As the primary subtype of thyroid cancer, ATC remains one of the deadliest diseases. Because of the variability in ATC genes, ATC can easily metastasize and has a lack of valid therapeutic targets. Therefore, at present, chemotherapy combined with thyroid surgery is the primary adjunct therapy for ATC. However, these therapies have some limitations and tend to result in poor prognosis. Therefore, it is necessary to reevaluate the etiology and molecular mechanisms of ATC as well as explore new potential diagnostic, therapeutic, and prognostic targets using bioinformatics analysis techniques.
Based on four GEO databases for gene expression, a total of 764 DEGs including 314 upregulated and 450 downregulated genes were screened using R integration analysis. In order to investigate the biological roles of these DEGs, we performed GO and KEGG pathway analyses. GO term enrichment analysis showed that upregulated DEGs mainly participated in cell division, while the downregulated DEGs were enriched in integral components of the membrane. KEGG pathway analysis indicated that upregulated DEGs mainly participated in ECM-receptor interaction and protein digestion and absorption, whereas the downregulated DEGs were enriched in thyroid hormone synthesis. Next, we constructed a PPI network using the DEGs and screened potential hub DEGs using MOCDE (MOCDE ≥ 7). After UALCAN (P < 0.05) screening and verification by the GEPIA website, we identified several hub DEGs including TYMS, FN1, CHRDL1, SDC2, ITGA2, COL1A1, COL9A3, and COL23A1. Finally, we used the KM plotter tool to predict the prognosis of hub DEGs in patients with ATC.
Based on the KM plotter, high expression levels of TYMS, FN1, CHRDL1, and SDC2 as well as low expression levels of ITGA2, COL1A1, COL9A3, and COL23A1 were associated with RFS in patients with ATC (Fig. 7); however, only TYMS, COL9A3, and COL23A1 were significant for OS (Fig. 8). Therefore, we found that a high expression level of TYMS and low expression levels of COL9A3 and COL23A1 were related to favorable prognostic factors in patients with ATC.
TYMS is a protein coding gene that contributes to the de novo mitochondrial thymidylate biosynthesis pathway [10–12]. This function maintains the thymidine-5-prime monophosphate pool critical for DNA replication and repair . Among its related pathways are one carbon pool by folate and the E2F transcription factor network . GO annotations related to this gene include protein homodimerization activity and mRNA binding. Microarray and immunohistochemical studies have shown that the expression of this enzyme is significantly upregulated in a variety of tumors, including breast, bladder, cervical, kidney, lung, and gastrointestinal cancers [13–17]. The high expression of TYMS was also associated with poor clinical prognosis of these cancers, suggesting that TYMS may act as an oncogene. In fact, ectopic expression of TYMS in xenograft models has been shown to confer transformed and tumorigenic phenotypes on normal cells. It is worth noting that the elevated expression level of TYMS also showed greater invasion and metastasis ability in these cells. Protein enzymes have been of interest as targets for cancer chemotherapeutic agents. ATC is also a metastatic and aggressive cancer. TYMS belongs to the hub DEGs in our ATC data module 1. Therefore, TYMS is likely to have a strong relationship with aggressive and metastatic ATC.
COL9A3 encodes one of the three alpha chains of type IX collagen, which is the major collagen component of hyaline cartilage . Diseases associated with COL9A3 include multiple epiphyseal dysplasia type 3 and intervertebral disc disease [19, 20]. Among its related pathways are the integrin pathway and gastric cancer network 2. COL9A3 also participates in the protein digestion and absorption pathway . Studies have shown that COL9A3 is highly expressed in the brain, retina, salivary glands, and thyroid gland but remains low in ATC [21, 22]. Therefore, we infer that COL9A3 has a certain relationship with ATC that can be used as a target for ATC diagnosis or prognosis prediction.
COL23A1 (Collagen Type XXIII Alpha 1 Chain) is a Protein Coding gene,which is a member of the transmembrane collagens, a subfamily of the nonfibrillar collagens that contain a single pass hydrophobic transmembrane domain. It is a new transmembrane collagen found in metastatic tumor cells and highly expressed in thyroid and cardiovascular systems [24, 25]. Among its related pathways are the integrin pathway and degradation of the extracellular matrix. From the analysis results in this study, it can be seen that COL23A1 is derived from module 2. COL23A1 has an important paralog of this gene is COL5A1.The KEGG results showed that COL5A1 is related to the ECM-receptor interaction, protein digestion and absorption, focal adhesion, amoebiasis, and phosphoinositide 3-kinase (PI3K)-protein kinase B (Akt) signaling pathways. These cellular pathways indicate that the body is in a state of accelerated metabolism, and the PI3K-Akt signaling pathway is a typical tumor metabolism pathway [26, 27]. In recent years, several studies have shown that COL23A1 is closely related to renal cell carcinoma, head and neck cancer, and so on [28, 29]. Based on the data mining results in this study, we believe that COL23A1 is closely related to the occurrence and prognosis of ATC.
Considering all the existing research results, we speculate that the eight hub DEGs have significant expression levels or distinct pro-cancer effects in different cancers. Detecting the levels of these genes may be useful in the early diagnosis or prognosis assessment of some cancers.
Based on multiple datasets and thorough bioinformatics analysis, an eight-gene cohort shows a superior prediction of prognosis and survival of ATC. However, there are limitations existing in the present study that require acknowledgment. Such as, lack of experimental verification, which will be the focus of our later work.”