Microenvironment matters: insights from the FOSTER consortium on microenvironment-driven approaches to osteosarcoma therapy

Abstract

Osteosarcoma (OS), a prevalent malignant bone tumor, has seen limited progress in treatment efficacy and patient outcomes over decades. Recent insights into the tumor microenvironment (TME) have revealed its crucial role in tumor progression and therapeutic resistance, particularly in OS. This review offers a comprehensive exploration of the OS microenvironment, meticulously dissecting its crucial components: the mesenchymal stromal TME, the immune microenvironment, hypoxia-induced adaptations, and the impact of the physical microenvironment. By demonstrating how these elements collectively drive tumor proliferation, immune evasion, and invasion, this review explores the intricate molecular and cellular dynamics at play. Furthermore, innovative approaches targeting the OS microenvironment, such as immunotherapies, are presented. This review highlights the importance of the TME in OS progression and its potential as a source of novel therapeutic strategies, offering new hope for improved patient outcomes.

Introduction

Osteosarcoma (OS) is a rare malignant primitive bone tumor that mainly affects children and young adults. Despite advances in cancer treatment, progress in the management of OS has been relatively limited. Traditional approaches, including surgery and chemotherapy, have remained the mainstays of treatment for decades, with modest improvements in survival rates. Targeted therapies in OS have had limited success, largely due to the complex and heterogeneous nature of the disease. The genomic complexity of osteosarcoma, resulting from chromotrypsis, leads to numerous somatic mutations and gene copy number alterations. This lack of significant progress underscores the complexity of OS biology and highlights the need for novel therapeutic strategies.

In recent years, there has been a growing recognition of the importance of the tumor microenvironment (TME) in driving tumor progression and therapeutic resistance, particularly in OS. The unique characteristics of the OS microenvironment, including alterations in the extracellular matrix (ECM), immune cell interactions, and hypoxic conditions, play critical roles in shaping tumor behavior and response to treatment. Therefore, understanding the intricate dynamics of the OS microenvironment is crucial for deciphering the complexities driving tumor development and progression.

This review focuses on four key aspects of the OS microenvironment: (i) the mesenchymal stromal TME, (ii) the immune microenvironment, (iii) the impact of hypoxia, and (iv) the influence of physical microenvironment, shedding light on their involvement in OS progression and resistance. The mesenchymal stromal TME orchestrates activities that sustain tumorigenesis and influences the synthesis of an altered ECM, fueling tumor progression and contributing to therapeutic resistance. Additionally, dysregulated interactions between tumor cells and immune subsets of the microenvironment play a pivotal role in OS progression, and disrupt the osteo-immunological balance, challenging traditional pro- and anti-tumor characteristics within these cellular subsets. Hypoxia is another critical determinant in the OS microenvironment that drives cellular adaptation, impacting tumor cell proliferation and influencing the immune TME. Hypoxia-induced neoangiogenesis also facilitates local and distant invasion of OS cells. Alterations in the physical microenvironment, such as changes in ECM composition and mechanical properties, significantly impact OS progression and chemotherapy efficacy. Finally, this review examines the immunotherapies that have been employed in OS clinical trials in recent years and discusses the potential of new treatments, particularly focusing on promising approaches such as anti-GD2 antibodies, T cells expressing chimeric antigen receptors (CAR-T), CD47 inhibition and indoleamine-2,3-dioxygenase (IDO). The review article emphasizes that a deeper understanding of the OS environment is essential for developing effective treatments and improving patient outcomes.

Role of mesenchymal stromal cells in osteosarcoma progression

The mesenchymal stromal TME is pivotal in the emergence and progression of OS. It refers to the specialized cellular microenvironment surrounding tumor cells and consisting of embryonic mesoderm-derived, non-malignant host cells such as mesenchymal stem cells (MSCs), cancer-associated fibroblasts (CAFs), osteoblasts, adipocytes, and endothelial cells [1]. These various cell types collectively elicit an extensive tumor-sustaining and tumor-promoting program of activities, including release of growth factors, exosome-encapsulated RNA and metabolites, and cytokines. These activities cause a plethora of responses, including enhanced blood vessel formation, tumor expansion, tumor invasion, and promotion of the cancer stem cell (CSC) phenotype (Fig. 1).

Fig. 1.

Signaling within the mesenchymal TME, highlighting the prominent contribution of MSC cells and the complex consequence of incoming and outgoing signalling cues on OS behaviour and TME architecture (created with Biorender.com)

In addition, TME-resident mesenchymal cells direct the synthesis of altered ECM, incorporating ECM protein isoforms rarely found in normal tissues, such as oncofoetal forms of fibronectin [2]. Such alternate ECM protein isoforms play a known role in promoting tumor progression [3] and could constitute an opportunity for site-selective delivery of anticancer therapeutics in OS [4, 5]. Thus, antibodies against these oncofoetal fibronectin could serve to direct the targeted delivery of anticancer agents, including cytotoxic prodrugs or nanoparticles with anticancer or immune-enhancing cargo.

Long-standing evidence supports the involvement of mesenchymal TME in pathological bone remodeling. This perturbation in the balance between bone formation and resorption is a hallmark of OS, leading to skeletal de-stabilization, local tumor expansion, and enhanced metastasis. Clinical investigation of bone injuries caused by this skeletal destabilization often prompts the diagnosis of OS. Strategies targeting bone resorption or osteoclasts activity have shown potential in inhibiting OS progression in preclinical studies, although clinical testing has not as yet demonstrated significant benefit [6–8].

Numerous studies emphasize the critical contribution of MSCs within the OS TME in influencing the fate and behavior of OS tumor cells [9, 10]. Several lines of evidence indicate that communication by MSCs involves the sharing of biomolecules via extracellular vesicles (EVs). EVs derived from bone marrow–derived MSCs promoted OS cell proliferation, invasion, and metastasis, in part due to the small noncoding microRNA miR-21-5p. This is responsible for the upregulation of the PI3K/AKT kinase and hedgehog pathways in OS, as well as upregulation of collagen beta (1-O) galactosyltransferase 2 protein (COLGALT2), which has a role in post-translational modification of collagen and other ECM proteins [11–13]. Systematic characterization of exosome cargo derived from bone marrow MSCs has identified a rich and complex mixture of non-coding RNAs, proteins and lipids, not all of which were consistently shared between different donors. This implies the possibility of patient-specific effects of EVs on cancer biology and behavior [14].

Other work describes the role of MSCs in supporting OS development and progression through the production of inflammatory cytokines. MSCs-conditioned medium promoted OS proliferation and metastasis by stimulating ERK and AKT signaling via IL-12, and migration inhibitory factor (MIF)–induced activation of the chemokine receptor CXCR4 [15], as well as activation of STAT3 via IL-6 [16, 17]. Furthermore, IL-6 secretion by MSCs reversed the inhibition of β2-adrenergic receptor (β2AR) by hypoxia, enabling β2AR agonist-mediated hypoxia-resistant proliferation of OS-derived cells [18]. Finally, co-culture of MSCs enhanced the resistance of OS cells to anoikis, via IL-8 secretion, activating the CXCR1/AKT pathway in OS cells and enabling anchorage-independent growth and pulmonary metastasis [19].

Significantly, the proximity of MSCs and OS cells supports the tumor’s energy requirements. Reactive oxygen production by OS cells leads to metabolic reprogramming of MSCs, enabling them to undertake anaerobic glycolysis with enhanced lactate production and excretion. This process, in turn, supplies neighboring OS cells with this energy-rich metabolite [20]. Intriguingly, tumor-derived MSCs share similar morphology and surface markers with normal tissue MSCs, but exhibit altered differentiation plasticity—including poor adipogenic potential and increased osteogenic potential compared to normal tissue MSCs [21].

Interactions with TME-resident MSCs has also been shown to affect the response of OS cells to chemotherapy. TGF-β production, activated by the peptide hormone leptin in tumor-resident MSCs, induces autophagy-linked platinum chemoresistance in OS [22]. These observations are consistent with earlier work associating high expression of leptin and sirtuin-1 in OS tumor tissue with poor prognosis in patients with OS [23], and evidence of TGF-β signaling axis leading to chemoresistance [24, 25]. Inhibitors of TGF-β signaling may therefore represent a route to enhance efficacy of chemotherapy in OS patients.

Other research indicates that OS cells actively shape the composition and direct activity within the mesenchymal stromal TME, thereby supporting their own proliferation and malignant progression. OS cells attract bone-derived MSCs through the secretion of C–C motif chemokine ligand 2 (CCL2)/monocyte chemoattractant protein 1 (MCP-1), growth-regulated oncogene alpha (GRO-α), and TGF-β1 and orchestrate their trans-differentiation to CAFs. These CAFs are distinct from native fibroblasts by the expression of alpha-smooth muscle actin and fibroblast surface protein as well as concomitant increase in CCL2, GRO-α, IL-6, and IL-8 secretion. This leads to a mesenchymal to amoeboid transition of the cancer cells, trans-endothelial migration, and metastasis [26]. An IL-8-driven reciprocal positive feedback interaction between OS and MSCs has also been shown to induce mobility and metastasis [27]. Co-culture of OS and MSCs, leading to IL-6 production, results in enhanced stemness in OS [28]. Furthermore, WNT signaling leading to TGF-β production in OS cells safeguards stemness in MSCs and blocks their osteogenic differentiation, while independently augmenting MSC-driven production of VEGF, involved in angiogenesis, and IL-6, involved in promoting OS stemness and proliferation [29].

Acidosis, caused by OS cell proliferation, causes activation of the transcription factor nuclear factor-kappa B (NFĸB) in MSCs, enhancing their paracrine inflammatory secretome activity. This includes the production of cytokines IL-6 and IL-8, granulocyte–macrophage colony-stimulating factor (GM-CSF) and granulocyte colony-stimulating factor (G-CSF) which enhances OS stemness, proliferation, and metastasis [30, 31]. Furthermore, this is linked to OS cell resistance to doxorubicin [32]. Finally, OS-derived EVs affect epigenomic changes in MSCs and influence the expression of genes related to bone microenvironment remodeling, including matrix metalloproteinase-1 (MMP1) expression, involved in extracellular matrix remodeling, and vascular endothelial growth factor (VEGF-A), involved in endothelial cell proliferation and the promotion of angiogenesis [33].

Long-standing evidence emphasizes considerable cellular complexity and inter-tumor heterogeneity within the OS mesenchymal TME, highlighting the variable presence of normal osteoblasts, undifferentiated osteoblastic and chondroblastic bone precursors, endothelial cells, and CAFs [34, 35]. Recent single-cell RNA sequencing (scRNAseq) studies are consistent with these views and provide unprecedented insight into the signaling interactions between different cell components. Work based on 11 OS samples identified 25 cell clusters, underscoring the high cellular heterogeneity in OS cancer tissue. Using cell lineage–specific markers, these clusters could be assigned to 11 different cell types, including chondroblastic cells, osteoblastic cells, pericytes, fibroblasts, endothelial cells, and MSCs, all representing cell components of mesenchymal origin [36]. Consistent with previous research, these mesenchymal cell types were present in all analyzed cancers. However, their relative abundance was highly variable between different samples [36].

Notably, communication network analysis based on receptor-ligand pairing confirmed the prominent role of MSCs in the communication with OS cells and other components of the TME [37]. Analysis of data from tumors with and without prior neoadjuvant therapy revealed intriguing communication network differences, including a substantial increase in receptor-ligand pairing involving osteoblasts and the surrounding stromal TME in patients who received chemotherapy [38].

Significantly, this work further raises the possibility of trans-differentiation of OS cancer cells, yielding malignant cell subpopulation with fibroblastic, chondroblastic, and myoblast features. A second independent study based on the analysis of treatment-naïve primary OS from six patients reported similar conclusions concerning cancer cell plasticity [39]. In addition, this study emphasized the considerable molecular heterogeneity within the fibroblast and osteoblast population [39] that may extend to other cell types.

Finally, a study comparing scRNAseq datasets of OS lung metastasis to datasets derived from normal lung highlighted overt molecular differences in the respective tissue-resident endothelial cells and further myeloid cells supporting the notion that tumors reprogrammed these cell types [40]. Other work indicates trans-differentiation of endothelial cells to chondroblast-like and fibroblast-like lineage [41].

Together, this work emphasizes the high level of complexity and cellular plasticity within the mesenchymal OS TME. Importantly, the existing and new data highlight and abundantly confirm a uniquely prominent role of MSCs in orchestrating signaling within this tissue environment.

Immune intrigue: macrophages, osteoclasts, and T cells in osteosarcoma

Over the past decade, the development of innovative high-throughput technologies and computational analyses tools has enabled considerable progress in describing and understanding the interactions between tumor cells and cells of the immune microenvironment, which can enable new therapeutic stratification of OS’s patients [42, 43]. The OS TME is the bone marrow, a highly dynamic environment composed of bone, immune, stromal, and vascular cells, embedded in a mineralized ECM [44–46]. In OS, the condition is characterized not only by molecular events that lead to the genetic instability of osteoblastic/mesenchymal derived cells but also by a disruption of osteoimmunological balance. This disruption involves deregulated interactions between tumor cells, immune cells, as well as osteoblastic and osteoclastic bone-remodeling cells [47]. Immune cells are schematically classified into myeloid and lymphoid compartments, with myeloid-derived cells being the predominant population in the OS microenvironment. These myeloid cells are particularly plastic and can exhibit either pro-tumorigenic or anti-tumorigenic characteristics, depending on the specific stimuli they encounter. However, this classification is somewhat theoretical, as the biomarkers commonly used to distinguish between cell populations are not always sufficiently discriminating. The different cellular subsets of the microenvironment are interconnected and plastic, with activities that evolve depending on stimuli and their degree of maturation [48–50].

In the last 10 years, the integrated analysis of clinical data and various immunohistochemical, transcriptomics, epigenomics, and genomics data have enabled us to better stratify OS patients [36, 39, 43, 51–64]. Recently, an integrated multi-omics analysis [43] of 121 OS biopsies, incorporating whole exome sequencing, bulk RNAseq, gene copy number, and DNA methylation profiles, identified four patient subgroups with distinct molecular characteristics and clinical prognoses. These subgroups are OS with significant MYC amplification or homologous recombination defects (the most aggressive); immune cold tumors (lacking active anticancer immunoreactivity); and immune hot tumors (showing active anticancer immunoreactivity). Importantly, the immune microenvironment of OS is rich in myeloid cells, such as monocytes, tumor-associated macrophages (TAMs), dendritic cells (DCs), and neutrophils, and is immunosuppressive. Conversely, it is deficient in tumor-infiltrating lymphocytes (TILs), including CD8 + , CD4 + , regulatory T cells (T-reg), and natural killer (NK) cytotoxic lymphocytes [58–60].

Transcriptomic approaches using scRNAseq, from eleven and six patients respectively [36, 39], confirmed the extreme inter- and intra-tumoral heterogeneity of OS cells and their immune microenvironment. These two studies demonstrated that in the myeloid compartment, monocytes and macrophages represented 70–80% of cells, DCs 5%, with neutrophils in the minority. Zhou Y et al. [36] identified three TAM subgroups, M1, M2, and M3. M1-TAMs expressed inflammatory factors, such as CCL2, CCL3, CCL4, CXCL2, and CXCL3, and activation of pro-inflammatory signaling pathways, including IFN-α, IFN-γ, IL-2/STAT5, and IL-6/JAK/STAT3. M1-TAMs also activated TGF-β and Hedgehog signaling pathways. It is known that M2-TAMs are the main tumor-associated anti-inflammatory macrophages, displayed relatively high expression of CD163, MRC1, MS4A4, and MAF, with a small proportion of M2-TAMs showing relatively high expression levels of M1-TAM marker genes, demonstrating the existence of a dynamic transformation between M1 and M2-TAMs in the OS microenvironment [36]. M3-TAMs reflected a FABP4 + subtype of alveolar macrophages with pro-inflammatory activity. These studies also showed that myeloid-derived CD1c + DCs, essential for the activation of CD8 T cells, could be used therapeutically [36] (discussed later in this article). The prognostic significance of cDC1 cells was confirmed by Liu W et al., who demonstrated that cDC1s are linked to a favorable prognosis [52]. This study also identifies a role for anti-phagocytic signals in immune evasion. By expressing anti-phagocytic signals, tumors can evade macrophage-mediated phagocytosis. CD47 is a classic antiphagocytic signal that binds to its receptor SIPRα on macrophages to protect cells from phagocytosis. Unfortunately, therapeutically blocking CD47 proved toxic, leading to the search for better candidates by which macrophage silencing could be reversed or prevented. More recently, CD24 (a CSC marker involved in self-renewal and differentiation of OS cells) was identified as a new antiphagocytic signal. CD24 has been shown to be highly expressed in high-grade OS cells, and its expression is associated with disruption of macrophage activation and phagocytosis, offering new therapeutic avenues [52].

Multiple studies agree on the deleterious role of osteoclasts, associating them with poor therapeutic response and prognosis [65]. A recent RNAseq bulk transcriptomic study [62] defined two groups of patients according to the molecular characteristics of the cells of their TME: one group with a favorable prognosis, associated with innate immunity, and a second group with an unfavorable prognosis, associated with angiogenesis and osteoclastic activity.

Immunohistochemistry and serum bioassay analyses conducted on 124 patients from the OS2006 cohort revealed that interactions between TAMs and osteoclasts negated the beneficial effect of zoledronate, a bisphosphonate with anti-osteoclastic activity [53]. This is likely due to zoledronate’s harmful impact on a subpopulation of pro-inflammatory macrophages, which are associated with favorable neoadjuvant therapeutic response and prognosis [53]. Similarly, a multiplex immunohistochemical approach showed a high proportion of cells co-stained with CD163 and CD68. These markers identify M1, M2 macrophages, and osteoclastic populations, highlighting the difficulty in clearly differentiating between pro-inflammatory M1 or pro-tumor M2 subgroups. This work led to the hypothesis of a subgroup of M0-polarized macrophagic cells, whose role could vary in response to different stimuli [54].

Zhou Y et al. [36] also noted a relatively lower proportion of mature OC cells in chondroblastic, lung metastatic, and recurrent lesions compared to primary OS lesions. They identified three distinct subclusters of osteoclasts. The first subgroup consisted of osteoclast progenitor cells expressing high levels of the myeloid markers CD74 and CD14, along with the osteoclastic markers cathepsin K (CTSK) and tartrate-resistant acid phosphatase 5 (ACP5), and exhibiting a hyperproliferative phenotype suggestive of stimulation by osteogenesis. The second group comprised immature osteoclasts co-expressing both myeloid and osteoclastic markers, while the third group included mature osteoclasts demonstrating high levels of CTSK and ACP5 and low CD74 and CD14. They also observed that the gene expression profiles of these cells evolved with maturation, undergoing a progressive transition from genes associated with antigenic presentation to those linked with bone resorption and remodeling [36]. In their quest for a prognostic signature from osteoclast populations, Shao H et al. [66] leveraged bulk RNA sequencing and scRNAseq data to identify eleven significant osteoclast differentiation–related genes (ODRGs) as key predictors of survival, including Tropomyosin 1 (TPM1), SERPINE2, and DCN. This study revealed that ORDGs in distinct differentiation states were enriched in distinct functions and pathways, revealing their complex role in OS progression [66].

The analysis of gene expression profiles of different TIL subgroups (CD4- CD8- T, CD8 + T, CD4 + T, T-reg, proliferating T cells, NK T cells, NK, CXCL13 + , B cells) revealed low lymphocyte infiltration in the OS microenvironment and higher expression of T cell depletion inhibitory receptors, such as T cell immunoreceptor with Ig and ITIM domains (TIGIT) on CD8, CD4, T-reg, and NK cells, offering new therapeutic perspectives with TIGIT receptor inhibitors [36].

The use of algorithms to deconvolve immune infiltration according to the gene expression profiles of tumor samples [63, 67–69] showed low lymphocyte immune infiltrate scores compared with other non-sarcomatous solid tumors or other sarcomas [63]. Similarly, Liu Y et al. [39] reported that certain macrophage subtypes, M2-polarising TXNIP + and pro-inflammatory IFIT1 + , could modulate regulatory T cells and contribute to CD8 + T cell depletion [39]. Moreover, T cell receptor (TCR) clonal diversity was found to be low in all OS tumors, regardless of being immune-hot or immune-cold, which may be due to adaptive immune resistance mechanisms, including an increase in cells expressing programmed cell death protein 1 (PD-1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), IDO, the presence of myeloid-derived suppressor cells (MDSCs), or lower levels of major histocompatibility class I genes (MHC-I) [63].

It has been reported that OS can evade immune surveillance and destruction by cytotoxic lymphocytes by suppressing MHC-I expression [52]. Liu et al. showed that MHC-I transcription and the interferon-gamma response are repressed in OS cells with high copy number variation [39]. Interestingly, neo-adjuvant chemotherapy can alter the local tumor immune landscape towards being richer in CD3 + T lymphocytes, CD8 + T lymphocytes, programmed death-ligand 1 + (PD-L1 +) immune cells, while reducing HLA-DR-CD33 + MDSCs. This suggests that tumors post-chemotherapy may be more sensitive to immune-activating therapies [63, 70]. Other interesting therapeutic strategies have been reported by Fujiwara et al. [71]. Their preclinical study showed that inhibition of Colony-stimulating factor 1 (CSF-1) depleted TAMs and FOXP3 + regulatory T cells and enhanced infiltration of CD8 + T cells into the microenvironments of both primary and metastatic osteosarcoma sites [71]. These results could lead to a new immunotherapy strategy for bone sarcomas.

Several studies have reported a positive effect of systemic infection on the growth of OS in dogs and humans, suggesting another avenue of therapeutic immunomodulation worth exploring [72]. For example, in a large cohort of OS patients, Jeys et al. observed better 10-year survival in patients who developed an infection within 1 year of orthopedic surgery (84.5% in the infected group vs 62.3% in the non-infected group) [73]. As bacterial proliferation activates Toll-Like Receptor 4 (TLR4) [70], this suggests a relationship between TLR4 and OS progression. Using a murine model, two studies reported that TLR4 activation by systemic administration of lipopolysaccharide (LPS) suppressed the progression of OS via stimulation of CD8 + T cells [72, 74]. These studies demonstrate the therapeutic potential of using TLR4 agonists in OS. Figure 2 summarizes the composition and signaling interactions within the OS immune TME.

Fig. 2.

Composition and signaling interactions of the osteosarcoma immune TME, highlighting the prominent role of macrophages and suppressor CD8 T in generating an immunosuppressive, immune-cold environment (created with Biorender.com)

Finally, natural killer (NK) cells are essential players in the innate immune system, known for their role in tumor surveillance. In OS, however, the study of NK cells within the tumor microenvironment is still in its early stages. Unlike more extensively researched cancers, such as breast cancer, the immune landscape of OS is less understood, with only a few studies focusing specifically on NK cells. Despite this, NK cells hold potential as important effectors in OS, possibly influencing tumor progression and emerging as targets for future therapeutic strategies.

The impact of neoangiogenesis and hypoxia in osteosarcoma

Hypoxia is a crucial factor in the interplay between OS cells and its TME. It is defined as a tissue oxygen concentration below the level needed for normal cell function. Physiological oxygen tension, termed physioxia, is defined as oxygen levels from 5 to 9% depending on body regions [75–77]. In bones, it is usually less than 9% [72, 74, 78]. Hypoxia arises from an imbalance between rapid proliferation of tumor cells and insufficient oxygen supply [79–82]. Evidence indicates that hypoxia-driven mechanisms are multi-faceted, impacting the OS cells themselves as well as broadly affecting components of the TME, including immune cells, angiogenic cells, fibroblasts, as well as the ECM.

The hypoxia-inducible factor (HIF) protein family was identified as key transcription factors that initiate cellular adaptation to hypoxia. The constitutively expressed subunit HIF1β dimerizes with one of two oxygen-dependent subunits, HIF1α, that accumulates under acute hypoxia, or HIF2α that accumulates under chronic hypoxia [80, 83]. Thereafter, the dimers formed directly bind hypoxia response elements (HREs) present on the regulatory region of hypoxia-regulated target genes [84], regulating a multitude of functional pathways that impact tumor activity. These include solute carrier family 2 member 1 (SLC2A1, better known as GLUT-1), a glucose transporter crucial in cell metabolism, neoangiogenesis (VEGF), along with WNT/β-catenin signaling, which leads to increased tumor cell resistance and invasiveness [79, 80, 83] (Fig. 3).

Fig. 3.

The molecular mechanisms by which hypoxia influences osteosarcoma progression through the hypoxia-inducible factor (HIF) protein family and associated pathways. Regulation of these pathways by HIF proteins under hypoxic conditions results in enhanced tumor progression, including increased glucose metabolism, angiogenesis, and invasiveness (created with Biorender.com). ARNT, aryl hydrocarbon receptor nuclear translocator; GLUT-1, glucose transporter 1; HRE, hypoxia response element; VEGF, vascular endothelial growth factor; VEGFR, receptors for vascular endothelial growth factor

Studies highlight the frequent overexpression of HIF family members and their targets as biomarkers of resistance and poor prognosis in OS [85–89]. Beyond its role in OS cell proliferation, HIF2α expression appears to induce tumor cell heterogeneity, resulting in less effective treatments. This clonal diversity within the tumor arises from genetic and epigenetic changes triggered by hypoxia, driving tumor evolution and resulting in varied patterns of cellular tumorigenicity, along with increased resistance and invasiveness [79, 83, 90]. The accumulation of multiple DNA abnormalities due to TME and hypoxia contributes to the complex karyotype typically seen in OS [62, 78, 89]. To maintain tumor cell growth and heterogeneity under hypoxic conditions, hypoxia-driven pathways activate the PI3K/PTEN/AKT pathway, which regulates cell survival, differentiation, and apoptosis [78, 80]. To enable cancer cells to cope with the subphysiological levels of oxygen, an adaptive change in metabolic activity is progressively implemented [90]. Single-cell approaches indicate that glycolysis and glutaminolysis are the primary metabolic pathways driving OS aggressiveness [39, 91, 92]. These metabolic patterns are also linked to specific immune environmental parameters, such as high levels of macrophages, CD8 + T cells, B cells, and MSCs, and are accompanied by reduced neoangiogenesis [92]. Several molecular players were shown to be involved, such as hexokinase, aldolase and GLUT1, but no specific metabolic signature has been identified that would allow for a transcriptomics-based metabolic classification of OS [39, 92].

The relationship between OS cells and immune TME seems to be centrally linked to hypoxia [39, 88, 92]. The presence of M1/M2 macrophage may be linked to the modulated expression of mTOR/HIF1 pathway [88]. Studies have identified a significant link between the enhanced expression of HIF1α and the presence of the pro-tumoral macrophages and osteoclasts [89, 92]. This cross-relationship positively influences the growth of OS and its invasiveness [93] and may be considered as a predictive marker of poor outcomes. This substantial correlation explains the increased presence of hypoxic pathways and osteoclastic phenotypes leading to T cell exhaustion in OS tumors [39, 53]. Recent single-cell approaches propose combined osteoclastic, immune, and hypoxic scores as a new method of classifying OS [39, 62, 92].

Other cellular interactions, including fibroblastic, endothelial, and osteoblastic cells, also respond to hypoxia. In addition to the immune and osteoclastic compartment, the bone TME is a complex environment where tumor tissue is surrounded by osteoblasts constructing normal bone. Hypoxia-induced modulation of matrix stiffness and angiogenesis helps nourish both normal and tumor cells within the TME and influence their biomechanical stimulation [91, 94]. Neoangiogenesis, which is primarily induced by hypoxic intratumoral niches with elevated HIF1α expression and further stimulated by the activation of the VEGF/VEGFR pathway [36, 92, 95, 96], is the key to enhancing local and distant OS invasion. The neovessels are characterized by interlaced appearance and chaotic growth. The primitive endothelial cells, mostly CD34 + , are then influenced by VEGF stimulation, supported by other pathways such as angiopoietins and TGFs [36, 89, 92, 95]. Many scRNAseq analyses [36, 91, 92, 97] highlight the role of this neoangiogenic TME, proposing a new ‘angiogenic score’ to supplement immune and hypoxic signatures and subclassify OS. Together, these studies enhance our understanding of the temporal and spatial impacts of hypoxia on OS progression. However, the absence of correlation between primary tumors and metastases in these studies complicates our comprehension of hypoxia’s role in the OS environment.

Together, these findings reveal that tumor hypoxia influences several TME factors, endowing OS cells with resistance to current treatments. This suggests that new strategic approaches will be important to counteract hypoxia-induced responses. For example, physiologically “reoxygenating” OS tissue through targeted therapeutic strategies could help enhance sensitivity to chemotherapy or treat disease relapse [92, 95, 98].

Osteosarcoma’s physical microenvironment

Bone is a dynamic tissue that undergoes constant remodeling by osteoblasts and osteoclasts throughout our lifetime [99]. Healthy bone tissue exhibits anisotropic properties; i.e., its properties depend on direction. The mechanical properties of cortical bone along its longitudinal direction differ from those along the radial direction, which typically correlates with the dominant orientation of Haversian canals in the osteons [100, 101].

In OS, rapid bone remodeling can lead to an abnormal highly isotropic (i.e., randomly oriented) structure due to changes in bone matrix, especially with regard to osteon and collagen fiber alignment [102]. These changes are heterogeneous, with skeletal changes being osteoblastic (bone-forming), osteolytic (bone-resorbing), or mixed (for a review, see Misaghi et al. [103]). Aggressive OS tends to be characterized by accelerated bone resorption and a discontinuous bone boundary [104]. This leads to substantial changes in the tumor physical microenvironment, including abnormal ECM characteristics, cellular mechanics, and mechanical forces.

The physical microenvironment and biomechanical properties of OS are emerging as vital regulators of tumor progression and chemo-resistance (Fig. 4). The ECM is highly dynamic and undergoes significant changes in OS, with the resultant biochemical and biomechanical cues influencing cell migration and proliferation. Changes to the ECM result from alterations in composition and/or how components are spatially arranged and cross-linked. Increased secretion of collagen results in the deposition of additional collagen into the bone matrix, which contributes to a mechanically stiff ECM. Increased ECM stiffness can also result from increased cross-links between collagen fibrils independent of collagen density [105]. ECM stiffening due to excess deposition and cross-linking of collagen is a key change in the physical microenvironment of tumors, which in turn influences matrix porosity and integrin-mediated cellular adhesion, and results in the modulation of cellular behavior and response to chemotherapy. An increasing number of clinical studies have highlighted altered collagen patterns as a prognostic factor in various cancers, including OS [106], revealing that various collagen types correlate with the clinical stage, metastasis, and response to therapy [107]. Mineralized ECM produced by OS cells display irregularly shaped, randomly distributed pores throughout the mineralized matrix, and ultrastructural analysis revealed differences in collagen fibril orientation, displaying low collagen density with short fiber length [108].

Fig. 4.

OS is affected by its physical microenvironment. The key aspects of the physical microenvironment that are altered in OS, each of which contributes to tumor development, progression, and response to treatment, are schematized (created with Biorender.com)

The proteoglycan composition of bone ECM is also altered in OS. Increased expression of versican, a modular proteoglycan with a highly glycosylated structure, has been found in OS tissues relative to normal tissues. Using stable knockdowns of versican (V0 and V1) in both MG63 and U2OS cell lines, it was found that its expression is upregulated by TGFß1 which enhances OS cell motility [109]. Furthermore, the overexpression of CD44 has been observed in metastatic and recurrent OS patient tumor specimens compared to primary tumors [110]. CD44 binds hyaluronic acid (HA), which is a key ECM component expressed by stromal and tumor cells and is the most specific ligand for CD44 activation. CD44-HA interaction has been reported to enhance metastasis to the lungs as well as chemoresistance in OS cells [111].

Tumor cells are characteristically known to cause the stiffening of their surrounding matrix via increased deposition of ECM proteins and cross-linking [112]. Dense ECM can impede the penetration of therapeutic agents into the tumor mass. Hence, an alternative therapeutic strategy is tumor blockade therapy, which involves creating an artificial ECM barrier at the tumor’s periphery to trap cancerous cells and hinder their proliferation and migration. The polymer 1,2-distearoyl-sn-glycero-3-phosphoethanolamine–N-poly(ethylene glycol)-alendronate (DPA) has been used to initiate biomineralization around Oss [113]. After peri-tumoral injection into mice bearing subcutaneous human 143B OS xenograft, the bisphosphonic acid groups served as biomineralization-initiating sites to recruit calcium and phosphate ions, forming a mineral barrier to inhibit cancer cell proliferation and migration. Additionally, DPA acted as an anti-osteoclastic agent to synergistically alleviate bone destruction to slow down OS progression. This was examined using tartrate-resistant acid phosphatase staining of the tibial platforms of mice bearing orthotopically xenografts after treatment, where the DPA groups displayed fewer osteoclasts [113].

Matrix stiffness is known to impact cellular responses in various biological contexts. While increased tissue stiffness is a typical characteristic of solid tumors, OS tissues are weaker and more compliant than the rigid bone tissue in which these cancers originate [114]. Healthy bone tissue exhibits stiffness values typically in the gigapascals (GPa) range, reflecting its rigid nature. To provide physiologically relevant insights into the mechanical characteristics of bone tissue, an optimal 3D substrate to recapitulate healthy trabecular bone would ideally exhibit a Young’s modulus of 2–10 GPa [115]. Unfortunately, there is a lack of experimental data available pertaining to the characterization of the mechanical properties of clinical OS samples. Furthermore, the use of engineered substrates with clinically relevant stiffness for the evaluation of OS cell response is notably limited.

When cultured on substrates of varying stiffnesses, OS cells exhibit varied responses. U2OS OS cells cultured on stiffer substrates (i.e., 50 kPa) displayed optimal spheroid growth and increased expression of CSC markers relative to the softer (2–25 kPa) substrates [116]. High matrix stiffness regulates cell morphology and promotes epithelial-mesenchymal transition (EMT) and migration in MG63 cells in vitro. Matrix stiffness has been shown to promote actin polymerization, cytoskeletal remodeling, and the translocation of myocardin-related transcription factor A in OS cells [117]. OS cells respond to decreasing 3D substrate stiffness by increasing nuclear localization of the mechanoresponsive proteins YAP and TAZ, along with the downregulation of expression of total YAP. Decreasing 3D substrate stiffness also led to the downregulation of active insulin-like growth factor 1 receptor (IGF-1R) and mTOR, with significant impact on downstream effectors of the IGF-1R/mTOR pathway [118]. Interestingly, this pathway has been a target of various clinical trials in sarcoma [119–121].

OS cells cultured within 3D electrospun scaffolds displayed increased resistance to combination chemotherapy compared to conventional monolayer cultures [118]. Using soft (4 kPa) and rigid (40 kPa) hydrogel culture models to study rigidity-sensing dysfunction in OS, expression of sensor proteins crucial for cells to sense their physical environment were quantified, including cytoskeletal proteins, e.g., Actinin Alpha 1 (ACTN1) [122]. Transformed OS cells exhibited resistance to anoikis, which was mediated by a downregulation of rigidity-sensing components, e.g., TPM1. Moreover, a TP53 mutation (missense mutation R156P) in transformed OS cells resulted in a gain of function to inhibit rigidity sensing [122]. Together, these findings suggest that the loss of rigidity-sensing of the physical microenvironment is a critical influence in sustaining the tumorigenicity of OS cells.

The topographical features of the ECM represent another key structural feature of the physical microenvironment. Contact guidance is a matrix topography-driven cellular response, whereby the orientation of cultured cells is influenced by the geometrical features of the matrix, such as micro- or nano-grooves [123]. Along with the loss of tissue architecture that occurs as OS progresses, OS cells appear to be less sensitive to the topographical changes accompanying this, and do not align their nucleus along elongated features as healthy osteoblasts do [124]. Remarkably, SaOs-2 OS cells remodel their nuclei to fit within the available spaces between 3D micron-sized pillars ex situ, adjustable by varying pillar size and spacing [125]. This nuclear deformation has been linked to actomyosin contractility and nucleo-cytoskeletal connections [125]. The significant reduction in nuclear deformation after vimentin knockdown also confirmed the role of intermediate filaments in supporting this nuclear deformation [126]. It has also been previously reported that higher substrate roughness enhances MG63 human OS cell adhesion, but negatively impacts cell spreading [127].

Current in vitro methods for investigating OS biology rely heavily on monolayer cell culture, which does not recapitulate the key mechanical cues provided by the physical TME. The development of engineered scaffolds with bone-mimicking composition has been a major focus of tissue engineering research, offering valuable in vitro models for investigating OS biology. Collagen-based scaffolds are highly tunable in terms of collagen content, stiffness, and porosity [128, 129] and have also been mineralized with various forms of calcium phosphate, such as hydroxyapatite [130], to better mimic bone ECM composition and improve osteogenesis. The mechanical properties of such mineralized collagen scaffolds (typically ~ 0.1 to 10 kPa) are typically orders of magnitude lower than that for the bone tissue in which OS initiates [131, 132]. To better reflect the stiffness of the bone niche, the mechanical strength of these mineralized scaffolds has been improved by the incorporation of polylactic acid fibres [133] or zinc [134]. The improvement of the mechanical properties of mineralized collagen scaffolds to enhance bone tissue biomimicry is an area of active research.

Despite the extensive use of OS cell lines in basic discovery research, differences in the mechanical phenotypes between OS cells and associated cell types, such as osteoblasts, remain understudied. Using real-time deformability cytometry, which employs microfluidics and high-speed imaging for mechanical phenotyping of cells, the deformability of STRO-1 positive (STRO-1 +) MSC population and MG-63 OS cells (a STRO-1 + OS cell line) was reported to be similar [135]. However, when unselected MSC populations were tested using the same technique, MSCs were found to be stiffer than OS-derived cells and MSCs enriched for STRO-1 + expression [135]. Aggressive OS has been reported to display a fibroblastic morphology with reduced focal adhesions [136]. Atomic force microscopy has been used to measure the stiffness of the cytoskeleton of murine OS cells in presence and absence of cytochalasin, an inhibitor of actin polymerization [137]. In LM8 cells, a highly metastatic murine OS cell line, actin levels were significantly lower than in the murine Dunn cells from which LM8 cells are derived. The cytoskeleton of OS cells was also significantly softer than that of normal osteoblasts, which can enhance their invasive capabilities [137]. A tendency towards increased deformability of highly metastatic OS cells, e.g., LM5, has also been described relative to the less metastatic counterparts SaOs-2 and HuO9 [138], highlighting cell mechanics as a potential prognostic marker of circulating tumor cells in OS patients.

Three differentiation trajectories from a CSC-like subset in OS have been identified, allowing the classification of OS into three groups with different prognoses [139]. Cancer cells in each group showed distinct interactions with other normal cells in their microenvironment, highlighting the complex interplay between OS cells and their surroundings. Collagen Type I Alpha 1 Chain (COL1A1) and alkaline phenyl phosphatase (ALP) were found to be upregulated in OS cells relative to normal osteoblasts, as well as increased expression of COL1A1 and Collagen Type I Alpha 2 Chain (COL1A2) in CSC-like cells relative to MSCs, signifying enhanced ECM formation ability and/or abnormal osteogenesis in OS cells [139]. Wu et al. established an OS TME-based prognostic index (TMEindex) to provide estimates about patient survival and individual response to checkpoint inhibitor therapy [140] using publicly available transcriptomics datasets for over 200 OS samples, identifying key TME-related genes. KEGG pathway analysis of differentially expressed genes between OS samples with high and low TMEindex revealed enrichment of signatures closely related to immune system-related biological processes and ECM [140].

Along with intrinsic cellular mechanical properties, extrinsic physical forces within the TME can also direct OS progression. Using cells from spontaneous OS tumors in mice with bone-specific knockout of pRb-1 and p53 in the osteoblastic lineage, silencing the stemness-related factor Sox2 significantly reduced YAP expression levels. This leads to a decrease in osteogenic differentiation and an increase in tumorigenicity. These effects were reversed by fluid flow shear stress by increasing YAP levels, while substrate stiffness did not cause significant effects on YAP levels [141]. This highlights the modulation of fluid movement through OS tumors as a potential novel therapeutic avenue.

While an increasing number of studies provide evidence for the potential of targeting elements of the mechanical OS microenvironment as a novel therapeutic strategy, many questions remain regarding the interplay between the various features of the physical microenvironment and biochemical factors in the OS niche. Cell–matrix interactions can be engineered in the laboratory to regulate key signaling pathways involved in OS, e.g., Hedgehog signaling, offering a step-change in our ability to recapitulate OS in vitro [142, 143]. Improved in vitro and in vivo models that accurately recreate the mechanical microenvironment of human OSs will be critical to evaluate new treatment approaches that may influence OS cell proliferation and metastasis to overcome mechanisms of chemoresistance, ultimately improving patient outcomes.

Innovative approaches in osteosarcoma therapy: manipulating the tumor microenvironment and immune response

The search for new therapeutic strategies for OS has increasingly focused on manipulating the tumor microenvironment and its immune components, aiming to enhance treatment efficacy and improve patient outcome. There is a long history of targeting the immune microenvironment in the treatment of soft tissue and bone sarcomas, including OS. Despite indications of positive responses, most immunotherapy trials to date have been unsuccessful.

Initial strategies targeting the immune microenvironment aimed to stimulate the immune system through the use of interferon‐alpha [144]. EURAMOS-1, an international randomized controlled trial, investigated maintenance therapy with pegylated interferon alfa-2b (IFN-α−2b) in patients whose OS showed good response to chemotherapy. Interferons have antiangiogenic, direct anti-tumor, and immunostimulating properties [145]. In addition, several preclinical studies and sporadic clinical evidences have provided the rationale for the use of IFN-α−2b in OS [146]. However, the approach was not successful and the addition of this immunomodulator did not influence patient survival [144]. Following the same idea of stimulating innate immunity, muramyl tripeptide-phosphatidyl ethanolamine (MTP-PE) [147] was added to chemotherapy, resulting in a statistically significant improvement in overall survival and a trend towards better event-free survival [148–150]. These pivotal clinical trials led to the approval of liposomal formulation of MTP-PE (L-MTP-PE; mifamurtide) by the European Medicines Agency for the treatment of newly diagnosed, high-grade non-metastatic OS in patients under 30 years of age. A recent clinical study (Clinicaltrial.gov: NCT01459484) supported the use of mifamurtide combined with high-dose ifosfamide as a salvage treatment for cases with an inadequate histologic response to induction chemotherapy [151]. The rationale for using MTP-PE derived from circumstantial evidence showing that postoperative bacterial infection improves OS prognosis, with a 10-year survival of 84.5% compared to 62.2% in patients with no infection [73]. One mechanistic justification is the potential anti-tumor activity of bacterial cell wall muramyl dipeptides stimulating innate immune cells. Kleinerman et al. demonstrated that monocytes from OS patients could be rendered tumor-cytotoxic by in vitro incubation with MTP-PE or by intravenous administration [152]. Preclinical studies in dogs with spontaneously arising OS confirmed the ability of L-MTP-PE to control microscopic metastatic disease [153]. Thus, it was postulated that L-MTP-PE could stimulate innate immune cells, leading to activation of macrophages and monocytes and favoring the local release of anti-tumor cytokines. This was indeed demonstrated in early-phase clinical studies [154–157]. However, the correlation between patient outcomes and the effects of mifamurtide on monocyte/macrophages and related systemic cytokine release remains unclear.

Monocytes and macrophages, accounting for 70–80% of total myeloid cells in OS [36], are the critical abundant components of the OS microenvironment. At present, there is no consensus on the impact of M1 and/or M2 macrophages in OS [158]. The indiscriminate activation of macrophages with MTP-PE may be one of the reasons why the drug demonstrated little evidence for response and impact on disease control in metastatic OS [159]. More detailed evaluation of OS immune infiltration is necessary to better define the functions and efficacy of mifamurtide, which will be investigated in the SARC13/OS 2016 trial currently underway in France.

Strategies involving immune checkpoint inhibitors have received great attention in sarcomas due to their success against several metastatic solid tumors, such as melanoma and renal carcinoma (for a review, see Sharma et al. [160]). These strategies are based on the removal of the “brakes” of the immune system, such as CTLA-4 or the interaction of PD‐1 with PD‐L1 or PD‐L2, with subsequent activation of T cells. Current immunotherapy trials are targeting these interactions with monoclonal antibodies. Several clinical studies have been registered to investigate the use of monoclonal anti‐CTLA‐4 antibodies tremelimumab and ipilimumab or the anti‐PD‐1 antibodies nivolumab and pembrolizumab in sarcomas, including OS. Unfortunately, results in sarcomas have been disappointing, very likely because OS tumors lack sufficient immune response (“cold tumors”). Therefore, simply removing the “brakes” is insufficient to elicit antitumor immunoreactivity. Combinations of classic cytotoxic chemotherapeutic agents, such as doxorubicin or cisplatin, with immune checkpoint inhibitors was found to enhance the efficacy of immunotherapy in several tumor types [161, 162], and may represent an interesting strategy for future testing in OS. Preclinical studies have shown that the alkylating chemotherapeutic agent, trabectedin, recruits T cells to the OS tumor site [163], indicating that this drug could be exploited to convert a “cold” tumor type, such as OS, into a “hot” one that is richer in T cells, making it amenable to checkpoint inhibitor–based immunotherapies. In tumors with insufficient immune T cell infiltration, the immune system needs to be reprogrammed to develop effective antitumor T cell strategies. Combining these approaches with chemotherapy or radiotherapy may represent an important therapeutic opportunity.

In addition, an antitumor immune response against OS may be achieved using sarcoma‐directed vaccines. Vaccines directed against OS have been explored for decades, with the first OS vaccine introduced in 1970 [164]. OS contains genetic abnormalities and/or chromosomal translocations that generate neoantigens and express tumor‐specific or differentiation antigens that are not expressed on most normal tissues, including MAGE‐1, disialogangliosides (GD2 and GD3), and NY‐ESO‐1, all serving as possible targets for vaccine development. However, so far, clinical trials using vaccines (for a review, see Nathenson et al. [165]) have demonstrated limited benefits.

Another strategy for inducing tumor immune responses involves the ex vivo expansion of lymphocytes, including T cells or NK cells, which are the main effectors of the adaptive immune response. TILs extracted from OS were found to be cytotoxic against allogeneic tumor cells [166] and the presence of cytotoxic tumor-infiltrating CD8 + T cells (CD8 + TILs) in pretreated OS was found to be associated with better prognosis [60]. The feasibility and safety of producing large numbers of autologous antitumor‐specific, cytotoxic T lymphocytes in patients with solid tumors have been demonstrated [167], and a phase 2 clinical trial is recruiting to evaluate the efficacy of autologous cytotoxic T lymphocytes in treating patients with unresponsive or relapsed tumors (Clinicaltrial.gov: NCT03449108).

A further extension of this technique is chimeric antigen receptor (CAR)–expressing T cells, with a genetically modified TCR to a specific tumor‐associated antigen. The generation and safety of CAR-T cells has been demonstrated, and treatment with CAR-T cells has produced remarkable clinical responses with certain subsets of B-cell leukemia or lymphoma [168–170]. More recently, immunotherapy with CAR-T cells that target the disialoganglioside GD2 expressed on tumor cells have been suggested as a therapeutic option for patients with high-risk neuroblastoma [171]. GD2 is also expressed at high level in OS [172]. A phase I trial assessing the feasibility and safety of T cells expressing an anti-GD2 chimeric antigen receptor in children and young adults with GD2 + solid tumors, including OS, following cyclophosphamide-based lymphodepletion was performed (Clinicaltrial.gov: NCT02107963). However, no results have been published yet. In preclinical studies, B7-H3 CAR T cells have been shown to mediate significant antitumor activity in vivo, causing regression of established solid tumors in xenograft models including OS [173, 174] but further studies are needed to establish the feasibility of this approach in OS. Specific cataloguing of OS-related neoantigens would be of particular interest as a basis for considering CAR‐related strategies.

Besides GD2 targeting CAR-T, anti-GD2 therapeutic approaches include disialoganglioside GD2 vaccines, immuno-cytokines, immunotoxins, antibody–drug conjugates, radiolabeled antibodies, targeted nanoparticles, and T cell engaging bispecific antibodies are being examined as additional modalities. Anti-GD2 antibodies target GD2-expressing tumor cells, leading to phagocytosis and destruction by means of antibody-dependent cell-mediated cytotoxicity, lysis by complement-dependent cytotoxicity, and apoptosis and necrosis through direct induction of cell death. In addition, anti-GD2 monoclonal antibodies may also prevent homing and adhesion of circulating malignant cells to the ECM. Thus, anti-GD2 therapy may represent a therapy with clinical potential for OS patients. However, a phase 2 study in recurrent OS patients receiving the recombinant GM-CSF sargramostim together with the anti-GD2 monoclonal antibody dinutuximab (clinicaltrial.gov: NCT02484443) was completed without indicating any improvements in disease control rate [175]. Other immunotherapies aim to harness the functions of human DCs; for example, CD1c + have been used as a source of vaccine immunotherapy and have yielded encouraging immunological and clinical results [176].

Overall, while extensive preclinical evidence supports immunotherapy in OS, early clinical trials have been disappointing. Most immunotherapies are based on T cells in the TME, but given that these tumors that have few infiltrating T cells, more attention may need to be paid to targeting macrophages rather than T cells and reprogramming a strong inhibitory immune microenvironment. Deeper knowledge on immune infiltration and on the impact of factors/cytokines that can modulate the tumor immune microenvironment is necessary. Actions to counteract signaling associated with a strong inhibitory immune microenvironment, including TGF-β, STAT3, VEGF, and IL-10 or support cytokines that regulate immune responses by stimulating immune therapeutic cells, such as IL-2, IL-12, and IL-15 should be considered. Reassuringly, a prospective study in primary metastatic OS patients administering chemotherapy and IL-2 showed a survival rate of over 40% of patients at 3 years [177].

Additional immunotherapeutic strategies of interest include CD47 inhibitors. CD47 is an anti-phagocytic signal that assists tumors escape phagocytosis by macrophages [178] while IDO is an intracellular enzyme, which leads to effector T cell anergy through downregulation of tryptophan [179]. These targets may represent an attractive immunotherapeutic strategy, but further research is needed to clearly define their impact in OS.

Conclusions

Long-standing evidence, together with new single-cell RNA data, emphasizes the complex composition of the OS TME, and the plethora of molecular interactions within. Together, this evidence draws attention to abundant signaling originating from the mesenchymal TME, a widely observed absence of T lymphocytes and extreme richness in myeloid cells in the immune TME, and the impacts of metabolic cues, including those emanating from reactive oxygen species and hypoxia, in molding cancer behavior and TME architecture. Ample evidence indicates that the TME composition and its interactions with resident cancer cells provides both prognostic indicators and druggable opportunities. This includes prognostic signatures informed by extracellular matrix components and, independently, the immune TME, potentially enabling novel and early risk stratification but also indicative that these TME features are important attributes in determining disease behavior. The OS physical microenvironment, characterized by dynamic changes in the ECM, also plays a critical role in tumor progression and resistance to therapy.

With the advancing understanding of the OS TME, evidence for novel therapeutic approaches is emerging. These stretch from targeting tumor-promoting cytokine networks within the mesenchymal TME to enhancing tissue stiffness, as well as targeting neoangiogenic cues emanating from hypoxia. Strong emerging evidence highlights an unusual immune environment characterized by a paucity of adaptive immunity, including T cells, and an abundance of myeloid cells. This unusual environment may explain the efficacy of therapies affecting myeloid cytotoxicity, such as muramyl tripeptide-phosphatidyl ethanolamine, and the limited benefit of strategies aimed at enhancing T cell function, including immune-checkpoint targeting. Given this unique immune-environmental architecture, approaches are needed that enhance T cell recruitment to the TME and strategies that enhance or enable the anticancer capabilities of myeloid cells could be exceptionally promising and should be prioritized.

While the path to effective therapies for OS remains challenging, ongoing research into the TME is steadily advancing, offering promising avenues to overcome current limitations and significantly improve outcomes for patients facing this aggressive cancer.

Abbreviations

ACP5

Tartrate-resistant acid phosphatase 5

ACTN1

Actinin Alpha 1

ALP

Alkaline phenyl phosphatase

β2AR

β2-Adrenergic receptor

CAF

Cancer-associated fibroblast

CAR-T

Chimeric antigen receptor (CAR)–expressing T cells

CCL2

C-C motif chemokine ligand 2

COL1A1

Collagen Type I Alpha 1 Chain

COL1A2

Collagen Type I Alpha 2 Chain

COLGALT2

Collagen beta(1-O) galactosyltransferase 2

CSC

Cancer stem cell

CTLA-4

Cytotoxic T-lymphocyte-associated protein 4

CTSK

Cathepsin K

DC

Dendritic cell

DPA

1,2-Distearoyl-sn-glycero-3-phosphoethanolamine–N-poly(ethylene glycol)-alendronate

ECM

Extracellular matrix

EMT

Epithelial-mesenchymal transition

EV

Extracellular vesicle

G-CSF

Granulocyte colony-stimulating factor

GM-CSF

Granulocyte-macrophage colony-stimulating factor

GRO-α

Growth-regulated oncogene alpha

HA

Hyaluronic acid

Hh

Hedgehog

HIF

Hypoxia-inducible factor

HREs

Hypoxia response elements

ICI

Immune checkpoint inhibitor

IDO

Indoleamine-2,3-dioxygenase

IFN-α-2b

Interferon alfa-2b

IGF-1R

Insulin-like growth factor 1 receptor

L-MTP-PE

Muramyl tripeptide-phosphatidyl ethanolamine encapsulated in liposomes (mifamurtide)

LPS

Lipopolysaccharide

MCP-1

Monocyte chemoattractant protein 1

MDSC

Myeloid-derived suppressor cell

MHC-I

Major histocompatibility class I genes

MIF

Migration inhibitory factor

MMP1

Matrix metalloproteinase-1

MCS

Mesenchymal stem cell

MTP-PE

Muramyl tripeptide-phosphatidyl ethanolamine

NK

Natural killer

OS

Osteosarcoma

PI3K/AKT

Phosphoinositide 3-kinase (PI3K)/B kinase protein

PD-1

Programmed cell death protein 1

PD-L1

Programmed death-ligand 1

scRNAseq

Single-cell RNA sequencing

SLC2A1

Solute carrier family 2 member 1

T-reg

Regulatory T cells

TAM

Tumor-associated macrophage

TCR

T cell receptor

TIGIT

T cell immunoreceptor with Ig and ITIM domains

TIL

Tumor-infiltrating lymphocyte

TLR4

Toll Like Receptor 4

TME

Tumor microenvironment

TPM1

Tropomyosin 1

VEGF-A

Vascular endothelial growth factor A

Author contribution

A.D., M.P., L.F., M.H.A., N.E., M.N., K.S., S.M. and A.G.M. wrote the review with equal contribution. M.P. and L.F. prepared the figures. All authors reviewed the manuscript prior to submission.

Funding

LF was supported by Bone Cancer Research Trust (BRCT) award, BCRT/6720, with additional support from a CRUK City of London Centre 2020 Development Fund Award and BCRT award BCRT/8021 to SM. No other fundings were obtained. The funders had no role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the script; or the decision to submit the manuscript for publication.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.