Genomic and transcriptomic analysis of 180 well differentiated thyroid neoplasms and 16 anaplastic thyroid carcinomas using massively parallel sequencing

Abstract

Well-differentiated thyroid carcinoma (WDTC) is the most prevalent human endocrine cancer with good prognosis. Recently, The Cancer Genome Atlas (TCGA) broadened the understanding of this disease by analyzing data for 496 papillary thyroid carcinomas (PTCs) which is the most common type of WDTC. However, in their analysis, follicular thyroid carcinoma (FTC) which accounts for 10% of thyroid malignancy was not included. Furthermore, the molecular pathogenesis from WDTC to anaplastic thyroid carcinoma (ATC) which is a very rare and one of the most lethal type of human malignancy is not fully elucidated until now. In the first chapter, the molecular characteristics of WDTCs were illustrated by large scale RNA-sequencing (RNA-seq) analysis on 180 WDTCs including 30 minimally invasive FTCs (miFTCs), 25 follicular adenomas (FAs), 48 follicular variant of PTCs (FVPTCs), and 77 classical PTCs (cPTCs). The mutation profiling suggested that the gene alterations in WDTC is very mutually exclusive and the type of driver mutation was tightly associated with the histology of thyroid cancer. BRAFV600E was only identified in PTC (71.43% and 25.00% for cPTC and FVPTC, respectively) and most fusion genes were discovered in PTC (17.60%) rather than miFTC or FA (only one PAX8-PPARG in miFTC). Meanwhile, H/K/NRAS mutations were found in FVPTC, miFTC, and FA with high incidence (47.92%, 50.00%, and 24.00%, respectively). Moreover, the alterations in DICER1, EIF1AX, EZH1, IDH1, PTEN, SOS1, and SPOP were frequently identified in miFTC and FA rather than PTC. Based on gene expression profiles, WDTCs were classified as three molecular subtypes, Non-BRAF-Non-RAS (NBNR) as well as BRAF-like and RAS-like regardless of their histological classification. Each molecular subtype represented the differential chromosomal aberration and regulation of signaling pathways such as MAPK, p53, cell-cell communication, and metabolism pathways. In the second chapter, to describe genomic, transcriptomic, and epigenomic features of ATC, whole-genome sequencing with HpaII digestion and RNA-seq were performed on 16 ATCs. Moreover, the integrative analysis with additional 324 WDTCs demonstrated the genomic, transcriptomic, and epigenomic changes during the progression of ATC from WDTC. TERT (75%), BRAF (43.75%) H/NRAS (31.25%) were the most recurrently altered genes in ATC. In addition, 75% of ATC had extra somatic mutation or deletion in EIF1AX, AKT1, PIK3CA, TP53, CDKN2A/CDKN2B, BRCA1, PTEN, TET1, LATS1/LATS2, ATRX, KMT2C, and U2AF1. The patients with ATC harbored TP53 germline mutations more frequently than WDTC patients (p = 0.002). Through gene expression analysis, ATCs were separated from WDTCs as a novel molecular subtype, ATC-like. Notably, the significant up-regulation of IL6, CD274 (PD-L1) and PDCD1LG2 (PD-L2) were observed which could be considered as potential immunotherapeutic targets. Interestingly, CD274 and PDCD1LG2 were usually up-regulated in ATCs with deep deletion of CDKN2A/CDKN2B and they represented the worst thyroid cell differentiation. From DNA methylation analysis, the global DNA methylation pattern of ATC was distinguished from WDTC and the more frequent hypermethylation of ATC, enriched in CpG island (p < 0.001), than WDTC was detected. These significant results broadened the current understanding of thyroid carcinoma hence allows more efficient diagnostic strategy and attractive targeted therapeutic targets in the future.