Identification and validation of tumor microenvironment remodeling markers associated with prognosis in...

Tumor microenvironment remodeling (TER) plays an important role in the progression, invasion, and metastasis of differentiated thyroid cancer (DTC). However, the prognostic relevance of TER-associated genes in DTC remains unclear. This study aimed to identify TER-related genes (TERRGs) and develop a prognostic model for DTC.

Consensus clustering analysis grouped DTC samples into distinct molecular subtypes. The key prognostic TER-related signatures were identified through an integrative approach combining gene set enrichment analysis (GSEA) algorithm, differentially expressed analysis, univariate Cox regression, and 101 machine learning (ML) algorithms, with predictive performance assessed by area under the curve (AUC) values. The expression levels of key signatures were determined using clinical DTC specimens by quantitative real-time polymerase chain reaction (qRT-PCR).

The DTC patients were classified into three molecular clusters. After evaluating 101 ML models, the random survival forest (RSF) model and Ridge regression model demonstrated great predictive performance and led to the identification of key signatures (CAMP, DDIT4L, LMX1B, NAT16, and CALN1). Key signatures demonstrated excellent predictive performance with AUCs of 0.96, 0.99, and 1.00 for 1-, 2-, and 3-year survival, respectively. Expression analysis revealed that CALN1 was upregulated in DTC tissues, while the others were significantly downregulated.

This study provides a comprehensive evaluation of the prognostic significance of TERRGs in DTC and identifies five key prognostic signatures, providing a new insight into the molecular mechanisms underlying TER.