Development of a Prognostic Model Based on SCISSOR+ Osteoblastic Osteosarcoma Cell-Associated Genes
Haiyan Xu, Yuanyuan GaoIntroduction:
This study aimed to establish a relevant risk model to improve the prognostic prediction for patients with osteosarcoma (OS).
Methods:
Bulk RNA-seq and single-cell RNA-seq data were integrated to perform functional enrichment analysis, cell-cell communication analysis, and prognostic modeling. Using the SCISSOR algorithm, a Riskscore model was developed based on prognostic differentially expressed genes (DEGs) associated with SCISSOR+ osteoblastic OS cells and subsequently its prognostic performance was evaluated. In addition, the associations of the Riskscore with immune infiltration and predicted drug response were explored.
Results:
The single-cell landscape exhibited a relatively high proportion of osteoblastic OS cells, which were further divided into six main cell clusters (C1-C6) with distinct pathway enrichment patterns. A higher proportion of SCISSOR+ osteoblastic OS cells in C1 and C4 was observed, with OS patients with a high score of SCISSOR+ cells showing worse survival. Cell-cell communication analysis demonstrated that SCISSOR+ cells communicated with other cells through the predicted MIF-(CD74+CXCR4) interaction, whereas other cells communicated with SCISSOR+ cells through the predicted PPIA- BSG interaction. A Riskscore model incorporating SCISSOR+ osteoblastic OS cell-associated prognostic genes was developed and stratified patients into distinct prognostic groups in retrospective cohorts. High-risk patients showed lower stromal, immune, and ESTIMATE scores, whereas multiple immune cell signatures were enriched in the lowrisk group. The Riskscore was negatively correlated with the predicted IC50 values of several candidate drugs.
discussion:
This study identified heterogeneous osteoblastic OS cell subclusters using GSE162454, constructed a validated Riskscore model via SCISSOR algorithm and 6 genes, and analyzed its correlation with OS prognosis, immune infiltration and drug sensitivity.
Discussion:
Based on the GSE162454 dataset, this study identified heterogeneous osteoblastic OS cell subclusters and developed a six-gene Riskscore model associated with SCISSOR+ osteoblastic OS cells, and retrospectively analyzed the associations of the Riskscore with the prognosis, immune infiltration, and predicted drug-response profiles in OS.
Conclusion:
The present findings preliminarily demonstrated the relationship between SCISSOR+ osteoblastic OS cell-associated prognostic genes and the prognosis of OS patients.