquisition and cross-presentation by bystander immune and non-immune cells may also act as a decoy to divert antitumor immunity from cancer cells. EwS EVs may be directly involved in generation of immature proinflammatory myeloid cells in local TME and systemic circulation. It was shown that EwS EVs induced secretion of IL-6, IL-8 and tumor necrosis factor (TNF) by primary CD33⁺ myeloid cells and CD14⁺ monocytes, and inhibited their maturation into antigen-presenting DCs ^[97][175]. In particular, CD14⁺ cells differentiated in the presence of EwS EVs exhibited a semi-mature phenotype and immunosuppressive activity, including reduced expression of co-stimulatory molecules CD80, CD86 and HLA-DR, activation of the innate immune response gene expression programs, and the ability to interfere with activation of CD4⁺ and CD8⁺ T cells. Therefore, EwS EVs may contribute to systemic inflammation and immunosuppression by skewing differentiation and maturation of blood-circulating and tumor-infiltrating myeloid cells. Mechanistically, induction of immunosuppressive myeloid cells is primarily mediated by various protein and RNA constituents present in tumor EVs ^[98][176]. EwS EVs carry multiple mRNAs encoding oncogenic drivers, including EWS-FLI1, EZH2 and stem cell-associated proteins ^[99][100][177,178], some of which can be transferred to the neighboring mesenchymal stem cells ^[101][179]. Whether or not EV-derived RNAs are actually capable of driving a sustainable protein expression in the recipient cells is an open question, given that the majority of these RNAs, including mRNAs and microRNAs, are severely fragmented and present in less than one copy per EV ^[102][103][180,181].