Endometrial cancer is a common gynecological malignant tumor. Although surgery, radiotherapy, chemotherapy and hormone therapy can significantly improve the treatment effect, the clinical prognosis of patients with advanced and recurrent endometrial cancer is relatively poor, and the 5-year survival rate of patients is low, which seriously threatens the life and health of women. In recent years, the role of metabolic reprogramming in tumors has been widely studied. Glycometabolism reprogramming is one of the characteristics that tumor cells are different from normal cells. Even under the condition of sufficient oxygen, tumor cells are more likely to use glycolysis for rapid energy supply. Therefore, studying the relationship between metabolic reprogramming and tumor development is becoming a new method for tumor diagnosis, prevention and treatment to elucidate the new mechanism of tumor development from the perspective of metabolic abnormalities and to intervene and correct the metabolic abnormalities of cells. At present, many studies have reported the relationship between glycolysis and endometrial cancer [21,22]. However, research on biomarkers related to glycolysis in endometrial cancer remains limited. It has been reported that clinical characteristics, such as age, stage, differentiation and lymph node metastasis, cannot accurately predict the prognosis of patients [23]. As a result, an increasing number of studies are exploring gene biomarkers, and many studies have found that developing multiple gene-related risk models can improve the prediction efficiency [24,25]. Therefore, the purpose of this study was to explore the risk biomarkers related to the prognosis of endometrial cancer and to analyze further their relationship with immune cell infiltration.
We first downloaded glycolysis-related gene sets from GSEA and screened differentially expressed genes in endometrial cancer and normal samples, including 128 upregulated genes and 28 downregulated genes. Furthermore, we used GO and KEGG enrichment analysis to verify the biological function and signaling pathways of differentially expressed genes. We found that these differentially expressed glycolysis genes are indeed mainly related to pyruvate metabolic processes, glycolytic processes and glycolysis/gluconeogenesis. Next, we used univariate Cox regression to initially screen genes related to the prognosis of endometrial cancer and further used LASSO regression analysis to screen and construct the prognostic gene signature. A total of ten mRNAs (PFKM, PSMC4, NUP85, PDHA1, CDK1, CLDN9, CENPA, GPI, NUP155 and GPCI) significantly related to the overall survival of endometrial cancer were identified to construct the prognostic gene signature. Based on this gene signature, patients were divided into a high-risk subgroup and a low-risk subgroup according to the risk score. K-M analysis showed that the overall survival rate of patients in the high-risk group was significantly lower than that in the low-risk group. It is suggested that it is of great clinical significance to evaluate the prognosis of patients by calculating the risk score of endometrial cancer patients based on the gene signature.
To further verify the value of the ten-gene signature in clinical prognosis, we analyzed the relationship between ten genes and clinical stage and found that the expression of PFKM, NUP85, PDHA1, CDK1, CLDN9, CENPA, GPI, NUP155 and GPCI increased with increasing clinical stage. To further evaluate whether the prognostic ability of the ten-gene signature is independent of other clinical parameters, including age, stage, grade, metastasis and lymph node metastasis, we conducted univariate and multivariate Cox regression analyses on the whole data set and found that the gene signature was an independent prognostic factor of endometrial cancer. Furthermore, ROC curve analysis was conducted to verify the prognostic and diagnostic value of the gene signature. The results showed that the diagnostic significance of the gene signature was the strongest (AUC = 0.730), which was better than that of age, stage, grade, metastasis and LNM. This result demonstrated that our gene signature has considerable prognostic and clinical diagnostic value. In addition, we integrated multiple predictors (including risk scores) through a nomogram to effectively predict the 1-, 3- or 5-year survival rate of patients, which may help to plan short-term follow-up of individualized treatment.
Our GSEA enrichment analysis revealed that many signaling pathways were significantly enriched in the high-risk subgroup, including pathways related to metabolism and metabolic diseases, such as pyruvate metabolism, glycolysis gluconeogenesis, inositol phosphate metabolism, insulin signaling pathway and type II diabetes mellitus. It has been reported that tumor metabolic reprogramming, or the Warburg effect, has been considered one of the ten characteristics of cancer. Both malignant transformation and tumor development, including invasion and metastasis, require metabolic reprogramming [26]. Metabolic heterogeneity is an important reason for the failure of treatment to produce the same effect on cancer cells [27]. It has been reported that diabetes was closely related to the increased risk of endometrial cancer (RR = 1.72) [28]. High glucose and insulin resistance are important characteristics of diabetic patients. It has been confirmed that high glucose could increase glucose uptake and glycolysis activity by regulating the AMPK/mTOR/S6 and MAPK pathways and promote the proliferation and invasion of endometrial cancer cells [29]. High insulin levels are an independent influencing factor of endometrial cancer. Increased insulin and IGF-1 could activate downstream signaling pathways by binding with IR and IGF-1 receptor to promote the proliferation of endometrial cancer cells [30]. These results suggested that the poor prognosis of patients in the high-risk subgroup might be closely related to tumor metabolic reprogramming and the activation of metabolic disease-related pathways. In addition, other pathways closely related to tumorigenesis and development were also significantly enriched in the high-risk subgroup, such as the cell cycle, endometrial cancer, the ERBB signaling pathway, the MAPK signaling pathway, the mTOR signaling pathway, and the Wnt signaling pathway. Taken together, these results show that our gene signature is closely related to metabolic imbalance and provide a potential molecular mechanism for elucidating the relationship between the gene signature and endometrial cancer progression.
In our gene signature, most genes have been reported to be closely related to the occurrence and development of cancer. PFKM, the second rate-limiting enzyme in the glycolysis pathway, has been shown to be closely related to the increased risk of breast cancer [31]. PSMC4 is a member of the proteasome complex, which is responsible for recognizing ubiquitin-labeled substrates and ingesting them into the proteasome (19S regulatory complex). The overexpression of PSMC4 promoted the degradation of some key cell regulatory proteins, such as tumor suppressors, and further promoted the progression of tumors [32]. Therefore, inhibition of the proteasome is a promising cancer treatment strategy. The nucleoporins NUP155 and NUP85 were reported to be upregulated in hepatocellular carcinoma, accompanied by TP53 silencing and overexpression of cell cycle-related genes [33]. PDHA1 is the main regulatory site of PDH activity. PDHA1 regulates the deactivation or activation of PDH through phosphorylation and dephosphorylation and then affects the mitochondrial tricarboxylic acid cycle and glycolysis metabolic flow. It has been reported that the expression of PDHA1 is abnormal in a variety of tumors, and it is closely related to tumor invasion, drug resistance and prognosis by affecting tumor cell glucose metabolism. The upregulation of PDHA1 could promote the metastasis of cholangiocarcinoma [34]. In contrast, another study reported that low expression of PDHA1 predicted poor prognosis in gastric cancer [35]. Several studies have shown that inhibition of CDK1 expression could significantly inhibit the proliferation and invasion of endometrial cancer cells [36-37]. CENPA and CDK1 were also identified as prognostic markers of lung cancer [38]. Overexpression of CLDN9 could promote tumor cell invasion through Tyk2/STAT3 signaling [39]. Upregulation of GPI-anchored proteins could promote tumor cell migration and progression by enhancing the ERBB signaling pathway [40]. GPCI has received growing interest in recent years due to its high capability of visualizing soft tissue, and GPCI has been reported to have potential value in the diagnosis of breast cancer [41]. We further verified the expression of the gene signature in endometrial carcinoma by using the immunohistochemistry results from the Human Protein Alts database. The results showed that the expression of PDHK1, NUP85, CDK1, CENPA, GPI, GPC1, PSMC4 and PFKM in endometrial carcinoma was higher than that in normal endometrium, which was consistent with the prediction of the gene signature. There were frequent genetic alterations of ten gene signatures, and amplification and increased mRNA were the most common changes. The results of coexpression analysis showed that NUP85, NUP155, CDK1 and CENPA had a strong correlation. Next, we analyzed the PPI network to explore the correlation between the ten genes and explore the potential interacting proteins. A total of 48 proteins participated in the network construction. Among these proteins, we focused on the interaction between BUB1 and most model genes. In our previous study, we screened 21 genes closely related to the prognosis of endometrial cancer patients, including BUB1, and established a prognostic model for early warning of endometrial cancer based on a low-density chip of 492 tumor-related genes [42]. Moreover, The WGCNA results indicated that the ten-gene signature was significantly associated with a functional gene module that were mainly enriched in cell division, regulation of cell cycle process and DNA replication. It is noteworthy that there was significant correlation between the gene module and clinical grade of patients. Taken together, the above results suggest that the ten-gene signature may play an important role in the occurrence and development of endometrial cancer.
In the past, many studies have focused on the role of glycolysis in tumors. It has been reported that aerobic glycolysis in tumors constantly produces lactic acid, which provides energy for the tumor, and the increased lactic acid in the microenvironment could also affect the immunotherapy effect [43]. It has also been found that antitumor metabolism therapy combined with immunotherapy can effectively inhibit tumor growth [44]. To further explore the relationship between the glycolysis-related gene signature and immune cell infiltration and immune function, we first analyzed the immune scores, stromal scores and ESTIMATE scores of patients in the high-risk subgroup and low-risk subgroup based on our gene signature. We found that the immune scores, stromal scores and ESTIMATE scores of patients in the high-risk subgroup were significantly lower than those of the low-risk subgroup. At the same time, the overall survival rate of patients with low immune scores and estimated scores was significantly higher than that of patients with high scores. Some studies have shown that the more immune cells enter the tumor metastasis, the higher the immune score is, the higher the survival rate is and the lower the recurrence rate is. Metastasis with the smallest number of immune cells entering represented the worst immune microenvironment, and immune escape was most likely to occur under this condition [45]. These results suggested that the poor prognosis of patients in the high-risk subgroup might be closely related to the low immune scores. However, whether the activation of glycolysis-related pathways affects the infiltration of immune cells warrants further investigation.
We further found that many immune cells, such as activated dendritic cells, M1 macrophages, M2 macrophages, memory activated T cells, and follicular helper T cells, were significantly higher in the high-risk subgroup than in the low-risk subgroup, while dendritic cell resting, memory resting T cells, and regulatory T cells (Tregs) were significantly lower in the high-risk subgroup. There was a positive correlation between the three immune cells and the overall survival rate of patients, including dendritic cell resetting, NK cell activation and T cell regulation (Treg). According to previous research, resting dendritic cells exist in most tissues and are activated to mature antigen-presenting cells under external stimulation. Antigen presentation by resting DCs could induce protective immunity [46]. Tregs play a key role in maintaining immune system homeostasis. Some studies have shown that the high density of Treg cells in tumors is related to the clinical prognosis of tumors, such as liver cancer and gastric cancer [47]. High proportions of Tregs among tumor-infiltrating CD4+ T cells were favorable [48]. In addition, we also found that the ten genes were negatively correlated with T cell regulation (Tregs), including CDK1 (r=-0.41), CDK1 (r=-0.34), and NUP155 (-0.22). Taken together, the above results suggest that the poor prognosis of patients in the high-risk subgroup may be closely related to the decrease in the proportions of dendritic cell resetting and T cell regulation (Tregs), while the glycolysis-related gene signature might be involved in the regulation of T cell regulation (Tregs), and the specific mechanism governing this phenomenon merits further investigation.