Targeting the osteosarcoma immune microenvironment for improved immunotherapy and translational applications

Abstract

Osteosarcoma (OS) is the most common primary malignant bone tumor in children and adolescents. Despite advances in surgery and chemotherapy, outcomes remain poor in metastatic cases, with five-year survival rates below 30%. This stagnation highlights the urgent need for novel therapeutic strategies. Growing evidence indicates that the tumor immune microenvironment (TIME) plays a central role in OS progression, metastasis, and resistance to treatment. Immunosuppressive cells, including tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), and regulatory T cells (Tregs), dominate the TIME, while cytotoxic T cells often exhibit exhaustion. Stromal barriers, hypoxia, and metabolic constraints further impair immune activity. Recent single-cell and spatial transcriptomic studies reveal that immune and stromal architectures strongly correlate with prognosis and therapeutic response. These features contribute to the limited efficacy of current immunotherapies, including immune checkpoint inhibitors (ICIs) and CAR-T cells. In this review, we summarize the cellular, molecular, and spatial components of the OS TIME, critically evaluate current immunotherapeutic strategies, and highlight emerging translational approaches aimed at overcoming immune resistance and improving clinical outcomes.

Introduction

Osteosarcoma (OS) is the most prevalent primary malignant bone tumor in children and adolescents [1]. Despite advances in surgery and multi-agent chemotherapy, survival outcomes remain poor, particularly in metastatic disease where five-year survival rates fall below 30% [2]. Notably, the lungs are the most common site for OS metastasis, and patients with pulmonary metastases exhibit a dismal five-year survival rate of only 20–30% [3]. Over the past four decades, conventional treatment regimens have reached a therapeutic plateau, underscoring the need for novel strategies that address the mechanisms driving OS progression and resistance.

Recent studies highlight the critical role of the tumor immune microenvironment (TIME) in shaping OS behavior and therapeutic response. The TIME in OS is characterized by a complex interplay of various immune cells, stromal components, and signaling molecules that collectively foster an immunosuppressive milieu [4] (Fig. 1). Tumor-associated macrophages (TAMs), particularly the M2 phenotype, are the most abundant immune cells within the OS microenvironment, accounting for approximately 50% of the immune cell population [5]. These M2-TAMs contribute to tumor progression by promoting angiogenesis, extracellular matrix (ECM) remodeling, and immune evasion through the secretion of anti-inflammatory cytokines such as interleukin-10 (IL-10) and transforming growth factor-beta (TGF-β) [6]. Myeloid-derived suppressor cells (MDSCs) and regulatory T cells (Tregs) further suppress cytotoxic T lymphocytes and natural killer (NK) cells, facilitating immune escape [7, 8]. Hypoxia and metabolic stress within the tumor niche amplify these effects and contribute to chemoresistance [9]. However, these findings often remain siloed, focusing on individual cell types or pathways in isolation, without a unifying framework that explains how the TIME shapes therapeutic resistance.

Fig. 1.

Schematic diagram of the osteosarcoma tumor microenvironment

In recent years, OS research is increasingly focused on overcoming TIME-associated barriers. These include immune checkpoint inhibitors (ICIs), Chimeric antigen receptor T (CAR-T) cells targeting OS-associated antigens (e.g., GD2, HER2), dendritic cell vaccines, macrophage modulators such as mifamurtide, and metabolic interventions targeting indoleamine-2,3-dioxygenase (IDO) or arginase pathways. However, clinical success has been limited, likely due to low immunogenicity and its suppressive immune landscape [10]. Nevertheless, ongoing research into the modulation of the OS TIME holds promise for enhancing the effectiveness of immunotherapeutic strategies [11]. Technological advances like single-cell RNA sequencing and spatial transcriptomics have begun to reveal the heterogeneity and spatial compartmentalization of the TIME, offering new insights into immune evasion and treatment failure.This review aims to provide a comprehensive overview of the current understanding of the OS TIME, highlighting the roles of various immune cell populations, the mechanisms underlying immune evasion, and the potential therapeutic targets within the microenvironment. By elucidating the complex interactions within the OS TIME, we hope to pave the way for the development of more effective immunotherapeutic approaches that can improve outcomes for patients afflicted with this aggressive malignancy.

The immunosuppressive architecture of the osteosarcoma TIME

OS typically develops in bone and is characterized by a profoundly immunosuppressive TIME that promotes tumor growth, metastasis, and therapeutic resistance. Unlike immune-inflamed tumors such as melanoma or non-small cell lung cancer, OS frequently exhibits “immune-excluded” or “immune-desert” phenotypes [12], where effector lymphocytes are restricted to stromal margins and fail to infiltrate tumor nests. Central to these architectures are immunosuppressive cellular neighborhoods composed of TAMs, MDSCs, Tregs, and exhausted cytotoxic T cells. These populations co-localize in spatial niches that suppress anti-tumor immunity via cytokine secretion (e.g., IL-10, TGF-β), metabolic deprivation, and immune checkpoint expression.

Myeloid cell-driven suppression

TAMs are among the most abundant immune cells in OS and are often skewed toward the M2-like phenotype, supporting angiogenesis, immunosuppression, and tissue remodeling [4]. Rather than functioning as antigen-presenting cells, M2-TAMs promote immune escape through secretion of IL-10, TGF-β, and VEGF [13]. They also contribute to stromal organization and indirectly recruit additional suppressive cells, including MDSCs and Tregs.

MDSCs are particularly enriched in OS and are recruited by tumor-derived factors such as GM-CSF and IL-6. They impair T cell responses through the production of reactive oxygen species, nitric oxide, and arginase [14], while also promoting angiogenesis and epithelial-to-mesenchymal transition via TGF-β and hepatocyte growth factor (HGF) [15]. Compared to many other cancers, OS harbors exceptionally high levels of MDSCs, particularly in metastatic lesions, where their presence correlates with poor survival. Targeting MDSC trafficking using CXCR4 inhibitors has shown preclinical promise, but clinical efficacy remains unproven [16, 17].

T and NK cell dysfunction

Effector T cells are present within OS lesions but exhibit dysfunctional or exhausted phenotypes. CD8 + cytotoxic T cells and CD4 + helper T cells tend to localize at tumor margins and periosteal regions [18], where they express high levels of inhibitory receptors such as PD-1, TIM-3, and CTLA-4, and demonstrate poor effector function [19]. Tregs (CD4 + FoxP3+) are concurrently enriched, secreting immunosuppressive cytokines (e.g., IL-10, IL-35), and expanding in response to cues from TAMs and MDSCs [20]. As a result, OS TIME features a high Treg/MDSC to CD8 + T cell ratio, indicative of a “cold” immune phenotype.

NK cells are detectable in OS but are also functionally inhibited [21]. Although activated NK cells demonstrate cytotoxicity against OS cells in vitro, tumor-derived TGF-β and other suppressive signals limit their activity. Interestingly, PD-1/PD-L1 blockade not only enhances T cell responses but also restores NK cell function [22], suggesting a therapeutic opportunity if inhibitory pathways can be overcome.

Stromal remodeling and physical barriers

Beyond immune cells, cancer-associated fibroblasts (CAFs) and mesenchymal stromal cells (MSCs) play a pivotal role in shaping the TIME [23]. These fibroblastic populations deposit dense ECM, generating a rigid osteoid scaffold that physically impedes immune infiltration [24]. Under hypoxic conditions, CAFs release IL-6 and TGF-β, which further polarize TAMs toward the M2 phenotype and expand Tregs. OS tumor cells also contribute by secreting IL-6 and IL-8, which recruit and condition immune and stromal cells. Notably, circulating tumor cells (CTCs) may actively remodel distant microenvironments, facilitating metastasis through IL-6/STAT3 signaling [25].

Signaling pathways sustaining TIME

At the signaling level, OS tumors hijack multiple inflammatory and immunoregulatory pathways to maintain the suppressive TIME. Constitutive STAT3 activation promotes IL-10 and VEGF production [26], while IL-6-driven JAK/STAT3 and MEK/ERK cascades enhance OS cell survival and migration [27]. NF-κB signaling further recruits and polarizes TAMs and MDSCs, while upregulating immune checkpoints such as PD-L1²⁸. Hypoxia-inducible factors (HIFs) induce VEGF-C/D and CXCL12 expression, stimulating angiogenesis and myeloid cell recruitment [29]. Chemokines like CXCL8 (IL-8), CXCL10, and CXCL12 coordinate the trafficking of immune and stromal cells, shaping both immune exclusion and metastasis [30]. Together, these pathways form positive feedback loops that reinforce immune dysfunction and therapeutic resistance.

Spatial and single-cell insights

Recent single-cell RNA sequencing and spatial transcriptomic analyses have begun to unravel the complex immune landscape of OS [31, 32]. These technologies reveal significant heterogeneity among immune populations, with diverse macrophage phenotypes, rare innate-like T cells, and dynamic changes in response to therapy [33]. Spatial imaging has identified distinct cellular neighborhoods—such as PD-1+ T cell-MDSC clusters and immune-cold tumor cores—that correlate with survival outcomes. For instance, higher spatial proximity between effector and suppressive cells predicts poorer prognosis, while an “OS Spatial Score” quantifying immune-tumor contact has been proposed as a prognostic marker. Another spatial-transcriptomic analysis of OS lymph node metastases showed that OS cells actively remodel the local TIME by engaging myeloid cells, CAFs, and lymphocytes to create a permissive niche [32]. These spatial and molecular insights underscore the importance of considering TIME architecture, not just composition, in designing effective immunotherapies for OS.

Immune checkpoint blockade in osteosarcoma

Biology and clinical relevance

Immune checkpoints are critical brakes on anti-tumor immunity, and their ligands/receptors are widely dysregulated in OS. PD-1 and its ligand PD-L1 are of primary importance. PD-1 is upregulated on tumor-infiltrating CD4⁺ and CD8⁺ T cells in OS patients, both at primary and metastatic sites [34]. PD-L1 is expressed on about 20–25% of primary OS tumor cells, and is often induced by IFN-γ or oncogenic signaling [35]. Notably, expression of PD-L1 in OS correlates with the presence of T cells, dendritic cells, and NK cells, suggesting that PD-L1 is upregulated as an adaptive response to immune pressure. High PD-1/PD-L1 in OS metastases strongly associates with metastasis and worse overall survival. Mechanistically, PD-1 engagement on T cells recruits SHP2 phosphatase to dampen TCR signaling, inhibits Akt and MAPK pathways, and ultimately blunts T-cell proliferation and cytokine secretion [36]. PD-1/PD-L1 interactions also impair NK cells: blocking this axis increases NK granule B release and OS killing efficiency. In addition to PD-L1, PD-L2 (another PD-1 ligand) is expressed in OS, especially in lung metastases. PD-L2 knockdown in OS cells reduced metastasis in models by inhibiting the RhoA/ROCK/LIMK pathway and promoting autophagy [37]. Thus, PD-L1/PD-1 and PD-L2/PD-1 represent parallel pathways by which OS evades cytotoxic lymphocytes.

CTLA-4 is another inhibitory receptor found on T cells (particularly Tregs) in OS patients. CTLA-4 binds CD80/CD86 on antigen-presenting cells with higher affinity than CD28, outcompeting the costimulatory signal and reducing IL-2 production. CTLA-4 also promotes Treg activity and IDO expression in dendritic cells, further restraining effector responses. Although CTLA-4 expression has not been extensively mapped in OS specifically, trials of ipilimumab (anti-CTLA-4) in pediatric OS showed it was safe but did not improve outcomes [38]. This suggests that CTLA-4 plays a role but is only one of many redundant inhibitory pathways in OS.

Emerging checkpoints such as TIM-3, LAG-3, and TIGIT have also been detected in OS. Multiple studies indicate that TIM-3 and LAG-3 are frequently co-expressed with PD-1 on TILs in sarcomas, including OS. In a cohort of OS patients, the co-expression of PD-L1, TIM-3 and LAG-3 was positively correlated: tumors with high PD-L1 also showed high TIM-3 and LAG-3 on infiltrating immune cells [39]. These receptors contribute to T-cell exhaustion: TIM-3 ligation by galectin-9 induces T-cell apoptosis, while LAG-3 binding to MHC-II or FGL1 inhibits TCR signaling [40]. TIGIT, expressed on Tregs in OS, engages CD155 on tumor cells or DCs and reinforces immune suppression. Notably, one study found that pulmonary OS metastases had significantly higher expression of IDO, PD-1, PD-L1, LAG-3, and TIM-3 than primary tumors, implying that these checkpoints converge to create an “immune desert” in metastases [41]. Beyond these classical checkpoints, additional immune modulators such as HHLA2 and B7-H3 (CD276) are consistently expressed on OS cell membranes, representing dominant mechanisms of immune escape. These molecules provide new rationales for immune-checkpoint inhibition, although clinical outcomes have been limited so far.

These findings underscore two key features of OS immune evasion: (1) checkpoint expression in OS reflects adaptive immune pressure despite overall low T-cell infiltration, and (2) multiple redundant pathways operate simultaneously, making single-target interventions unlikely to succeed. Understanding the functional interplay among these checkpoints—and their spatial context within the tumor—is essential for designing effective therapeutic strategies.

Clinical outcomes of checkpoint blockade

Despite encouraging preclinical evidence, ICIs have shown limited efficacy in OS patients. In murine models, PD-1 blockade reduced tumor volume, increased CD8⁺ T-cell infiltration, and decreased Tregs. Another study found anti-PD-1 therapy in mice increased NK and M1 macrophage infiltration and induced apoptosis of OS lung metastases. Combination approaches have shown promise: for example, in mice combined anti-PD-L1 with anti-CTLA-4 and radiotherapy led to enhanced CD8 + T cell infiltration and tumor control.

However, clinical trials have yielded disappointing outcomes. In a Phase II study (SARC028), pembrolizumab (anti-PD-1) was given to 22 patients with advanced OS-only 1 patient (5%) had an objective response [41]. A larger retrospective analysis similarly reported only occasional, short-lived responses. CTLA-4 blockade has also failed as monotherapy: ipilimumab in 17 pediatric OS patients yielded no clinical benefit. Nivolumab (anti-PD-1) in 13 children was well tolerated but similarly ineffective [38]. The anti-PD-L1 antibody atezolizumab in 12 relapsed OS patients was safe but yielded no responses [42, 43]. Collectively, these trials indicate that single-agent ICIs produce little tumor shrinkage in OS [44].

A meta-analysis encompassing 327 patients across 10 clinical trials revealed that while tyrosine kinase inhibitors (TKIs) achieved longer overall survival (11.67 vs. 6.37 months) and progression-free survival (PFS) (4.79 vs. 1.46 months) than ICIs, a subset of osteosarcoma patients exhibited stable disease following anti-CTLA-4 plus radiotherapy [45]. Despite overall modest response rates, these data affirm that ICIs remain biologically active in at least a fraction of OS cases.

The TIME may determine ICI responsiveness. Exploratory biomarkers such as the neutrophil-to-lymphocyte ratio, TIM-3, TP53 mutation status, LAG-3, IDO, and tertiary lymphoid structures (TLS) are being investigated for predictive value. Notably, tumor-associated TLS were detected in 17.3% of advanced OS cases, and patients harboring these TLS demonstrated improved median PFS (5.5 months) compared with those lacking TLS (3.0 months) [46, 47]. The PEMBROSARC trial (NCT02406781) evaluated PD-1 inhibition in TLS-positive soft-tissue sarcomas, including 33 osteosarcoma cases, reporting an ORR of 26.7% and a 6-month non-progression rate of 40%. These findings suggest that TLS presence may predict ICI benefit, and promoting TLS formation could enhance immunotherapy efficacy in OS [48].

Furthermore, combination strategies involving ICIs continue to evolve. In a phase II trial (NCT02815995) assessing durvalumab (anti-PD-L1) plus tremelimumab (anti-CTLA-4), the 12-week PFS rate reached 49%, with manageable toxicities (grade 3–4 AEs in ~ 37% of patients) [49]. The pembrolizumab + metronomic cyclophosphamide regimen also induced partial responses in a subset of OS patients (6.7%), despite negative PD-L1 expression [50]. These outcomes underscore the potential of combinatorial immunotherapy, particularly when guided by TIME biomarkers such as TLS and myeloid-cell signatures.

These results highlight a critical need to reframe expectations: in OS, ICIs alone are insufficient to remodel the TIME or overcome immune escape. Therapeutic efforts must therefore move beyond monotherapy to combination regimens informed by TIME biology.

Mechanisms of resistance to ICIs

Checkpoint expression heterogeneity is one barrier, as noted, PD-L1 is only present in a minority of OS tumors (~ 25%) [51], and PD-1/PD-L1 levels vary between primary and metastatic sites. Loss of MHC class I expression is frequently observed in OS, further impeding T-cell recognition of neoantigens [52]. OS has a moderate mutational burden with many structural variants, but paradoxically few strong neoantigens have been identified, so the neoantigen pool is relatively limited compared to highly mutated cancers. This antigen paucity renders many OS tumors “invisible” to T cells.

The immune-excluded TIME is another major mechanism. The dense osteoid-rich ECM physically impedes T-cell infiltration [53]. Tumor nests in OS often reside in low-oxygen, fibrotic zones where T cells cannot efficiently penetrate. High intratumoral pressure and abnormal vasculature further limit leukocyte trafficking. Even when T cells are present, immunosuppressive cells dominate: TAMs, MDSCs, and Tregs outnumber effectors and secrete suppressive mediators [54, 55]. Accumulation of metabolites like adenosine and lactate likely creates a metabolically hostile milieu for T cells. For example, OS cells and stromal cells upregulate IDO in response to inflammation [56]; in one study, high IDO in OS samples was linked to worse outcome. IDO catabolizes tryptophan into kynurenines, starving effector T cells and promoting Tregs [57].

Cytokine milieu also favors suppression: high intratumoral TGF-β drives fibrogenesis and directly inhibits cytotoxic lymphocytes and NK cells. OS CAFs secrete IL-6, which not only fuels tumor growth via STAT3 but also polarizes macrophages to an M2 state [58]. These combined factors, checkpoint heterogeneity, lack of cognate neoantigens, physical and cellular exclusion, and suppressive metabolites, create a formidable wall of resistance to checkpoint inhibition.

Strategies to improve ICI efficacy

Given the multifaceted resistance to ICIs, combination strategies are being developed to sensitize OS to immunotherapy (Fig. 2). Radiation and certain chemotherapies can convert tumors into immunogenic hubs. Radiotherapy enhances antigen release and type I interferon signaling, boosting dendritic cell maturation [58]. In preclinical OS models, combining radiotherapy with PD-1/CTLA-4 blockade significantly increased CD8⁺ infiltration and Treg depletion [58]. Chemo-immunotherapy regimens (e.g., doxorubicin + PD-1 blockade) have shown promise, potentially via immunogenic cell death and MHC upregulation [59]. Anti-angiogenic agents (e.g., VEGF inhibitors) may also enhance T-cell entry by normalizing vasculature and limiting MDSC infiltration [60]. Oncolytic viruses are another emerging adjuvant, by infecting OS cells, they stimulate innate immunity and can be engineered to express cytokines.

Fig. 2.

Summary of current treatment strategies for osteosarcoma

Reprogramming suppressive cells is also a key focus. Colony-stimulating factor 1 receptor (CSF1R) inhibitors (e.g., pexidartinib) can shift TAMs toward pro-inflammatory states. CXCR4 antagonists block MDSC trafficking, while ATRA induces MDSC differentiation. IDO inhibitors combined with chemotherapy have shown synergistic anti-tumor effects in OS models [61, 62]. Metabolic modulators are also under study: for example, combining an IDO inhibitor (1-MT) with gemcitabine chemotherapy reversed immune suppression and led to synergistic killing of OS cells in preclinical models [63]. These approaches aim to reduce the immunosuppressive barriers and allow ICIs to work.

Cancer vaccines and dendritic cell activation are additional avenues to boost anti-OS immunity. Personalized neoantigen or peptide vaccines could elicit new T-cell responses against OS-specific mutations. While OS has had few such trials, neoantigen prediction pipelines now suggest many OS tumors harbor candidate peptides. Dendritic cell vaccines using patient-derived tumor lysates have shown safety and some immune activation [64]. Combining DC vaccines with checkpoint blockade is an appealing concept to both prime T cells and then release them from inhibition. In some studies, oncolytic viruses or adjuvants (GM-CSF, TLR agonists) are used to mature DCs and improve antigen presentation in OS.

Neoantigen-targeted and metabolic reprogramming approaches are also under development. Given the genomic instability of OS [65], next-generation sequencing can identify neoepitopes unique to each tumor. Vaccines or TCR-engineered T cells against such neoantigens are a precision strategy, though still experimental in OS. On the metabolic front, blocking adenosine signaling (e.g., A2A receptor antagonists) may reverse T-cell suppression, as adenosine is known to accumulate in hypoxic TIME (an active area of research in sarcomas) [66]. Inhibiting glutamine or glycolysis pathways in OS could also alter the TIME acidity and nutrient landscape to favor immune cell function.

Altogether, these strategies aim to reshape the OS TIME to enable effective immune checkpoint blockade. Integrated approaches, guided by TIME profiling and spatial transcriptomics, may offer the best chance of overcoming resistance and achieving durable immunotherapy responses in OS.

Rewiring the osteosarcoma immune microenvironment

T cell-based therapeutic strategies

CAR T-cell therapies have been extended to OS antigens. Unlike T-cell checkpoint therapies, CAR-T cells can target tumor surface proteins without MHC. Several targets have been identified on OS cells, including ganglioside GD2, HER2, IL-1 receptor α, IGF-1R, EphA2, ROR1, CSPG4, and CD44v6 [67]. GD2 and HER2 CAR-T cells showed potent in vitro cytotoxicity against OS lines; a Phase I trial of GD2-CAR T cells (third-generation construct, NCT02107963) demonstrated safety in OS patients [68]. However, CAR-T efficacy in OS has been limited by poor trafficking to bone lesions and by antigen loss. Strategies to overcome this include CAR-T “armoring” with cytokine support or using bispecific constructs (e.g., CD3×GD2 bispecific antibodies). TCR-engineered T cells recognizing intracellular OS antigens (e.g., neoantigens or overexpressed peptides) are also under development but face HLA restriction and TME barriers.

T cells engineered to express membrane-anchored and tumor-targeted interleukin-12 (attIL12-T cells) not only target tumor cells but also disrupt the ECM and eliminate CAFs, thereby enhancing T-cell infiltration and anti-tumor activity [69]. Additionally, integrating CAR-T cell therapy with chemotherapy agents, such as low-dose doxorubicin, can modulate the TIME by reducing PD-L1 expression and facilitating T-cell infiltration.

Macrophage-based therapeutic strategies

Macrophage-directed therapy represents one of the few immune strategies in osteosarcoma that has reached phase III clinical testing. The prototypical agent, muramyl tripeptide phosphatidylethanolamine (MTP-PE; mifamurtide), activates monocytes via NOD2 signaling to selectively eliminate tumor cells while sparing normal counterparts [70, 71]. Although the pivotal trial combining mifamurtide with standard MAP chemotherapy for resectable non-metastatic overall survival was confounded by a factorial design that also tested ifosfamide, mifamurtide achieved a statistically significant improvement in OS despite not meeting the event-free survival (EFS) endpoint [72]. This survival benefit, particularly pronounced in younger and localized disease cohorts, led to EMA approval but not FDA endorsement, underscoring the divergent regulatory interpretations of marginal statistical gains versus clinical meaningfulness.

Real-world data and subsequent non-randomized studies have shown 1-year and 2-year OS rates of 71.7% and 45.9% in patients with resectable recurrent or metastatic disease [73]. While these findings are encouraging compared with historical controls, the absence of randomized confirmation limits their interpretability. Thus, the ongoing NCT03643133 trial is crucial to determine whether mifamurtide offers reproducible benefit when added to modern multimodal therapy. For now, the drug remains available in the U.S. only through compassionate-use programs, reflecting the regulatory and logistical challenges that continue to impede its broader clinical adoption.

In parallel, other macrophage-activating strategies such as interferon-α2b maintenance therapy (EURAMOS-1) failed to improve EFS or OS [74], highlighting that immune activation per se is insufficient without precise understanding of TME context and timing. The failure of IFNα2b, despite its immunostimulatory profile, underscores the importance of spatially and temporally coordinated macrophage reprogramming rather than systemic activation.

Beyond broad activation, precision macrophage modulation has emerged as a next-generation concept. The CD47–SIRPα axis represents a prototypical “don’t eat me” signal co-opted by OS cells to evade phagocytosis. Blockade of CD47 can enhance macrophage-mediated clearance of OS cells and suppress pulmonary metastasis formation in PDX models [75]. Clinically, the anti-CD47 antibody magrolimab has shown manageable safety and pharmacodynamic activity in early-phase trials across tumor types [75]. Although monotherapy responses were limited, its favorable tolerability provides a rationale for combination strategies, particularly with anti-GD2 antibodies that engage macrophages through FcγR signaling [76]. This synergy directly addresses the immune effector deficit characteristic of OS and could potentially augment antibody-dependent cellular phagocytosis in metastatic lesions.

From a metastasis-focused clinical perspective, lung macrophages play a pivotal role in establishing a permissive niche for OS dissemination. Tumor-cell expression of VCAM-1 facilitates adhesion to integrins on pulmonary macrophages, promoting metastatic seeding and outgrowth [77]. The development of anti-VCAM-1 monoclonal antibodies aims to disrupt this crosstalk and thereby prevent colonization of the pre-metastatic lung niche [78, 79]. Such approaches could have high translational value in the adjuvant or minimal-residual-disease setting, where metastatic establishment rather than bulk tumor growth dictates patient outcomes.

In summary, targeting macrophages in OS is transitioning from non-specific immune stimulation (mifamurtide, IFNα2b) to mechanistically precise modulation (CD47 blockade, VCAM-1 targeting). While mifamurtide remains the only macrophage-directed therapy with phase III evidence, its limited regulatory acceptance and modest absolute benefit reflect the need for better patient selection and combinatorial design. Future directions should integrate macrophage-modulating agents with checkpoint inhibitors or targeted therapies, guided by biomarkers of TAM density and polarization. Such integration, rather than additive combination, may finally leverage the macrophage axis toward clinically meaningful gains in OS survival.

DC-based therapeutic strategies

DCs are essential for initiating and regulating anti-tumor immune responses. Beyond DC vaccines, strategies to activate tumor-infiltrating DC subsets are being studied. Intratumoral administration of TLR or STING agonists can activate resident DCs, enhancing antigen presentation and T-cell priming [80]. In vitro-generated CD103⁺ conventional dendritic cells (cDC1s), when pulsed with tumor antigens and activated with poly I: C, have shown superior efficacy in inducing systemic and durable anti-tumor T-cell responses in osteosarcoma models. Combining CD103⁺ cDC1 vaccination with immune checkpoint blockade, such as CTLA-4 inhibition, has resulted in complete tumor regression in preclinical studies [81].

Boosting the immune infiltration

The immune cell composition of the osteosarcoma microenvironment is characterized by complex heterogeneity, involving both TAMs and TILs. Approximately 75% of osteosarcomas demonstrate infiltration by T cells and macrophages [24], suggesting a potentially immunogenic phenotype. However, this immune presence does not necessarily translate into an effective antitumor response. Increased TIL abundance often coincides with the upregulation of PD-L1 and the B7 family protein HHLA2, both of which are associated with unfavorable event-free survival [35]. In contrast, a higher ratio of cytotoxic CD8 + T cells to FOXP3 + regulatory T cells correlates with markedly improved outcomes, with nearly all patients above the median ratio surviving beyond 16 years, compared with less than 50% in those below the median [82]. These findings emphasize that the immune balance, rather than the absolute number of infiltrating cells, dictates tumor control in OS, highlighting immune evasion as a key pathogenic mechanism.

Efforts to harness this immune potential have centered on both antigen-agnostic and antigen-specific strategies. Adoptive transfer of expanded TILs harvested from surgical specimens represents a promising approach under clinical evaluation (NCT03449108) [82]. After ex vivo expansion and reinfusion, these lymphocytes are supported with IL-2 to promote in vivo activation, theoretically enabling them to overcome the highly suppressive osteosarcoma microenvironment. Although still experimental, this approach provides a proof of concept that immune effector dysfunction in OS can be reversed through mechanical and pharmacologic reprogramming.

In parallel, NK cell–based therapies are emerging as an alternative to T cell–directed immunotherapy. NK cells, equipped with activating and inhibitory receptors, can recognize stress-induced ligands and mediate cytotoxicity independent of antigen specificity. Preclinical studies have confirmed NK cell cytotoxicity against OS cells both in vitro and in vivo [83–85], and ongoing clinical trials (NCT02100891) aim to assess their safety and efficacy in relapsed or refractory disease. More recently, CAR-engineered NK cells have extended this concept by combining innate cytotoxicity with antigen-specific targeting, paralleling the success of CAR-T therapy in hematologic malignancies [86]. These next-generation NK approaches hold the potential to integrate both innate and adaptive immunity, particularly in tumors with low MHC expression or poor T cell infiltration.

The challenge remains to define which subsets of patients harbor immunogenic tumors amenable to such strategies and to identify biomarkers that reflect functional, rather than quantitative, immune infiltration. Future research should integrate genomic and spatial transcriptomic profiling to refine patient selection, elucidate resistance mechanisms, and design combinatorial therapies that can meaningfully extend survival in osteosarcoma.

Adjuvant and neoantigen-based approaches

Neoantigen vaccines and adjuvant strategies to reshape the TIME are similarly being studied. Oncolytic bacteria (e.g. Listeria expressing tumor antigens) can trigger innate immunity and prime T cells [87]. Checkpoint modulation of innate cells (e.g. agonists of STING or TLRs delivered intratumorally) can induce a cascade of type I interferons, enhancing antigen presentation [88]. Such adjuvant therapies essentially convert non-T-cell elements into allies against OS. Moreover, a phase I clinical study (NCT01241162) involving patients with neuroblastoma and sarcoma evaluated the combination of the DNA methyltransferase inhibitor decitabine with a dendritic cell vaccine targeting tumor-associated antigens MAGE-A1, NY-ESO-1, and MAGE-A3. The regimen was well tolerated, demonstrating favorable safety with no dose-limiting toxicities. However, its therapeutic efficacy remained uncertain, likely owing to the limited sample size and early-phase nature of the trial [89].

In summary, these multifaceted strategies targeting T cells, TAMs, DCs, and the ECM/CAF components of the TIME hold promise for enhancing the efficacy of immunotherapies in osteosarcoma. Representative clinical trials of therapeutic modalities are summarized in Table 1, which outlines key ongoing or completed clinical studies. Continued research into these combinatorial approaches is essential for developing effective treatments for this aggressive malignancy.

Table 1.

Clinical trials targeting key components of the osteosarcoma tumor immune microenvironment

Targets	Treatments	Conditions	Phase	Status	Allocation	Clinicaltrials.gov identifier
Immune checkpoint blockade	Pembrolizumab	Advanced Sarcomas	II	Completed	Non-Randomized	NCT02301039.
	Ipilimumab	Advanced and/or refractory solid tumors in Children	I	Completed	Non-Randomized	NCT01445379
	Atezolizumab	Solid Tumors in Pediatric and Young Adult	I/II	Terminated	N/A	NCT02541604
	Avelumab	Recurrent or Progressive Osteosarcoma	II	Completed	N/A	NCT03006848
	Nivolumab + Azacitidine	Recurrent, Resectable Osteosarcoma	Ib/II	Active, not recruiting	N/A	NCT03628209
	Pembrolizumab + Metronomic Cyclophosphamide	Advanced Sarcomas	II	Recruiting	Non-Randomized	NCT02406781
	Tremelimumab + Durvalumab	Multiple Sarcoma Subtypes	II	Completed	Non-Randomized	NCT02815995
	Oleclumab + Durvalumab	Recurrent, Refractory, or Metastatic Sarcoma	II	Active, not recruiting	N/A	NCT04668300
T Cell	Anti-GD2-CAR engineered T cells	Children and Young Adults With GD2 + Solid Tumors	I	Completed	Non-Randomized	NCT02107963
Macrophage	Mifamurtide + Post-operative Chemotherapy	Newly Diagnosed High Risk Osteosarcoma	II	Recruiting	Active, not recruiting	NCT03643133
	Combination Chemotherapy, PEG-Interferon Alfa-2b, and Surgery	Resectable Osteosarcoma	III	Completed	Randomized	NCT00134030
	Magrolimab + Dinutuximab	Relapsed Osteosarcoma	I	Completed	Non-Randomized	NCT04751383
TILs	LN-145 or LN-145-S1	Osteosarcoma	II	Active, not recruiting	Non-Randomized	NCT03449108
	Haploidentical allo-HSCT + donor NK cells	High-risk solid tumors	II	Terminated	N/A	NCT02100891

Translational impact and future directions

The multi-omics dissection of the OS TIME is fundamentally reshaping therapeutic paradigms. High-dimensional studies have revealed patient subgroups (e.g. “immune-inflamed” vs. “immune-desert” tumors) that could be stratified for different treatments. For example, patients whose tumors have a high “OS Spatial Score” (dense immune infiltrates contacting tumor) might benefit more from ICIs alone, while “immune-cold” patients may need ECM-targeting or myeloid modulators first [90]. Integration of single-cell RNA-seq, spatial proteomics, and TCR/BCR repertoire sequencing allows mapping of antitumor clonotypes and exhaustion states—this could guide personalized immunotherapies (e.g. engineering T cells from existing clones).

As a predictor of therapeutic efficacy, biomarkers such as the “OS Spatial Score” should be regarded not merely as descriptive metrics but as decision rules that can be prospectively operationalized. Practically, the score can be quantified using multiplex immunofluorescence by enumerating CD8⁺ T cells (and optionally GZMB⁺/PD-1⁺ subsets) within a defined interaction radius (e.g., 20–30 μm) of cytokeratin- or osteosarcoma-marker–positive tumor cells per mm², weighted by direct contact frequency. An analogous estimate can be derived from spatial transcriptomics through spot deconvolution and nearest-neighbor graph modeling. To minimize sampling bias in calcified or necrotic bone, multi-regional sampling (≥ 3 representative ROIs) and reporting of both the mean and dispersion of the score are recommended, as high spatial heterogeneity itself may predict unstable responses to ICIs. Importantly, the Spatial Score should be treated as a dynamic biomarker rather than a static baseline measure. Reassessment can help determine whether immune ingress has occurred before committing to prolonged checkpoint blockade or targeting therapy.

Mechanistic studies have also uncovered novel therapeutic vulnerabilities the NF-κB pathway in OS is emerging as a vulnerability—preclinical work suggests combining NF-κB inhibitors with checkpoint blockade could overcome resistance by simultaneously preventing PD-L1 induction and TAM/MDSC recruitment [28]. Multi-omics also highlight metabolic axes (e.g. IDO-kynurenine, adenosine) as key nodes; clinical trials of IDO inhibitors (e.g. epacadostat) in OS are being considered in combination with anti-PD-1²⁸. Spatial transcriptomics may further reveal microregional niches (such as the “immune-cold parenchyma” or “macrophage-rich corridors”) that can be targeted by local therapies (e.g. intratumoral cytokine injection or engineered oncolytic viruses).

Despite these advances, several challenges remain. Translating multi-omics insights into actionable clinical biomarkers is still in its infancy. The rarity and heterogeneity of OS complicate patient stratification and limit trial scalability. Furthermore, while precision combination therapies appear promising in preclinical settings, their safety, sequencing, and synergistic windows in humans are largely undefined.

Looking forward, precision immunotherapy design in OS will likely involve profiling each tumor’s TIME, mutanome, and stromal composition. This could enable rational combinations: for instance, an OS patient whose tumor shows high CXCL12 and MDSCs might receive CXCR4 inhibitors plus ICIs; another with abundant suppressive macrophages might get CSF1R blockade. Adoption of advanced technologies (spatial ATAC-seq, multiplex imaging, organoid co-cultures) will facilitate the translation of these insights. Ultimately, the convergence of genomics, immunology, and bioengineering promises to transform OS therapy by turning the immune microenvironment from foe to friend, and by delivering multi-modal, mechanism-based, and personalized treatments [32].

Author contributions

Jinlin Cai: Conceptualization, Writing—original draft, Visualization, Writing—review & editing.Shijie Qiu: Writing—review & editing.Biao Sun: Writing—review & editing.Jianbin Ge: Writing—review & editing.Zhe Yu: Supervision, Project administration, Writing – review & editing.Chao Wang: Supervision, Conceptualization, Writing—review & editing.

Funding

This research received no external funding.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

The manuscript has been approved for publication by all authors.

Competing interests

The authors declare no competing interests.