Answers are grounded in this paper and cite their sources.
Background Considerable suspicious thyroid nodules still cannot be diagnosed after the preoperative fine needle aspiration, thereby novel diagnostic tools are imperative for clinical practice. Methods 884 thyroid cancer patients were enrolled from eight independent datasets and 29,561 cells were obtained from a single-cell RNA dataset. 20 published transcriptome signatures were retrieved. Results We comprehensively identified the significantly increased proportion of T cells in thyroid cancer via single-cell RNA analysis. Combined with the bulk expression data, 17 T cell-related genes were screened out. The thyroid diagnostic model (TDM), a consensus machine-learning-derived model, was determined and compared with 20 published transcriptome signatures. TMD displays stable and powerful performance with excellent AUCs in seven cohorts (1.000, 1.000, 1.000, 1.000, 1.000, 0.926, 0.904). Notably, the high-risk group is typically featured by high-immune states and cell invasiveness. Besides, the tumor immune microenvironment characterized high-risk group with high infiltration of antigen presentation-related cells, increased expression of antigen presentation-related molecules, and some co-inhibitor molecules, indicating enhanced immune activation and sensitivity to immunotherapy. Conclusion TDM provided an attractive potential approach for identifying thyroid cancer at high risk in an early stage and deciphering its immune microenvironment to optimize clinical management for patients with thyroid cancer.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2023) studied this question.