MGST1 drives lymph node metastasis in papillary thyroid carcinoma via mitochondrial metabolic reprogramming and immune suppression

BackgroundWhile papillary thyroid carcinoma (PTC) generally exhibits a favorable prognosis, a subset of patients experiences poorer outcomes, with lymph node metastasis (LNM) significantly increasing the risk of recurrence and correlating with a worse prognosis. Mitochondrial metabolic reprogramming and immune evasion play pivotal roles in driving lymph node metastasis; however, the specific actionable targets within these processes remain largely unexplored.MethodsWe constructed a multi-omics discovery framework by integrating TCGA/GTEx transcriptomics, a large-scale independent clinical cohort, and single-cell RNA sequencing.

Both mitochondrial metabolic signatures and immune evasion landscapes were investigated to characterize their roles in driving LNM. A consensus machine learning framework was deployed to isolate core drivers, which were rigorously validated through genetic manipulation and pharmacological assays.ResultsUnsupervised clustering identified a high-risk “Mito-high” subtype, characterized by distinct metabolic reprogramming and an immunosuppressive microenvironment, in which CD8+ T cell depletion coincided with Treg enrichment.

Machine learning prioritized MGST1 as the core predictor, consistently ranking as a top feature in both the integrated TCGA/GTEx dataset and our independent cohort. The MGST1-based model demonstrated robust predictive performance, achieving an AUC of 0.833 in external validation.