Identification of JAML as an Immune-Associated Prognostic Marker in Non-Small Cell Lung Cancer Authors: Qin Tian, Fucheng Zhang, Chunyan Han Published: 2026 ABSTRACT AbstractIntroductionNon-small cell lung cancer (NSCLC) remains a major cause of cancer-related mortality worldwide, and the identification of novel prognostic biomarkers associated with tumor immunity is urgently needed. Junctional adhesion molecule-like (JAML), a member of the junctional adhesion molecule family, participates in leukocyte adhesion, migration, and T-cell activation. Although JAML has been implicated in immune regulation and tumor progression in other cancers, its expression pattern, prognostic significance, and association with the immune microenvironment in NSCLC remain unclear. This study aimed to investigate the clinical and immunological significance of JAML in NSCLC.MethodsTranscriptomic and clinical data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases were analyzed to evaluate JAML expression patterns in NSCLC subtypes. The prognostic value of JAML was assessed using Kaplan–Meier survival analysis and Cox regression models. The association between JAML expression and immune cell infiltration was investigated using TIMER2.0, CIBERSORT, and TISIDB analyses. Functional enrichment analyses were performed to explore potential biological pathways associated with JAML expression. In addition, JAML expression was validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR) in paired NSCLC and adjacent normal tissues.ResultsJAML expression was significantly decreased in NSCLC tissues compared with normal tissues (P < 0.005), with the lowest expression observed in lung squamous cell carcinoma (LUSC) and reduced expression in lung adenocarcinoma (LUAD). Survival analysis demonstrated that patients with high JAML expression had significantly improved overall survival compared with those with low expression (univariate HR = 0.68, 95% CI: 0.54–0.86, P = 0.001; multivariate HR = 0.76, 95% CI: 0.57–1.00, P = 0.049). Immune infiltration analysis revealed that JAML expression was significantly associated with multiple immune cell populations, including CD8+ T cells (r = 0.42, P < 0.001), suggesting a close relationship between JAML expression and the tumor immune microenvironment. qRT-PCR validation confirmed that JAML expression was approximately 2.3-fold higher in adjacent normal tissues than in NSCLC tissues (P < 0.05).ConclusionJAML is downregulated in NSCLC and its high expression is associated with favorable overall survival and distinct immune infiltration patterns. These findings indicate that JAML may serve as a potential prognostic biomarker and provide insights into the relationship between JAML expression and the tumor immune microenvironment in NSCLC.Keywords: non-small cell lung cancer, JAML protein, human, prognosis Introduction Lung cancer remains the leading cause of cancer-related mortality worldwide, with non-small cell lung cancer (NSCLC) comprising 80–85% of all cases.1,2 NSCLC is categorized into squamous cell carcinoma, large cell carcinoma, and adenocarcinoma—the most prevalent NSCLC type, accounting for approximately 40% of lung cancers.3,4 Recent advancements in radiotherapy, chemotherapy, targeted therapies, and immunotherapies have significantly improved the assessment and treatment of lung cancer.1,5 However, the prognosis of NSCLC continues to be poor.6,7 In the context of lung cancer progression and immunotherapy, junctional adhesion molecules (JAMs), a family of immunoglobulin-like proteins expressed by various cell types, including leukocytes, platelets, epithelial cells, and endothelial cells, play important roles in regulating cell polarity, epithelial barrier integrity, and leukocyte migration.8 JAML (junctional adhesion molecule-like protein), in particular, functions as an adhesion molecule involved in leukocyte transendothelial migration and serves as a costimulatory receptor for epithelial T-cell activation.9–11 By regulating cellular adhesion and migration, JAML contributes to inflammatory responses and immune cell trafficking.9,10,12,13 Previous studies have also suggested that JAML may participate in tumor immunity.8 For example, JAML overexpression has been reported to promote tumor progression in gastric cancer.14 In contrast, recent studies in NSCLC have indicated a potential antitumor role of JAML through modulation of immune cell functions. Hao et al demonstrated that JAML-mediated activation of polyfunctional CD8+ tissue-resident memory T cells (CD8+ TRM cells) enhanced antitumor immunity and was associated with favorable outcomes in NSCLC patients.15 Furthermore, Eschweiler et al reported that JAML expression on CD8+ and γδ T cells was associated with improved survival and enhanced tumor immune responses.16 However, the comprehensive expression pattern of JAML, its prognostic significance, and its association with the broader tumor immune microenvironment in NSCLC remain insufficiently characterized. Therefore, further investigation is required to clarify the clinical and immunological relevance of JAML in NSCLC. In this study, we integrated large-scale datasets from TCGA and GEO cohorts, including 1237 tumor/normal samples, together with qRT-PCR validation in clinical tissues, to systematically investigate JAML expression patterns and clinical significance in NSCLC and its major subtypes, including lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC). Furthermore, we explored the relationship between JAML expression, immune cell infiltration, immune-related pathways, and patient prognosis. This study provides additional evidence regarding the potential role of JAML as an immune-associated prognostic biomarker in NSCLC. Materials and Methods Data Acquisition RNA-seq data and clinical information of patients with TCGA–NSCLC were obtained from the Genomic Data Commons (GDC) Xena database (https://portal.gdc.cancer.gov/). A total of 1133 samples, including 1025 NSCLC tumor samples and 108 normal control samples, were included, with available clinicopathological characteristics and survival information. The TCGA cohort was selected as the discovery cohort due to its large sample size, comprehensive genomic profiles, and well-annotated clinical information. To assess the reproducibility of JAML expression patterns and prognostic associations, independent GEO cohorts were included as external validation cohorts. The TCGA dataset was further divided into two NSCLC subtypes: TCGA–LUAD (524 LUAD samples and 59 normal controls) and TCGA–LUSC (501 LUSC samples and 49 normal controls), which were analyzed separately to evaluate the expression characteristics of JAML across different NSCLC subtypes. External validation was performed using GEO datasets, including GSE37745 (196 NSCLC samples, with 195 samples retaining available survival information) and GSE19188 (156 samples, comprising 72 NSCLC samples [45 LUAD and 27 LUSC] and 65 normal controls). These independent cohorts were used to confirm the consistency of JAML expression patterns and its association with clinical outcomes across different platforms. LinkedOmics Database Analysis The LinkedOmics database (http://www.linkedomics.org) is a multi-omics data analysis platform used for comparing multi-omics data within and between tumor types.9 The LinkFinder module was employed to screen differentially expressed genes associated with JAML, and their correlation was tested using the Pearson correlation coefficient. Gene Set Enrichment Analysis (GSEA) of Gene Ontology Biological Process (GO_BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was conducted in the LinkInterpreter module.2.3 GeneMANIA database analysis GeneMANIA (http://genemania.org) is a tool for gene function analysis and predicting protein–protein interaction (PPI) networks.10,11 Based on the input gene list, GeneMANIA utilizes genomics and proteomics data to identify genes with similar functions. In this study, GeneMANIA was used to predict the PPI network of JAML. GeneMANIA Database Analysis GeneMANIA (http://genemania.org) is a tool for gene function analysis and prediction of protein-protein interactions (PPI) networks.10 Based on the input gene list, GeneMANIA utilizes genomics and proteomics data to identify genes with similar functions. In this study, GeneMANIA was used to construct the PPI network of JAML. TIMER2.0 Database Analysis TIMER2.0 (http://timer.cistrome.org/) is a public platform for analyzing and visualizing the infiltration of immune cells in tumors.12 TIMER2.0 was used to assess the correlation between JAML expression and the increased infiltration of immune cells in LUAD. TISIDB Analysis The TISIDB (http://cis.hku.hk/TISIDB/index.php) database provides a comprehensive analysis of tumor–immune cell interactions.13 This study used TISIDB to validate the results of TIMER2.0 analysis and further explore the association between JAML expression and immune cell characteristics. Human Protein Atlas (HPA) Database Analysis The HPA (https://www.proteinatlas.org/) provides protein expression data in tumor and normal tissues. This study retrieved the immunohistochemical data regarding JAML expression in NSCLC tissues from the HPA database. RNA Extraction and qRT-PCR RNA was extracted from tumor and peritumoral tissues surgically resected from patients with NSCLC at Shandong Provincial Third Hospital, China. All patients were pathologically diagnosed with NSCLC. The study was approved by the Ethics Committee of Shandong Provincial Third Hospital (approval number: KYLL-2023019) and adhered to the Declaration of Helsinki. Informed consent was obtained from all participants, and their privacy was ensured. RNA extraction from rapidly frozen tissues was performed by using the TRIzol™ reagent (Ambion, 155596–026). The RNA concentration and purity were measured with a NanoDrop spectrophotometer. cDNA was synthesized using Hiscript® III RT SuperMix (Vazyme, R323-01), and qRT-PCR was conducted on a Bio-Rad CFX Connect™ real-time system, the primer sequences are presented in Table 1. Table 1Primer Sequences Used for Quantitative Real-Time PCR (qRT-PCR)Target GenePrimer OrientationSequence (5′–3′)JAMLForwardGCAGCATGTTTTGCCCACTGJAMLReverseAGGCGGGGAAACATTCAAGTGAPDHForwardCGAAGGTGGAGTCAACGGATTTGAPDHReverseATGGGTGGAATCATATTGGAACAbbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; JAML, junctional adhesion molecule-like. Primer Sequences Used for Quantitative Real-Time PCR (qRT-PCR) Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; JAML, junctional adhesion molecule-like. The 2-ΔΔCT method was used to calculate the fold change in gene expression. The PCR conditions are listed in Table 2. Melting curve data were collected to ensure PCR specificity. The inclusion criteria were as follows: (1) patients with histologically confirmed NSCLC; (2) patients who underwent surgical resection; (3) availability of paired tumor and adjacent normal tissues. The exclusion criteria included: (1) patients with other malignant tumors; (2) insufficient tissue samples; and (3) incomplete clinical information. Table 2Reverse-Transcription Reaction Conditions for Quantitative Real-Time PCR (qRT-PCR)Temperature (°C)Duration255 min5015 min855s4HoldNote: The reaction was maintained at 4 °C after completion. Reverse-Transcription Reaction Conditions for Quantitative Real-Time PCR (qRT-PCR) Note: The reaction was maintained at 4 °C after completion. Statistical Analysis For variables exhibiting a normal distribution, Student’s t-test was used to analyze measurement data, whereas the Mann–Whitney U-test was applied for variables that did not follow a normal distribution. The distribution of continuous variables was assessed before statistical testing. Comparisons among multiple groups were performed using the Kruskal–Wallis test. The primary endpoint was the association between JAML expression and patient survival outcomes, including OS and PFS. Secondary endpoints included JAML expression differences, clinicopathological associations, and immune infiltration characteristics. Receiver operating characteristic (ROC) curves were utilized to evaluate the discrimination ability of JAML expression between LUAD and normal tissues. For survival analyses, patients were stratified into high- and low-JAML expression groups according to the median JAML expression value within each cohort. Kaplan–Meier survival analysis and Log rank tests were performed to evaluate the association between JAML expression and OS/PFS. Cox regression analysis was conducted to determine the association between JAML expression and survival outcomes. The correlation between JAML expression and immune cell infiltration was assessed using Spearman correlation analysis. The potential influence of multiple comparisons was considered when interpreting immune infiltration analyses. All analyses were performed using R statistical software (version 3.6.3) and SPSS software (version 24.0). Statistical significance was set at P < 0.05, and all P-values were two-sided. Because this study was based on retrospective public datasets and clinical tissue validation, randomization and blinding procedures were not applicable. Results JAML Expression in NSCLC and ROC Curve Analysis The baseline characteristics of patients with NSCLC are presented in Table 3. Compared to normal tissue samples in the TCGA database, JAML expression was significantly reduced in NSCLC samples (P < 0.05). As shown in Figure 1, the mRNA expression level of JAML was significantly higher in normal tissues than it was in NSCLC tissues. The Kruskal–Wallis test (a non-parametric test suitable for multiple comparisons) was employed to analyze the differences in JAML expression in control, LUSC, and LUAD samples. The results revealed that the JAML expression was the highest in the control group, lowest in the LUSC samples, and intermediate in the LUAD samples, with statistically significant differences between the groups (P < 0.05). Additionally, the ROC curve was utilized to evaluate the sensitivity and specificity of JAML in diagnosing NSCLC in the TCGA and GSE datasets (Figure 1). In both datasets, the area under the ROC curve (AUC) for JAML was greater than 0.9, indicating high diagnostic efficacy in distinguishing NSCLC tissues from the normal ones. The consistent trend of reduced JAML expression in NSCLC was observed across both TCGA and GEO cohorts, supporting the robustness of the findings across independent datasets. Table 3Baseline Clinicopathological Characteristics of Patients with Non-Small Cell Lung CancerVariableCategoryn (%)SexFemale404 (40.24)Male600 (59.76)Age<65 years554 (58.32)>65 years396 (41.68)T categoryT1284 (28.37)T2560 (55.94)T3116 (11.59)T441 (4.10)N categoryN0649 (65.82)N1221 (22.41)N2109 (11.06)N37 (0.71)M categoryM0746 (96.01)M131 (3.99)Tumor stageI519 (51.9)II278 (27.8)III163 (16.3)IV32 (3.2)JAML expression statusHigh502 (50.0)Low502 (50.0)Note: Data are presented as n (%). Percentages are reproduced from the source table.Abbreviations: JAML, junctional adhesion molecule-like; NSCLC, non-small cell lung cancer. Baseline Clinicopathological Characteristics of Patients with Non-Small Cell Lung Cancer Note: Data are presented as n (%). Percentages are reproduced from the source table. Abbreviations: JAML, junctional adhesion molecule-like; NSCLC, non-small cell lung cancer. Differential expression and diagnostic performance of JAML in NSCLC. (A) Boxplot showing JAML mRNA expression in lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and normal lung tissues from the TCGA dataset. (B) Boxplot showing JAML mRNA expression in lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and normal lung tissues from the GEO dataset. (C) Receiver operating characteristic (ROC) curve evaluating the diagnostic performance of JAML for distinguishing NSCLC tissues from normal lung tissues in the TCGA cohort. (D) Receiver operating characteristic (ROC) curve evaluating the diagnostic performance of JAML for distinguishing NSCLC tissues from normal lung tissues in the GEO cohort. The corresponding area under the curve (AUC) values are indicated in each panel. Data were analyzed using the Wilcoxon rank-sum test. Relationship Between JAML Expression and Clinical Characteristics of NSCLC Tissues To evaluate the impact of JAML expression on patient survival, the expression levels of JAML, survival times, and survival status in different samples from the TCGA–NSCLC dataset were analyzed. Kaplan–Meier curves were used to compare the survival differences between the high- and low-JAML expression groups. To further explore the clinical significance of JAML, subgroup analyses were performed within the TCGA–NSCLC dataset using clinical characteristics such as age, gender, and tumor stage. The Kaplan–Meier curves were also employed to assess the impact of JAML expression on prognosis in different subgroups (Figure 2). Association between JAML expression and clinicopathological characteristics in NSCLC. (A) Violin plot showing JAML expression according to age in patients with NSCLC. (B) Violin plot showing JAML expression according to sex in patients with NSCLC. (C) Violin plot showing JAML expression according to pathological stage in patients with NSCLC. (D) Volcano plot illustrating genes differentially expressed between the high- and low-JAML expression groups. (E) Sankey diagram summarizing the relationships among JAML expression level, clinicopathological characteristics, and histological subtype. Statistical significance was determined using the Wilcoxon rank-sum test. Correlation Between JAML Expression and Prognosis in NSCLC Tissues To assess the impact of JAML expression on patient survival, patients with NSCLC were placed into high- and low-JAML expression groups based on the optimal cutoff value (cutoff = 0.3, where the P-value was minimized), and a Kaplan–Meier survival analysis was carried out. As shown in Figure 3A and B, the survival rate of patients in the low-JAML expression group was significantly lower than that in the high-expression group. Consequently, a higher number of deaths occurred in the low-expression group, which suggests that low JAML expression is associated with poor prognosis. To further explore the clinical value of JAML, subgroup analyses were conducted within the TCGA–NSCLC dataset based on clinical characteristics such as age, gender, tumor stage (T), lymph node metastasis (N), and the overall tumor stage. Kaplan–Meier curves were used to assess the impact of JAML expression on prognosis in these subgroups. As depicted in Figure 3C and D, in the high-expression group, the P-values for age, tumor stage (T), lymph node metastasis (N), and the overall tumor stage were all less than 0.05, indicating a significant correlation between JAML expression and these clinical features. Notably, within the tumor stage subgroup, patients with stage I cancer had the highest survival probability, while those with stage IV cancer had the lowest. Prognostic significance of JAML expression in patients with NSCLC. (A) Kaplan–Meier overall survival (OS) curve comparing patients with high and low JAML expression in the GEO cohort. (B) Kaplan–Meier overall survival (OS) curve comparing patients with high and low JAML expression in the TCGA cohort. (C) Kaplan–Meier survival analysis stratified by tumor stage in patients with high JAML expression to evaluate the prognostic significance of JAML expression. (D) Kaplan–Meier survival analysis stratified by tumor stage in patients with low JAML expression to evaluate the prognostic significance of JAML expression. Survival differences were assessed using the Log rank test. Survival Analysis of JAML Expression Univariate and multivariate Cox regression analyses were performed to evaluate the impact of JAML expression levels and clinical characteristics (including age, gender, and tumor stage) on the OS rates, and hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated. The results are presented in Table 4 (Supplementary Figure 1A and B). Based on the outcomes of univariate and multivariate Cox regression analyses and in conjunction with clinical features (such as the age of the patient and T stage of the tumor) and JAML expression levels, a composite nomogram was constructed to predict the survival rates of the patients at different time points (Supplementary Figure 1C and D). The C-index of this nomogram was 0.701, indicating good predictive performance. Table 4Univariable and Multivariable Cox Regression Analyses of Prognostic Factors for Overall Survival in Patients with Non-Small Cell Lung CancerPredictorUnivariable HR (95% CI)P valueMultivariable HR (95% CI)P valueSex1.1489 (0.9360–1.4101)0.18431.0512 (0.8252–1.3391)0.6861Age1.0113 (1.0002–1.0224)0.04561.0141 (1.0012–1.0272)0.0315Tumor stage1.4793 (1.3306–1.6446)<0.0011.0975 (0.8234–1.4630)0.5256M category2.2694 (1.4392–3.5783)<0.0011.4863 (0.7100–3.1111)0.2931N category1.4083 (1.2393–1.6003)<0.0011.2290 (0.9524–1.5861)0.1130T category1.4250 (1.2555–1.6173)<0.0011.2195 (1.0113–1.4706)0.0377JAML expression status0.6830 (0.5425–0.8599)0.00120.7579 (0.5749–0.9991)0.0492Note: The coding scheme and reference category for each predictor should be reported in the manuscript or table footnote.Abbreviations: CI, confidence interval; HR, hazard ratio; JAML, junctional adhesion molecule-like; NSCLC, non-small cell lung cancer. Univariable and Multivariable Cox Regression Analyses of Prognostic Factors for Overall Survival in Patients with Non-Small Cell Lung Cancer Note: The coding scheme and reference category for each predictor should be reported in the manuscript or table footnote. Abbreviations: CI, confidence interval; HR, hazard ratio; JAML, junctional adhesion molecule-like; NSCLC, non-small cell lung cancer. Immune Infiltration Alterations in the immune system play a pivotal role in disease initiation and progression. To investigate differences in immune cell composition across samples, we carried out immune cell profiling based on gene expression levels. Immune cell infiltration analysis of the TCGA–NSCLC dataset was performed using the ssGSEA method, using lollipop plots to illustrate the correlation between JAML expression and 28 immune cell subtypes. As shown in Figure 4, under the threshold of |cor|>0.3 and P < 0.05, 24 immune cell types were identified, 23 of which exhibited significant positive correlations with JAML expression (with MDSC showing the strongest association). In contrast, CD56bright natural killer cells demonstrated a significant negative correlation with JAML expression. Further comparative analysis using the Wilcoxon test (a non-parametric method) revealed differences in the 28 immune cell subtypes between JAML high- and low-expression groups. Of these 28 immune cell subtypes, 27 showed statistically significant differences between the two groups at a significance threshold of P < 0.05. Notably, the consistent increase in multiple immune cell populations in the high-JAML expression group may reflect a generally enhanced immune-active tumor microenvironment rather than a specific regulatory effect of JAML on each individual immune cell population (Figure 4). To evaluate the relationship between differentially infiltrated immune cells and patient survival in the JAML-expression subgroups, survival analyses were separately conducted for the high- and low-expression groups. As depicted in Figure 4, under P < 0.05, 13 immune cell types (eg, natural killer T cells) in the high-expression group and 20 (eg, macrophages) in the low-expression group were found to display significant associations between infiltration levels and prognosis. Supplementary Figure 5 presents differences among the top five immune cells with significant differences, which can be used to illustrate the correlation between the expression level of JAML in immune cells such as T cells and B cells and lung cancer metastasis. Association between JAML expression and immune cell infiltration in NSCLC. (A) Correlation analysis between JAML expression and the relative abundance of 28 tumor-infiltrating immune cell populations estimated by single-sample gene set enrichment analysis (ssGSEA). (B) Bubble plot showing correlations between JAML expression and immune cell infiltration levels. Bubble size represents the absolute Spearman correlation coefficient, whereas color indicates the corresponding P value. (C) Kaplan–Meier survival analysis evaluating the prognostic significance of natural killer T (NKT) cell infiltration in patients with high JAML expression. (D) Kaplan–Meier survival analysis evaluating the prognostic significance of natural killer T (NKT) cell infiltration in patients with low JAML expression. Statistical significance was assessed using the Log rank test. ns, not significant; ***P < 0.001; **P < 0.0001. GSEA and JAML Co-Expressed Gene Analysis (1) Gene Set Enrichment Analysis (GSEA) of JAML in the TCGA–NSCLC dataset revealed significantly enriched pathways associated with its expression. The gene was enriched in 3073 Gene Ontology (GO) terms and 77 signaling pathways. The top five pathways with the smallest P-values were as follows: cell cycle checkpoint, G2/M transition of mitotic cell cycle, cell cycle, chemokine signaling pathway, and cytokine–cytokine receptor interaction (Figure 5). Functional enrichment analysis and identification of JAML-associated genes. (A) Gene Set Enrichment Analysis (GSEA) showing significantly enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with JAML expression. (B) Gene Set Enrichment Analysis (GSEA) showing significantly enriched Gene Ontology (GO) biological processes associated with JAML expression. Running enrichment scores, ranked gene lists, and leading-edge subsets are shown. (C) Heatmap showing the expression profiles of the top 200 genes positively correlated with JAML. (D) Volcano plot illustrating differentially expressed genes associated with JAML expression. Red and blue dots represent positively and negatively correlated genes, respectively, whereas gray dots indicate genes without significant differential expression. (E) Gene Ontology (GO) enrichment analysis of the top 200 JAML co-expressed genes. (F) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of the top 200 JAML co-expressed genes. Chord diagrams illustrate the relationships between enriched biological terms/pathways and the corresponding genes. (2) To identify the genes co-expressed with JAML, Spearman correlation analysis was performed on the TCGA–NSCLC dataset. Genes with |cor|>0.3 and P < 0.05 were considered statistically significant. The top 200 positively correlated genes were selected as co-expressed genes for downstream analyses (Figure 5C–F). Functional enrichment analysis of these genes using the R package clusterProfiler led to the identification of 457 GO terms and 21 KEGG pathways. The results were visualized with the ggplot2 package, highlighting key biological processes and pathways linked to JAML activity. PPI and GGI Network Construction for JAML GeneMANIA-Based GGI Network GeneMANIA identifies and generates a list of genes functionally similar to the target gene and constructs interactive networks. A gene–gene interaction (GGI) network of JAML was generated using GeneMANIA, as shown in Supplementary Figure 2A. PPI Network Analysis PPI networks elucidate functional associations between genes. To construct the PPI network for JAML, protein interaction data were retrieved from the STRING database, a comprehensive resource for known and predicted PPIs. The resulting PPI network for JAML comprised 11 proteins and 15 interaction edges (Supplementary Figure 2B). ceRNA Network Construction MicroRNAs (miRNAs) regulate gene silencing by binding to mRNAs. This process is counteracted by competing endogenous RNAs (ceRNAs), such as lncRNAs and circRNAs, which competitively bind to shared miRNA response elements (MREs). Here, we constructed an lncRNA–miRNA–mRNA ceRNA network to investigate JAML-associated regulatory mechanisms in NSCLC. Potential ceRNA interactions involving JAML were predicted using bioinformatics databases. Intersection analysis identified eight overlapping candidates. Correlations between JAML and targeting miRNAs in the TCGA–NSCLC dataset were analyzed, and two miRNAs (hsa-miR-1238-3p and hsa-miR-873-5p) were identified that exhibit significant negative correlations (P < 0.05). Further predictions made using the LNCBOOK and starBase databases revealed 2500 and 133 lncRNAs targeting hsa-miR-1238-3p and hsa-miR-873-5p, respectively, with 39 overlapping candidates. The ceRNA regulatory network was constructed by integrating JAML-associated miRNAs and key lncRNAs, which finally led to the identification of 8 core genes and 13 interaction nodes (Supplementary Figure 3). Predictive Value for Immunotherapy This study utilized the TISIDB database to analyze the relationship between the JAML-expression levels and lymphocytes, immunomodulators, and chemokines. A significant association was considered when P < 0.05 and “rho” values were greater than 0.2 or less than −0.2. The expression of JAML (AMICA1) showed a significant positive correlation with most lymphocytes. It demonstrated the strongest association with macrophages in LUAD and with MDSCs in LUSC (Figure 6). In addition, it exhibited a significant positive correlation with most immunoinhibitors; HAVCR2 was the strongest immunoinhibitor in LUAD and CSF1R in LUSC. For immunomodulator–immunostimulators, the expression of JAML (AMICA1) showed a significant positive correlation with most immunostimulators. CD80 was identified as the strongest immunostimulator in LUAD and CD48 in LUSC. The expression of AMICA1 had a significant positive correlation with most of the immunomodulator–MHC molecules, with DPB1 being recognized as the strongest immunomodulator in both LUAD and LUSC. Additionally, the expression of AMICA1 was found to exhibit a significant positive correlation with most chemokines, with CCL23 demonstrating the strongest correlation in both LUAD and LUSC. AMICA1 also exhibited a significant positive correlation with most of the chemokine receptors, with CCR6 exhibiting the strongest correlation in LUAD and CCR5 in LUSC. Clinical samples, including five NSCLC cancer tissues and five paired adjacent non-cancerous tissues, were collected. The expression of the JAML gene in these tissues was subsequently analyzed using qPCR, which confirmed its higher expression levels in normal tissues compared to those in adjacent cancer tissues (Figure 7 and Supplementary Figure 4). Correlation of JAML expression with immune-related molecules in LUAD and LUSC. (A) Correlation between JAML expression and lymphocyte abundance in LUAD. (B) Correlation between JAML expression and lymphocyte abundance in LUSC. (C) Correlation between JAML expression and immunoinhibitory molecules in LUAD. (D) Correlation between JAML expression and immunoinhibitory molecules in LUSC. (E) Correlation between JAML expression and immunostimulatory molecules in LUAD. (F) Correlation between JAML expression and immunostimulatory molecules in LUSC. (G) Correlation between JAML expression and chemokine receptor expression in LUAD. (H) Correlation between JAML expression and chemokine receptor expression in LUSC. Correlation analyses were performed using the TISIDB database based on Spearman correlation analysis. Validation of JAML expression in clinical NSCLC tissues. Relative JAML mRNA expression was determined in paired NSCLC tissues and adjacent non-tumorous lung tissues by quantitative real-time PCR (qRT-PCR). GAPDH was used as the endogenous reference gene. Data are presented as the mean ± standard deviation (SD). Statistical significance was determined using Student’s t-test. **P < 0.01. Discussion This study investigated the impact of JAML expression on the progression and prognosis of NSCLC using multi-omics data from the TCGA and GEO databases. Across the TCGA and GEO cohorts, JAML expression was consistently lower in NSCLC tissues than in normal lung tissues, with particularly low expression in LUSC. Moreover, reduced JAML expression was associated with poorer overall survival. These findings suggest that loss of JAML expression may accompany NSCLC progression and that JAML may have a potential protective or tumor-suppressive role in this disease. Given the established functions of JAML in cell adhesion, leukocyte migration, and immune-cell activation, its downregulation may disrupt normal cell–cell interactions and weaken antitumor immune surveillance, thereby contributing to a less favorable clinical outcome. The prognostic relevance of JAML also appeared to vary according to the clinicopathological stage of NSCLC. The association between higher JAML expression and more favorable survival was more evident in patients with earlier-stage disease, whereas advanced-stage tumors remained associated with poor outcomes. One possible explanation is that the adhesion- and immune-related functions of JAML may contribute more effectively to tumor restraint during the early stages of disease, but may become insufficient as tumor heterogeneity, metastatic dissemination, and immune evasion increase during progression. Multivariate Cox regression analysis further supported the prognostic relevance of JAML after adjustment for established clinical variables, although its predictive value should be validated in independent cohorts. Taken together, these findings support a potential protective role for JAML in NSCLC rather than establishing it as a definitive tumor suppressor. This interpretation is consistent with previous bioinformatics studies that identified JAML as a favorable prognostic gene and associated its reduced expression with an increased risk of lung cancer.14,15 Our analysis extends these observations by showing that the prognostic significance of JAML may be influenced by disease stage and other clinicopathological characteristics. However, the role of JAML in the tumor immune microenvironment remains unclear. Prior studies highlight its involvement in γδ T-cell activation and tissue homeostasis,16,17 as well as its critical role in dendritic cell (DC) migration, which may enhance responses to DC-based cancer immunotherapy.18 An analysis of the LinkedOmics database revealed that most of the JAML-co-expressed genes are enriched in immune-related pathways, consistent with the GeneMANIA network findings. GSEA further identified key pathways, including cell cycle, chemokine signaling, and cytokine–cytokine receptor interactions. Immune infiltration analysis demonstrated significant positive correlations between JAML expression and 23 immune cell types—most strongly with myeloid-derived suppressor cells (MDSCs)—while CD56bright natural killer cells showed a significant negative correlation. These results underscore the potential dual role of JAML in modulating both tumor progression and immune regulation, warranting further mechanistic exploration.19 To compare the differences in immune infiltration cells between the high- and low-JAML expression groups, 28 types of immune cells were analyzed. At a significance threshold of P < 0.05, significant differences were observed in 27 types of immune cells between the two groups, with higher infiltration levels in the high-expression group for all immune cells. To evaluate the relationship between the differentially infiltrated immune cells and patient survival time in the JAML-expression groups, survival analyses were separately performed for the high- and low-expression groups. At a significance threshold of P < 0.05, the infiltration levels of 13 immune cells (such as natural killer T cells) in the high-JAML expression group were found to be significantly associated with prognosis, whereas in the low-expression group, 20 immune cells (such as macrophages) showed significant associations with prognosis. The top 200 genes co-expressed with JAML were selected and identified as genes involved in processes such as cell adhesion molecules and phagocytosis. A PPI network and a GGI network for JAML were constructed. Through the PPI network for JAML, the 11 most relevant nodes and 15 associated networks were identified. Finally, a competing endogenous RNA (ceRNA) network for JAML in NSCLC was predicted and constructed using bioinformatics databases. ceRNAs, a recent research focus, can regulate gene expression by competitively binding microRNAs. Correlation and bioinformatics prediction analyses identified hsa-miR-1238-3p and hsa-miR-873-5p as candidate microRNAs potentially associated with JAML regulation. However, these analyses do not establish direct binding or functional regulation. Previous studies have shown that JAML promotes tumor progression in gastric cancer in vitro.20 The distinct roles of JAML reported in different malignancies highlight the context-dependent nature of its biological functions. Although JAML has been suggested to promote tumor progression in gastric cancer, our findings indicate that higher JAML expression is associated with favorable prognosis and immune-enriched characteristics in NSCLC. Several factors may contribute to this discrepancy, including differences in tissue lineage, tumor-specific signaling networks, genetic background, and the composition and activation status of the tumor immune microenvironment. In addition, JAML expression in bulk tumor datasets may partly reflect differences in immune-cell abundance rather than tumor-cell intrinsic expression alone. Therefore, the biological role of JAML may depend on the cellular context and tumor type. Future functional studies, including JAML gain- and loss-of-function experiments, immune-cell interaction assays, and in vivo models, are required to determine the molecular mechanisms underlying its context-dependent effects. In addition, previous studies have reported that JAML expression was elevated in NSCLC tissues and associated with tumor metastatic progression.21,22 In contrast, our analysis based on TCGA and GEO cohorts showed that JAML expression was reduced in NSCLC tissues and that higher JAML expression was associated with favorable clinical outcomes. These discrepancies may be attributable to differences in patient cohorts, database sources, tumor subtypes, disease stages, sample composition, ethnic backgrounds, and analytical approaches. Furthermore, as bulk transcriptomic datasets reflect the combined signals from tumor cells and the tumor microenvironment, variations in immune-cell composition may also influence the observed association between JAML expression and clinical characteristics. More broadly, the present bioinformatics findings provide associative rather than mechanistic evidence. The current analyses cannot determine whether altered JAML expression directly affects NSCLC-cell behavior or primarily reflects changes in the abundance and activation status of infiltrating immune cells. Therefore, gain- and loss-of-function experiments are required to clarify the biological effects of JAML. JAML overexpression and knockdown, followed by assessments of cell proliferation, apoptosis, migration, invasion, and cell-cycle progression, would help determine whether JAML directly influences malignant phenotypes. Rescue experiments would further be necessary to confirm the specificity of the observed effects and to establish the direction of the regulatory pathway. The predicted upstream and downstream relationships also require direct validation. In particular, dual-luciferase reporter assays, RNA immunoprecipitation, microRNA overexpression or inhibition, and rescue experiments are needed to determine whether hsa-miR-1238-3p and hsa-miR-873-5p directly regulate JAML. Similarly, the enrichment of JAML-associated genes in cell-cycle, chemokine-signaling, and cytokine–cytokine receptor-interaction pathways should be validated by directly measuring key pathway proteins and their activation states. Immune-cell co-culture systems and in vivo tumor models would also help clarify whether JAML influences immune-cell recruitment, activation, or suppressive function. In conclusion, this study demonstrated that JAML expression is significantly lower in NSCLC tissues than in normal lung tissues, and that higher JAML expression is associated with more favorable overall survival. These findings suggest that JAML may have a potential protective role in NSCLC and may serve as an immune-associated prognostic biomarker. JAML expression was also associated with distinct patterns of immune-cell infiltration and immune-related signaling, suggesting its potential involvement in the tumor immune microenvironment, particularly in LUAD. In addition, bioinformatics analyses identified hsa-miR-1238-3p and hsa-miR-873-5p as candidate microRNAs potentially involved in JAML regulation, providing a basis for further investigation of the ceRNA regulatory network in NSCLC. However, the present study provides correlation-based evidence, and further functional experiments are required to determine the precise biological mechanisms underlying the role of JAML in NSCLC progression and immune regulation. Several limitations should be acknowledged. First, this study primarily relied on public transcriptomic datasets, computational analyses, and preliminary tissue validation; therefore, the observed associations between JAML expression, immune infiltration, and prognosis do not establish direct causal or mechanistic relationships. In addition, the clinical validation cohort used for qRT-PCR analysis was relatively small (five paired NSCLC tumor and adjacent normal tissues), and further validation in larger independent clinical cohorts is required to confirm the expression pattern and clinical significance of JAML. Functional studies, including gain- and loss-of-function experiments, rescue assays, pathway validation, and in vivo models, are required to elucidate the underlying mechanisms. Second, immune infiltration analyses based on computational algorithms may be influenced by tumor purity and other microenvironmental factors. Finally, the predicted ceRNA regulatory network and the clinical significance of JAML require further validation in larger independent clinical cohorts. Funding Statement This work was supported by grants from the Key Specialty Fund of Shandong Provincial Third Hospital [grant number KYLL-2023019] and the Zhong guan cun Precision Medicine Foundation [grant number GXZDH28]. Abbreviations NSCLC, Non-Small Cell Lung Cancer; JAML, Junctional adhesion molecule-like protein; LUSC, Lung Squamous Cell Carcinoma; LUAD, Lung Adenocarcinoma; TCGA, The Cancer Genome Atlas; GEO, Gene Expression Omnibus; JAMs, Junctional Adhesion Molecules; MDSCs, Myeloid-Derived Suppressor Cells; qRT-PCR, Quantitative Real-Time Polymerase Chain Reaction; ROC, Receiver Operating Characteristic; AUC, Area Under the Curve; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; PPI, Protein–Protein Interaction; GGI, Gene–Gene Interaction; ceRNA, Competing Endogenous RNA; miRNA, MicroRNA; lncRNA, Long Non-Coding RNA; MRE, MicroRNA Response Element; ssGSEA, Single-sample Gene Set Enrichment Analysis; HSA, Homo sapiens (human); AUC, Area under the Curve (used in ROC analysis); TISIDB, Tumor–Immune System Interaction Database; HPA, Human Protein Atlas; AMI, Adhesion Molecule Involved in Chemotaxis. Data Sharing Statement All data generated or analyzed during this study are included in this published article. Ethics Approval and Consent to Participate This study was approved by the Medical Ethics Committee of Shandong Provincial Third Hospital (Ethics number: KYLL-2023019). Informed consent was obtained from all individual participants included in the study. Author Contributions All authors made substantial contributions to the conception and design of this study. Qin Tian was responsible for data acquisition, bioinformatics analysis, and drafting the manuscript. Fucheng Zhang contributed to data analysis, interpretation of results, and manuscript preparation. Chunyan Han contributed to study conception, supervision, critical revision of the manuscript, and intellectual content improvement. All authors critically reviewed the manuscript, approved the final version for publication, and agreed to be accountable for all aspects of the work, and agreed on the journal to which the article was submitted. Disclosure The authors declare that they have no competing interests. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this published article.