Macrophage-Derived Transcriptional Signatures Predict Prognosis and Drug Sensitivity in Thyroid Cancer: Integrative...

Thyroid cancer is the most common malignancy of the endocrine system. Tumor-associated macrophages (TAMs) play a pivotal role in modulating the tumor microenvironment and promoting tumor progression. However, the prognostic implications of macrophage heterogeneity in thyroid cancer remain unclear.

Single-cell RNA-seq analysis was conducted to identify tumor-enriched macrophage subpopulations and hdWGCNA was used to define related gene modules. A prognostic model was built using 117 machine learning algorithm combinations and validated by Kaplan-Meier analysis. A nomogram combining clinical features and risk scores was established. Genomic alterations, immune profiles, and treatment responses were compared between risk groups using TCGA and GDSC2 datasets. In vitro experiments were performed to validate the role of SMYD3 in tumor progression and drug sensitivity.

Single-cell analysis identified a tumor-enriched macrophage subset with distinct functional states. hdWGCNA revealed macrophage-related gene modules linked to poor prognosis, and a machine learning-based model effectively stratified patient risk. High-risk patients had worse survival, older age, advanced stage, and lower BRAF mutation frequency but more diverse oncogenic alterations. They also showed enhanced T-cell exclusion and altered immune infiltration. Drug prediction analysis indicated greater sensitivity to Rapamycin, BDP-00009066, AZD5363_1916, and Cediranib_1922 in the high-risk group. Functional assays confirmed that SMYD3 knockdown suppressed proliferation and migration, and reduced sensitivity to AZD5363_1916, highlighting its role in modulating therapeutic response.

This study identifies a tumor-enriched macrophage subpopulation with prognostic relevance and develops a robust machine learning-based model for risk stratification in thyroid cancer. SMYD3 is implicated in both tumor progression and drug sensitivity, offering a potential biomarker for precision treatment strategies.