An Auto-RS Signature for Prognostic Stratification and Drug Sensitivity Prediction in Osteosarcoma

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Osteosarcoma (OS) metastasis and poor chemotherapy response impede therapeutic progress.
• Early detection of OS metastasis is crucial for survival but lacks effective diagnostic tools.
• Tumor microenvironment transcriptional states offer insights into metastasis but are underexplored.

Purpose of the Study:

• To identify molecular markers associated with osteosarcoma metastasis.
• To develop a computational signature, Auto-RS, for early metastasis detection and risk stratification.
• To predict drug sensitivity in osteosarcoma patients based on the Auto-RS signature.

Main Methods:

• Analysis of transcriptomic and clinical data from 139 osteosarcoma patients.
• Application of a LASSO-Cox framework to derive the Auto-RS gene expression-based risk score.
• Integration of Auto-RS with the Precily algorithm for drug sensitivity prediction and single-cell analysis for model interpretability.

Main Results:

• The Auto-RS signature, comprising four autophagy genes (BNIP3, MYC, PEA15, SAR1A), independently predicted overall survival in osteosarcoma.
• Auto-RS demonstrated superior accuracy (AUC=0.88) as a single predictor compared to metastasis status, sex, and age.
• Auto-RS improved prognostic model discrimination and aligned with metastatic trajectories at single-cell resolution, identifying gemcitabine and cytarabine as potential therapies for high-risk patients.

Conclusions:

• Autophagy-mediated transcriptional 'stress fingerprints' are strongly linked to osteosarcoma metastasis.
• The Auto-RS signature enables early therapeutic stratification of osteosarcoma patients, irrespective of overt metastatic status.
• Auto-RS provides insights into osteosarcoma metastasis mechanisms at single-cell resolution and can guide therapeutic strategies.