Spatial ligand-receptor circuits are reshaping how osteosarcoma immune resistance is read
Osteosarcoma has resisted simple immunotherapy narratives for years. Single-agent checkpoint blockade has shown limited activity, and this mini-review argues that the problem is not only weak immune activation, but also where immune cells sit, what they can physically reach, and which local signals shut them down before they make contact with tumor cells.
The review centers on osteosarcoma in two anatomically distinct settings: primary bone tumors and pulmonary metastases. Those niches do not behave like interchangeable containers of malignant cells. Bone brings remodeling, matrix density, osteoclast biology and marrow-derived myeloid traffic; the lung adds fibrocollagenous encapsulation, compartmentalized chemokine signaling and a different set of access barriers for cytotoxic lymphocytes. In that setting, the same immune cell type can be present yet functionally irrelevant if it remains on the wrong side of a stromal boundary.
A one-size-fits-all immune checkpoint model cannot explain that spatial biology.
Bone and lung niches impose different rules on immune access
Human tissue profiling in the review separates tumor interiors from myeloid-rich interfaces and fibrocollagenous borders. That distinction matters because a ligand–receptor pair can appear active in transcriptomic data yet have very different meaning depending on whether the interacting cells are in direct contact, separated by matrix, or confined to opposite sides of a capsule. The authors repeatedly distinguish spatial association from functional causality, and that separation gives the piece its clinical value: expression alone is not enough to call a therapeutic vulnerability.
The review also makes a practical point about specimen choice. Primary tumors, mixed lesions, resected lung metastases and pleural-fluid samples are not equivalent sources of evidence. Pleural-fluid cells may reveal immune programs, but they cannot reconstruct the tissue architecture needed to infer a physical exclusion boundary. Likewise, post-treatment specimens can be enriched for viable tumor, histiocytes, osteoclasts and extracellular bone matrix in ways that alter interpretation of any ligand–receptor readout.
Macrophage checkpoints dominate the most mature mechanistic data
Several of the most developed circuits involve tumor cells suppressing macrophage engulfment. GD2-linked suppression of phagocytosis emerges as one of the clearest examples. In the human setting, the receptor appears to be SIGLEC7; in mice, the orthologous pathway involves Siglec-E. Downstream, SHP2 signaling links receptor engagement to impaired engulfment. The biologic logic is direct: if a tumor cell can decorate itself with a glycoconjugate signal that engages inhibitory myeloid receptors, it can blunt one of the innate immune system’s oldest clearance mechanisms before adaptive immunity is even relevant.
CD47 and CD24 fit the same broad theme, though the evidence base is not identical. CD47 has antibody-associated phagocytosis data in osteosarcoma models, while CD24 knockdown increased engulfment in culture and in mice. For CD24, the human receptor assignment remains less secure in osteosarcoma than the functional phenotype itself, which is exactly the kind of gap translational readers need to see. Expression, receptor assignment and in vivo relevance are related, but they are not the same claim.
Fibroblasts are not passive scaffolding in lung metastasis
The most striking stromal circuit in the review links CXCL14 to integrin α11β1. In that pathway, stem-like osteosarcoma cells signal to fibroblasts, promoting fibroblast activation and pulmonary niche formation. The message here is not simply that fibroblasts are abundant. It is that they can become active participants in building a metastatic habitat that helps malignant cells persist after arrival in the lung.
A second stromal circuit is even more metabolically layered. Oxidized-lipoprotein uptake by CD36-positive cancer-associated fibroblasts leads to ANGPTL4 secretion, which then engages integrin α5β1 and contributes to CD8 T-cell dysfunction. The biology is plausible and clinically unsettling: lipid handling by stromal cells can be converted into an immunosuppressive secretome, and exhausted T cells may be the end product of a metabolic conversation they never initiated. The review is careful, though, not to overstate site specificity. This circuit was studied in treated human osteosarcoma and in subcutaneous and paratibial mouse models, not in a proven pulmonary capsule. Evidence for function is real; claims about anatomical exclusivity should wait.
A tumor can build its own immune weather if fibroblasts are recruited into the wrong metabolic state.
Chemokine axes, bone remodeling and vascular signals widen the barrier logic
The review groups several additional pathways as microenvironmental barriers rather than interchangeable immune checkpoints. CCL2–CCR2 supports monocyte recruitment, with cross-species evidence and canine metastatic data suggesting that stromal trafficking is part of the disease biology. CXCL12–CXCR4 appears in human pediatric lung metastases and patient-derived cultures, where spatial profiling and pharmacology suggest a role in local survival programs and perhaps immune exclusion. Yet the review resists a simplistic conclusion: reduced viability after CXCR4/MIF inhibition does not automatically prove that chemokine blockade will restore T-cell access in intact tissue.
Bone remodeling adds another layer. RANKL–RANK/OPG biology governs osteoclast lineage activity and osteolysis, but the review treats it as a structural barrier with immune consequences rather than a direct immune checkpoint. That framing is clinically useful. Bone destruction, matrix turnover and altered vascular behavior can change how immune cells move, where they accumulate and whether they function, even if the pathway was not originally defined in immunology. Similarly, vascular routes involving VEGF, VEGFR and angiopoietin-TIE2 biology may affect immune trafficking, but multikinase activity in metastatic osteosarcoma does not by itself identify the immune mechanism responsible for benefit.
Biomarker development will depend on anatomy, not expression alone
The review’s most useful translational message is that spatial measurements may become biomarkers only after prospective validation. CellPhoneDB, NicheNet and CellChat can generate candidate interactions from transcriptomic data, but those predictions still need protein detection, receptor activation and perturbation studies before they can guide treatment choice. For now, cell-to-tumor distance, compartment-specific receptor expression and stromal architecture are candidates for biomarker development, not accepted selection criteria.
That restraint is appropriate because several circuits remain partially mapped. The review notes that receptor dependencies inferred from transcriptomic data, especially for CD24 and macrophage migration inhibitory factor signaling, need more precise attribution. Human and murine receptor assignments do not always line up cleanly, and some functional studies have been done in engineered systems that do not fully reproduce lung metastasis or bone tumor architecture.
For clinicians and translational researchers, the message is straightforward: osteosarcoma immune resistance is spatially organized. The next useful experiments will not just ask whether a pathway exists, but where it acts, which cell compartment owns the interaction, and whether disrupting it changes phagocytosis, T-cell function or metastatic persistence inside the real tissue niche.