Transcriptome Analyses Identify a Metabolic Gene Signature Indicative of Dedifferentiation of Papillary Thyroid Cancer

Abstract

Abstract Context Metabolic reprogramming is a common feature of tumorigenesis. It remains unknown concerning the expression pattern of metabolism-associated genes in dedifferentiated thyroid cancer (DDTC). Objective This study aimed to identify a useful signature to indicate dedifferentiation of papillary thyroid cancer (PTC). Design and Setting We used one discovery and two validation cohorts to screen out aberrant metabolic genes in DDTC, and further used The Cancer Genome Atlas (TCGA) cohort to search for independent risk factors for the low-differentiated phenotype of PTC as a signature of dedifferentiation. The prediction of the signature for DDTC was validated in the TCGA cohort and the combined Gene Expression Omnibus cohort. We also analyzed the correlations of the signature risk score with clinicopathological features of PTC. Gene set enrichment analyses were performed in the TCGA cohort. Results Significant enrichment of metabolic pathways correlated with differentiation status of PTC. A signature of metabolic genes including LPCAT2, ACOT7, HSD17B8, PDE8B, and ST3GAL1 was discovered and validated across three cohorts. The signature was not only predictive of DDTC but also significantly associated with BRAFV600E mutation (P < 0.001), T3/T4 stage (P < 0.001), extrathyroidal extension (P < 0.001), lymph node metastasis (P < 0.001), and tumor/lymph node/metastasis III/IV stage (P < 0.001) in PTC. Downregulations of LPCAT2 expression (P = 0.009) and ST3GAL1 expression (P = 0.005) increased risks of decreased disease-free survival for patients. Furthermore, the signature was implicated in a number of oncogenic biological pathways. Conclusions Our findings suggest that metabolic deregulations mediate dedifferentiation of PTC, and that the metabolic gene signature can be used as a biomarker for DDTC.