Constructing a bladder cancer prognostic model related to exosome using machine learning and identifying THBS1 as a...

Area of Science:

• Oncology
• Molecular Biology
• Bioinformatics

Background:

• Exosomes play a crucial role in bladder cancer (BCa) development.
• Current prognostic systems for BCa lack exosome-based approaches.
• There is a need for novel prognostic models and therapeutic targets in BCa.

Purpose of the Study:

• To develop an exosome-related prognostic model for BCa using machine learning.
• To identify potential therapeutic targets for BCa.
• To improve personalized treatment strategies for bladder cancer.

Main Methods:

• Utilized ten machine learning algorithms on BCa transcriptomic data.
• Validated the prognostic model using survival analysis and Cox regression.
• Explored biological mechanisms via immune microenvironment evaluation and functional enrichment analysis.
• Identified and validated potential therapeutic targets using molecular docking and experimental assays.

Main Results:

• Developed a machine learning-derived prognostic model (MLDPM) for BCa.
• The MLDPM effectively stratified patients into distinct risk categories with differential survival outcomes.
• Identified THBS1 as a potential biomarker and therapeutic target, showing significant inhibition of BCa cell migration and invasion upon suppression.
• Computational docking revealed interactions between THBS1/MMP9/CXCL12 and chemotherapeutic compounds.

Conclusions:

• Established the first machine learning-based exosomal prognostic system for BCa.
• Identified THBS1 as a promising biomarker and therapeutic target for bladder cancer.
• The study offers a novel computational and experimental approach for personalized BCa treatment.