Single-cell and bulk transcriptomic characterization of the osteosarcoma microenvironment reveals lineage-specific programs and inflammatory differences associated with metastasis

Osteosarcoma has long frustrated oncologists. It is the most common primary malignant bone tumor, striking hardest in adolescents, and while roughly sixty percent of younger or non-metastatic patients survive, outcomes for metastatic disease remain grim. Survival gains have plateaued despite decades of multimodal therapy, and even the promise of immunotherapy has faltered: pembrolizumab showed no clinically significant antitumor activity in a phase 2 trial of advanced osteosarcoma, echoing the low response rates seen in the SARC028 study. Now, a research team from Southeast University and Southern Medical University in China has taken a cell-by-cell look at why this tumor may be so resistant to immune attack, and their findings are as sobering as they are illuminating.

The study, published in Frontiers in Genetics, deployed single-cell RNA sequencing on tumor specimens from six treatment-naive osteosarcoma patients, then cross-checked the results against an independent bulk transcriptomic cohort of fifty-three patients with known metastatic status and survival data. After rigorous quality control, the researchers retained 44,189 cells carrying 25,473 detected genes. Computational clustering using the Leiden algorithm resolved twenty-three transcriptional clusters, which the team annotated into eleven distinct cell types spanning malignant, immune, stromal, and bone-lineage populations.

The cellular composition tells its own story. Myeloid cells dominated the landscape, accounting for 34.0 percent of all cells, followed by osteoblasts at 16.6 percent, osteoclasts at 9.9 percent, T cells at 9.7 percent, and fibroblasts at 8.4 percent. Every lineage was recovered in every patient, but osteoblasts and myeloid cells showed the largest inter-patient variance in proportional representation, hinting at substantial biological heterogeneity from one tumor to the next.

Program-score analysis revealed a striking lineage structure. Osteoblasts scored highest for the osteogenic program, consistent with their bone-forming identity and the established roles of the transcription factors RUNX2 and SP7 in driving osteoblast differentiation. Myeloid cells, by contrast, showed the highest scores for both inflammatory and immune-evasion programs, and these two programs were positively correlated with each other (r = 0.66) while the osteogenic program was negatively correlated with the inflammatory one (r = -0.42). In other words, the cells most capable of mounting an inflammatory response are also the ones most enriched for transcriptional states associated with dodging immune surveillance.

To dig deeper into the stromal compartment, the team reclustered osteoblasts, osteocytes, and mesenchymal stromal cells into seven subclusters with heterogeneous dysfunction and inflammatory scores. But here the data issued a caution. Pairwise Spearman correlations between representative osteogenic and inflammatory genes were generally weak at the single-cell level. RUNX2 and IL6 showed essentially no co-expression (r = -0.01), and SOST and DKK1 barely correlated (r = 0.12). The researchers explicitly declined to interpret these subclusters as evidence for a discrete immunosuppressive osteoblast population or a self-amplifying inflammatory-Wnt circuit, noting that transcriptional programs vary across cells and are only partially coordinated.

Ligand-receptor analysis, computed as the product of mean ligand expression in sender cells and mean receptor expression in receiver cells across twenty pairs, identified MIF-CD74 (score 0.98), SPP1-CD44 (0.77), and SPP1-ITGAV (0.31) as the strongest communication axes. Fibroblasts emerged as the strongest overall sender of signals, followed by osteoclasts and myeloid cells, while myeloid cells were the strongest overall receiver, followed by osteoclasts and B cells. These expression-based scores flag plausible signaling highways but, as the authors stress, they cannot demonstrate physical interaction, signaling flux, or functional dependence without perturbation experiments and spatially resolved measurements.

The independent bulk cohort provided a crucial reality check on generalizability. Differential expression between metastatic and non-metastatic tumors identified 333 genes at a false discovery rate below 0.05, and the inflammatory module was significantly lower in metastatic tumors (9.02 versus 9.32; Mann-Whitney p = 0.0078). Yet the osteogenic module showed no significant difference between groups, and key osteogenic genes such as RUNX2 and SP7 were not significantly altered in the metastatic contrast. More strikingly, the single-cell osteoblast-versus-rest effect showed only modest concordance with the bulk metastatic effect across 12,831 shared genes (Spearman rho = 0.254), with direction concordance nearly split down the middle. A twenty-gene osteoblast signature projected onto the bulk cohort failed to stratify overall survival (log-rank p = 0.968), arguing against presenting osteoblast-centered transcriptional signatures as established prognostic biomarkers.

To validate computational predictions with experimental evidence, the team exposed primary human osteoblasts to conditioned medium from MG-63 osteosarcoma cells and measured four key transcripts by RT-qPCR. The results were dramatic. RUNX2 expression plummeted to 0.38-fold of control levels (p < 0.001), and SP7 fell to 0.41-fold (p < 0.001), confirming that osteosarcoma-derived paracrine signals are sufficient to suppress the master osteogenic transcription factors in normal osteoblasts. Conversely, SOST, which encodes sclerostin, a Wnt pathway antagonist that inhibits bone formation, surged to 3.12-fold (p < 0.001), and RANKL, a pro-osteoclastic cytokine, rose to 2.87-fold (p < 0.01). Tumor-conditioned osteoblasts thus acquire a phenotype that would destabilize peritumoral bone integrity and potentially reshape the local niche.

The authors are careful to note that these observations are not contradictory even though single-cell osteoblasts within tumors expressed high levels of osteogenic markers while conditioned-medium-treated normal osteoblasts suppressed them. The single-cell comparison is a lineage-versus-rest contrast within tumors, whereas the in vitro experiment tests the response of normal osteoblasts to a defined exposure. Soluble factors from MG-63 cells are sufficient to alter selected osteogenic and bone-remodeling markers under experimental conditions, but the assay did not measure IL6, CXCL8, or DKK1, used only one osteosarcoma cell line, and therefore should not be considered independent validation of a broader immunosuppressive transcriptional state.

Taken together, the study delivers a reproducible descriptive framework of the osteosarcoma microenvironment while modeling the kind of analytical restraint that cross-sectional transcriptomic data demand. The findings are best viewed as hypothesis-generating. Priority next steps include replication in an independent osteosarcoma single-cell cohort, spatial localization of stromal and myeloid states, formal functional testing of candidate ligand-receptor pathways such as MIF-CD74 and SPP1-related interactions, expansion of the RT-qPCR panel to inflammatory markers, and use of additional osteosarcoma cell lines. Clinical biomarker claims, the authors conclude, must await cohorts with sufficient outcome annotation and cell-type-aware validation before any therapeutic implications are drawn. For a tumor that has resisted both chemotherapy refinements and immunotherapy breakthroughs, that kind of disciplined map of who is talking to whom inside the microenvironment may prove to be exactly the foundation future progress requires.

Subject of Research: Single-cell and bulk transcriptomic characterization of the osteosarcoma microenvironment reveals lineage-specific programs and inflammatory differences associated with metastasis

Article Title: Single-cell and bulk transcriptomic characterization of the osteosarcoma microenvironment reveals lineage-specific programs and inflammatory differences associated with metastasis

Article References: Zhou, Z., & Li, Z. (2026). Single-cell and bulk transcriptomic characterization of the osteosarcoma microenvironment reveals lineage-specific programs and inflammatory differences associated with metastasis. Frontiers in Genetics, 17, Article 1927885. https://doi.org/10.3389/fgene.2026.1927885

Image Credits: AI Generated

DOI: 10.3389/fgene.2026.1927885

Keywords: osteosarcoma, single-cell RNA sequencing, tumor microenvironment, osteoblasts, myeloid cells, metastasis, bulk transcriptomics, ligand-receptor interactions, RUNX2, SOST, RANKL, immune evasion