Construction of a multigenic diagnostic, prognostic, and immune infiltration model with methylation-associated regulators in esophageal squamous cell carcinoma

Introduction

The latest epidemiological report on cancer reveals a global incidence of 510,716 cases of esophageal cancer, with 445,129 deaths, highlighting the exceptionally high mortality rate associated with this malignancy (1). In China, a recent report identified 224,000 cases of esophageal cancer, representing approximately 43.9% of the global total (2). In contrast to cases in higher-income, Western nations, approximately 90% of esophageal cancer cases in China are diagnosed as esophageal squamous cell carcinoma (ESCC) (3,4). Furthermore, significant variations in the incidence of ESCC have been observed across different genders and regions within China (4). The incidence of ESCC is influenced by a variety of factors, including nitrosamines, dietary habits, smoking, alcohol consumption, obesity, oral infections, human papilloma virus infection, and gastroesophageal diseases (4-8). The diverse pathogenic mechanisms associated with these factors present significant challenges for clinical treatment and the prediction of prognosis. Our study aimed to establish a highly practical prognostic and diagnostic model to provide a reliable and widely applicable framework for the clinical diagnosis and prognostic analysis of ESCC.

There is an urgent clinical need for reliable prediction models for ESCC to accurately assess patient prognosis and evaluate the efficacy of chemotherapy and immunotherapies. Various prognostic models for ESCC, which focus on ferroptosis, necroptosis, programmed cell death, immune modulators, immune checkpoint inhibitors, epithelial–mesenchymal transition molecular heterogeneity, and noncoding RNA, are currently in development (9-13). Furthermore, numerous methylation-related genes have been identified that are associated with the diagnosis and prognosis of patients with ESCC (14-17). For example, whole-genome methylation analysis has identified the methylated genes potassium voltage-gated channel subfamily A member 3 (KCNA3) and otopetrin 2 (OTOP2) as potential diagnostic markers for ESCC (14). In ESCC, protein arginine methyltransferase 5 (PRMT5) is upregulated, and its elevated levels are linked to metastasis, suggesting it as an independent prognostic factor for overall survival (OS) (16).

In this study, we investigated the association of methylation-related genes with the prognosis of patients with ESCC. Using The Cancer Genome Atlas (TCGA) database, we finally identified the diagnostic efficiency and immune infiltration associated with four genes in ESCC: homocysteine-inducible ER protein with ubiquitin-like domain 1 (HERPUD1), trans-2,3-enoyl-CoA reductase (TECR), melanoma antigen gene A11 (MAGEA11), and NOP2/Sun RNA methyltransferase 6 (NSUN6). We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-341/rc).

Methods

Prognostic factor screening

Prognostic genes were selected using TCGA database (https://portal.gdc.cancer.gov), with 82 patients with ESCC being categorized into two groups: a good prognosis group (n=41) and a poor prognosis group (n=41). Good prognosis and poor prognosis were classified according to the median of OS of ESCC patients. We identified 967 prognostic genes with a significance threshold of P<0.05. The “survival package” (version 3.3.1) in R software (version 4.2.1; The R Foundation for Statistical Computing, Vienna, Austria) was used for batch-fitting survival regression models. Nonzero coefficients were subsequently screened using least absolute shrinkage and selection operator (LASSO) regression. Sixteen nonzero coefficients were then used in a Cox proportional hazards model, with four genes ultimately being identified as the independent prognostic factors for patients with ESCC. HERPUD1, TECR, MAGEA11, and NSUN6 were subsequently employed to construct both a nomogram model and a risk scale model. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Gene expression analysis

The gene expression of HERPUD1, TECR, MAGEA11, and NSUN6 were analyzed with TCGA database and three Gene Expression Omnibus (GEO) datasets (GSE161533, GSE45670, and GSE20347) via the “ggplot2” package (version 3.3.6), the “stats” package (version 4.2.1), and the “car” package (version 3.1-0) in R software (version 4.2.1).

Survival analysis in patients with ESCC

Risk scores of HERPUD1, TECR, MAGEA11, and NSUN6 associated with OS, disease-specific survival (DSS), and progression-free survival (PFS) were used to evaluate the proportional risk hypotheses via the “survival” package (version 3.3.1) and to conduct fitted survival regression analyses. The outcomes of these analyses were subsequently visualized with the “survminer” (version 0.4.9) and “ggplot2” (version 3.3.6) packages in R software (version 4.2.1).

Diagnostic analysis

Using data from the TCGA (n=82) and the Genotype-Tissue Expression (GTEx) portal (n=653), we extracted the gene expression data for HERPUD1, TECR, MAGEA11, and NSUN6 for the analysis of diagnostic efficacy in patients with ESCC. Receiver operator characteristic (ROC) curve analysis was performed on the data using the “pROC” package (version 1.18.0), and the results were visualized using “ggplot2” package (version 3.4.4) in R software (version 4.2.1).

Association between prognostic factors and immune infiltration

The single-sample gene set enrichment analysis (ssGSEA) algorithm was used to investigate the association between HERPUD1, TECR, MAGEA11, and NSUN6 and immune infiltration based on markers for 24 immune cells in patients with ESCC. The algorithm was implemented via the “GSVA” package in R version 1.46.0.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis

Firstly, the TCGA database was utilized to identify differentially expressed genes (DEGs) associated with HERPUD1 and TECR. Subsequently, immune infiltration-related genes (score ≥10) were screened using the GeneCards database (https://www.genecards.org/). Finally, overlapping genes between the two gene sets were selected for further analysis. Enrichment analysis results were visualized using R (version 4.2.1) and the ggplot2 package (version 3.4.4).

Statistical analysis

The data are presented as the mean ± standard deviation. LASSO and Cox proportional hazards regression were employed to select prognostic indicators with the “glmnet” package (version 4.1.7), “survival” package (version 3.3.1), and “rms” package (version 6.3-0) in R software version 4.2.1. Spearman correlation coefficient analysis was used to assess correlations. The t-test or Wilcoxon rank-sum test was used for comparing the differences between groups via the “ggplot2” package (version 3.3.6), “stats” package (version 4.2.1), and “car” package (version 3.1-0) in R. The R “survival” package (version 3.3.1) was used to select the prognostic factors in patients with ESCC. Kaplan-Meier survival analysis was used to evaluate the association of prognostic indicators with OS, DSS, and PFS via the “survival” (version 3.3.1), “survminer” (version 0.4.9), and “ggplot2” (version 3.3.6) packages. Time-dependent area under the curve (AUC) analysis was used to evaluate the prognostic values of HERPUD1, TECR, MAGEA11, and NSUN6 at various time points via the “timeROC” (version 0.4) and “ggplot2” (version 3.4.4) packages in R software (version 4.2.1). Nomogram and calibration related models were built and visualized using the R “rms” package (version 6.3-0). During data analysis, missing values were handled using imputation methods, and normalization was performed using the normalizeBetweenArrays function from the limma package (version 3.52.2). A P value <0.05 indicated statistical significance.

Results

Screening of prognostic factors in ESCC

We first analyzed the prognostic genes using TCGA database, categorizing 82 patients with ESCC into two groups: a good prognosis group (n=41) and a poor prognosis group (n=41). We identified 967 prognostic genes with a significance threshold of P<0.05. GO analysis identified methylation-associated signaling pathways, such as RNA methylation, RNA methyltransferase activity, S-adenosylmethionine-dependent methyltransferase activity, and macromolecule methylation as potentially linked to prognosis (Figure 1A). Based on these findings, we hypothesized that methylation-related genes could serve as prognostic indicators for patients with ESCC. We used GeneCards to screen 2,336 methylation-related genes and intersected them with prognostic factors for ESCC, identifying 158 methylation-related prognostic factors (Figure 1B). Finally, we performed LASSO regression analysis with cross-validation curves (Figure 1C) and LASSO coefficient path diagrams (Figure 1D) to identify nonzero coefficients. We ultimately identified 16 nonzero coefficients. Based on 16 nonzero coefficients, we screened four independent prognostic genes using Cox proportional hazards regression analyses: HERPUD1 [hazard ratio (HR) =5.345, 95% confidence interval (CI): 1.297–22.024; P=0.02], TECR (HR =5.162, 95% CI: 1.379–19.325; P=0.02), MAGEA11 (HR =0.037, 95% CI: 0.006–0.217; P<0.001), and NSUN6 (HR =0.196, 95% CI: 0.054–0.705; P=0.01) (Table 1).

Figure 1 Screening of prognostic factors in ESCC. (A) ESCC-related prognostic factors were selected from TCGA database, and then functional enrichment analysis of prognostic factors was performed via GO analysis. (B) Methylation-related prognostic factors were screened with the GeneCards database. (C) LASSO cross-validation curves and (D) LASSO coefficient path diagrams were drawn to identify the nonzero coefficients. ESCC, esophageal squamous cell carcinoma; TCGA, The Cancer Genome Atlas; GO, Gene Ontology; LASSO, least absolute shrinkage and selection operator.

The association of HERPUD1, TECR, MAGEA11, and NSUN6 with clinical parameters in patients with ESCC

The expression level of HERPUD1 was significantly higher in the high clinical stage (III–IV) group compared to the low clinical stage (I–II) group. Additionally, the expression levels of TECR, MAGEA11, and NSUN6 were not significantly correlated with pathological stage, histological grade, primary therapy outcomes, alcohol history, or reflux history (Figure 2).

Figure 2 The association of HERPUD1, TECR, MAGEA11, and NSUN6 with clinical parameters in patients with ESCC. The expression level of HERPUD1, TECR, MAGEA11, and NSUN6 was used to analyze their correlation with clinical stage, pathological stage, histological grade, primary therapy outcomes, alcohol history, and reflux history. *, P<0.05. HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; TPM, transcripts per million; ESCC, esophageal squamous cell carcinoma.

Verification of HERPUD1, TECR, MAGEA11, and NSUN6 expression in the GEO and TCGA databases in patients with ESCC

Our results showed that HERPUD1, TECR, and MAGEA11 were significantly upregulated in ESCC tissues as compared to normal tissues. In contrast, NSUN6 was significantly downregulated in ESCC tissues as compared to normal tissues (Figure 3A). We then used the GEO database to validate the expression levels of HERPUD1, TECR, MAGEA11, and NSUN6 in ESCC tissues. Our findings indicated that MAGEA11 was significantly upregulated in ESCC tissues as compared to adjacent noncancerous tissues in three GEO datasets: GSE161533, GSE45670, and GSE20347. However, the expression levels of HERPUD1, TECR, and NSUN6 did not differ significantly between ESCC tissues and adjacent noncancerous tissues in the three GEO datasets (Figure 3B).

Figure 3 Verification HERPUD1, TECR, MAGEA11, and NSUN6 expression with the GEO and the TCGA databases in patients with ESCC. (A) HERPUD1, TECR, MAGEA11, and NSUN6 expression levels were evaluated with TCGA database. (B) GEO database was used to validate the expression levels of HERPUD1, TECR, MAGEA11, and NSUN6 in ESCC tissues with three GEO datasets: GSE161533, GSE45670, and GSE20347. ***, P<0.001. TPM, transcripts per million; HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; GEO, Gene Expression Omnibus; TCGA, The Cancer Genome Atlas; ESCC, esophageal squamous cell carcinoma.

The association of HERPUD1, TECR, MAGEA11, and NSUN6 with OS, DSS, and PFS in patients with ESCC based on the TCGA dataset

Patients with ESCC and high HERPUD1 expression had a worse prognosis than did those with low expression, as indicated by shorter OS (HR =3.27, 95% CI: 1.35–7.94; P=0.009) and DSS (HR =6.05, 95% CI: 1.73–21.10; P=0.005). Similarly, patients with ESCC with high TECR expression experienced shorter OS (HR =2.54, 95% CI: 1.11–5.79; P=0.02) as compared to those with low expression. TECR expression did not correlate with DSS (HR =2.33, 95% CI: 0.89–6.09; P=0.08) or PFS (HR =1.65, 95% CI: 0.86–3.17; P=0.13). In contrast, patients with high MAGEA11 expression had better OS (HR =0.29, 95% CI: 0.12–0.70; P=0.006), DSS (HR =0.27, 95% CI: 0.10–0.78; P=0.01), and PFS (HR =0.45, 95% CI: 0.23–0.87; P=0.01) than did those with low expression. Similarly, patients with high NSUN6 expression showed better OS (HR =0.19, 95% CI: 0.07–0.52; P=0.001), DSS (HR =0.21, 95% CI: 0.07–0.65; P=0.006), and PFS (HR =0.43, 95% CI: 0.22–0.83; P=0.01) than did those with low expression (Figure 4).

Figure 4 The association of HERPUD1, TECR, MAGEA11, and NSUN6 with OS, DSS, and PFS in patients with ESCC. The correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with OS, DSS, and PFS in patients with ESCC was evaluated using Kaplan-Meier survival analysis. HR, hazard ratio; HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; OS, overall survival; DSS, disease-specific survival; PFS, progression-free survival; ESCC, esophageal squamous cell carcinoma; CI, confidence interval.

The establishment of a prognostic nomogram model with HERPUD1, TECR, MAGEA11, and NSUN6 in patients with ESCC

First, we used time-dependent AUCs to evaluate the prognostic values of HERPUD1, TECR, MAGEA11, and NSUN6 at various time points. Our results showed that all four genes had strong predictive value at different time points. An AUC value greater than 0.5 indicates that the gene is a negative prognostic factor, while an AUC value less than 0.5 suggests that it is a positive prognostic factor. Consequently, our findings suggest that HERPUD1 and TECR may serve as negative prognostic factors for patients with ESCC, whereas MAGEA11 and NSUN6 may act as positive prognostic indicators (Figure 5A). The AUC (95% CI) of combined model for predicting the OS for patients with ESCC was 0.881 (0.821–0.963) at 1-year, 0.860 (0.807–0.941) at 3-year, and 0.944 (0.901–0.989) at 5-year (Table 2). We created a prognostic nomogram based on four genes to predict the OS of patients with ESCC. We applied a weighted risk score derived from these four genes. This allowed us to calculate a total score to estimate survival probabilities at 1, 3, and 5 years (Figure 5B). The prognostic calibration curves showed the differences between the predicted probabilities from our nomogram for the OS of patients with ESCC at 1, 3, and 5 years and the observed probabilities. Our statistical analyses indicated that the prognostic model had a good fit, with a concordance index (C-index) of 0.839 (95% CI: 0.802–0.876), a likelihood ratio test value of 37.69 (P<0.001), and a Wald test result of 22.94 (P<0.001) (Figure 5C).

Figure 5 The establishment of a prognostic nomogram model with HERPUD1, TECR, MAGEA11, and NSUN6 in patients with ESCC. (A) Time-dependent AUCs were used to evaluate the prognostic values of HERPUD1, TECR, MAGEA11, and NSUN6 at various time points. HERPUD1, TECR, MAGEA11, and NSUN6 were used to establish the (B) nomogram model and (C) calibration curves for patients with ESCC. HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; AUC, area under the curve; ESCC, esophageal squamous cell carcinoma.

Risk score model of HERPUD1, TECR, MAGEA11, and NSUN6 for patients with ESCC

We used risk factor graphs to visually display the trends of prognostic factors and risk scores. We found that higher risk scores were associated with shorter survival times (Figure 6A). Subsequently, we quantitatively illustrated this relationship using the Kaplan-Meier plotter, and the results indicated that high risk scores were significantly correlated with shorter survival times (HR =5.77, 95% CI: 2.13–15.58; P<0.001) (Figure 6B).

Figure 6 Risk scoring model of HERPUD1, TECR, MAGEA11, and NSUN6 in patients with ESCC. (A) Risk factor graph of the trends of prognostic factors and risk scores. (B) The relationship of risk score with OS was analyzed using a Kaplan-Meier plotter. “0” represents survivor; “1” represents death. HR, hazard ratio; CI, confidence interval; HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; ESCC, esophageal squamous cell carcinoma; OS, overall survival.

The diagnostic efficacy of HERPUD1, TECR, MAGEA11, and NSUN6 in patients with ESCC

To evaluate the diagnostic performance of HERPUD1, TECR, MAGEA11, and NSUN6, we collected gene expression data from the TCGA and the GTEx datasets. Specifically, the GTEx datasets provided gene expression levels from normal tissues, while the TCGA datasets provided gene expression levels from tumor tissues. The biomarkers HERPUD1 (Figure 7A and Table 3), TECR (Figure 7B and Table 3), MAGEA11 (Figure 7C and Table 3), and NSUN6 (Figure 7D and Table 3) were evaluated in terms of their ability to diagnosis ESCC, and their AUC values were 0.778 (95% CI: 0.734–0.821), 0.673 (95% CI: 0.638–0.708), 0.665 (95% CI: 0.618–0.712), and 0.584 (95% CI: 0.537–0.632), respectively. Although our results showed that individual biomarkers did not improve the diagnosis of ESCC, the combined analysis of all four biomarkers significantly improved diagnostic performance, achieving an AUC of 0.911 (95% CI: 0.888–0.935) (Figure 7E and Table 3).

Figure 7 The diagnostic efficacy of HERPUD1, TECR, MAGEA11, and NSUN6 in patients with ESCC. The ability of biomarkers (A) HERPUD1, (B) TECR, (C) MAGEA11, (D) NSUN6, and (E) a combined model in diagnosing ESCC as the ROC AUC. HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; TPR, true positive rate; FPR, false positive rate; AUC, area under the curve; CI, confidence interval; ESCC, esophageal squamous cell carcinoma; ROC, receiver operator characteristic.

Correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with immune infiltration in patients with ESCC

To comprehensively determine the correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with immune infiltration, we used the ssGSEA algorithm to calculate their respective relationships with the enrichment scores of 24 types of immune cells (Figure 8A). Our results indicated that HERPUD1 was significantly positively correlated with various immune cells, while TECR and NSUN6 showed significant negative correlations with multiple immune cell enrichment scores. MAGEA11 appeared to have a weak correlation with immune cell enrichment scores. Subsequently, the correlation between HERPUD1 and TECR with immune cell enrichment scores was validated at the gene level. HERPUD1 demonstrated a significant positive correlation with key immunomodulators, including immunoinhibitors (Figure 8B), major histocompatibility complex (MHC) molecules (Figure 8C), and immunostimulators (Figure 8D), whereas TECR was significantly negatively correlated with major immunomodulators. These results suggest that different prognostic factors may exhibit completely opposite correlations with immune infiltration.

Figure 8 Correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with immune infiltration in patients with ESCC. (A) The ssGSEA algorithm was used to calculate the relationship of HERPUD1, TECR, MAGEA11, and NSUN6 with the enrichment scores of 24 types of immune cells. The correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with (B) immunoinhibitors, (C) MHC molecules, and (D) immunostimulators was measured with Spearman correlation analysis. *, P<0.05. HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; MAGEA11, melanoma antigen gene A11; NSUN6, NOP2/Sun RNA methyltransferase 6; MHC, major histocompatibility complex; ESCC, esophageal squamous cell carcinoma; ssGSEA, single-sample gene set enrichment analysis; Cor, correlation.

GO and KEGG analysis of HERPUD1 and TECR-related immune infiltration pathway enrichment in patients with ESCC

Venn diagram represents the gene intersection (n=25) of immune infiltration-related genes and HERPUD1-related DEGs (Figure 9A). Subsequently, GO and KEGG analyses were employed to investigate the immune infiltration-associated signaling pathways mediated by HERPUD1 (Figure 9B). HERPUD1 may regulate immune infiltration through humoral immune response, mononuclear cell differentiation, and B cell activation. Venn diagram represents the gene intersection (n=48) of immune infiltration-related genes and TECR-related DEGs (Figure 9C). On the other hand, TECR potentially influences immune infiltration via humoral immune response, positive regulation of T cell activation, and negative regulation of cytokine production (Figure 9D). These findings suggest that HERPUD1 and TECR may exhibit divergent signaling mechanisms in modulating immune infiltration, despite sharing some overlapping biological processes.

Figure 9 GO and KEGG analysis of HERPUD1 and TECR-related immune infiltration pathway enrichment in patients with ESCC. (A) Venn diagram represents the gene intersection of immune infiltration-related genes and HERPUD1-related DEGs; (B) GO and KEGG analysis of the gene intersection of immune infiltration-related genes and HERPUD1-related DEGs; (C) Venn diagram represents the gene intersection of immune infiltration-related genes and TECR-related DEGs; (D) GO and KEGG analysis of the gene intersection of immune infiltration-related genes and TECR-related DEGs. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; HERPUD1, homocysteine-inducible ER protein with ubiquitin-like domain 1; TECR, trans-2,3-enoyl-CoA reductase; DEGs, differentially expressed genes; BP, biological process; CC, cellular component; MF, molecular function; adj, adjusted.

Discussion

In this study, we constructed a prognostic model using methylation-related regulators HERPUD1, TECR, MAGEA11, and NSUN6 to evaluate the prognosis of patients with ESCC. A higher prognostic risk score was associated with shorter OS in patients with ESCC (HR =5.77, 95% CI: 2.13–15.58; P<0.001). Time-dependent AUC analysis revealed that HERPUD1, TECR, MAGEA11, and NSUN6 had high prognostic predictive value at different time points. Furthermore, our analysis revealed that combined diagnostic model based on HERPUD1, TECR, MAGEA11, and NSUN6 demonstrated excellent diagnostic efficacy for ESCC (AUC =0.911; 95% CI: 0.888–0.935). Finally, through immune enrichment analysis, we discovered that HERPUD1 was significantly positively correlated with immune infiltration at both the cellular and genetic levels, while TECR showed a significant negative correlation with immune infiltration levels. Overall, we found that the prognostic model constructed with the methylation-related regulators HERPUD1, TECR, MAGEA11, and NSUN6 could effectively predict the prognosis of patients with ESCC, provide good diagnostic efficacy for ESCC, and reflect the immune cell infiltration of the immune microenvironment in patients with ESCC. These genes may serve as diagnostic and prognostic biomarkers as well as therapeutic targets for the diagnosis, treatment, and prognosis of ESCC.

Beyond the prognostic and diagnostic models we developed, our analysis of the TCGA and three GEO databases revealed that MAGEA11 was consistently upregulated in patients with ESCC across all datasets. Furthermore, MAGEA11 was identified as a protective factor for OS, DSS, and PFS in patients with ESCC. It was also found that MAGEA11 had the best predictive efficacy for the 5-year OS of patients with ESCC.

The MAGE family is a well-known tumor-associated antigen composed of more than 60 genes. MAGEA11 is a type I MAGE and is rarely expressed in normal adult tissues but is highly expressed in various cancers, including ESCC (18). In cancer cells, DNA methylation plays a primary role to mediate MAGEA11 gene silencing (18). Xiao et al. (19) reported that the upregulation of MAGEA11 in gastric cancer is significantly correlated with shorter survival and metastasis by immune infiltration. Mohsenzadegan et al. (20) found that a significant and positive correlation was found between both nuclear and cytoplasmic expressions of the MAGEA6 protein as well as expression of cytoplasmic MAGEA11 protein with histological grade, lamina propria invasion, and muscularis involvement in patients with bladder cancer; this suggested that the coexpression of cancer-testis antigens of MAGEA6 and MAGEA11 is associated with tumor aggressiveness. Kalvapudi et al. (21) indicated that the MAGE gene family, including MAGEA1, MAGEA3, MAGEA4, MAGEA1, and MAGEA12, exhibited the highest expression levels in gastroesophageal adenocarcinoma (GEAC) and may be associated with poor prognosis in GEAC, offering potential therapeutic targets for the treatment of GEAC. Given these aforementioned findings, MAGEA11 may be a molecule associated with poor prognosis in bladder cancer and gastric cancer, which is completely contrary to our results. However, we also found that the expression of the MAGEA11 gene is significantly upregulated in patients with ESCC. Furthermore, we did not find a significant association between MAGEA11 and immune cell infiltration or immunomodulators. Therefore, we can infer that the prognostic significance of MAGEA11 and its regulatory mechanisms may vary across different tumors.

Recent studies have shown that HERPUD1 maintains mitochondrial function and energy metabolism, along with autophagy, while inhibiting apoptosis and ferroptosis to sustain the malignant characteristics of cancer cells (22,23). However, there are no relevant reports on the role and clinical significance of HERPUD1 in regulating the tumor biology of ESCC. Based on bioinformatics results, we found that the high expression of HERPUD1 is associated with poor prognosis in patients with ESCC and is related to the enrichment of various immune cells, including B cells, eosinophils, natural killer cells, and T effector memory (TEM) cells. The correlation of HERPUD1 with immune cell infiltration and its regulatory mechanisms have not been validated by in vivo and in vitro experiments. Our findings also indicated that compared to the other three prognostic factors, HERPUD1 (AUC =0.778; 95% CI: 0.734–0.821) had the strongest diagnostic efficacy for ESCC. Meanwhile, HERPUD1 had a high predictive value for the 5-year OS of patients with ESCC.

In esophageal cancer, TECR has been identified as a differentially methylated gene, which may be associated with cancer stem-like cells (24). In our study, we found that the high expression of TECR was associated with shorter OS in patients with ESCC, despite its limited utility as a diagnostic marker. Our results also revealed that TECR was negatively correlated with the infiltration of various immune cells and immunomodulators. Notably, it was significantly negatively correlated with the enrichment scores of macrophages and TEM cells.

HERPUD1 and TECR exhibit divergent correlations with immune infiltration in our analyses. However, to date, no experimental evidence has been reported in the literature to elucidate the biological mechanisms underlying their association with immune infiltration. The observed correlations are currently supported only by bioinformatics algorithms (25,26), which suggest potential, but unproven, relationships. It is important to note that genes directly linked to immune infiltration typically include immunoinhibitors, MHC molecules, and immunostimulators. While other genes (e.g., HERPUD1 and TECR) may show associations with immune infiltration, these relationships could be indirect. In our study, we present these findings solely to hypothesize that the prognostic value of HERPUD1 and TECR might involve interactions with immune-related molecules (e.g., via immunoinhibitory/stimulatory pathways or antigen presentation) to influence ESCC outcomes.

NSUN6 exhibits strong substrate specificity for messenger RNA (mRNA), primarily targeting the 3'UTR region of the common sequence motif CTCCA located within the stem-loop structure (27). The 5-methylcytosine (m5C) modification mediated by NSUN6 can enhance mRNA abundance and translation efficiency and promote the methylation of cytosine at position 72 (C72) in the 3' end receptor of tRNACys and tRNAThr, with target recognition dependent on the 3'-CCA tail (27). NSUN6 has also been shown to inactivate macrophage stimulating 1 (MST1) through m5C modification and to activate the target gene of yes-associated protein (YAP) in breast cancer, thereby triggering osteoclast differentiation and bone metastasis (28). NSUN6 mediates m5C modification of multiple genes to deteriorate the progression of cervical cancer, lung cancer, and colon adenocarcinoma (29-31). Fang et al. (32) found that NSUN6 expression is significantly increased in colorectal cancer tissues and closely related to B cells and CD8⁺ T cells. In our study, NSUN6 was significantly downregulated in ESCC tissues, and its high expression was strongly associated with a favorable prognosis. However, its diagnostic value remained relatively limited. Moreover, NSUN6 showed only a weak correlation with a few immune cells, and its correlation with immune cells was notably weaker than that of HERPUD1 and TECR in patients with ESCC.

The prognostic model based on HERPUD1, TECR, MAGEA11, and NSUN6 demonstrates high predictive performance, with AUC values no lower than 0.860 at various time points. For comparison, Cao et al. (9) constructed a combined cell death index (CCDI) model, which showed the highest reported accuracy in their study (AUC =0.834). Notably, our model achieves a maximum AUC of 0.944, markedly surpassing the predictive capability of the CCDI model. Based on single-cell RNA sequencing (scRNA-seq) and bulk RNA analysis of dendritic cells, Shi et al. (33) found a prognostic model that demonstrated good predictive performance with AUC values of 0.81, 0.76, and 0.74 for 1-, 3-, and 5-year survival predictions, respectively. Using the anoikis-related genes, Cao et al. (34) established a prognostic risk model for ESCC with AUC for 1- and 3-year OS of 0.684 and 0.627 respectively. Overall, our ESCC prognostic model, based on methylation-associated regulators, exhibits obviously enhanced predictive performance compared to existing models.

Although our study constructed a prognostic and diagnostic model based on HERPUD1, TECR, MAGEA11, and NSUN6 for the diagnosis and prognostic prediction of ESCC, some limitations to the study should be noted. Firstly, the expression of HERPUD1, TECR, and NSUN6 varies across different datasets, and we were able to account for this due to the different sample sizes of different datasets. Perhaps the expression of these genes needs to be validated in a study with a larger sample size. Secondly, there were differences in the correlation of HERPUD1, TECR, MAGEA11, and NSUN6 with immune cells. We only predicted this relationship through bioinformatics without validating it in in vivo experiments. Thirdly, our study did not collect external clinical samples to validate the model. We plan to address this gap in future research by expanding the validation cohort. Specifically, we will collect peripheral blood samples and tumor tissue samples from additional ESCC patients to rigorously assess the robustness and generalizability of the model. Fourthly, the prognostic model comprising HERPUD1, TECR, MAGEA11, and NSUN6 demonstrates significant predictive value in ESCC. However, translating theoretical findings into clinical practice remains challenging. Currently, evaluating these genes requires obtaining tumor tissue samples from ESCC patients, a process that is invasive, resource-intensive, and burdensome for patients, which limits its clinical feasibility. To address this, our future work will focus on non-invasive validation by investigating the consistency of HERPUD1, TECR, MAGEA11, and NSUN6 expression between peripheral blood samples and tumor tissues, as well as validating the prognostic model’s stability across these biospecimens. Developing an online platform to calculate prognostic probability using the following formula: Risk Score = Intercept + α × [HERPUD1 (low =0 or high =1)] + β × [TECR (low =0 or high =1)] + λ × [MAGEA11 (low =0 or high =1)] + δ × [NSUN6 (low =0 or high =1)]; parameters: intercept =0; α=1.12; β=1.19; λ=−2.50; δ=−1.90.

Conclusions

We developed a robust prognostic model incorporating HERPUD1, TECR, MAGEA11, and NSUN6, revealing that a higher risk score was significantly associated with worse OS. We have also shown that the model combining HERPUD1, TECR, MAGEA11, and NSUN6 had high diagnostic efficacy for ESCC. Overall, the diagnostic and prognostic model based on HERPUD1, TECR, MAGEA11, and NSUN6 may serve as a biomarker for improving the clinical outcomes of patients with ESCC. Our findings need to be validated in larger studies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-341/rc

Peer Review File: Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-341/prf

Funding: This work was supported by the Shaanxi Provincial Health Research Fund Project of China (No. 2022C007).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-341/coif). S.M. reports grants from Ipsen biopharmaceuticals; consulting fees from Merck, Eisai, BeiGene, BMS; payment or honoraria from Integrity CE, LLC, and Academy for Continued Healthcare Learning; outside the submitted work; and serves as board member of Esophageal Cancer Action Network and panel member of NCCN. A.S. reports consulting or advisory board role with AstraZeneca, Bristol-Myers Squibb, Merck, Exelixis, Pfizer, Xilio therapeutics, Taiho, Amgen, Autem therapeutics, KAHR medical, Arcus therapeutics, Regeneron, Replimune and Daiichi Sankyo; institutional research funding from AstraZeneca, Bristol-Myers Squibb, Merck, Clovis, Exelixis, Actuate therapeutics, Incyte Corporation, Daiichi Sankyo, Five prime therapeutics, Amgen, Innovent biologics, Dragonfly therapeutics, Oxford Biotherapeutics, Replimune, Phanes therapeutics, Arcus therapeutics, Regeneron and KAHR medical, outside the submitted work. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022. J Natl Cancer Cent 2024;4:47-53. [Crossref] [PubMed]
3. Rogers JE, Sewastjanow-Silva M, Waters RE, et al. Esophageal cancer: emerging therapeutics. Expert Opin Ther Targets 2022;26:107-17. [Crossref] [PubMed]
4. Feng Y, Wu N. Comments on National guidelines for diagnosis and treatment of esophageal carcinoma 2022 in China (English version). Chin J Cancer Res 2022;34:633-4. [Crossref] [PubMed]
5. Gao H, Li L, Zhang C, et al. Systematic Review with Meta-analysis: Association of Helicobacter pylori Infection with Esophageal Cancer. Gastroenterol Res Pract 2019;2019:1953497. [Crossref] [PubMed]
6. Wang SM, Freedman ND, Katki HA, et al. Gastroesophageal reflux disease: A risk factor for laryngeal squamous cell carcinoma and esophageal squamous cell carcinoma in the NIH-AARP Diet and Health Study cohort. Cancer 2021;127:1871-9. [Crossref] [PubMed]
7. Simba H, Kuivaniemi H, Abnet CC, et al. Environmental and life-style risk factors for esophageal squamous cell carcinoma in Africa: a systematic review and meta-analysis. BMC Public Health 2023;23:1782. [Crossref] [PubMed]
8. Niu C, Liu Y, Wang J, et al. Risk factors for esophageal squamous cell carcinoma and its histological precursor lesions in China: a multicenter cross-sectional study. BMC Cancer 2021;21:1034. [Crossref] [PubMed]
9. Cao K, Zhu J, Lu M, et al. Analysis of multiple programmed cell death-related prognostic genes and functional validations of necroptosis-associated genes in oesophageal squamous cell carcinoma. EBioMedicine 2024;99:104920. [Crossref] [PubMed]
10. Li C, Song W, Zhang J, et al. Single-cell transcriptomics reveals heterogeneity in esophageal squamous epithelial cells and constructs models for predicting patient prognosis and immunotherapy. Front Immunol 2023;14:1322147. [Crossref] [PubMed]
11. Guo D, Sheng K, Zhang Q, et al. Single-cell transcriptomic analysis reveals the landscape of epithelial-mesenchymal transition molecular heterogeneity in esophageal squamous cell carcinoma. Cancer Lett 2024;587:216723. [Crossref] [PubMed]
12. Wang X, Peng W, Zhao Y, et al. Immune cell related signature predicts prognosis in esophageal squamous cell carcinoma based on single-cell and bulk-RNA sequencing. Front Oncol 2024;14:1370801. [Crossref] [PubMed]
13. Wei QY, Jin F, Wang ZY, et al. MicroRNAs: A novel signature in the metastasis of esophageal squamous cell carcinoma. World J Gastroenterol 2024;30:1497-523. [Crossref] [PubMed]
14. Bian Y, Gao Y, Lu C, et al. Genome-wide methylation profiling identified methylated KCNA3 and OTOP2 as promising diagnostic markers for esophageal squamous cell carcinoma. Chin Med J (Engl) 2024;137:1724-35. [Crossref] [PubMed]
15. Fan Z, Hao J, He F, et al. Novel DNA methylation markers for early detection of gastric cardia adenocarcinoma and esophageal squamous cell carcinoma. Sci China Life Sci 2024;67:2701-12. [Crossref] [PubMed]
16. Zou X, Lin Y, Ren H, et al. PRMT5-PAK1 Signaling Participates in Metastasis and Is Associated With Poor Prognosis in Human Esophageal Carcinoma. Anticancer Res 2024;44:593-604. [Crossref] [PubMed]
17. Xi Y, Lin Y, Guo W, et al. Multi-omic characterization of genome-wide abnormal DNA methylation reveals diagnostic and prognostic markers for esophageal squamous-cell carcinoma. Signal Transduct Target Ther 2022;7:53. [Crossref] [PubMed]
18. Lian Y, Meng L, Ding P, et al. Epigenetic regulation of MAGE family in human cancer progression-DNA methylation, histone modification, and non-coding RNAs. Clin Epigenetics 2018;10:115. [Crossref] [PubMed]
19. Xiao C, Yang L, Jin L, et al. MAGEA11 as a STAD Prognostic Biomarker Associated with Immune Infiltration. Diagnostics (Basel) 2022;12:2506. [Crossref] [PubMed]
20. Mohsenzadegan M, Razmi M, Vafaei S, et al. Co-expression of cancer-testis antigens of MAGE-A6 and MAGE-A11 is associated with tumor aggressiveness in patients with bladder cancer. Sci Rep 2022;12:599. [Crossref] [PubMed]
21. Kalvapudi S, Pachimatla AG, Seager RJ, et al. Cancer/testis antigen expression and co-expression patterns in gastroesophageal adenocarcinoma. Med Oncol 2024;41:227. [Crossref] [PubMed]
22. Paredes F, Navarro-Marquez M, Quiroga C, et al. HERPUD1 governs tumor cell mitochondrial function via inositol 1,4,5-trisphosphate receptor-mediated calcium signaling. Free Radic Biol Med 2024;211:24-34. [Crossref] [PubMed]
23. Peng Y, Li N, Tang F, et al. Corosolic acid sensitizes ferroptosis by upregulating HERPUD1 in liver cancer cells. Cell Death Discov 2022;8:376. [Crossref] [PubMed]
24. Yu X, Teng Y, Jiang X, et al. Genome-Wide DNA Methylation Pattern of Cancer Stem Cells in Esophageal Cancer. Technol Cancer Res Treat 2020;19:1533033820983793. [Crossref] [PubMed]
25. Wen J, Huang Q, Yao J, et al. Focusing on scRNA-seq-Derived T Cell-Associated Genes to Identify Prognostic Signature and Immune Microenvironment Status in Low-Grade Glioma. Mediators Inflamm 2023;2023:3648946. [Crossref] [PubMed]
26. Li D, Chen T, Li QG. Identification of a m(6)A-related ferroptosis signature as a potential predictive biomarker for lung adenocarcinoma. BMC Pulm Med 2023;23:128. [Crossref] [PubMed]
27. Li M, Tao Z, Zhao Y, et al. 5-methylcytosine RNA methyltransferases and their potential roles in cancer. J Transl Med 2022;20:214. [Crossref] [PubMed]
28. Li C, Wang S, Xing Z, et al. A ROR1-HER3-lncRNA signalling axis modulates the Hippo-YAP pathway to regulate bone metastasis. Nat Cell Biol 2017;19:106-19. [Crossref] [PubMed]
29. Yu M, Ni M, Xu F, et al. NSUN6-mediated 5-methylcytosine modification of NDRG1 mRNA promotes radioresistance in cervical cancer. Mol Cancer 2024;23:139. [Crossref] [PubMed]
30. Lu Z, Liu B, Kong D, et al. NSUN6 Regulates NM23-H1 Expression in an m5C Manner to Affect Epithelial-Mesenchymal Transition in Lung Cancer. Med Princ Pract 2024;33:56-65. [Crossref] [PubMed]
31. Cui Y, Lv P, Zhang C. NSUN6 mediates 5-methylcytosine modification of METTL3 and promotes colon adenocarcinoma progression. J Biochem Mol Toxicol 2024;38:e23749. [Crossref] [PubMed]
32. Fang X, Miao C, Zeng T, et al. Role of m(5) C RNA methylation regulators in colorectal cancer prognosis and immune microenvironment. J Clin Lab Anal 2022;36:e24303. [Crossref] [PubMed]
33. Shi M, Zhang H, Ma L, et al. Innovative prognostic modeling in ESCC: leveraging scRNA-seq and bulk-RNA for dendritic cell heterogeneity analysis. Front Immunol 2024;15:1352454. [Crossref] [PubMed]
34. Cao S, Li M, Cui Z, et al. Establishment and validation of the prognostic risk model based on the anoikis-related genes in esophageal squamous cell carcinoma. Ann Med 2024;56:2418338. [Crossref] [PubMed]

(English Language Editor: J. Gray)