Endometrial carcinoma (EC) is the most common gynecologic malignancy,is an increasing public health concern worldwide [2]. With the continuous emergence of new therapeutic targets, the mining of prognostic biomarkers has become a research hotspot. These biomarkers can be used to predict the risk of patients and guide personalized adjuvant therapy for early high-risk patients [21]. Autophagy genes can play a vital role as biomarkers in many cancers, such as melanoma [22], gastric cancer [23] and non-small cell lung cancer [24], etc. But the prognostic research of differentially expressed autophagy-related genes (DEARGs) in endometrial cancer is lacking. Therefore, it is necessary to understand the autophagy molecular mechanism and identification of new biomarkers for the development of endometrial cancer.
In this study, a total of 45 differentially expressed autophagy-related genes were identified based on the TCGA database, including 23 up-regulated DEARGs and 22 down-regulated DEARGs. Moreover, we predicted the biological functions and enrichment pathways of these genes using bioinformatics methods. GO and KEGG analysis showed that these DEARGs were significantly enriched in autophagic biological pathways, suggesting that dysregulation of autophagy played a key role in the development and progression of EC.
We also performed PPI network analysis and screened out three modules and six key genes (CASP3, GAPDH, MYC, MAPK3, FOXO1, CDKN2A), of which CASP3 was the most important. Co-expression analysis showed that co-expressed genes with CASP3 were mostly enriched in cell cycle, cell division and Metabolism of RNA. This indicates that CASP3 plays a central role in executing apoptosis and thus in the carcinogenic process.
A basic study shows that Caspase-3 (CASP3) plays an important role in promoting colon cancer cell invasion and metastasis and therefor sever as a potential target for colon cancer therapy [25]. Jia Lin et al. found that CASP3 polymorphisms confer to the lung cancer susceptibility [26]. In human cancer treatment, several studies reported that patients with higher levels of procaspase-3 or active caspase-3 had a worse prognosis than patients with lower levels of procaspase-3 or active caspase-3 [27-30]. In this study, CASP3 as a central gene also showed its abnormal expression in EC development, and this finding can guide future exploration of EC mechanisms.
Next, we distinguished a prognostic signature consisting of three identified DEARGs (CDKN2A, PTK6, GRI2D), and most of them are harmful factors of the EC. The 3-DEARG signature divided the OS of EC into high- and low-risk subgroups. Kaplan–Meier analyses indicated that high-risk subgroups with poor survival. Notably, the univariate and multivariate regression analysis indicated that the risk score system of 3-DEARG signature could be regarded as an independent prognostic model to provide a more accurate assessment of OS in EC. Analysis of the ROC curve shows that the risk score is a better prognostic indicator compared with other clinical characteristics (age, grade) used for risk stratification in EC patients. Therefore, our biomarker may provide a simple and accurate prediction for the prognosis of endometrial cancer.
For clinical correlation analysis, we found that risk score increases with age and pathological grade of the patient and the expression of CDKN2A, PTK6, and GRI2D increases with age. GRI2D expression is higher in the high pathological grade group. High age and histopathological grade predicted poor prognosis for endometrial carcinoma. These results indicated that CDKN2A, PTK6, and GRI2D were highly EC-prognosis-related.
Using the online database, we verified that the expression of CDKN2A and PTK6 in tumors was higher than normal samples, and they were negatively correlated with the overall survival of EC patients. Besides, both have higher expression levels in different subtypes of EC tissues. Their expression levels also increased with EC stage. We further found that CDKN2A and PTK6 were widely expressed in a variety of tumor tissues, whereas GRID2 expression was negative. GSEA enrichment analysis showed that CDKN2A and PTK6 were mostly associated with Metabolism of xenobiotics by cytochrome P450 and Nitrogen metabolism, which may provide a new direction for EC research.
The CDKN2A is a gene type of cyclin-dependent kinase inhibitor, its mutations have been related to breast cancer, melanoma and squamous cell carcinoma [31-33]. Recent findings show that polymorphisms in the CDKN2A gene are associated with EC in postmenopausal women [34]. As for PTK6, an intracellular tyrosine kinase, in tumors overexpressing PTK6, this unusual soluble tyrosine kinase is emerging as a mediator of cancer cell phenotypes, including increased proliferation, survival, and migration [35]. In breast cancer cells, oncogenic PTK6 functions to protect cells from autophagic induced death under anchorage-independent conditions [36]. GRID2, human glutamate receptor delta-2, is a relatively new member of the family of ionotropic glutamate receptors which are the predominant excitatory neurotransmitter receptors in the mammalian brain. Recent research has demonstrated that GRID2 connected to several diseases including ataxia, inherited disease and cancer [37, 38]. In short, almost all DEARGs in the signature are closely related to cancer. Therefore, there is reason to believe that these three DEARGs can be used in the clinic as a reliable and repeatable prognostic biomarker.
To our knowledge, this is the first landscape study on the prognosis of DEARGs signals in endometrial cancer based on large clinical data sets. However, there are some limitations in this study. For example, our research is actually an analysis based on previous data, lacking in vivo and in vitro experimental studies. Therefore, further investigations are still needed to verify the accuracy for estimating prognoses and to test its clinical utility in patient management.