Integrated multi-omics and single-cell analyses identify metabolic heterogeneity and therapeutic vulnerabilities in medullary thyroid cancer.

BACKGROUND

Medullary thyroid cancer (MTC) is a heterogeneous and aggressive malignancy with limited therapeutic options. Metabolic reprogramming, a hallmark of cancer, may offer a promising avenue for understanding and managing MTC.

METHODS

RNA sequencing data of 101 MTC samples were obtained from a published dataset PRJCA008783, and untargeted metabolomic profiling was performed on 51 paired samples. Metabolic subtypes were identified using clustering analyses and validated using immunohistochemistry (47 cases), multiplex immunofluorescence (12 cases), and a previously published single-cell RNA sequencing dataset (7 cases derived from PRJCA021386). Deep learning-based approaches were applied to develop prognostic models.

RESULTS

Three metabolic subtypes were identified. The M3 subtype, associated with poor prognosis, was characterised by upregulated glycosaminoglycan (GAGs) biosynthesis, particularly chondroitin sulfate, and elevated expression of CHSY1, a key GAGs biosynthetic enzyme. M3 tumours displayed enhanced epithelial-mesenchymal transition (EMT) signatures. Multi-omic analyses implicated CHSY1 may promote EMT through interactions with myofibroblasts, which was supported by immunohistochemistry and immunofluorescence. Two prognostic classifiers, the 8 Metabolites Model and the 28 Metabolic Genes Model, effectively stratified patients by recurrence risk, with predictive power largely driven by GAGs-associated metabolism.

CONCLUSIONS

Our study reveals substantial metabolic heterogeneity in MTC and proposes a novel metabolic classification system, offering mechanistic insights and supporting metabolite-driven prognostication for precision management of MTC.