Development of a Chemoresistant Risk Scoring Model for Prechemotherapy Osteosarcoma Using Single-Cell Sequencing

Area of Science:

• Oncology
• Genomics
• Biotechnology

Background:

• Chemoresistance significantly hinders osteosarcoma treatment success.
• Predicting chemoresistance before treatment is a critical unmet need.
• Single-cell sequencing offers a promising avenue for evaluating chemoresistance.

Purpose of the Study:

• To establish an effective chemoresistance risk scoring model for pre-chemotherapy osteosarcoma.
• To leverage single-cell RNA sequencing (scRNA-seq) for identifying chemoresistant tumor cell markers.
• To validate the model's performance in distinguishing chemoresistant from chemosensitive osteosarcoma.

Main Methods:

• Comprehensive analysis of osteosarcoma bulk and scRNA-seq datasets (TARGET-OS, GSE162454).
• Identification of chemoresistant tumor clusters using enrichment analysis and AUCell scoring.
• Development of a ten-gene signature and a risk scoring model via LASSO regression, trained on TARGET-OS and validated on GSE33382.

Main Results:

• Osteosarcoma chemoresistant clusters (C14, C25) exhibited osteogenic cancer stem cell patterns and promoted angiogenesis, bone remodeling, and immunosuppression.
• A ten-gene signature identified chemoresistant osteosarcoma, leading to a risk scoring model with high discriminatory ability (AUC=0.82 training, AUC=0.84 validation).
• The scRNA-seq-based model demonstrated superior performance compared to a bulk RNA-based model in the validation set (AUC=0.84 vs AUC=0.54).

Conclusions:

• The study provides insights into the characteristics of chemoresistant osteosarcoma tumor cells.
• A novel chemoresistance risk scoring model based on scRNA-seq was successfully established.
• The model offers a feasible and accurate method for evaluating chemoresistance in osteosarcoma prior to treatment.