iMLGAM: Integrated Machine Learning and Genetic Algorithm-driven Multiomics analysis for pan-cancer immunotherapy...

Area of Science:

• Oncology
• Immunology
• Bioinformatics

Background:

• Immune checkpoint blockade (ICB) therapy effectiveness varies significantly among patients.
• Predictive biomarkers are crucial for optimizing ICB treatment strategies.

Purpose of the Study:

• To develop an advanced multi-omics data integration tool for predicting ICB therapy outcomes.
• To identify novel molecular targets for enhancing ICB efficacy.

Main Methods:

• Development of the integrated Machine Learning and Genetic Algorithm-driven Multiomics analysis (iMLGAM) R package.
• Validation of iMLGAM scores across independent patient cohorts.
• Clustered regularly interspaced short palindromic repeats (CRISPR) screening to identify key regulatory molecules.

Main Results:

• iMLGAM scores demonstrated superior predictive performance for ICB therapy response compared to existing biomarkers.
• Lower iMLGAM scores were significantly associated with enhanced therapeutic responses and favorable tumor immune microenvironments.
• Centrosomal Protein 55 (CEP55) was identified as a key mediator of tumor immune evasion.

Conclusions:

• The iMLGAM package provides a robust tool for personalized cancer immunotherapy prediction.
• CEP55 represents a promising therapeutic target for overcoming resistance to ICB therapy.
• These findings offer new strategies to improve patient outcomes in cancer immunotherapy.