thyroid-cancer

Eight cancer genes linked to epithelial-mesenchymal transition predict prognosis and therapy response in papillary thyroid cancer.

Epithelial-mesenchymal transition (EMT) enables cancer cells to become invasive and metastasize. This study used single-cell RNA sequencing combined with 101 machine learning algorithm combinations to build an EMT-based prognostic model for papillary thyroid carcinoma (PTC).

Dan Manning

Dan Manning

— 1 min read
Share
Eight cancer genes linked to epithelial-mesenchymal transition predict prognosis and therapy response in papillary thyroid cancer.

Eight cancer genes linked to epithelial-mesenchymal transition predict prognosis and therapy response in papillary thyroid cancer.

Epithelial-mesenchymal transition (EMT) enables cancer cells to become invasive and metastasize. This study used single-cell RNA sequencing combined with 101 machine learning algorithm combinations to build an EMT-based prognostic model for papillary thyroid carcinoma (PTC).

Eight EMT-related prognostic genes were identified: TYRO3, E2F1, TNFSF15, TGFBR3, PTX3, FHL2, SNAI1, and WT1. High-risk patients had significantly worse survival, different immune infiltration profiles, and distinct drug sensitivities. Single-cell analysis identified fibroblasts as key cellular mediators, with FHL2, PTX3, and TGFBR3 active during critical differentiation stages. In vitro experiments validated expression patterns.

The 8-gene risk model could guide prognostic evaluation and treatment selection in PTC.

Key Findings

  • 8 EMT-related prognostic genes identified: TYRO3, E2F1, TNFSF15, TGFBR3, PTX3, FHL2, SNAI1, WT1
  • High-risk group (HRG) had significantly worse OS than low-risk group (LRG)
  • HRG showed different immune infiltration and drug sensitivity profiles
  • Fibroblasts identified as key cell population in EMT-related PTC progression
  • In vitro experiments confirmed bioinformatics-predicted expression patterns

Implications

This EMT-based gene signature could be used for prognostic stratification in PTC, identifying patients who may need more intensive follow-up or adjuvant treatment. Drug sensitivity differences suggest potential for personalized therapy based on risk group.

Caveats

Bioinformatics + cell line validation; abstract-only. External clinical validation in large prospective cohorts needed. PTC generally has good prognosis—clinical utility of additional prognostic tools needs demonstration in high-risk subgroups. Machine learning model may overfit.

Source: Cancer medicine — 2026-04-01

Read more