PDX药物筛选揭示结直肠癌中的精准联合治疗机会

Colorectal cancer (CRC) exhibits extensive molecular heterogeneity driven by oncogenic alterations such as KRAS, NRAS, and BRAF mutations. Despite the promise of targeted therapies, clinical responses are variable and often limited by resistance, underscoring the need for improved molecular stratification and rational combination strategies. We conducted large-scale in vivo drug screening using 140 patient-derived xenograft (PDX) models representing key CRC molecular subtypes: KRAS G12C (n=40), RAS/RAF wild-type (n=40), and BRAF V600E (n=60). Models were treated with standard-of-care agents, including anti-EGFR, BRAF/EGFR, and KRAS G12C/EGFR inhibitors. Tumor responses were assessed alongside molecular profiling via whole-exome sequencing, RNA sequencing, and Reverse Phase Protein Array. Integrated multi-omics analyses identified biomarkers of response and resistance mechanisms. Resistant tumors frequently exhibited MAPK reactivation and RTK/PI3K-AKT pathway activation. BRAF V600E models showed heterogeneous responses to BRAF/EGFR inhibition (40% disease control), which improved with co-targeting of PI3K or downstream MAPK components. RAS/RAF wild-type models responded to anti-EGFR therapy (100% in treatment-naïve; 40% in pretreated), with resistance driven by RTK signaling mitigated via MAPK inhibition. KRAS G12C models demonstrated modest responses to KRAS/EGFR inhibitors (70% disease control), enhanced by SOS1, SHP2, or AKT inhibition. Additional genotype-specific vulnerabilities, including WEE1 and EZH2 sensitivity in KRAS G12C tumors, support rational combination strategies. This comprehensive PDX-based screen delineates actionable, genotype-specific vulnerabilities in CRC and provides a preclinical framework for precision combination therapies to guide future clinical trials.