A nine-gene signature with potential targets for predicting the prognosis of patients with esophageal cancer

Introduction

Globally, esophageal cancer ranks as the seventh most prevalent cancer (1). Esophageal squamous cell carcinoma (ESCC) is more common in East Asian countries. It has a poor prognosis, and it is one of the threatening factors to public health (2). The lack of effective early diagnosis biomarkers and reliable therapeutic targets is the main reason for the late hospital visits and high mortality rates among clinical esophageal cancer patients. Consequently, it is critically important to comprehend the molecular workings of ESCC to pinpoint novel targets for prognosis and treatment, and to enhance the prognosis and survival duration for these patients.

To date, many features have been identified as prognostic elements of ESCC, and these include age, gender, tumor stage, smoking, and alcohol consumption (3-5). However, these clinical features do not effectively distinguish between high and low survival rates, and they are not able to identify patients who are prone to benefit from treatment. Recently, with the advancement of microarray and sequencing of dysregulated RNA, long non-coding RNAs (lncRNAs) together with messenger RNAs (mRNAs) have been linked to human cancer. Some of these molecules are crucial in the diagnosis and prognosis (6-10), suggesting that the prognostic model related to RNA-Seq expression profiles can be used as prognostic signatures. Therefore, in the era of genomics, identifying a prognostic and remedial biomarker that can predict the overall survival (OS) of individuals suffering from ESCC is urgently needed. Unfortunately, although a number of studies have constructed prognostic signatures associated with lncRNAs and mRNAs to predict patients’ survival in various cancers (11-15), prognostic signatures for ESCC are rare.

Our findings reveal a nine-gene signature with significant predictive significance for ESCC, potentially augmenting conventional clinical prognostic indicators. The prognostic risk model was constructed with training datasets, and we further identified the reliability of the model through the utilization of test datasets and entire datasets. We showed that this nine-gene marker was a standalone prognostic element with good resistance to chemotherapy changes in high-risk patients in the light of the signature. The quantitative reverse transcription polymerase chain reaction (qRT-PCR), immunohistochemistry (IHC) staining assays and survival analysis verified the differential expression of FABP3 in ESCC and FABP3-associated prognosis. In summary, our work provides an independent and excellent prognostic biomarker for patients with ESCC. Overall, our study pointed out a new set of ESCC-associated signatures, which may contribute to developing an approach to treating patients with ESCC. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-146/rc).

Methods

Patients and specimens

Corresponding clinical data and the RNA Seq data of patients with esophageal carcinoma were obtained from the TCGA. Specifically, our collection contains 170 samples, including 159 tumor tissue specimens and 11 normal ones from the TCGA database (website: https://portal.gdc.cancer.gov/analysis_page?app=CohortBuilder&tab=general; accessed on June 12, 2022). In addition, we collected primary ESCC tissues that were surgically excised from 57 patients and 44 normal tissues adjacent to esophageal cancer at Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School from 2021 to 2024. Every surgical specimen was taken from the specimens that remained after diagnostic sampling. Table S1 details the clinical and pathological characteristics of the 57 ESCC patients. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethics committee of Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School (No. 2023-648-02) and informed consent was taken from all the patients.

Verification of genes with varied expression levels

We used Perl and R scripts for the entirety of data processing. In addition, we analyzed the lncRNA and mRNA expression profiles and screened for the genes with differential expression between ESCC and normal specimens. The identification of dysregulated genes utilized the edgeR package of the R language in the two groups (16-18). The study utilized preprocessed RNA-seq data from TCGA [normalized to fragments per kilobase of transcript per million mapped reads (FPKM) values]. Differential expression analysis was performed using the R package “DESeq2”. Gene identifier conversion was conducted using Perl scripts, and data adjustment was carried out with the R package “limma”.

We designated |log₂fold change (log₂FC)| >1.5 and adjusted P<0.05 as the benchmark for distinguishing their differences.

Identification and confirmation of the prognosis risk pattern

There were three phases in identifying and validating the prognostic gene signature (see the study flowchart in Figure S1A). In this study, the dataset was randomly partitioned into a training set (52.2%, n=83) and a testing set (47.8%, n=76). The univariate Cox proportional-hazard regression method was applied to estimate the relevance of differentially expressed genes (DEGs) and patients’ OS in the training group. Nineteen prospective genes (P<0.01) were designated to participate in the stepwise multivariate Cox regression fitting. The stepwise multivariate Cox regression model was adjusted for the other OS-related genes. Additionally, we recognized the genes that conformed to the model and showed an independent connection with OS. Lastly, on the basis of the median score of risk, the participants were grouped into the high-risk population and the low-risk population for comparison, and the difference in survival time between the two populations was probed by applying the log-rank test.

Correlations between clinical characteristics and the risk score

Patients suffering from ESCC were categorized according to gender, age, tumor grade, tumor-node-metastasis (TNM) stage, radiation, pharmaceutical, alcohol consumption, and risk score. Then, we computed the risk score between the two groups and observed the differences between the factors.

Independent prognostic characteristics of the nine-gene marker

The prognostic marker that might be independent of clinical parameters, such as gender, tumor grade, TNM stage, age, radiation, pharmaceutical, alcohol, and risk score were validated, then the Cox regression model method through the application of the forward stepwise technique was used to analyze univariate and multivariate. The nomogram was constructed based on a Cox regression model incorporating influential factors including age, sex, disease stage, grade, and a nine-gene risk score. The C-index with its 95% confidence interval (CI) was calculated using the bootstrap method. The R package of survival receiver operating characteristic (ROC) was utilized to process all data. Statistical significance was considered to be reached when the P value was <0.05.

qRT-PCR and IHC

Formalin-fixed, paraffin-embedded (FFPE) tissue blocks from 57 ESCC cases and 44 adjacent normal tissues were used to prepare paraffin rolls and construct tissue microarray (TMA) sections for subsequent PCR amplification and IHC analysis. The steps of PCR experiment are as follows. Total RNA was extracted using the RNA extraction kit (Accurate Biology, Guangzhou, China, AG21024) according to the manufacturer’s protocol. Complementary DNA (cDNA) was synthesized from the extracted RNA using the Evo M-MLV Reverse Transcription Premix Kit (Accurate Biology, AG11728). qPCR analysis was performed using the SYBR Green Pro Taq HS qPCR Kit (Accurate Biology, AG11702) on a LightCycler^® 480 detection system (Roche, Basel, Switzerland). The following primers targeting FABP3 were used: FABP3-F: 5'-GGCACCTGGAAGCTAGTGG-3'; FABP3-R: 5'-CTGCCTGGTAGCAAAACCC-3'. The steps of IHC experiment are as follows. The antigen repairing procedure involved Tris-ethylenediaminetetraacetic acid (Tris-EDTA) buffer (pH 9.0) of high-pressure heat recovery for 8 min. Endogenous peroxidase was inhibited by 3% H₂O₂ for 10 min and washed in phosphate-buffered saline for 5 min three times. Using the EnVision technique (DAKO, Glostrup, Denmark), IHC was employed to stain the tissue sections. The tissues were mixed with the primary anti-FABP3 rabbit polyclonal antibody (1:200 dilution, 10676-1-AP; Proteintech Group, Wuhan, China) and incubated for 8 h at 4 ℃. Then, the sections were incubated in the second EnVision antibody at 37 ℃ for 30 min. Finally, 3,3'-diaminobenzcematoxylin were applied to detect proteins. The standardized scoring system for FABP3 immunohistochemical evaluation is outlined as follows: the staining intensity was graded on a scale of 0–3 (0= negative, 1= weak, 2= moderate, 3= strong), while the percentage of positive cells was scored from 0–4 (0=0%, 1=1–25%, 2=26–50%, 3=51–75%, 4=76–100%). The final IHC score (range, 0–12) was calculated by multiplying the intensity score by the proportion score. In cases of discordant evaluations between two pathologists, a third senior pathologist performed the definitive assessment.

Data statistics and bioinformatics analysis

The statistical evaluation was conducted utilizing R software (version 3.6.1) and GraphPad Prism (version 8.2.1). For comparing groups, Wilcoxon, Chi-squared tests were employed, and survival graphs were produced using The Kaplan-Meier plotter (www.kmplot.com) to assess prognosis. The nomogram was constructed based on a Cox regression model incorporating influential factors including age, sex, disease stage, grade, and a nine-gene risk score. The C-index with its 95% CI was calculated using the bootstrap method. The analysis in Bioinformatics employed databases like The Cancer Genome Atlas (TCGA), Gene Set Enrichment Analysis (GSEA), and Kyoto Encyclopedia of Genes and Genomes (KEGG).

Results

Screening for DEGs between ESCC and adjacent tissues

There were 5,634 DEGs (log₂|FC| >1.5 and adjusted P<0.05) between 159 ESCC and paracancerous tissues; among them, 2,521 were overexpressed and 3,113 were underexpressed lncRNAs and mRNAs (Figure S1B).

Identification of prognostic gene signature

Using univariate analysis, we screened 19 candidate genes to identify genetic prognostic markers. The significant prognostic genes were included in multivariate analyses. Among 19 candidate variables, the nine-gene prognosis prediction model (Table 1) was screened out and was determined to have an independent correlation with prognosis. Based on Cox proportional risk regression model, the risk score of risk of each gene was subsequently determined: risk score = (−0.5904 × expressionRP11-51F16.1) + (−0.7608 × expressionGNG12-AS1) + (−0.5340 × expressionPRICKLE2-AS1) + (−0.3496 × expressionAC132825.1) + (0.5264 × expressionFABP3) + (0.7333× expressionRP11-4K3-A.3) + (0.4370 × expressionRP11-2N1.2) + (0.2262 × expressionKB-1254G8.1) + (0.2599 × expressionRP11-294N21.3). According to the model, for the training set, participants with ESCC were categorized into high-risk (n=41) and low-risk (n=42) populations on the basis of the median score of risk. Individuals in the high-risk population had considerably worse OS than those in the low-risk population (P=4.964e−05, log-rank test, Figure 1A). In the high-risk population, the median survival time was 1.104 years, compared with 1.2726 years in the low-risk population.

Figure 1 In the training populations, nine genetic signatures that were noteworthily related to survival in patients with esophageal squamous cell carcinoma were identified. (A) Overall survival of the two groups of patients through the Kaplan-Meier curve analysis. (B) Time-dependent ROC curve analysis. (C) Based on the nine-gene signature, the distribution of the risk score, survival condition, and the expression status of lncRNA in two types of populations. AUC, area under the curve; lncRNA, long non-coding RNA; ROC, receiver operating characteristic.

Additionally, the survival rates were 19.5% and 40.4% in the high-risk and low-risk populations, separately. The ROC curve of time-dependent results indicated that these nine genetic signatures strongly forecast the OS in patients with ESCC [area under the curve (AUC) =0.852; Figure 1B]. Following that, we sequenced the training cohort according to an ascending order of the risk score, and the follow-up period and genetic heat map of the populations were presented based on this standard (Figure 1C). Among these nine genes, the coefficients of RP11-51F16.1, GNG12-AS1, AC132825.1, and PRICKLE2-AS1 were negative, indicating that they may have the function of promoting survival; in the training populations, comparing the high-risk population with the low-risk population, the remaining five genes appear to be hazardous elements and showed overexpression in the high-risk population.

Validation of the nine-gene signature of ESCC

To further confirm the precision of the model, we verified the markers in the test data groups and whole data groups. The survival risk score of every participant was evaluated according to the formula shown above for the test group (n=76). We utilized the optimal cut-off point to divide the dataset into the high-risk set (n=45) and the low-risk set (n=31). The results of the Kaplan–Meier curve suggested that a considerable discrepancy in the prognosis existed between the two sets (P=3.815e−06; log-rank, Figure 2A). The ROC curve results indicated that the nine gene markers reliably forecasted the OS of patients with ESCC (AUC =0.853, Figure 2B). The test group nine-gene expression heatmap sorted on the basis of the risk score is displayed in Figure 2C. In the entire TCGA dataset, the Kaplan-Meier curves suggested that the survival time within the high-risk group (n=86) was significantly different from the time of the low-risk group (n=73, P=7.585e−10, log-rank test, Figure 3A). The AUC was 0.912 (n=159, Figure 3B). The entire TCGA dataset nine-gene expression heatmap sorted on the basis of the risk score is displayed in Figure 3C.

Figure 2 Independent verification of the nine-gene signature in the independent test populations. (A) OS of patients with ESCC of the two populations through the Kaplan-Meier curve analysis. (B) Time-dependent ROC curve analysis. (C) Based on the nine-gene signature, the distribution of the risk score, survival condition, and the expression status of lncRNA in two types of populations. AUC, area under the curve; ESCC, esophageal squamous cell carcinoma; lncRNA, long non-coding RNA; OS, overall survival; ROC, receiver operating characteristic.

Figure 3 Performance evaluation of the nine-gene signature in the entire The Cancer Genome Atlas cohort. (A) OS of patients with ESCC of the two populations through the Kaplan-Meier curve analysis. (B) Time-dependent ROC curve analysis. (C) Based on the nine-gene signature, the distribution of the risk score, survival condition, and the expression status of lncRNA in two types of populations. AUC, area under the curve; ESCC, esophageal squamous cell carcinoma; lncRNA, long non-coding RNA; OS, overall survival; ROC, receiver operating characteristic.

Independent prognostic characteristics of the nine-gene marker

A total of 159 participants with aggregate clinical characteristics, such as gender, tumor grade, radiation, TNM stage, pharmaceutical, alcohol, age, and risk score, were recruited for detailed analysis. The basic clinicopathological characteristics of ESCC training and testing sets originating from TCGA are presented in Table 2. The correlation analysis revealed no correlations regarding the risk score. The relationship ranging from the risk score to additional clinical and pathological aspects and the OS of patients with ESCC is shown in Figure 4. Using the univariate and multivariate analyses, we showed that the risk score of the nine-gene signature was an independent prognostic indicator of OS [hazard ratio (HR) =1.202, 95% CI: 1.084–1.334, P<0.001] (Figure 5).

Figure 4 The relationship between the risk score and other clinicopathological elements of patients with ESCC. ESCC, esophageal squamous cell carcinoma; G, grade.

Figure 5 Clinical characteristics and the risk score of clinical significance in the whole populations. CI, confidence interval; M, metastasis; N, node; T, tumor.

Clinical tissue validation of FABP3

In comparison with normal esophageal epithelium, both FABP3 mRNA and protein were significantly overexpressed in ESCC (Figure 6A,6B). We further analyzed the clinical data of 57 ECSS individuals and found that the patients characterized by high expression (IHC score ≥8) of FABP3 had a shorter OS (P=0.04; Figure 6C). This was further verified through the Kaplan-Meier plotter (P=0.005; Figure 6D). FABP3 staining was detected in the cytoplasm of ESCC tissues (Figure 6E). Moreover, the upregulation of FABP3 in patients with ESCC significantly correlated with grade (P=0.04; Table 3).

Figure 6 The Kaplan-Meier plotter and immunohistochemical analysis of FABP3 expression in clinical samples of esophageal carcinoma. (A) Quantitative reverse transcription polymerase chain reaction results showed that FABP3 was highly expressed in ESCC. (B) Immunohistochemistry results showed that FABP3 was highly expressed in ESCC. (C) Immunohistochemistry scores of patients with ESCC from TMA analyzed using the Kaplan-Meier survival analysis. (D) Relationship between expression of FABP3 and prognosis verified by the Kaplan-Meier plotter. (E) The FABP3 expression in ESCC tissues by immunohistochemistry. Original magnification ×60 and ×200. **, P<0.01; ****, P<0.0001. CI, confidence interval; ESCC, esophageal squamous cell carcinoma; HR, hazard ratio; NT, normal tissue; TMA, tissue microarray.

Construction of a nomogram integrating the nine-gene signature and clinical characteristics

To propose an effective method to predict the recurrence risk and thus promote personalized management of patients with ESCC, we constructed a combined nomogram of our nine-gene signature and clinical characteristics (gender, age, stage, tumor grade, and the score of risk) for predicting 1-, 2-, and 3-year OS (Figure 7). The C-index was 0.8567 (95% CI: 0.8136–0.8998), which also indicated the high predictive accuracy.

Figure 7 A nomogram including clinical features and the risk score. G, grade.

Functional analysis

To demonstrate the potential molecular mechanisms of prognostic expression of this nine-gene signature, we executed Pearson’s correlation analyses among the nine genes and protein-coding genes in the TCGA. The protein-coding genes related to at least one of the nine genes (Pearson’s coefficient >0.4, P<0.001) included 988 correlated genes. As depicted in Figure 8, Gene Ontology (GO) analysis demonstrated that the top 16 GO functional groupings (adjusted P<0.001) were clustered in the metabolic and adenosine triphosphate (ATP)-dependent processes (Figure 8A). The functional analysis of KEGG suggested that the top 19 functional categories of pathways (adjusted P<0.05) were largely gathered in the cell cycle, including dilated cardiomyopathy, arrhythmogenic right ventricular, and renin secretion (Figure 8B).

Figure 8 Functional analysis of this nine-gene signature. (A) The drawing of the nine-gene functional enrichment; (B) KEGG pathways markedly relevant to the related protein-coding genes. (C) High-risk and low-risk sets of markedly enriched gene ontology terms; (D) significantly relevant to high-risk and low-risk sets by KEGG pathways. KEGG, Kyoto Encyclopedia of Genes and Genomes.

Interestingly, GSEA functional enrichment results also suggested that the high-risk sets defined by signatures were significantly enriched in fatty acid metabolism, glycolysis, and ATP metabolic process (P<0.05; Figure 8C,8D). Hence, the impact of these nine genes on patients with ESCC, with poor prognosis, may be mediated through the above pathways and biological processes.

The signature of therapeutic responses in patients with ESCC

To explore whether this signature was of therapeutic worth, we used the resistance to chemotherapy changes. We found that 25 kinds of chemotherapy and targeted medicines have considerable differences by evaluating the half maximal inhibitory concentration (IC₅₀) between the high-risk and the low-risk populations, and the high-risk populations have higher IC₅₀ values (P<0.05; Figure 9). Hence, these markers may play a crucial role that the performance substantiated in the prognosis and treatment of tumor.

Figure 9 The signature of therapeutic responses in patients with ESCC. ESCC, esophageal squamous cell carcinoma; IC₅₀, half-maximal inhibitory concentration.

Discussion

ESCC is a malignancy among others that threatens human health (19-21). Up to this point, there has been no solid evidence that biomarkers can predict the survival of patients with ESCC. In addition, it is significantly challenging to predict the prognosis of this disease using clinical parameters, and the effect is unsatisfactory. Therefore, finding prognostic biomarkers is an urgent need. Many prognostic studies on single molecular markers have been performed over the past few decades; however, the results varied significantly between different datasets (22-24). Multiple genes and their potential for use as clinical markers to evaluate patients’ prognosis are receiving much attention. For instance, Lv et al. established a prognostic model containing seven endocrine-related genes in acute myeloid leukemia (AML), which can effectively improve the accuracy of predicting AML survival (25). He et al. established a prognostic model containing twelve genes in bladder cancer, which can effectively predict its occurrence (26).

In our work, we successfully constructed a nine-gene signature to prognose the viability of patients with ESCC in the training populations and confirmed it on the testing populations and entire populations as well. Additionally, when contrasted with other clinical indicators, the nine-gene-based risk score had a higher weight and a well-performed prognostic prediction based on various methods available. We also verified the relationship between FABP3 and clinicopathological data and prognosis in our clinical samples. More importantly, the resistance to chemotherapy changes indicated that 25 kinds of chemotherapy and targeted medicines have considerable differences through evaluating IC₅₀ between the high-risk and the low-risk populations based on the patterns, confirming that the signature has an efficient role in the diagnosis and therapy for ESCC.

Analysis of the correlations between the level of gene expression and viability showed that five of the nine genes negatively correlated with survival time. To put it another way, genes with elevated-level expression indicated comparatively unfavorable prognosis, while the other four genes exerted a beneficial influence on patient survival. Within the five risk genes, FABP3, referred to as fatty acid binding protein 3, is located on human chromosome 1p35.2. On the one hand, it has been reported to be upregulated and serves as an adverse prognostic factor in a wide variety of diseases (27-31). More importantly, by constructing a multigene prognostic risk scoring system, Wang et al. (32) and Dai et al. (33) also validated FABP3 as a risk indicator in esophageal cancer. Our IHC results demonstrated that FABP3 in 57 ESCC tissues was upregulated and was closely correlated with poor prognosis, acting as an indicator of risk. This fact suggested that FABP3 could function as an oncogene in ESCC. Therefore, we recommend FABP3 as an autonomous prognostic factor in individuals with ESCC. Among the remaining protective genes, the study demonstrated that the GNG12-AS1 is downregulated in breast cancer, along with the DIRAS3 (34). Thus, GNG12-AS1 may extend the tumor suppressor function. This result is consistent with our findings, suggesting that GNG12-AS1 is likely to be a protective marker. Additionally, apart from those discussed above, other genes have not been reported so far. However, their high or low expression within ESCC had a negative or positive influence on the survival of patients.

Functionally, it is apparent that there is a need for further investigation. As far as we know, the functions of these nine genes have rarely been reported (35-38). The functional analysis results indicated that the protein-coding genes, which are associated with these nine genes, are mainly involved in metabolic and ATP-dependent processes. KEGG functional analysis revealed that these genes are mostly related to dilated cardiomyopathy, arrhythmogenic right ventricular, and renin secretion. Interestingly, GSEA analysis showed that metabolism-related functions, such as fatty acid metabolism, glycolysis, and ATP metabolic process, were also mainly clustered in the high-risk group, suggesting that the involvement of the nine genes in patients with ESCC, with poor prognosis, may be mediated via these pathways; nonetheless, additional research is required to examine and confirm their roles.

However, our study had some limitations. First, limited data were available for performance evaluation, and more datasets on the robustness of this signature need to be collected in the future. Second, we did not use assays to detect the mechanism behind the nine-gene signature. To sufficiently understand their biological functions in ESCC, it is necessary to carry out experimental research on these genes. Ultimately, given that our research was retrospective, future clinical trials are crucial for assessing the model’s effectiveness and applicability in forecasting prognosis.

Conclusions

We screened nine prognosis-related genes by using the data from the TCGA database and the method of multivariate analysis, and then constructed a prediction model. This model performed well in predicting high-risk patients with ESCC. Further, through qRT-PCR and IHC experiments, it was verified that the high expression of FABP3 in ESCC was closely related to the poor prognosis of ESCC, to confirm that the nine prognosis-related genes prediction model has high accuracy in the diagnosis of ESCC.

Acknowledgments

We are grateful to the patients who participated in this study for their support.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-146/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-146/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-146/prf

Funding: This study was supported by grants from the National Natural Science Foundation of China (No. U1903305 & 82160542), Key areas of the XPCC science and technology project (No. 2020AB022), International Science and Technology Cooperation Project of Science and Technology (No. 2022BC006), and Achievement transformation and Application promotion project of Shihezi University (No. CGZH202006).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-146/coif). The 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 and its subsequent amendments. The study was approved by the ethics committee of Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School (No. 2023-648-02) and informed consent was taken from all the patients.

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. Jubashi A, Kotani D, Kojima T, et al. Current landscape of targeted therapy in esophageal squamous cell carcinoma. Curr Probl Cancer 2024;53:101152. [Crossref] [PubMed]
2. Shen S, Liu B, Guan W, et al. Advancing precision medicine in esophageal squamous cell carcinoma using patient-derived organoids. J Transl Med 2024;22:1168. [Crossref] [PubMed]
3. Cao J, Zhang Q, Xuan Y, et al. The expression and prognostic value of IFIT3 in esophageal squamous cell carcinoma. Transl Cancer Res 2024;13:6219-34. [Crossref] [PubMed]
4. Ren ZT, Kang M, Zhu LY, et al. Long-term survival and risk factors in esophageal squamous cell carcinoma: A Kaplan-Meier and cox regression study. World J Gastrointest Surg 2024;16:3772-9. [Crossref] [PubMed]
5. Minami R, Noma E, Moriguchi Y, et al. Differences in the Microvascular Arrangement Lead to Improved Clinical Diagnostics of Esophageal Neoplasms: A Single-Center Retrospective Study. Diagnostics (Basel) 2024;14:2852. [Crossref] [PubMed]
6. Qiu Y, Tian Z, Miao TY, et al. The METTL3-m(6)A-YTHDC1-AMIGO2 axis contributes to cell proliferation and migration in esophageal squamous cell carcinoma. Gene 2024;908:148281. [Crossref] [PubMed]
7. Zhou Q, Ye W, Yu X, et al. A pathway-based computational framework for identification of a new modal of multi-omics biomarkers and its application in esophageal cancer. Comput Methods Programs Biomed 2024;247:108077. [Crossref] [PubMed]
8. Wang HF, Zhou XF, Zhang QM, et al. Involvement of circRNA Regulators MBNL1 and QKI in the Progression of Esophageal Squamous Cell Carcinoma. Cancer Control 2024;31:10732748241257142. [Crossref] [PubMed]
9. Yang F, Dan M, Shi J, et al. Efficacy and safety of PD-1 inhibitors as second-line treatment for advanced squamous esophageal cancer: a systematic review and network meta-analysis with a focus on PD-L1 expression levels. Front Immunol 2024;15:1510145. [Crossref] [PubMed]
10. Chen Z, Wang Y, Chen J, et al. Identification of biomarkers for tumor regression grade in esophageal squamous cell carcinoma patients after neoadjuvant chemoradiotherapy. Front Oncol 2024;14:1426592. [Crossref] [PubMed]
11. Shah M, Sarkar D. HCC-Related lncRNAs: Roles and Mechanisms. Int J Mol Sci 2024;25:597. [Crossref] [PubMed]
12. Zhang RN, Fan JG. Lipid metabolism-related long noncoding RNAs: A potential prognostic biomarker for hepatocellular carcinoma. World J Gastroenterol 2024;30:3799-802. [Crossref] [PubMed]
13. Zhang M, Zhang C, Zhou F, et al. LINC02154 Promotes Esophageal Squamous Cell Carcinoma Progression via the PI3K-AKT-mTOR Signaling Pathway by Interacting With IGF2BP2. Mol Carcinog 2025;64:985-96. [Crossref] [PubMed]
14. Xiao H, Zhou T, Yang Y, et al. LncRNA-DANCR Promotes ESCC Progression and Function as ceRNA to Regulate DDIT3 Expression by Sponging microRNA-3193. Cancer Sci 2025;116:1324-38. [Crossref] [PubMed]
15. Ma Y, Qian L, Wang D, et al. LncRNA XIST Promotes Proliferation, Migration and Invasion of Esophageal Squamous Cell Carcinoma Cells via Regulation of miR-186-5p/ZEB1. Anticancer Res 2025;45:897-908. [Crossref] [PubMed]
16. Wang H, Wang J, Chen Y, et al. Global research progress and trends in traditional Chinese medicine for chronic kidney disease since the 21st century: a bibliometric analysis. Front Med (Lausanne) 2024;11:1480832. [Crossref] [PubMed]
17. Xie Y, Zhai Y, Lu G. Evolution of artificial intelligence in healthcare: a 30-year bibliometric study. Front Med (Lausanne) 2024;11:1505692. [Crossref] [PubMed]
18. Kang J, Jiang N, Shataer M, et al. Research progress of breast cancer surgery during 2010-2024: a bibliometric analysis. Front Oncol 2024;14:1508568. [Crossref] [PubMed]
19. Mou X, Peng Z, Yin T, et al. Non-endoscopic Screening for Esophageal Squamous Cell Carcinoma: Recent Advances. J Gastrointest Cancer 2024;55:118-28. [Crossref] [PubMed]
20. Xu SJ, Luo YF, Huang J, et al. Prognostic value of immunosuppression scores in patients with esophageal squamous cell carcinoma: a multicenter study. Front Immunol 2024;15:1517968. [Crossref] [PubMed]
21. Yang F, Xiao H, Dai X, et al. Impact of APOBEC3s on the occurrence, development and prognosis of esophageal squamous cell carcinoma. Future Oncol 2025;21:117-25. [Crossref] [PubMed]
22. Yang H, Chen Y, Huang X, et al. Bioinformatics Analysis Reveals a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma. Int J Med Sci 2024;21:1213-26. [Crossref] [PubMed]
23. Chen K, Lin Z, Shen Y, et al. A novel amino acid metabolism-related gene signature to predict the overall survival of esophageal squamous cell carcinoma patients. J Thorac Dis 2024;16:3967-89. [Crossref] [PubMed]
24. Lin F, Zhu LX, Ye ZM, et al. Computed Tomography-Based Intratumor Heterogeneity Predicts Response to Immunotherapy Plus Chemotherapy in Esophageal Squamous Cell Carcinoma. Acad Radiol 2024;31:4886-99. [Crossref] [PubMed]
25. Lv W, Wang Y, Hu F, et al. A Prognostic Survival Model Based on Endocrine-Related Gene Expression in Acute Myelogenous Leukemia. Acta Haematol 2025; Epub ahead of print. [Crossref]
26. He Y, Xiang L, Yuan J, et al. Lactylation Modification as a Promoter of Bladder Cancer: Insights from Multi-Omics Analysis. Curr Issues Mol Biol 2024;46:12866-85. [Crossref] [PubMed]
27. Lee SM, Kwak JY, Ryu D, et al. High glucose induces FABP3-mediated membrane rigidity via downregulation of SIRT1. Biochim Biophys Acta Gen Subj 2025;1869:130802. [Crossref] [PubMed]
28. Feng R, Ma S, Bai R, et al. Establishment and characterization study of ovine mammary organoids. BMC Vet Res 2025;21:184. [Crossref] [PubMed]
29. Wang Q, Zhang C, Yu B, et al. FABP3 promotes cell apoptosis and oxidative stress by regulating ferroptosis in lens epithelial cells. Free Radic Res 2025;59:250-61. [Crossref] [PubMed]
30. Ivarsson Orrelid C, Rosberg O, Weiner S, et al. Applying machine learning to high-dimensional proteomics datasets for the identification of Alzheimer's disease biomarkers. Fluids Barriers CNS 2025;22:23. [Crossref] [PubMed]
31. Azad TD, Ran KR, Materi JD, et al. A multi-analyte blood test for acute spinal cord injury. J Clin Invest 2025;135:e185463. [Crossref] [PubMed]
32. Wang L, Wei Q, Zhang M, et al. Identification of the prognostic value of immune gene signature and infiltrating immune cells for esophageal cancer patients. Int Immunopharmacol 2020;87:106795. [Crossref] [PubMed]
33. Dai J, Reyimu A, Sun A, et al. Establishment of prognostic risk model and drug sensitivity based on prognostic related genes of esophageal cancer. Sci Rep 2022;12:8008. [Crossref] [PubMed]
34. Niemczyk M, Ito Y, Huddleston J, et al. Imprinted chromatin around DIRAS3 regulates alternative splicing of GNG12-AS1, a long noncoding RNA. Am J Hum Genet 2013;93:224-35. [Crossref] [PubMed]
35. Michler S, Schöffmann FA, Robaa D, et al. Fatty acid binding to the human transport proteins FABP3, FABP4, and FABP5 from a Ligand's perspective. J Biol Chem 2024;300:107396. [Crossref] [PubMed]
36. Záveský L, Jandáková E, Weinberger V, et al. Long non-coding RNAs PTENP1, GNG12-AS1, MAGI2-AS3 and MEG3 as tumor suppressors in breast cancer and their associations with clinicopathological parameters. Cancer Biomark 2024;40:61-78. [Crossref] [PubMed]
37. Ao S, Liang L, Peng L, et al. Identification and validation of an m5C-related lncRNA signature for predicting prognosis and immune response in clear cell renal cell carcinoma. Discov Oncol 2025;16:227. [Crossref] [PubMed]
38. Dwivedi A, Thippana M, Khammampalli S, et al. Unraveling the gender-specific molecular landscape of lung squamous cell carcinoma progression. J Biomol Struct Dyn 2025; Epub ahead of print. [Crossref]