EC is one of the most common gynecologic malignant cancers worldwide and the incidence rate is increasing every year while its age of onset is decreasing (1). The International Federation of Gynecology and Obstetrics (FIGO) staging system, together with the histological grade and the classification type are nowadays the main factors for classifying EC patients with different prognosis (2). In stage I and II EC patients, the 5-year survival rates are about 80–90% and 70–80%, respectively, while the one of stage III and IV disease is 20–60% (3). Multiple factors including genomic and clinical factors play an essential role in the development and prognosis of EC, but the current classification system cannot achieve a comprehensive and accurate prediction of survival outcomes in EC patients (4). Therefore, it is of great significance to further explore the prognostic factors and establish a more precise prognostic model combining gene expression profiles and traditional clinical features.

The relationship between metabolic dysfunction and EC has been extensively reported and studied (5). Prior studies have shown that the obesity or diabetes related metabolic syndrome is closely associated with the occurrence and poor prognosis of EC (6). The potential mechanism may involve insulin resistance, glucose disorder and lipid metabolism (7). It has been reported that the glycolysis ability is significantly increased in EC cells, which can be promoted by the abnormal gene expressions resulting in the acceleration of the progression of EC (8). Hence, the screening of metabolic related genes associated with the EC progress is crucial for the prognosis of EC patients as well as for choosing a targeted therapy to improve their outcomes.

The progression of cancer cells is associated with their surroundings and with the tumor microenvironment (TME). TME is composed of immune cells, extracellular matrix, mesenchymal cells and inflammatory mediators, which turn out to have impacts on tumor growth, metastasis, and clinical survival outcomes (9). A previous study has demonstrated that immune-related genes may contribute to the development of new prognostic biomarkers for EC patients (10). The Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data (ESTIMATE) algorithm (9) and the Tumor Immune Estimation Resource (TIMER) algorithms (11) have also provided an extensive analysis of immune cells and TME associated genes of cancer cells, and have found the relationship between the omics data and the prognosis of patients.

In the present study, we aimed to analyze the potential metabolism-related gene biomarkers associated with the prognosis of EC, and explore the correlation between EC and immune infiltration in order to clarify the molecular mechanism underlying the tumorigenesis of metabolism-induced EC.

The mortality of EC has been doubled during the past 20 years. Although the 5-year survival rate of early EC patients is greater than 85%, about 13–25% of EC patients (initially considered to have a good outcome) present recurrence and metastasis (1). Thus, accurate prognostic indicators are required to assist clinicians into conducting more accurate clinical evaluation. Database-based bioinformatics analysis has been increasingly used to screen out target biological molecules with prognostic potential. But previous studies were mainly based on either genomic factors or clinical factors (24, 25), both of which played a vital role in tumorigenesis and prognosis of EC. A metabolism-related risk signature for EC was introduced by the present study using large-scale cancer datasets, and then an OS prediction nomogram was created by integrating the risk signature and clinicopathological features to accurately predict the outcomes of the individual EC patients. This model was validated by an external clinical patient cohort from PKUPH. CIBERSORT and TIMER algorithms were used to conduct an integrative analysis of immune scores and immune cells in the different risk groups characterized by the risk signature in EC patients, aiming to study the molecular mechanism underlying the tumorigenesis of the metabolism-induced EC.

Obesity is more closely related to the development of EC than any other cancer types in women (26). It is also related to worse outcomes with a 2.6 to 4.7-fold increase in EC risk and doubled risk in diabetic patients (27). In the present study, 61 metabolism-related overlapping genes were detected and they were mostly enriched in metabolic processes and in amino acid/glucose metabolism according to GO and KEGG enrichment analyses. Extensive metabolic dysregulation occurs during the process of obesity such as hyperglycemia, insulin insensitivity, abnormal metabolites, and high-density lipoprotein (28). The underlying mechanisms may be the cancer proliferation caused by glucose uptake and fatty acid synthase (29). Increased blood glucose levels may contribute to the development of EC (30). In EC cells, the glycolysis rate is higher and glucose oxidation is lower (31).

Then the six metabolism-related genes were selected by performing LASSO regression analysis, including CA3, HNMT, PHGDH, CD38, PSAT1, and GPI. It is noteworthy that some of these genes have been reported in earlier studies associated with cancer. CA3 and CD38 were down-regulated in EC samples in both TCGA datasets and clinical cohort from PKUPH. The CA3 gene (carbonic anhydrase III) is a member of a multigene family which are mainly isozymes encoding carbonic anhydrase (32). In human hepatocellular carcinoma, this gene is down-regulated, and it may be involved in the process of apoptosis or programmed cell death (33). CD38 has been linked to the inhibition of the metabolism and the proliferation in prostate cancer (34). PHGDH, PSAT1, and GPI were up-regulated in EC samples in both cohorts. Human 3-phosphoglycerate dehydrogenase (PHGDH) is an important enzyme in the process of serine synthesis, and the serine and glycine has offered sufficient energy and metabolites accelerating the proliferation of cancer cells (35). At present, the expression of PHGDH has been found increased in multiple types of cancers, including breast cancer, cervical cancer, glioma, melanoma, pancreatic cancer, and colon cancer (36). Moreover, the elevation of PHGDH expression is usually related to cancer progression and poor prognosis (37). Currently, PSAT1 has mainly been explored in non‐small‐cell lung cancer, colon cancer, esophageal cancer, and breast cancer (38). The over-expression of PSAT1 is a marker of poor prognosis in cancer patients and even increases the resistance of chemotherapy (39). Histamine N-methyltransferase (HNMT) is one of the main enzymes catabolizing histamine in humans. Glycosylphosphatidylinositol-anchored proteins (GPI-APs) are vital for a variety of cell functions, and some of them have been reported to make contribution to the tumor occurrence and progression (40). Thus, our study has demonstrated that these genes could perform as potential prognostic biomarkers for EC and the prognostic function of these six genes manifests in its overall effects as a whole risk signature, but not a single gene or the correlation among these genes.

Afterwards, a risk signature model was established based on genes expression profiles and coefficients of their association with survival, and the AUC of the ROC curves of the whole cohort based on this model was higher than 0.7 at 3-, 5-, and 7-year of OS. GSEA analysis identified the “fatty acid metabolism,” “chemokine signaling pathway,” “insulin signaling pathway,” and “PI3K-AKT-MTOR signaling” as the potentially relevant pathways in the high-risk group. While “inflammatory response,” “JAK-STAT signaling pathway,” “KRAS signaling,” “T cell receptor signaling pathway,” and “apoptosis” were enriched pathways in the low-risk group. The relationship between metabolic dysfunction and the prognosis of EC has been extensively reported and studied, and the potential mechanisms are diverse. The PI3K/AKT/mTOR signaling pathway is overactivated in many tumors and participates in cancer invasion (41). The mTOR activation via CCL18 leads to cell migration in tumors such as EC (42). Through fatty acids metabolism (FAS), cancer cells can form cell membrane, store energy and produce signal molecules (43). It has been found to be up-regulated in many cancers including EC and it is also a reliable marker of increased risk of recurrence in EC. Growth factor receptors interact and activate downstream PI3K/AKT/mTOR axis with subsequent transcriptional activation of FAS expression and it is critical for aerobic glycolysis and tumor growth (44). The specific inhibition of FAS gene could lead to apoptosis of tumor cells (45). Chemokine signaling is found to participate in mTOR/KIF5B-mediated epithelial to mesenchymal transition and neoplastic metastasis by the PI3K/AKT/mTOR signaling pathway in EC (42). Insulin acts as a growth factor that promotes EC cell proliferation and inhibits apoptosis via PI3K/AKT/mTOR pathways (46). Insulin signaling pathway, especially insulin resistance, is positively correlated with the aggressiveness of EC and local tumor dissemination (47, 48). The promotion of EC cell proliferation involves activation of STAT3 and ERK2 signaling pathways in previous studies and blocking the JAK-STAT signaling pathway could be a rational therapeutic strategy for EC (49).

A nomogram was generated based on multiple clinical features and the risk signature to allow its application in clinical practice. This model provided a high accuracy for the prediction of the OS in the discovery and validation cohorts, suggesting the excellent predictive capability of this risk model in the prognosis of EC patients. Meanwhile, the analysis results of the subgroups with diverse characteristics showed that this nomogram could not only accurately differentiate patients in the whole groups, but could also accurately predict the OS in different clinicopathological subgroups. Several studies have suggested various prognostic biomarkers for EC based on mRNA or miRNA expression profiles (50). However, hub genes previously identified usually contained only biomarkers or several genomic markers without clinical features, restricting the clinical applicability of these biomarkers and decreasing the specificity of the overall prognostic models. Even though a similar metabolism-related risk model including nine genes associated with prognosis of EC was constructed in one study recently (51). There were still many differences between the two studies. Our study obtained the metabolic gene list from MSigDB and enrolled more DEGs in the LASSO analysis from two databases, of which the intersection made the results more credible. Besides, we further investigated the main metabolic pathways of the prognosis-related gene set through GSEA analysis in our study and the metabolic pathways corresponding to different outcomes were elucidated above. Additionally, we validated the risk model and nomogram internally from multiple aspects and externally with patients from our own center. With sufficient evidence, we constructed an integrative nomogram to allow for better prediction of prognosis and precision medicine in patients with EC.

TME is responsible for the tumor development and progression and the activation of immune system, which is a decisive factor during tumorigenesis and progression of the tumor (52). The metabolic pathways are activated in immune cells. Immune system needs to produce many metabolites to provide energy for various functions of immune cells (53). Moreover, there are quite a number of immune cells involving in metabolic pathways (54, 55). Our findings imply that the immune, stromal, and total scores are all negatively related to the metabolism-related risk of EC patients, proposing that more immune cells in the TME is associated with good prognosis of low-risk EC patients. In another study, lower immune and stromal scores are also found in high grade and invasive subtypes (10). The metastatic foci with the least amount of immune cells infiltration represents the worst immune microenvironment and the immune escape is most likely to occur (56). In breast cancer, CD8+ T cells, CD4+ T cells, M0 and M2 macrophages showed lower infiltrating concentration in the high tumor-associated immune genes groups, which suggested an adverse association between them (56). Similar to above, the results suggested that the immune cells have a significantly lower infiltrating density in the high-risk group. Among them, high numbers of CD8+ T-lymphocytes is an independent positive predictive factor for OS in EC patients (57). In addition, our data showed that high immune score patients have longer OS, presenting that the TME composition affects the final clinical outcomes of EC patients. Nevertheless, further studies are needed to clarify the mechanisms related to these immune microenvironments.

Up to now, the six-gene based predictive model has not been previously published, and our study has investigated the relationship between aberrant metabolism-related genes and the prognosis of EC patients. Moreover, the revealed biosignature is easy to test routinely, which can provide a cost-effective and accurate prognosis of EC in clinical practice. Nevertheless, there are some limitations existing in this study. First, metabolism is quite variable with tumor stage, between primary tumor and metastases, and likely even with circadian rhythm. Given its complexity and variability, larger sample sizes and broader perspectives are needed to explore the relations and associations between metabolism and cancers. Second, the risk model based on these six genes and the nomogram are all potential signature, and its clinical action needs to be further validated in more clinical centers. What’s more, in vivo and in vitro basic experimental verification is also needed to elucidate the molecular mechanisms and increase the persuasiveness and accuracy of these results.

Above all, we constructed a prognostic predictor aggregating six metabolism-related genes (HNMT, PHGDH, PSAT1, GPI, CA3, and CD38) and an integrative nomogram that could accurately and effectively predict the likelihood of OS and serve as a predictive tool for clinical prognosis and for guiding personalized anticancer treatment in EC patients. The function of immune cells infiltration in TME was also explored, which promotes the understanding of the underlying mechanisms involved in the development and prognosis of EC.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Peking University People’s Hospital (Approval number: 2020PHB063-01). The patients/participants provided their written informed consent to participate in this study.

YF and XL designed and executed the study, and analyzed the data. YF drafted all manuscripts, and all authors contributed to the critical discussion. JW revised the article. All authors contributed to the article and approved the submitted version.

This study is supported by National Key Technology R&D Program of China (Grant Nos. 2019YFC1005200 and 2019YFC1005201), and the National Natural Science Foundation of China (Grant Nos. 81874108, 81672571.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.