UALCAN (http://ualcan.path.uab.edu) is a database that uses The Cancer Genome Atlas (TCGA) database to collect clinical data from 31 cancer types. EFNA3 gene expression information and basic clinical features were obtained from TCGA (https://portal.gdc.cancer.gov) (13). UALCAN can analyze the relative expression of a gene across tumor and normal samples, as well as across various tumor subgroups that are based on the cancer stage, tumor grade, race, body weight, or other clinicopathological features. This resource serves as a platform for in silico validation of target genes and for identifying tumor subgroup-specific candidate biomarkers (14). Data in the UALCAN database was examined and filtered for the EFNA3 gene in BLCA using the following criteria: i) ‘gene symbol: EFNA3’ and ii) ‘TCGA dataset: Bladder urothelial carcinoma’. Then the following conditions were selected: ‘Expression’; ‘Survival’; ‘Correlation’; and ‘Pan-cancer view’.

LinkedOmics (http://www.linkedomics.org/login.php) uses the data of 32 cancer types from TCGA. These data can be used to analyze the relationship between mRNA gene expression and features such as methylation and mutation sites. In the present study, LinkedOmics was used to obtain information on EFNA3 expression in BLCA by setting the following filter conditions: i) ‘Gene: EFNA3’; ii) ‘Analysis Type: Cancer vs. Normal analysis’; iii) ‘Data Type: mRNA’; iv) ‘Cancer Type: Bladder urothelial carcinoma’; v) ‘Gene Summary: P-value<0.05, fold change=all, gene rank=top 10%’; and vi) ‘Statistical method: Pearson correlation test’. Following this, ‘Gene Set Enrichment Analysis (GSEA)’ tools were selected. Then the following conditions were selected: ‘GO analysis (biological process)’; ‘GO analysis (cellular component)’; ‘GO analysis (molecular function)’; and ‘KEGG pathway’.

GSEA is a computational method used to determine whether a predefined gene set exhibits statistically significant differences between two biological states. In the present study, GSEA was used to identify potential pathways associated with EFNA3 expression and prognosis in BLCA. For this, data associated with BLCA were downloaded from TCGA (https://portal.gdc.cancer.gov/), and GSEA software (version 4.1.0; Broad Institute and the University of California) was used for analysis. The gene expression profiles of patients with BLCA were divided into high and low-expression groups based on the median value of EFNA3 expression. Gene sets were considered significantly enriched when the false discovery rate (FDR) <0.25 and P. adjust <0.05.

The interacting proteins associated with EFNA3, and the interaction network was analyzed using the STRING search tool (http://string-db.org/). The necessary data were obtained by searching for protein names, species and other necessary information.

The present study was a retrospective study in which bladder cancer pathological specimens were obtained. A retrospective analysis of tissue sections stored in the pathological database was performed. A total of 491 samples of BLCA tissues were collected from patients undergoing surgical resection at The Zhejiang Provincial People's Hospital (Hangzhou, China) between January 1998 and December 2011. The patient cohort consisted of 432 males and 59 females, aged 35–79 years old (median, 63.2 years old). No patients had undergone radiotherapy or chemotherapy prior to surgical resection. Moreover, 80 control samples were collected from adjacent tissues located >5 cm from the tumor edge, which is usually defined clinically as normal tissue >5 cm from the tumor margin (samples were collected between January 1998 and December 2011) (Table I). The samples were subsequently used to prepare tissue microarrays (TMAs), which were constructed by Shanghai Xinchao Biological Technology Co., Ltd. The study was a retrospective analysis, approved by the Ethics Committee of Zhejiang Provincial People's Hospital and an informed consent waiver was obtained from the Ethics Committee of Zhejiang Provincial People's Hospital (approval no. QT2022423). The study was performed in accordance with the Declaration of Helsinki (15) and all patient details were removed to protect patient privacy throughout the process.

The changes in EFNA3 protein expression were studied using the TMA consisting of the 491 human BLCA samples and 80 non-cancerous human bladder tissue samples. Briefly, slides were incubated at 68°C for 2 h, followed by deparaffinization, dehydration in xylene, and rehydration. Subsequently, the sections were immersed in antigen retrieval buffer and boiled at 120°C in a pressure cooker for 3 min. The sections were then treated with 3% H₂O₂ for 15 min to quench endogenous peroxidase activity and 1% bovine serum albumin to prevent non-specific binding. Then, the sections were incubated with rabbit anti-EFNA3 antibody (1:500; cat. no. PA5-86397; Thermo Fisher Scientific, Inc.) at 4°C overnight, washed three times with PBS, incubated with biotin-labeled secondary antibody for 20 min at room temperature and then horseradish peroxidase-conjugated antibody for another 20 min. Finally, tissue sections were stained with 3,3-diaminobenzidine, counterstained with hematoxylin for 8 min at room temperature, dehydrated, washed, and mounted.

Immunostaining was assessed and scored according to the staining intensity by two independent observers blinded to the clinical and pathological data. EFNA3-positive expression was categorized into 4 classes based on staining intensity which were assigned as follows: 0, no staining; 1, light yellow (weak staining); 2, yellowish-brown (moderate staining); and 3, brown (strong staining). The proportion of stained cells was scored as follows: 0, <5% cells stained; 1, 6–25% cells stained; 2, 26–50% cells stained; and 3, ≥50% cells stained. The staining index was calculated by multiplying the intensity and proportion scores. A staining index ≥4 was considered to reflect high EFNA3 expression and an index of <4 was considered low EFNA3 expression.

SPSS (version 26.0; IBM Corp.) was used to perform all statistical analyses. Categorical data were assessed for statistical significance of differences using a χ² test or Fisher's exact test to compare the groups. Logistic regression analyses were performed to determine the effects of the EFNA3 gene and clinical factors on patient prognosis. Survival analysis was performed using the Kaplan-Meier (KM) method combined with log-rank test. KM analysis was the basis for plotting overall survival (OS) curves. Patients were grouped according to EFNA3 expression and assessed for survival status and OS time. KM survival curves were plotted using GraphPad Prism version 10 (GraphPad Software, Inc.). In addition, univariate and multivariate Cox regression analyses were used to determine the relationship between EFNA3 gene expression and clinicopathological characteristics of 491 patients with BLCA. All tests were two-tailed statistical tests. P<0.05 was considered to indicate a statistically significant difference.

EFNA3 expression level differences in various tumor and normal tissue samples were analyzed using TCGA data from UALCAN. Upregulated EFNA3 expression was observed in a number of cancer types, particularly in cervical squamous cell carcinoma and esophageal cancer, while significant differences were observed between the EFNA3 expression levels in bladder cancer and normal tissues (Fig. 1A). After analyzing 19 normal bladder and 408 BLCA tissues in TCGA, the results demonstrated that EFNA3 was expressed at high levels in patients with advanced tumor-node-metastasis (TNM) stages and non-papillary carcinomas (Fig. 1B-F).

IHC was then used to assess the EFNA3 protein expression levels in BLCA and normal bladder tissues from patients. The results demonstrated that non-tumor tissues had little EFNA3 expression and BLCA tissues had markedly high expression (Fig. 2A-I). EFNA3 protein expression was detected in 282 out of 491 (57.4%) BLCA samples and 25 out of 80 (31.3%) normal bladder tissues, indicating a significant upregulation of expression in BLCA tissues. In addition, a positive association between high EFNA3 expression and tumor size, invasion depth, lymph node metastasis, distant metastasis, vascular invasion, and histological grade was found, while there was no significant association with age and sex (Table II).

KM survival data demonstrated that the overall survival time was shorter in the high EFNA3 expression group compared with the low expression group (Fig. 3A and B). An analysis of the clinical sample data demonstrated that the mean survival time of patients with high EFNA3 expression was 38±1.03 months, which was significantly shorter than that of the low expression group (45±1.21 months) (Fig. 3C). ln addition, a univariate analysis of factors affecting survival was conducted and demonstrated that survival time was associated with tumor size (P=0.026), lymph node metastasis (P<0.01), vascular invasion (P<0.01), depth of invasion (P<0.01), distant metastasis (P=0.009), histological grade (P<0.01), and EFNA3 expression (P<0.01) (Table III). After entering these factors into a Cox proportional risk regression model, the results demonstrated that vascular invasion, histological grade, and depth of invasion were independent factors affecting the prognosis of patients with BLCA (Table III). Although EFNA3 expression was not an independent factor, it contributed to the development of BLCA, which may be one of the adverse factors affecting patient prognosis. The EFNA3 gene was also subjected to binary logistic regression analysis and the results revealed that EFNA3 (P=0.036) had some therapeutic value in the early diagnosis of bladder cancer.

Using the LinkedOmics web tool, co-expression of the EFNA3 gene in 408 BLCA samples from TCGA database was investigated. Genes positively correlated with EFNA3 (red spots; FDR=0.05) and genes negatively correlated with EFNA3 (green spots; FDR=0.05) are shown in the volcano plot in Fig. 4A. The top 50 genes positively and negatively correlated with EFNA3 are shown in the heat map in Fig. 4B and C. The top three positively associated genes were ephrinA4 (EFNA4), ADAM metalloprotease domain 15 (ADAM15) and serine protease 27 (PRSS27) (Fig. 4D-F), and the top three negatively associated genes were centrosomal protein 120, DENN domain-containing 4A and KIAA1109 (Fig. 4G-I). The results of the GO enrichment analysis of the LinkedOmics data demonstrated that genes co-expressed with EFNA3 were located mostly in a mitochondrial protein complex, respiratory chain, anchored component of membrane, Sm-like protein family complex or the cytosol (Fig. 5A). The genes were involved in mitochondrial gene expression, NADH dehydrogenase complex assembly, the nucleoside monophosphate metabolic process, protein localization in the endoplasmic reticulum and the ephrin receptor signaling pathway (Fig. 5B). This co-expression of genes plays an important role in ribosomal RNA binding, ephrin receptor binding, unfolded protein binding, threonine-type peptidase activity, and oxidoreductase activity, acting on a heme group of donors (Fig. 5C). The KEGG pathway analysis revealed that the genes co-expressed with EFNA3 were primarily enriched in the ribosome, Parkinson's disease, the proteasome, carbon metabolism, fructose and mannose metabolism, the spliceosome and systemic lupus erythematosus (Fig. 5D). These findings indicated that EFNA3 had a broad transcriptome impact.

To investigate the potential functional processes of EFNA3 in BLCA, TCGA data were used to perform a GSEA to locate KEGG pathways that were enriched in samples with high EFNA3 expression. Significantly enriched pathways were chosen based on their normalized enrichment score. The results demonstrated that ‘REPRODUCTION’, ‘M_PHASE’, ‘NEURONAL_SYSTEM’, ‘EPIGENETIC_REGULATION_OF_GENE_EXPRESSION’, ‘HCMV_EARLY_EVENTS’ and ‘CHROMOSOME_MAINTENANCE’ were mostly enriched in samples with high EFNA3 expression (Table IV and Fig. 6A-F). These data implied that EFNA3 may aid in the advancement of BLCA by participating in a number of cancer-related signaling pathways.

The STRING database was used to generate PPI networks of EFNA3 in BLCA. The results demonstrated that EFNA3 interacted with Eph receptor A1 (16), EPHA2, EPHA3, EPHA4, EPHA5, EPHA7, EPHA10, EPHB1, EPHB3, and phospholipase C γ1 (Fig. 7A). In addition, the protein interactions were analyzed using the GeneMANIA tool and the data demonstrated that EPHA1, EPHA3, EPHA5, EFNA4, Ras P21 protein activator 1, EPHA2, EPHA10, EPHA4, EFNA5, EFNB3, EFNB2, EFNB1, and PIK3R2 interacted with EFNA3 (Fig. 7B).