整合生物信息学与机器学习方法识别奥沙利铂耐药结直肠癌的新型诊断特征

Oxaliplatin resistance significantly impairs therapeutic outcomes in colorectal cancer. However, reliable diagnostic markers for early detection of resistance remain limited. This study aimed to identify novel diagnostic signatures through integrative bioinformatics and machine learning approaches.

We performed comprehensive bioinformatics analyses combining transcriptomics data from multiple cohorts. The diagnostic signatures were identified and validated using machine learning algorithms. Weighted gene co-expression network analysis (WGCNA) was employed to explore resistance-associated gene modules. Multiple computational methods including functional enrichment, protein-protein interaction networks, and immune infiltration assessment were conducted to comprehensively characterize the molecular features of oxaliplatin resistance.

Through integrative analysis and machine learning, we identified an 8-gene diagnostic signature (CHFR, TGFBRAP1, RPS4Y1, CYP26B1, NR4A2, FLJ20021, TNFSF9, CAV2) that demonstrated robust performance in distinguishing resistant cases (AUC = 0.868). Functional characterization revealed significant enrichment in metabolic reprogramming, DNA repair mechanisms, and immune modulation pathways. Systematic evaluation of tumor-immune interactions demonstrated distinct patterns of immune cell infiltration between resistant and sensitive groups, particularly in Natural killer cells and Activated CD8 T cells. Computational drug screening identified Glycidamide and orciprenaline as promising candidates, with favorable binding profiles against key resistance-associated targets.

Our study establishes a novel multi-gene diagnostic signature for oxaliplatin resistance through integrative bioinformatics and machine learning approaches. The comprehensive molecular characterization and identification of potential therapeutic candidates provide new insights into resistance mechanisms and clinical management strategies for oxaliplatin-resistant colorectal cancer.