Lactylation-related risk model for prognostication and therapeutic responsiveness in uterine corpus endometrial carcinoma

Abstract

Background

Uterine corpus endometrial carcinoma (UCEC) is a prevalent gynecological cancer characterized by varied clinical outcomes and responses to treatment. Developing effective prognostic models is essential for guiding clinical decision-making. Recent research indicates that lactylation—a process impacting gene expression and immune responses—can affect tumor growth, metastasis, and immune evasion through histone modification. This study introduces a lactylation-related risk model aimed at predicting UCEC prognosis and providing insights into treatment efficacy.

Methods

We analyzed transcriptomic data from The Cancer Genome Atlas (TCGA) for UCEC patients and identified two distinct lactylation-related patterns using consensus clustering. A risk model developed using Cox and Lasso regression has been studied for its ability to predict prognosis, immune cell infiltration, and treatment response. Additionally, we investigated the relationship between IGSF1 gene expression and clinical features. Gene Set Enrichment Analysis (GSEA) was performed to explore the function of the IGSF1 gene.

Results

Two distinct lactylation-related clusters were identified, along with 156 differentially expressed genes between these clusters that are associated with the prognosis of UCEC. A risk model was developed based on three genes: IGSF1, ZFHX4, and SCGB2A1. This model effectively predicts clinical characteristics of UCEC patients, including immune cell infiltration, genetic variations, drug sensitivity, and response to immunotherapy. Notably, IGSF1 is linked to poor prognosis and is associated with immune activity, tumorigenesis, and cancer metabolism.

Conclusions

This study demonstrates that the lactylation-related risk model plays a crucial role in predicting prognosis and the efficacy of immunotherapy in UCEC, offering valuable insights for personalized treatment approaches.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-02524-0.

Introduction

Uterine corpus endometrial carcinoma (UCEC) is a significant contributor to gynecological cancer morbidity and mortality worldwide [1]. The heterogeneity of UCEC poses substantial challenges in prognostication and the development of targeted therapeutic strategies [2]. Traditional prognostic factors, while informative, often lack the specificity and sensitivity required to tailor treatment plans effectively [3]. Consequently, there is a critical need for innovative prognostic tools that can capture the multifaceted nature of UCEC.

There is mounting evidence that tumor metabolism is crucial to the initiation and development of malignancies as well as affecting immune cells through the release of metabolites [4, 5]. Lactate was once considered only a metabolite of glycolysis and the final product of the Warburg effect. In recent years, the field of cancer biology has witnessed a paradigm shift with the recognition of post-translational modifications, such as protein lactylation, as regulators of cellular processes and drivers of oncogenesis [6]. Lactylation has been implicated in the modulation of gene expression and the regulation of immune responses [7–10]. It plays a pivotal regulatory role across a spectrum of diseases, encompassing developmental anomalies, neurodegenerative disorders, inflammation, and cancer [11]. The potential involvement of lactylation in cancer progression and immune regulation indicates that lactylation gene signatures may offer novel insights into the prognosis and therapy of UCEC.

This study seeks to establish a gene signature associated with lactylation that improves prognostic accuracy and provides insights into the immunogenic characteristics and therapeutic susceptibilities of UCEC. By integrating gene expression data with lactylation profiles, we aim to develop a comprehensive signature that serves as both a prognostic and predictive biomarker. The aims of this research include enhanced patient stratification, informed personalized treatment strategies, and ultimately improved clinical management of UCEC.

Materials and methods

Data acquisition and consensus clustering

Public gene expression data and corresponding clinical information were retrieved from TCGA (https://cancergenome.nih.gov/) databases using the “TCGAbiolinks” R package. The gene sets related to lactylation were obtained from previously published studies [12, 13]. A total of 308 lactylation-related genes were involved in subsequent analysis. The UCEC cohorts were classified into distinct clusters by consensus clustering algorithm using the “ConsensuClusterPlus” package. This clustering was based on the expression profiles of the 308 lactylation-related genes.

Identification of differential expressed genes (DEGs) and gene set variation analysis

DEGs between cluster C1 and C2 were determined utilizing the “limma” R package. The significance threshold was set at an adjusted p-value of < 0.05 and an absolute log2 fold change of > 1 to determine the DEGs. Functional annotation of the DEGs was performed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses. To identify specific biological pathways enriched in different clusters, Gene Set Variation Analysis (GSVA) was conducted using the “clusterProfiler” and “GSVA” R packages. KEGG pathways sourced from the Molecular Signatures Database (MSigDB) were used as the gene sets for GSVA.

Construction of risk model and survival analysis

Prognosis-related DEGs were identified through univariate Cox regression analysis. These DEGs were then subjected to least absolute shrinkage and selection operator (LASSO) regression analysis, followed by Cox regression analysis. LASSO regression analysis was employed to identify the model genes associated with prognosis. The coefficients obtained from the Cox regression analysis were used to calculate a risk score for each patient. The risk score was calculated using the following formula:

$$\text{Risk score}={\sum }_{\text{i}=1}^{\text{n}}(\text{Expi}\ast \text{Coefi})$$

Expi and Coefi represent the risk coefficient and the expression value of model genes, respectively. Patients from the TCGA UCEC dataset were divided into high- and low-risk groups using the median risk score. Kaplan–Meier survival analysis was used by the “survminer” R package to predict patient survival, followed by the generation of receiver operating characteristic (ROC) curves in order to predict performance of the risk model.

Cox regression analysis and nomogram construction

Univariable and multivariate Cox regression analysis was conducted to determine whether risk score was an independent prognosis factor for predicting the survival of UCEC patients. To further enhance the predictive ability of the risk model, a nomogram was developed using the “rms” R package to incorporate independent prognostic criteria such as the risk score, stage, age, and grade. This nomogram calculates the probabilities of overall survival at 1-year, 3-year, and 5-year intervals.

Enrichment analysis

To explore the underlying molecular mechanisms associated with the risk score and IGSF1, Gene Set Enrichment Analysis (GSEA) was performed. The analysis was conducted using the “GSEABase” R package, with the KEGG and HALLMARK gene sets retrieved from the Molecular Signatures Database (MSigDB). The GSEA threshold was set to an absolute normalized enrichment score (|NES|) > 1, a p-value < 0.05, and a false discovery rate (FDR) < 0.25. Spearman rank correlation analysis was performed to evaluate the correlation between the risk score and the enrichment scores of the KEGG and HALLMARK gene sets.

Tumor microenvironment landscape analyses and immunotherapy response prediction

The single sample gene set enrichment analysis (ssGSEA) was employed to analyse the enrichment of infiltrating immune cells and immune function [14, 15]. The correlation between six categories of immunomodulators: major histocompatibility complex (MHC), cytokines, chemokine receptors, immune stimulators, immune inhibitors, and immune checkpoints [16] and the model genes expression was calculated using Spearman’s rank correlation. The ESTIMATE algorithm was utilized to compute the ESTIMATE score, immune score, and stromal score. Data on the Tumor Immune Dysfunction and Exclusion (TIDE) scores, T cell dysfunction, T cell exclusion, and Microsatellite Instability (MSI) were obtained from the TIDE website (http://tide.dfci.harvard.edu/) [17]. The immunophenoscore (IPS) of UCEC patients in TCGA was acquired from the Cancer Immunome Atlas (https://tcia.at/home) [18]. IMvigor210 cohort (advanced urothelial cancer treated with anti-PD-L1 antibody atezolizumab) was used to validate the prediction of immunotherapy efficacy using the risk model [19].

Genetic variation analysis

The tumor mutation burden (TMB) was calculated from the somatic mutation data of TCGA by"Maftools"R package. We divided UCEC samples into high- and low-TMB groups based on the median TMB. The correlation between the TMB and the risk score was demonstrated using scatter plot. The copy number variation (CNV) and the somatic mutations of UCEC were obtained from TCGA database. The mutation annotation format (MAF) files from TCGA were generated using the “maftools” R package and the somatic mutations in low- and high-risk groups were visualized using waterfall plot.

Drug susceptibility analysis

The sensitivity analysis of 198 drugs from the Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Therapeutics Response Portal (CTRP) databases was performed using “oncoPredict” R package. The correlation between the IC50 values and the expression of model genes was calculated using Spearman’s rank correlation. The IC50 values of high- and low-risk groups were represented using violin plots to explore the drug susceptibility of different groups (Threshold: p-value < 0.01 and Mean IC50 < 5).

Statistical analyses

The statistical analysis in this study was performed using R software version 4.3.1. The statistical significance of normally distributed quantitative variables was assessed using the Student's t-test, while non-normally distributed variables were analyzed using the Wilcoxon rank-sum test. The Kruskal–Wallis test and one-way analysis of variance (ANOVA) were employed as non-parametric and parametric methods, respectively, for making comparisons among multiple groups.

Results

Landscape of genetic variation of lactylation-related genes in UCEC

A total of 332 lactylation-related genes, as identified in previously published studies, were included in this research [12] (Table S1). We ultimately found 308 lactylation-related genes included in UCEC data from TCGA. Subsequently, we examined the frequencies of somatic mutations and the burden of somatic copy number variations in lactylation-related genes using the TCGA-UCEC dataset. The genes exhibiting a high mutation frequency included ARID1 A (46%), TP53 (41%), CHD4 (23%), AHNAK (19%), PRKDC (18%), MKI67 (17%), RANBP2 (16%), and EP300 (14%) (Fig. 1A). Notably, 94.63% of the samples exhibited alterations. The frequencies of copy number variations are reported, with several genes demonstrating a significant burden of gains (Fig. 1B). Altogether, these data indicate that lactylation-related genes harbor somatic mutations and copy number alterations. Predominantly implicated in cancer progression, these genes may serve as potential biomarkers for the classification of UCEC.

Fig. 1.

Characterization of lactylation-related genes in UCEC. A Single nucleotide variant waterfall of lactylation-related genes. B Copy number variation of lactylation-related genes

Generation of lactylation-related clusters and functional enrichment analysis

A total of 553 tumor samples from patients with UCEC in the TCGA database were subjected to clustering based on lactylation-related genes. Utilizing the R package ConsensusClusterPlus, we conducted an unsupervised clustering analysis and determined k = 2 as optimal based on empirical cumulative distribution function (CDF) plots (Fig. 2A). These plots indicated that k = 2 exhibited the highest within-group correlations and the lowest between-group correlations compared to alternative values. Two distinct lactylation-related gene expression patterns, lactylation clusters C1 and C2, were observed (Fig. 2B). To evaluate survival outcomes, we performed Kaplan–Meier survival analysis for the two clusters. This analysis revealed poor OS in patients in cluster C1 (Fig. 2C).

Fig. 2.

Identification of two clusters in TCGA UCEC patients. A The cumulative distribution function (CDF) curve when k varies from 2 to 10. B Consensus clustering matrix for k = 2. C Kaplan–Meier survival curves of two clusters

As reported by GSVA, cluster C1 was significantly enriched in various pathways and processes, including spliceosome, mismatch repair, cell cycle, DNA replication, and oocyte meiosis, while cluster C2 was significantly enriched in phenylalanine metabolism, arachdonic acid metabolism, tyrosine metabolism, and ether lipid metabolism (Fig. 3A). To further elucidate the potential biological behaviors of each cluster, we conducted differential analyses. 1035 DEGs were selected based on threshold criteria (The adjusted p-value < 0.05 and |log2 FoldChange|> 1). Subsequently, we employed GO and KEGG analyses to identify the molecular functions of the DEGs (Fig. 3B, C). The DEGs are related to immune responses, including immunoglobulin complex and IL-17 signaling pathway. They are also associated with mitochondrial function and energy production processes, including mitochondrial respirasome, ATP biosynthetic process, aerobic respiration, and oxidative phosphorylation.

Fig. 3.

The functional annotation of DEGs. A Heatmap of GSVA analysis of two clusters. B KEGG enrichment analysis of DEGs between two clusters. C GO enrichment analysis of DEGs between two clusters

Based on the results, we identified two clusters with distinct immunological and metabolic features, indicating that lactylation may influence the initiation and progression of UCEC by modulating its immune microenvironment and metabolic pathways.

Constructing prognostic risk model

Univariate Cox regression analysis was conducted to identify DEGs with significant prognostic value (p-value < 0.05) for further investigation. A total of 156 DEGs associated with the prognosis of UCEC were finally identified. Subsequently, LASSO regression analysis was performed to eliminate redundant genes (Fig. 4A), resulting in the selection of five genes. Among these, the non-coding gene “AC092969.1” and the pseudogene “PPIAP6” were excluded from further analysis because they do not encode functional proteins and, therefore, do not directly contribute to cellular processes or functions. The remaining three genes—IGSF1, ZFHX4, and SCGB2 A1—were validated using Cox regression analysis. The final risk score was based on these three gene signatures. The correlation coefficients are presented in Fig. 4B and Table S2. Additionally, a heat map illustrates the expression levels of the three genes in relation to the risk model (Fig. 4C).

Fig. 4.

Construction of a prognostic risk model. A Lasso regression analysis of prognosis-related DEGs. B The correlation coefficient of model genes (the left side indicates positive correlation; the left side indicates negative correlation). C The survival status for each patient and the predictive value of risk score

Prognostic value and clinical implications of the risk model

The risk model demonstrated significant clinical relevance (Fig. 5A). Additionally, our analysis revealed a strong correlation between the risk score and the CNV high molecular subtype in UCEC (Fig. 5B). This molecular subtype, characterized by high levels of CNV, has been reported to have the poorest prognosis among patients with UCEC [20]. To further explore the characteristics of patients with varying risk scores, individuals from the TCGA UCEC dataset were stratified into high- and low-risk groups based on the median risk score. The concordance of this risk model with survival outcomes was notably strong, as evidenced by a c-index of 0.70. Kaplan–Meier survival analysis revealed that patients in the high-risk group exhibited significantly worse prognoses compared to those in the low-risk group (Fig. 5C). Furthermore, ROC analysis demonstrated that the risk model possesses robust predictive accuracy for 1-year, 3-year, and 5-year survival rates (Fig. 5D). To evaluate whether the risk model could serve as an independent prognostic factor, univariate and multivariate Cox regression analyses were performed on various clinical characteristics, including age, FIGO stage, and grade (Table S3). The results confirmed that the risk score is an independent prognostic indicator. Additionally, the nomogram further validated the utility of the risk model as a significant predictor of clinical outcomes (Fig. 5E).

Fig. 5.

The clinical characteristics of the risk model. A The relationship between risk scores and patient age, stage, and grade. B Distribution of risk scores across different molecular subtypes of UCEC. C Kaplan–Meier survival curves of two risk groups. D ROC of the risk model. E Nomogram of the risk score and clinical information

Transcriptome traits in different risk groups

To elucidate the functional characteristics of the risk groups, we performed GSEA on both high- and low-risk cohorts. The differential pathway enrichment observed between these groups provides insights into the underlying mechanisms of carcinogenesis that may impact patient prognosis. Our analysis revealed notable differences in enrichment scores between the high-risk and low-risk groups (Fig. 6A, B). Specifically, pathways such as cytokine-cytokine receptor interaction, oxidative phosphorylation, and antigen processing and presentation were significantly upregulated in the low-risk group, suggesting a potential negative correlation with immune activity. Conversely, various oncogenic pathways, including G2M checkpoint, MYC targets, KRAS signaling, cell cycle, and DNA replication, were activated in the high-risk group, indicating a potential positive correlation with carcinogenesis.

Fig. 6.

Gene set enrichment analysis of risk model. A GSEA based on the KEGG gene sets showed the states of biological processes in low- and high-risk groups. B GSEA based on the KEGG gene sets showed the states of biological processes in low- and high-risk groups

To further explore the relationship between the risk score and tumor hallmarks, we performed Spearman rank correlation analysis. Tumor-specific hallmarks such as KRAS signaling, MYC targets, and the G2M checkpoint pathway demonstrated significant correlations with the risk score (Fig. S1). Additionally, we analyzed the correlation between the risk score and KEGG pathway enrichment scores (Fig. S2).

Differences in genetic mutations

The initiation and progression of cancer are heavily influenced by genetic mutations [21]. Consequently, a deeper understanding of genetic alterations in UCEC could greatly facilitate the development of targeted therapies and novel tumor treatment strategies. In this study, we analyzed the 20 genes with the highest mutation frequencies in patients stratified into high-risk and low-risk groups, subsequently visualizing the results using waterfall plots. The findings revealed significant differences in gene mutation frequencies between the two groups (Fig. 7A, B). Additionally, we calculated the TMB and categorized patients into high-TMB and low-TMB groups. Kaplan–Meier survival analysis indicated that patients in the high-TMB group had significantly better prognoses (Fig. 7C). Moreover, scatter plot analysis revealed a significant negative correlation between the risk score and TMB (Fig. 7D; R = − 0.14, p = 0.0022).

Fig. 7.

Genetic variation in different risk groups. A Waterfall plot of single nucleotide variants in patients from the low-risk group. B Waterfall plot of single nucleotide variants in patients from the low-risk group. C Kaplan–Meier survival curve in patients from high- and low-TMB groups. D The correlation between risk score and TMB by scatter diagram

Tumor immune microenvironment in the risk model

To further explore the relationship between risk score and the tumor microenvironment (TME), we analyzed the differential abundance of immune-infiltrating cells and immune functions to characterize the TME landscape. Immune signatures were assessed using the ssGSEA algorithm, revealing that the low-risk group exhibited a significantly higher abundance of various immune cell types associated with antigen presentation, processing, and tumor cytotoxicity. These immune cells included activated CD8 + T cells, Th17 cells, activated B cells, CD56 dim natural killer (NK) cells, activated dendritic cells, and eosinophils (Fig. 8A). Consistently, the low-risk group demonstrated enhanced activity of pathways involved in antigen recognition, processing and presentation, and antitumor immunity, such as cytolytic activity, T cell co-stimulation, human leukocyte antigen (HLA) signaling, CC chemokine receptor (CCR) activity, and type II interferon (IFN) response (Fig. 8B). To further assess the relationship between risk score and immune signature, Spearman’s rank correlation analysis was performed, confirming significant associations (Fig. 8C, D). Collectively, these findings reveal a distinct immune signature between the high- and low-risk groups, suggesting that immune cell infiltration in the TME is closely associated with risk levels.

Fig. 8.

Characteristic of TME between low- and high-risk group. A ssGSEA enrichment analyses based on the infiltrating immune cells signature. B ssGSEA enrichment analyses based on the immune function signature. C Spearman’s rank correlation analysis between risk groups and infiltrating immune cells signature. D Spearman’s rank correlation analysis between risk groups and immune function signature. E ESTIMATE score, immune score and stromal score between low- and high-risk group. *Indicates p ≤ 0.05, **indicates p ≤ 0.01, ***indicates p ≤ 0.001, and ****indicates p ≤ 0.0001

To complement this analysis, the ESTIMATE algorithm was applied to calculate the ESTIMATE score, immune score, and stromal score. Higher ESTIMATE scores are indicative of lower tumor purity and better prognostic outcomes. Our analysis demonstrated a significant elevation in ESTIMATE scores, as well as stromal and immune scores, in the low-risk group compared to the high-risk group, indicating a greater abundance of infiltrating immune cells in the low-risk group (Fig. 8E). These results suggest that the low-risk group is characterized by a “hot” immune phenotype, which is marked by an increased presence of antitumor immune cells and upregulated antitumor pathways.

Moreover, immune checkpoints, which are molecules expressed by immune cells to regulate immune activation [22], were also investigated. We examined six categories of immunomodulators—MHC, cytokines, chemokine receptors, immune stimulators, immune inhibitors, and immune checkpoints—and their correlations with the expression levels of three model genes. The analysis revealed strong associations between these three model genes and both immune checkpoints and MHC molecules, further underscoring the critical role of immune regulation in the tumor microenvironment (Fig. S3).

Prediction of drug sensitivity and immunotherapy response

Based on the expression data of TCGA UCEC samples, we predicted the IC50 values of 198 drugs from the GDSC and CTRP databases. Our analysis revealed significant differences in drug sensitivity between the low- and high-risk groups (Threshold: p-value < 0.01 and Mean IC50 < 5). Specifically, 36 drugs exhibited lower IC50 values in the low-risk group, suggesting increased sensitivity to these agents. In contrast, 5 drugs showed lower IC50 values in the high-risk group (Table S4). Notably, patients in the high-risk group had significantly elevated IC50 values for chemotherapeutic agents commonly used in clinical practice, such as teniposide, topotecan, epirubicin, vincristine, docetaxel, camptothecin, dactinomycin, and paclitaxel, compared to those in the low-risk group (Fig. 9A). These elevated IC50 values in the high-risk group imply a reduced response to these drugs, which may adversely affect chemotherapy efficacy in these patients.

Fig. 9.

Prediction of the drug susceptibility and immunotherapy efficacy of patients with UCEC by the risk model. A The difference in the distribution of IC50 values of eight chemotherapeutic drugs between the high-risk and low-risk groups. Red represents the high-risk group, and blue represents the low-risk group. B Kaplan–Meier survival curve in patients from high- and low-risk groups in anti-PD-L1 cohort. C The cumulative distribution of immunotherapy response in patients from the high-risk and low-risk groups. D The distribution of risk score in different immunotherapy response groups. E TIDE score, Dysfunction score, and Exclusion score between low- and high-risk group. F IPS with CTLA4+ and PD-1+, CTLA4− and PD-1+, CTLA4+ and PD-1− and CTLA4− and PD-1− between low- and high-risk group. *Indicates p ≤ 0.05, **indicates p ≤ 0.01, ***indicates p ≤ 0.001, and ****indicates p ≤ 0.0001

Furthermore, our findings demonstrated that patients in the low-risk group were more sensitive to targeted small-molecule inhibitors, specifically rapamycin and dactolisib, known inhibitors of the PI3 K/AKT/mTOR signaling pathway. This suggests that low-risk patients may benefit from therapeutic interventions targeting the PI3 K/AKT/mTOR pathway (Fig. S4). Additionally, we calculated the correlation between the expression of the three model genes and the IC50 values of chemotherapeutic drugs. The results highlighted strong correlations between these model genes expression and drug IC50 values (Fig. S5). Together, these results suggest that the risk score has the potential to serve as a predictor of sensitivity to anticancer therapies.

To further assess the model’s ability to predict responses to immunotherapy, we stratified the IMvigor210 dataset into high- and low-risk groups based on the median risk score. In the anti-PD-L1 cohort (IMvigor210) [19], patients in the low-risk group exhibited significant clinical benefits and markedly prolonged survival (Fig. 9B). We compared the distribution of immune responses between samples with complete or partial responses (CR + PR) and those with stable or progressive disease (SD + PD) using cumulative distribution maps (Fig. 9C). However, no significant differences in risk scores were observed among these groups (Fig. 9D).

TIDE is widely used to evaluate responsiveness to immunotherapy, with higher TIDE values correlating with more favorable therapeutic outcomes [17]. Patients in the low-risk group exhibited higher levels of T cell dysfunction and MSI, along with lower levels of T cell exclusion and TIDE scores, compared to those in the high-risk group (Fig. 9E). Additionally, higher IPS values were observed in the low-risk group, indicating enhanced sensitivity to immunotherapy in this cohort (Fig. 9F).

IGSF1 as a potential target for immunotherapy in UCEC

To further explore the function of the risk model, we conducted Kaplan–Meier survival analysis on the three genes included in our model. Among them, IGSF1 and ZFHX4 were associated with poorer prognoses, whereas SCGB2 A1 exhibited an opposing trend, correlating with improved outcomes (Fig. 10A).

Fig. 10.

The clinical characteristics of IGSF1. A Kaplan–Meier survival curve of three model gene in UCEC. B The age, stage, and grade of UCEC patients from the high- and low-expression groups of IGSF1. C The expression of IGSF1 in different molecular subtypes of UCEC

The candidate gene IGSF1 encodes a membrane glycoprotein that is proposed to regulate thyroid function. Recent studies have identified IGSF1 as a promising immunosuppressive therapeutic target [23]. We further investigated the correlation between IGSF1 expression and various clinical characteristics, revealing significant associations with patient age, tumor grade, and stage (Fig. 10B). Moreover, our analysis uncovered a strong link between high IGSF1 expression and the CNV high molecular subtype in endometrial carcinoma (Fig. 10C).

To better understand the molecular mechanisms underlying IGSF1, we performed GSEA using the TCGA UCEC dataset. The results indicated that IGSF1 is closely related to immune activity, tumorigenesis, and cancer metabolism (Fig. 11A–C). The GSEA results are detailed in Supplementary Table 5.

Fig. 11.

GSEA enrichment analysis of IGSF1. A The enrichment of tumorigenesis-related pathways in UCEC. B The enrichment of immune activity related pathways in UCEC. C The enrichment of cancer metabolism-related pathways in UCEC

In conclusion, our findings suggest that IGSF1 plays a critical role in the progression of UCEC and may serve as a potential therapeutic target for immunotherapy.

Discussion

In recent years, accumulating evidence has underscored the pivotal role of lactate in tumor biology. Beyond serving as the primary metabolic fuel for neoplastic cells, lactate profoundly influences tumor progression, proliferation, metastasis, chemoresistance, and immunomodulation [24–26]. These effects are mediated through diverse mechanisms, including acidification of the immune microenvironment and upregulation of tumor resistance proteins [11]. Interestingly, lactate also serves as the substrate for lactylation, a newly identified type of post-translational modification that occurs on lysine residues of histones and other proteins. This modification links metabolic status to epigenetic and functional regulation. Notably, the lactylation modification of histone lysines has been shown to be widespread across human and mouse cells, playing a critical role in the regulation of cellular functions [6]. However, comprehensive multi-omics studies investigating this modification in UCEC remain limited.

To address this gap, we defined two clusters within the TCGA UCEC cohort using consensus clustering analysis of 308 lactylation-related genes. The resulting clusters exhibited distinct prognostic, immunological, and metabolic characteristics. GO and KEGG pathway enrichment analyses of differentially expressed genes between the two clusters highlighted significant enrichment in pathways related to immune activity and cellular metabolism. To enhance clinical utility, we developed a lactylation-related risk model based on prognosis-associated DEGs. This model incorporates the expression patterns of three genes—IGSF1, ZFHX4, and SCGB2 A1—and stratifies patients into high- and low-risk groups using median risk scores. The ROC curve demonstrated robust predictive performance, while survival analysis revealed significantly worse outcomes for patients in the high-risk group compared to those in the low-risk group.

To investigate the prognostic disparity between the high- and low-risk groups, differential pathway enrichment analysis was performed. Pathways related to cytokine-cytokine receptor interactions, oxidative phosphorylation, tyrosine metabolism, bile acid metabolism, and fatty acid metabolism were significantly upregulated in the low-risk group, suggesting a potential negative correlation between the risk score and immune activity/metabolism. Tumor cells rely heavily on glycolysis to sustain growth, producing lactate that contributes to the formation of a high-lactate microenvironment, which in turn inhibits T cell activity [27]. Histone lactylation, acting as a biological signaling molecule, modulates immune responses within the tumor microenvironment (TME) and impacts tumor growth and immune evasion mechanisms [28].

Immune signature analysis using ssGSEA revealed differential immune cell infiltration between the risk groups. The high-risk group exhibited elevated levels of central memory CD8 + T cells, effector memory CD4 + T cells, activated CD4 + T cells, and regulatory T cells (Tregs). In contrast, the low-risk group displayed higher levels of activated CD8 + T cells, Th17 cells, activated B cells, CD56 dim and CD56 bright natural killer cells, activated dendritic cells, and eosinophils. Tregs are known to suppress antitumor immunity, thereby impairing effective immune responses [29]. Conversely, CD8 + T cells are recognized for their cytotoxic activity against tumor cells [30], while Th17 cells contribute to inflammation and antitumor immune responses [31]. Activated B cells and dendritic cells play pivotal roles in antigen presentation and immune activation [32, 33], and eosinophils have also been implicated in antitumor immunity [34]. Furthermore, the low-risk group exhibited active signaling pathways related to cytolytic activity, T cell co-stimulation, HLA, CCR, and the type II IFN response, which enhance immune surveillance and bolster antitumor immunity. These findings suggest that the low-risk group possesses a more favorable immune microenvironment, characterized by enhanced immune cell infiltration, antigen presentation, and cytotoxic immune responses, thereby contributing to better tumor control and improved clinical outcomes.

Additionally, the low-risk group demonstrated significantly higher ESTIMATE scores, stromal scores, and immune scores, further supporting the characterization of the low-risk group as having a “hot” immune phenotype. This phenotype, associated with increased tumor-infiltrating immune cells, has been shown to respond favorably to immunotherapy [35]. Consistent with this, our analysis of immune checkpoints, TIDE scores, IPS, and immunotherapy cohorts validated the enhanced immunotherapy sensitivity of the low-risk group. In the anti-PD-L1 cohort (IMvigor210), patients in the low-risk group exhibited significant clinical benefits and prolonged survival. Elevated TIDE scores and IPS in the low-risk group further corroborate their improved sensitivity to immunotherapeutic interventions.

TMB is another critical biomarker for predicting response to immune checkpoint inhibitors, as high TMB correlates with increased neoantigen production and better immunogenicity [36, 37]. In line with previous studies, we found that high-TMB patients had improved prognoses. Moreover, a negative correlation was observed between risk score and TMB, suggesting that patients in the low-risk group may derive greater benefit from immunotherapy.

Lactate has also been implicated in tumor drug resistance, with lactate concentration and glycolysis rates serving as potential indicators of chemotherapy sensitivity [7]. Recent studies further highlight the role of lactylation in regulating DNA damage repair and contributing to chemotherapy resistance via NBS1 lactylation [38]. Our drug sensitivity analysis revealed that low-risk patients demonstrated heightened sensitivity to commonly used chemotherapeutic agents, underscoring the potential of the risk model as a predictor of anticancer therapy efficacy.

The three genes incorporated into our risk model warrant further investigation. ZFHX4 is involved in epithelial-mesenchymal transition (EMT) and extracellular matrix (ECM) remodeling, suggesting its role in promoting metastasis in ovarian cancer [39]. SCGB2 A1, a novel tumor antigen, has shown significant differential expression across major histological types of ovarian cancer [40] and is a prognostic marker associated with chemoresistance and radioresistance in colorectal cancer [41]. In UCEC, SCGB2 A1 has been linked to high levels of infiltrating CD8 + T cells and prognosis [42]. Lastly, IGSF1, an immune suppressor, is emerging as a potential target for cancer immunotherapy. Preclinical studies have demonstrated that IGSF1-targeting antibodies exhibit significant anticancer activity both as monotherapy and in combination with anti-PD-1 inhibitors [23]. Our findings further corroborate the association of IGSF1 with poor prognosis, immune activity, tumorigenesis, and cancer metabolism in UCEC.

In conclusion, our study highlights the critical role of lactylation modification in UCEC progression and immune modulation. The lactylation-based risk model demonstrates favorable prognostic performance and provides insights into immune phenotypes and therapeutic strategies, paving the way for precision medicine in UCEC. In addition, the three genes identified in the risk model still offer significant potential for further exploration within the context of endometrial cancer, particularly in understanding their biological mechanisms and therapeutic implications.

Limitation

The prognostic risk model described in this study is based on a cohort of UCEC patients from the TCGA database. Given the retrospective nature of this analysis, it is crucial to rigorously validate the accuracy of this model by incorporating larger sample sizes and prospective studies. Additionally, empirical evidence is essential to explore the regulatory roles of the three genes within the model in the initiation and progression of UCEC.

Author contributions

Y.Y and M.L developed the concept and designed the study. Y.Y analyzed the data and interpreted the results. Y.Y and M.L drafted the manuscript. All authors critically reviewed the manuscript and approved the final version of the manuscript.

Data availability

The data that support the findings of this study are available from TCGA database (URL:https://www.cancer.gov/ccg/research/genome-sequencing/tcga).

Declarations

Competing interests

The authors declare no competing interests.