(P 392) IMMUNOSUPPRESSIVE TUMOR MICROENVIRONMENT OF OSTEOSARCOMA

Objective: Osteosarcoma (OS) is the most common malignant bone tumor in the pediatric population, accounting for 6% of all pediatric tumors. It is characterized by a high degree of genomic instability, resulting in numerous copy-number alterations and genomic rearrangements but with no disease-defining recurrent activated mutations. Despite multiple clinical trials using various combination chemotherapy, there has been no significant improvement in prognosis for nearly four decades. Given the diverse genomic landscape of OS and the difficulty of identifying druggable therapeutic targets, the use of immunotherapy appears to be a reasonable alternative approach. However, results from recent clinical trials based on immune checkpoint inhibitors (ICIs) were disappointing. This suggests that there may be additional immune suppressive mechanisms in the tumor microenvironment (TME) in OS other than the PD-1/PD-L1 mechanism that could be interfering with the effects of immunotherapy. Therefore we carried out a detailed analysis of the TME of a small number of OS cases using a combination of single cell RNA sequencing and spatial transcriptomics techniques.

Methods: To better understand the immune microenvironment in OS and guide future therapeutic development, we performed single-cell RNA sequencing (scRNA-seq) using the 10x Genomics Chromium platform on six fresh tumor biopsy samples from pediatric OS patients. Raw data was processed using 10x CellRanger to produce transcript read counts for each cell. After further data filtering based on mitochondrial/ribosomal RNA content and doublet removal using DoubletFinder, counts were normalized using Seurat, and cells were integrated across samples with Harmony. Data was combined with a previously-published OS scRNA-seq cohort of six samples (GSE162454). Two additional OS samples were profiled using 10x Genomics Visium spatial transcriptomics for validation of discovered subtypes and to add spatial context.

Results: Initial clustering identified 16 major cell types based on high expression of canonical cell markers, including macrophages, OS cells, T cells, and granulocytes. Copy number changes in OS cells (identified with inferCNV) revealed multiple subclones in each tumor. Several immunosuppressive cell types were identified via subclustering of major cell types, including neutrophil myeloid-derived suppressor cells (MDSCs), regulatory T-cells, exhausted T-cells, and LAMP3+ dendritic cells. Markers for the cell types found in OS were identified for further validation and cohort expansion using imaging techniques, including Visium spatial transcriptomics. We discovered 9 total spot clusters in the Visium data, and performed deconvolution using the scRNA-seq cell identities to examine colocalization of discovered cell types. Many of the discovered clusters were common between patients and showed consistent cell type proportions. However, we did discover patient-specific differences in the frequency of some cell types, with one sample showing a higher proportion of T-cells along with increased presence of colocalized IFN-stimulated macrophages, and the other sample with a greater presence of neutrophils and MDSCs.

Conclusion: Using single-cell transcriptomics, we were able to discover the presence of multiple distinct immunosuppressive cell subtypes of neutrophils, T-cells, and dendritic cells. Additionally, spatial transcriptomics revealed multiple similar clusters between samples, and common colocalization of the discovered cell types within those clusters. However, differences in T-cell presence and interferon induction may be indicative of patient-specific immunogenicity of OS. Taken together, these preliminary results confirm the hypothesis of significant immunosuppression in the TME of OS caused by different mechanisms. These findings suggest potential novel immunotherapy targets that could be considered in future clinical trials. Functional validation of some of these targets are in progress.