NRF2通路激活预测喉癌的不良放疗反应：来自一项大型蛋白质基因组学表征研究的见解

Patients with advanced laryngeal cancer often experience poor outcomes, and current clinical guidelines lack reliable predictive biomarkers. Preclinical cancer studies have implicated NRF2, a transcription factor that promotes antioxidant defense and metabolic reprogramming, in resistance to radiation therapy (RT). This study investigates whether activation of the NRF2 pathway influences clinical outcomes in larynx cancer patients receiving RT or surgery. We identified 135 larynx cancer patients who were treated at Siteman Comprehensive Cancer Center with either surgery and RT or surgery alone. Tumor samples were obtained from the WashU Head & Neck Tumor Center tissue bank and characterized using RNA sequencing, targeted exon capture sequencing, and mass spectrometry-based proteomics. Gene and protein expression profiling were used to independently calculate an NRF2 activity score, allowing the classification of tumors along a continuum of NRF2 activity. Kaplan-Meier analysis and Cox proportional hazards modeling were employed to compare disease-free survival (DFS) and locoregional failure (LRF) rates between NRF2-high and NRF2-low groups. In patients treated with surgery and RT, differential gene and protein expression analysis identified numerous NRF2-induced targets that were associated with locoregional failure. In those treated with radiation, NRF2-high patients had significantly worse outcomes than NRF2-low, with 5-year LRF rates of 51.0% and 17.1% (HR = 7.2, 95% CI: 2.0-25.7), and 5-year DFS rates of 25.4% and 50.7% (HR = 2.9, 95% CI: 1.3-6.2), respectively. Notably, NRF2-activity status was not significantly prognostic in the surgery alone cohort, indicating that NRF2 activation is specifically associated with radiation resistance. Ontological enrichment analyses identified upregulation of pathways associated with antioxidant response and dsDNA break repair in NRF2 high tumors, suggesting possible mechanisms of resistance. Discovery-based proteomics aligned closely with the gene expression data. Exon capture sequencing revealed 10 patients harboring driver mutations in NRF2 or KEAP1, 9 of which were categorized as NRF2 high from expression data. Logistic regression identified NRF2 score as the most predictive variable associated with outcomes for radiation-treated patients, followed by nodal involvement. These findings suggest NRF2 activation is associated with worse outcomes in patients receiving surgery and radiation, but not surgery alone. Future implications include validating NRF2 as a biomarker across data sets and the development and testing of NRF2-targeted therapies.