Background: Thyroid carcinoma (THCA) is the most common endocrine-related malignant tumor. Despite the good prognosis, some THCA patients may deteriorate into more aggressive diseases, leading to poor survival. This may be alleviated by developing a novel model to predict the risk of THCA, including recurrence and survival. Ferroptosis is an iron-dependent, oxidative, non-apoptotic form of cell death initially described in mammalian cells, and plays an important role in various cancers. To explore the potential prognostic value of ferroptosis in THCA, ferroptosis-related long non-coding RNAs (FRLs) were used to construct model for risk prediction of THCA.
Methods: RNA-sequencing data of THCA patients and ferroptosis-related genes were downloaded from The Cancer Genome Atlas (TCGA) and FerrDb, respectively. A total of 502 patients with complete data were randomly separated into a training cohort and a validation cohort at the ratio of 2:1. The Pearson correlation coefficients were calculated to determine the correlation between ferroptosis-related genes (FRGs) and the corresponding lncRNAs, and those meeting the screening conditions were defined as FRLs. Gene Expression Omnibus (GEO) database and qRT-PCR were used to verify the expression level of FRLs in THCA tissues. Univariate and multivariate cox regression analysis were performed to construct a FRLs signature based on lowest Akaike information criterion (AIC) value in the training cohort, then further tested in the validation cohort and the entire cohort. Gene set enrichment analysis (GSEA) and functional enrichment analysis were used to analyze the biological functions and signal pathways related to differentially expressed genes between the high-risk and low-risk groups. Finally, the relative abundance of different tumor-infiltrating immune cells were calculated by CIBERSORT algorithm.
Results: The patients were divided into high-risk group and low-risk group based on a 5-FRLs signature (AC055720.2, DPP4-DT, AC012038.2, LINC02454 and LINC00900) in training cohort, validation cohort and entire cohort. Through Kaplan-Meier analysis and area under ROC curve (AUC) value, patients in the high-risk group exhibited worse prognosis than patients in the low-risk group. GEO database and qRT-PCR confirmed that LINC02454 and LINC00900 were up-regulated in THCA. Univariate and multivariate cox regression analyses showed that the risk score was an independent prognostic indicator. GSEA and functional enrichment analysis confirmed that immune-related pathways against cancer were significantly activated in the low-risk THCA patients. Further analysis showed that the immune cells such as plasma cells, T cells CD8 and macrophages M1, and the expression of immune checkpoint molecules, including PD-1, PD-L1, CTLA4, and LAG3, were remarkably higher in the low-risk group.
Conclusion: Our study used the TCGA THCA dataset to construct a novel FRLs prognostic model which could precisely predict the prognosis of THCA patients. These FRLs potentially mediate anti-tumor immunity and serve as therapeutic targets for THCA, which provided the novel insight into treatment of THCA.
Keywords: Ferroptosis; Prognostic signature; Thyroid carcinoma; Tumor immune microenvironment; lncRNA.
© 2022. The Author(s).