We investigated ITGA4 expression at the single-cell transcriptomic level in healthy infants and adults using the Human Transcriptome Cell Atlas (HTCA) database (https://www.htcatlas.org/) (18). For examining the expression differences of ITGA4 between pan-cancer and normal tissues, we obtained pan-cancer datasets from the Xena database (https://xenabrowser.net/) (19), which include transcriptome sequencing data (TPM format) and clinical information from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) and the Gene Type-Tissue Expression (GTEx, https://www.gtexportal.org/) (20) projects. Cancer types with fewer than three samples were excluded, and gene expression data were standardized using log₂(x + 0.001). Data processing and visualization were performed using R packages “AnnotationDbi”, “org.Hs.eg.db”, “stringr”, “stringi”, “ggplot2”, and “RColorBrewer”. Additionally, the “CPTAC” section of the UALCAN database (https://ualcan.path.uab.edu/index.html) (21) was employed to analyze the ITGA4 protein levels. The correlation between ITGA4 expression and tumor grade and stage was analyzed using the “limma” and “stringr” packages. Finally, the TISIDB database (http://cis.hku.hk/TISIDB/) (22) was used to analyze ITGA4 expression patterns across different molecular and immune subtypes.

To assess ITGA4’s diagnostic value, we utilized the R package “pROC” to plot receiver operating characteristic (ROC) curves. An area under the curve (AUC) above 0.7 indicated diagnostic value, above 0.8 indicated good accuracy, and above 0.9 indicated excellent diagnostic value. Subsequently, the “survival” and “forestplot” packages were used to generate forest plots for Cox proportional hazards regression models (23). Kaplan-Meier (KM) curves for overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) were plotted using the “survminer” and “ggplot2” packages. High and low ITGA4 expression groups were determined based on the median ITGA4 expression value. We then used the “timeROC” and “ggplot2” packages to create time-dependent ROC curves, evaluating ITGA4’s ability to predict 1-, 3-, and 5-year survival rates. We randomly selected 70% of the TCGA pan-cancer samples as a training set (6830/9784) and used the “rms” and “survival” packages to construct a nomogram model for predicting patient prognosis. Model accuracy was verified with calibration plots. Finally, we evaluated the model’s clinical decision-making value and predictive accuracy for 1-, 3-, and 5-year survival rates using time-dependent ROC curves with the “timeROC” package and decision curve analysis (DCA) with the “ggDCA” package for both the training set and validation set (2954/9784).

We constructed protein-protein interaction (PPI) networks for ITGA4 using the GeneMANIA (http://www.genemania.org) (24), STRING (https://cn.string-db.org/) (25), and Cytoscape software. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses for these proteins were performed and visualized using the R packages “clusterProfiler”, “org.Hs.eg.db”, “tidyr”, “ggplot2”, and cnetplot function (26). Additionally, we explored the potential functions of ITGA4 at the single-cell level using the CancerSEA database (http://biocc.hrbmu.edu.cn/CancerSEA/) (27). We identified hub proteins by intersecting the top 70 ITGA4-associated proteins from both databases. The results were visualized with Venn diagrams using “VennDiagram” and “ggplot2” packages and a correlation heatmap of ITGA4 and hub gene expression was created using “ComplexHeatmap” package (28). Functional enrichment analysis of the hub genes was subsequently performed using the GSCA database (https://guolab.wchscu.cn/GSCA/#/) (29).

We analyzed the Spearman correlation between ITGA4 and three TME scores (StromalScore, ImmuneScore, and ESTIMATEScore) using the R package “ESTIMATE”. Spearman correlation analysis between ITGA4 expression and immune cell infiltration was assessed using five algorithms—CIBERSORT, xCELL, TIMER, EPIC, and MCPCounter—via the “IOBR” and “psych” package. Correlation heatmaps were generated using “ggplot2” to visualize these results. We also used the TISIDB database (http://cis.hku.hk/TISIDB/index.php) (30) to examine the relationships between ITGA4 and various tumor-infiltrating lymphocytes (TILs), immunoregulatory genes (immunostimulators, immunoinhibitors, and major histocompatibility complex (MHC) genes), chemokines, and chemokine receptors across cancers (31, 32).

We visualized the correlation between ITGA4 and four key MMR genes (MLH1, MSH2, MSH6, and PMS2) using heatmaps generated with the “ggplot2” package. Utilizing the “TMB” and “inferHeterogeneity” functions in the “maftools” package, and referencing previous studies (33, 34), we obtained various tumor heterogeneity parameters, including TMB, mutational and clonal intratumoral heterogeneity (MATH), MSI, neoantigen load (NEO), tumor purity, ploidy, homologous recombination deficiency (HRD), and loss of heterozygosity (LOH). Additionally, we sourced tumor stemness scores such as RNAss, EREG.EXPss, DNAss, DMPss, ENHss, and EREG-METHss from existing research (35). “ggplot2” was used to create correlation heatmaps. Furthermore, lollipop plots illustrating TMB and MSI’s correlations with ITGA4 were generated with “ggplot2” package. To further explore the predictive role of ITGA4 and other biomarkers on immune checkpoint blockade (ICB) therapy responsiveness, we employed the “Biomarker Evaluation” and “Regulator Prioritization” modules of the Tumor Immune Dysfunction and Exclusion (TIDE) database (http://tide.dfci.harvard.edu). We also analyzed ITGA4 expression in various immunosuppressive datasets. Finally, we assessed ITGA4 expression changes pre- and post-ICB therapy in tumor models and pre- and post-cytokine therapy in tumor cell lines using the TIMSO database (http://tismo.cistrome.org/) (36).

We used the cBioPortal database (https://www.cbioportal.org/) (37) to investigate ITGA4 mutation frequency, types, distribution, specific sites, and the relationship between ITGA4 expression and copy number alterations (CNA). We then examined the top ten genes most likely to mutate in the ITGA4-mutant group compared to the non-mutant group. The SMART database (http://www.bioinfo-zs.com/smartapp/) (38) was used to analysis ITGA4 methylation levels and its 18 specific methylation sites, as well as the correlations between them in cancer and normal tissues.

We obtained therapeutic sensitivity data for anti-cancer drugs from CellMiner (https://discover.nci.nih.gov/cellminer/) (39). Using the R packages “impute,” “limma,” “ggplot2,” and “ggpubr,” we analyzed and plotted the correlation between drug sensitivity and ITGA4 expression levels. The 3D structures of the ITGA4 protein and chemotherapeutic drugs were downloaded from the AlphaFold Protein Structure Database (https://www.alphafold.ebi.ac.uk/) (40, 41) and the PubChem platform (https://pubchem.ncbi.nlm.nih.gov/), respectively. Molecular docking and visualization of the results were performed using AutoDockTool (version 1.5.7) and the Pamon (version 3.0.3).

Transcriptome data (TPM format) and prognostic information for the TCGA-STAD cohort were obtained from the Xena database. KM curves were plotted to compare the OS differences between patients with the highest 20% and lowest 20% ITGA4 expression levels, following the method outlined in the above sections. Patients were divided into high and low ITGA4 expression groups based on the median expression level. Differentially expressed genes (DEGs) between these groups were identified using “DESeq2,” “edgeR,” and “ggplot2” packages (|log₂FC| > 1, adjusted P < 0.05) and visualized with a volcano plot (42, 43). GO and KEGG enrichment analyses of these DEGs were also performed using the method described in the above sections. Cancer reference gene sets (c2.cp.kegg.v2022.1.Hs.symbols.gmt) were obtained from the Molecular Signatures Database (MsigDB, http://www.gsea-msigdb.org/gsea/). After ID conversion using the “org.Hs.eg.db” package, DEGs were ranked by log2FC, and Gene Set Enrichment Analysis (GSEA) (44) was conducted using the “clusterProfiler” package, with results visualized by “ggplot2.” We also analyzed the correlation between ITGA4 expression and three TME scores, as well as immune cell infiltration in GC, following the method from the “Correlation analysis of ITGA4 and cancer immunity” section. The single-cell RNA sequencing data and annotation files from the GSE134520 and GSE167297 datasets were obtained from the Tumor Immune Single-cell Hub (TISCH, http://tisch.comp-genomics.org/home/) (45) and were analyzed using the “MAESTRO” and “Seurat” packages with t-SNE for cell clustering.

Human GC cell lines MKN45, HGC27, AGS, KATO III, N87, and gastric mucosal epithelial cell line GES-1, authenticated by short tandem repeat (STR) analysis, were obtained from the Digestive System Tumor Prevention and Treatment and Translational Medicine Engineering lnnovation Center of Lanzhou University. The cells were cultured in RPMI-1640 medium (Cat: 11875101, Gibco) containing 10% fetal bovine serum (FBS, Cat: G8002, Servicebio) and 1% penicillin/streptomycin (Cat:C0222, Beyotime) at 37°C in a humidified incubator (POG-150, BOLV INSTRUMENT) with 5% CO₂.

Protein levels of ITGA4 were analyzed in six pairs of GC and adjacent tissues collected from gastrectomy patients at the Second Hospital of Lanzhou University (January-December 2023, with ethical approval from the Ethics Committee of the Second Hospital of Lanzhou University and informed consent from patients), and in the six cell lines mentioned earlier. The staging of all six GC clinical samples was classified as stage III. Tissues and cells were lysed on ice using a lysis buffer with a PSMF (Cat: P0100, Solarbio):RIPA (Cat: P0013B, Beyotime) ratio of 1:100 to extract proteins. Equal amounts of protein samples were separated using 10% SDS-PAGE gels and transferred to PVDF membranes (Cat: IPVH00010, Millipore). Membranes were blocked with 5% skim milk (Cat: D8340, Solarbio) for 1 hour, then incubated overnight at 4°C with primary antibodies against ITGA4 (1:1000, Cat: 19676-1-AP, Proteintech) and GAPDH (1:10000, Cat: 60004-1-Ig, Proteintech). Afterward, membranes were incubated with HRP-conjugated secondary antibodies (1:10000, Cat: SA00001-2, Proteintech) for 1 hour at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) imaging system (JP-600Plus, JIAPENG) and quantification was performed using ImageJ software.

Cells were seeded into a six-well plate (2×10⁵ cells/well). After attachment, cells were transfected with siRNA (20 μM) and Lipo6000 (Cat: C0526-0.5ml, Beyotime) following the manufacturer’s protocol (GENERAL BIO). Specifically, 3 μL siRNA and 5 μL Lipo6000 were diluted in 250 μL Opti-MEM (Cat: 31985062, Thermo) each, mixed, and added to 1500 μL 1640 complete medium. The transfection mixture was applied to the cells and incubated at 37°C with 5% CO₂ for 6 hours before replacing with fresh 1640 medium. ITGA4 expression was analyzed by WB 48 hours post-transfection. The siRNA primer sequences for ITGA4 (human) were: 5’-CGAACAGAACUGAGUAAAA(dT)(dT)-3’ and 5’-CCUACAACGUGGACACUGA(dT)(dT)-3’.

Cells were cultured in 6-well plates with RPMI-1640 medium (10% FBS) until confluent. A sterile scraper created a scratch, washed with PBS (Cat: F211131, BasalMedia), and incubated at 37°C, 5% CO2. Scratch width was imaged at 0 and 24 hours and analyzed with ImageJ.

Cells were washed with cold PBS 3 times, resuspended in 100 µL binding buffer, and incubated with 5 µL AV/APC or 7-AAD solution for 15 minutes in the dark (Cat: AP105, Multi Sciences). After adding 400 µL binding buffer, apoptosis rate was assessed by flow cytometry (BD Biosciences).

With approval from the Ethics Committee of the Second Hospital of Lanzhou University, we collected 80 pairs of paraffin-embedded GC and adjacent tissue samples from June 2020 to December 2023. After sectioning, samples were deparaffinized, rehydrated, and immersed in antigen retrieval solution (Cat: 005000, Thermo) under high pressure (150-200 kPa) for 10 minutes. The sections were then incubated with 3% hydrogen peroxide (Cat: 88597, Merck) for 20 minutes, followed with 10% BSA (Cat: 37520, Thermo) at room temperature for 60 minutes. Next, the sections were incubated with primary anti-ITGA4 antibody (1:100, Cat: 19676-1-AP, Proteintech) overnight at 4°C, then with secondary antibody (1:2000, Cat: A-11008, Thermo) for 1 hour. Thereafter, DAB staining (Cat: PR30010, Proteintech) and hematoxylin counterstaining (Cat: PR30004, Proteintech) were performed. ITGA4 expression was quantified using a digital imaging system (3DHISTECH, Hungary). Staining intensity was scored from 0 to 3 (0 for no staining, 1 for pale yellow, 2 for light brown, and 3 for dark brown), and the percentage of positive cells was divided into four equal grades from 0-100%, corresponding to scores from 1 to 4 in ascending order. The final score was the product of the intensity and positive area scores.

We analyzed the correlation between ITGA4 expression and various clinicopathological features, as well as serum tumor and immune markers, in the previously mentioned 80 GC cases. These entries included prognostic and diagnostic value, gender, age, TNM stage, pathological stage, perineural invasion, vascular invasion, gastric mucosal ulceration/bleeding, and serum tumor markers and immune cell levels (detailed in the “Results” section).

Statistical analysis and visualization were conducted using R software (version 4.2.3). Spearman correlation analysis was utilized to evaluate correlations. Results were based in triplicate experiments and were presented as means ± standard deviation (SD). Statistical significance was assessed using log-rank test, student’s t-test, Mann–Whitney test, Welch’s t test, Wilcoxon test, Chisq test, Yates’ correction and two-way ANOVA. P-values less than 0.05 considered statistically significant.

We first examined the expression of ITGA4 in normal tissues. HTCA database analysis showed low ITGA4 expression in most adult and infant tissues ( Supplementary Figure S1A ), indicating its potential as a therapeutic target with low tissue toxicity. We then analyzed ITGA4 mRNA levels using the TCGA database, observing significant expression differences in most cancer types (18 of 24) ( Figure 1A ). Further analysis combining TCGA and GTEx databases confirmed these variations in 29 of 34 cancer types ( Figure 1B ). Notably, cancers like GBM, BRCA, ESCA, STES, KIPAN, STAD, HNSC, KIRC, and CHOL consistently showed higher ITGA4 expression in individual and combined databases, while LUAD, KIRP, LUSC, BLCA, READ, and KICH consistently exhibited lower expression. Additionally, analysis from the CPTAC database revealed that ITGA4 protein expression was higher in tumor tissues compared to normal tissues in COAD, OV, KIRC, PAAD, HNSC, and GBM, while it was lower in LIHC, LUAD, and BRCA ( Figure 1C ).

Our results further demonstrated that ITGA4 levels were elevated in higher stages compared to lower stages in ACC, BLCA, KICH, KIRP, READ, and STAD, whereas a reverse trend was observed in SKCM and THCA ( Figures 1D–K ). Moreover, higher grades of BLCA, HNSC, LGG, and STAD displayed increased ITGA4 expression compared to their lower counterparts, with notable elevations in both higher grades and stages for STAD and BLCA ( Figures 1L–O ). Using the TISIDB database, we explored potential correlations between ITGA4 expression and the molecular and immune subtypes of various cancers, noting variable ITGA4 expression in molecular subtypes within BRCA, HNSC, KIRP, LGG, LIHC, OV, PCPG, STAD, and UCEC—for example, elevated expression levels in STAD’s EBV subtype and HNSC’s Mesenchymal subtype ( Supplementary Figure S1B ). Concurrently, ITGA4 expression varied across different immune subtypes in 23 cancer types ( Supplementary Figure S1C ). These observations underscore the complex expression patterns of ITGA4 in tumors, suggesting its significant role in tumor progression and potential utility in tailoring clinical strategies for different cancer subtypes.

We evaluated the diagnostic value of ITGA4 in distinguishing tumor from normal tissues using ROC curves. ITGA4 demonstrated diagnostic potential in 20 cancer types (AUC>0.7). Specifically, PAAD, SARC, and TGCT showed AUC values exceeding 0.9, with LAML, whose AUC is 1, demonstrating perfect diagnostic accuracy ( Supplementary Figures S2A–C ).

Subsequently, using univariate Cox regression forest plots and KM survival curves, we assessed the prognostic implications of ITGA4 expression across different cancers. Forest plots for OS revealed that ITGA4 is a risk factor in KIRP, LGG, and UVM, but acts as a protective factor in KIRC, LAML, LUAD, SKCM, and THYM ( Figure 2A ). These results were validated by KM survival curves, which demonstrated that high ITGA4 expression was associated with lower OS in LGG and UVM, while it correlated with higher survival probabilities in KIRC, LUAD, SKCM, and HNSC ( Supplementary Figure S3A ). Further exploration showed that elevated ITGA4 expression could negatively influence DSS in KIRP, LGG, SARC, and UVM, but it lowered the risk of adverse DSS outcomes in KIRC, LUAD, and THYM ( Figure 2B ). KM curves for DSS in KIRC, UVM, LGG, and LUAD corroborate these findings, and high ITGA4 expression also correlated with poorer DSS in KICH and UCEC, whereas it prolonged DSS in HNSC ( Supplementary Figure S3B ). In terms of DFI, high ITGA4 expression significantly correlated with shorter DFI in ESCA and KIRP ( Figure 2C ). KM analysis further indicated that ITGA4 might serve as a prolonging factor for DFI in CHOL ( Supplementary Figure S3C ). PFI analysis showed that ITGA4 as a protective factor in CHOL, KIRC, and UCEC, but demonstrated an adverse effect in LGG and UVM ( Figure 2D ), a finding supported by KM analysis ( Supplementary Figure S3D ).

To evaluate ITGA4’s predictive value for 1-, 3-, and 5-year survival, we produced time-dependent ROC curves. ITGA4 demonstrated strong predictive capabilities, such as an AUC of 0.784 for 5-year DLBC survival, 0.966 for 1-year KICH survival, and over 0.7 for both 3-year and 5-year TGCT survival ( Figures 3A–E ). We then developed a nomogram using TCGA data based on ITGA4 expression, patient age, and cancer type to predict survival probabilities ( Figure 3F ), which showed strong predictive performance in both training and validation datasets with AUC exceeding 0.7 for 1-, 3-, and 5-year survival ( Figures 3G, J ), validated by accurate calibration curves ( Figures 3H, K ). DCA on both training and testing datasets showed that models like All-1825 and All-1095 provided higher net benefits at low to medium thresholds, demonstrating their predictive value. Conversely, the All-365 model displayed limited utility at higher thresholds ( Figures 3I, L ). These findings emphasize the importance of ITGA4 expression-model selection based on clinical context to optimize patient outcomes, underscoring the value of tailored clinical decision-making.

We used STRING and geneMANIA databases to identify the top 20 and 50 proteins associated with ITGA4 respectively, with PPI networks constructed ( Figures 4A, B ). The intersection of these datasets revealed nine hub genes, with a pan-cancer heatmap confirming their positive correlation with ITGA4 ( Figures 4C, D ). Functional enrichment analysis of these 70 genes identified KEGG categories like ECM-receptor interaction, the PI3K-Akt signaling pathway, cell adhesion molecules, leukocyte transendothelial migration, and the Rap1 signaling pathway ( Figure 4E ). Correspondingly, GO enrichment analysis highlighted the involvement of ITGA4 in cell-substrate adhesion, integrin-mediated cell adhesion, extracellular matrix binding, leukocyte migration, and cellular extravasation ( Figures 4F, G ). Data from the CancerSEA database showed at the single-cell level that ITGA4 may impact cellular quiescence, differentiation, apoptosis, and processes related to cancer metastasis, invasion, and angiogenesis ( Supplementary Figures S4A–F ). To augment our comprehension of ITGA4’s roles in tumor progression, analyses of the aforementioned nine hub genes via the GSCA database indicated their potential roles in regulating apoptosis, the cell cycle, DNA damage repair, epithelial-mesenchymal transition (EMT), and key oncogenic signaling pathways including PI3K-AKT, RAS-MAPK, RTK, and TSC-mTOR pathways ( Figure 4H ). These analyses deepen our understanding of ITGA4’s role in tumor biology and microenvironments, offering crucial insights for developing targeted therapies and predicting tumor behaviors.

As TME has an important impact on tumor activity and response to treatment (46), we analyzed the correlation between ITGA4 expression and TME scores, revealing significant positive correlations with StromalScore, ImmuneScore, and ESTIMATEScore ( Figures 5A–C ). These findings suggest that tumors with high ITGA4 expression may have a rich stromal component, extensive immune cell infiltration, and high tumor purity. This indicates a potential pivotal role of ITGA4 in modulating the TME, particularly in the interactions between tumor cells and the surrounding stroma and immune cells.

We then conducted a detailed analysis using five distinct algorithms to elucidate the relationship between ITGA4 expression and immune cell infiltration across various cancer types. Overall, except for UCS, ITGA4 expression showed significant correlations with the infiltration of multiple immune cell types, especially macrophages, CD8⁺ T cells, and CD4⁺ T cells, in most cancers ( Figures 5D–H ). Specifically, results from the CIBERSORT algorithm revealed significant positive correlations between ITGA4 and both M1 and M2 macrophage types, which are antagonistic. Notably, the association with M2 macrophages generally exceeded that with M1 macrophages. This pattern could potentially be explained by the significant negative correlation of CD4⁺ Th1 cells with ITGA4 across multiple cancers using the XCELL algorithm (whereas Th2 cells, which are antagonistic to Th1 cells, showed significant positive correlations with ITGA4), since a decrease in Th1 cell infiltration could diminish the activation of M1 macrophages through reduced interferon-gamma (IFN-γ) secretion. We hypothesize that abnormal ITGA4 expression may foster an adverse inflammatory environment in tumors, leading to decreased M1 macrophage activation while facilitating their polarization towards the tumor-promoting M2 phenotype. The recruitment of CD4⁺ and CD8⁺ T cells to tumor sites is likely influenced by this inflammatory milieu and regulated by ITGA4. Furthermore, regulatory T cells (Tregs), cancer-associated fibroblasts (CAFs), and endothelial cells, also demonstrated significant positive correlations with ITGA4 in multiple cancers. We further analyzed the association of ITGA4 with TILs, immunoregulators, chemokines, and their receptors using the TISIDB database. Consistent with prior results, ITGA4 significantly correlated with many immune cells across different tumors ( Supplementary Figure S5A ). Moreover, ITGA4 exhibited significant associations with a broad range of Immune regulatory genes, including immunostimulators, immunoinhibitors, and MHC molecules ( Supplementary Figure S5B–D ). Additionally, ITGA4 showed substantial positive correlations with several chemokines and their receptors, notably including multiple members of the CCL, CXCL, CCR, and CXCR gene families ( Supplementary Figure S5E, F ). It’s worth noting that TME score, immune cell infiltration, and immune regulation all showed weak or inconsistent correlations with ITGA4 expression in LAML and UCS, warranting further investigation. The above findings underscore the complex role and potential importance of ITGA4 in regulating the tumor immune microenvironment, providing a theoretical foundation for future therapies targeting ITGA4.

Given the significant impact of tumor cell heterogeneity and stemness on malignancy and immunotherapy sensitivity, we explored the relationship between ITGA4 and related indices across multiple cancers. We initially investigated the association between ITGA4 and four key MMR genes (MLH1, MSH2, MSH6, PMS2) and found a significant positive correlation in 30 cancer types, excluding ACC, ESCA, and LUSC ( Figure 6A ). Further analysis revealed that ITGA4 was significantly negatively correlated with MSI in seven cancer types and with TMB in ten types ( Figures 6B, C ). This suggests that elevated ITGA4 expression could decrease MSI and TMB mediated by MMR, potentially complicating tumor responsiveness to immunotherapy. Intriguingly, in GBMLGG, LGG, and OV, a positive correlation between ITGA4 and TMB was observed, likely due to unintended mutations in MMR genes. Significant correlations were also noted between ITGA4 and other tumor heterogeneity markers such as HRD, LOH, MATH, NEO, ploidy, and tumor purity across multiple cancer types ( Supplementary Figure S6A ). Moreover, ITGA4 showed a consistent negative correlation with RNAss in most cancers, and its relationship with other stemness scores (EREG.EXPss, DNAss, DMPss, ENHss, EREG-METHss) varied in direction and degree ( Supplementary Figure S6B ), potentially affecting tumor cell drug resistance.

We subsequently evaluated the predictive efficacy of ITGA4 for immunotherapy outcomes in cancer patient cohorts treated with ICB. ITGA4 achieved an AUC over 0.5 in 13 cohorts, comparable to MSI.Score, and outperformed in predicting NSCLC (Ruppin2021_PD1_NSCLC) and melanoma (Riaz2017_PD1_Melanoma_Ipi.Prog, Gide2019_PD1_Melanoma) immunotherapy outcomes ( Figure 6D ). Additionally, ITGA4 expression varied among multiple Immunosuppressive datasets, showing high levels in METABRIC, TCGA Melanoma, ICB_Nathanson2017_CTLA4, and ICB_VanAllen2015_CTLA4 cohorts, but low in GSE12417_GPL570, Patel2017 1, and Shifrut 2018 pilot Average cohorts ( Figure 6E ). Furthermore, analysis using the TIMSO database revealed that ITGA4 could predict immunotherapy response in eight in vivo tumor killing assays (primarily involving anti-PD1, anti-CTLA4, and their combination, as well as anti-PDL1 and anti-PDL2) and three in vitro cytokine (primarily IFNg) killing assays ( Figures 6F, G ). These findings underscore the potential role and impact of ITGA4 in modulating cancer responses to immunotherapy.

Analysis based on the cBioPortal database revealed ITGA4 alterations in 22 cancer types, with mutations most common, especially in melanoma ( Figure 7A ). Of the 303 somatic mutation sites identified in ITGA4, 266 were missense mutations ( Figure 7B ). Supplementary Figure S7A further illustrates the distribution of ITGA4 mutation types across different cancers. Additionally, pan-cancer analysis showed a low correlation between CNAs and ITGA4 expression levels, with no significant differences across CNA types, suggesting minimal impact of CNA on ITGA4 expression in cancer ( Figures 7C, D ). Figure 7E displays the top ten genes more prone to mutations in the ITGA4 altered group compared to the non-altered group within the pan-cancer cohort, including TTN, TP53, MUC16, CSMD3, LRP1B, and SOCS2, all known to play roles in tumor progression (47–52).

Methylation plays a critical role in cancer progression (53). Analysis from the SMART database revealed elevated ITGA4 methylation levels in BLCA, BRCA, CHOL, COAD, ESCA, HNSC, LUAD, PAAD, PRAD, READ, STAD, and THCA, but lower levels in LIHC compared to normal tissues ( Figure 7F ). Among the 18 ITGA4 methylation sites analyzed, except for cg05246303 and cg25515269, the remaining 16 sites showed significant variations in over ten types of cancer ( Supplementary Figure S7B ). Furthermore, methylation at cg04531425, cg11947981, cg16057262, cg25024074, cg21995919, cg10965575, cg17265419, cg25515269, and cg05246303 were found to inhibit ITGA4 expression in various cancers. Meanwhile, methylation levels was negatively correlated with ITGA4 expression in CESC, CHOL, ESCA, GBM, MESO, TGCT, THCA, and UVM, and positively correlated in LAML, LIHC, and PCPG ( Figure 7G ). These findings emphasize the potential impact of ITGA4 methylation status on its expression across different cancers, providing insights for new cancer diagnostic or therapeutic strategies.

Figures 8A–P presents the top 16 drugs from the CellMiner database whose IC50 values are most significantly correlated with ITGA4 expression. Specifically, Fluorouracil and t-dcyd exhibit negative correlations with ITGA4 expression levels, with protein docking free energies of -4.53 kcal/mol and -5.47 kcal/mol respectively ( Figures 8Q, R ), indicating that cancers with higher ITGA4 may respond better to these drugs. Conversely, the IC50 values of the other 14 drugs are positively correlated with ITGA4 expression, suggesting potential resistance in tumors with elevated ITGA4 levels. These results underscore ITGA4’s potential impact on drug sensitivity in cancer, proposing it as a viable target for therapeutic intervention.

To validate the pan-cancer findings of ITGA4, we investigated its role in GC using the TCGA-STAD cohort. We observed a poorer prognosis in patients with top 20% ITGA4 expression compared to those with the bottom 20% ( Figure 9A ), correlating with findings from pan-cancer studies linking higher ITGA4 expression to advanced GC stages and grades. Further, DEGs between high and low ITGA4 expression groups were used for enrichment analysis ( Supplementary Figure S8A ). GO results indicated that ITGA4 primarily contributes to cell adhesion, immune infiltration, and immune regulation in GC, involving processes such as immune response-activating signal transduction, protein complexes involved in cell adhesion, chemokine binding, and MHC protein complex binding ( Figure 9B ). Meanwhile, KEGG and GSEA analyses suggested that ITGA4 might regulate crucial physiological processes beyond the TME of GC, such as ECM receptor interaction, cell adhesion, apoptosis, antigen processing and presentation, and oxidative phosphorylation. Moreover, ITGA4 impacts several classic cancer-related signaling pathways including PI3K-Akt, JAK-STAT, NF-kappa B, and Toll-like receptor pathways ( Figure 9C, D ). Additionally, our investigation into ITGA4’s immunological relevance in GC revealed significant positive correlations between its expression and three TME scores ( Figure 9E ). Also, three different algorithms showed significantly higher immune cell infiltration, such as macrophages, CD4⁺ and CD8⁺ T cells, in the high ITGA4 expression group compared to the low expression group ( Figures 9F, G ; Supplementary Figures S8B–E ). ITGA4 were also noted primarily expressed in CD8⁺ T cells, DC cells, and plasma cells in two GC single-cell sequencing cohorts (GSE134520 and GSE167297) ( Figure 9H ; Supplementary Figures S8F, G ), suggesting the expression of ITGA4 in these cells may potentially regulate the TME. These findings are generally consistent with the results from pan-cancer analyses and highlight a unique strong immunological correlation associated with ITGA4 in GC.

WB analysis revealed that ITGA4 expression was significantly higher in GC cell lines AGS, MKN45, and HGC27 compared to the normal GES-1 cell line ( Figures 10A, B ). Similarly, ITGA4 protein levels were elevated in GC tissues relative to adjacent normal tissues ( Figures 10C, D ). IHC analysis of 80 GC cases confirmed these results ( Figures 10E, F ) and showed that ITGA4 protein expression was higher in the low stage GC than in the high stage ( Supplementary Figures S9A, B ). Knockdown of ITGA4 attenuated the GC cell line MKN45 migration and enhanced its apoptosis ( Supplementary Figures S9C–H ). Additionally, ITGA4 expression demonstrated strong diagnostic value for these GC cases (AUC = 0.808), and cases with high ITGA4 expression were associated with poorer OS ( Figures 10G, H ). Furthermore, patients with high ITGA4 expression exhibited more advanced N and pathological stages, increased perineural and vascular invasion ( Table 1 ), along with elevated Ki-67 expression ( Supplementary Table S1 ). ITGA4 expression was also correlated with blood levels of neutrophils, lymphocytes, and basophils ( Supplementary Table S2 ). Moreover, there was a positive correlation between ITGA4 expression and serum levels of C-reactive protein (CRP) and carbohydrate antigen 125 (CA125) ( Figures 10I, J ).