RNAseq data from 33 tumor types were downloaded from the TCGA database (https://portal.gdc.cancer.gov) and processed using the STAR pipeline. The data were extracted in TPM format, with matched paraneoplastic and carcinoma samples corresponding to numbered pairs. Analyses were performed using R software (version 4.2.1), with relevant packages including ggplot2 [3.3.6], stats [4.2.1], and car [3.1–0]. Data processing was conducted using the log2 (value + 1) transformation. Statistical analyses included the Wilcoxon rank-sum test and the Wilcoxon signed-rank test. Tumor cell line data were obtained from the Cancer Cell Line Encyclopedia (CCLE), and normal cell line data were retrieved from the BioGPS database (http://biogps.org).

To evaluate differences in BIRC5 protein expression, immunohistochemistry (IHC) images of BIRC5 expression in both normal and tumor tissues were downloaded from the Human Protein Atlas (HPA) (http://www.proteinatlas.org). This analysis included various cancer types, such as lymphoma, skin cancer, liver cancer, breast cancer, and lung cancer.

The correlation between BIRC5 expression and various molecular and immune subtypes across different cancers was examined using the TISIDB database (http://cis.hku.hk/TISIDB/). Data on BIRC5 expression across different molecular and immune subtypes were downloaded from the “Gene” section of the database, and correlations between BIRC5 expression and specific cancer subtypes were subsequently analyzed.

To identify potential proteins interacting with BIRC5, the STRING database (https://string-db.org/) was utilized, and the top 20 proteins with the highest interaction probabilities were selected. The resulting protein-protein interaction network was visualized using Cytoscape software (version 3.9.0).

GO and KEGG enrichment analyses were conducted using the clusterProfiler package [4.4.4], following ID conversion of the input gene list via the org. Hs.e.g.,.db R package.

The diagnostic potential of BIRC5 in pan-cancer was assessed through receiver operating characteristic (ROC) curve analysis. RNAseq data from the pan-cancer STAR project were downloaded from the TCGA database, and data not corresponding to clinical information were excluded. The analysis was conducted using the pROC package [1.18.0], and the results were visualized using ggplot2 [3.3.6].

Kaplan-Meier plots were generated to evaluate the relationship between BIRC5 expression and cancer prognosis, including overall survival (OS), progression-free interval (PFI), and disease-free survival (DSS). Proportional hazards hypothesis testing and survival regressions were conducted using the survival package [3.3.1], with visualization performed using the survminer package [0.4.9] and ggplot2 [3.3.6]. Cox regression analysis was applied, and a significance threshold of p < 0.05 was set.

RNAseq data from the TCGA-LUAD project were downloaded, processed in TPM format, and paired with corresponding clinical data. Cox regression analysis was performed using the survival package [3.3.1] to test the proportional hazards hypothesis and identify significant variables for inclusion in the multivariate Cox model.

To predict patient survival, a nomogram was constructed, incorporating clinicopathological factors such as age, sex, disease type, stage, smoking status, alcohol consumption, BMI, and IRSS, using the rms package. ROC curves were employed to evaluate the predictive accuracy of the model. LASSO (least absolute shrinkage and selection operator) regression was performed to refine the model by reducing overfitting and improving interpretability. Additionally, one-way Cox regression identified 9 genes associated with lung cancer prognosis, which were incorporated into the final risk factor model.

Normal lung epithelial cells BEAS-2B and two lung cancer cell lines (A549, H1299) were cultivated in DMEM high-sucrose medium containing 10% FBS at 37°C and 5% CO₂. Adherent cells were trypsinized when reaching 80%–90% confluence, counted, passaged, and replated. Suspended cells were directly resuspended, collected, centrifuged, and then the cell suspension was uniformly distributed before plating.

We used RIPA buffer (1% PMSF) to lysed the cells, and BCA method was used to determine the concentration of proteins in cells or tissues. Proteins were separated by 7.5% SDS-PAGE, transferred to a PVDF membrane, and incubated with primary and secondary antibodies against the corresponding proteins. Band analysis was conducted post-exposure under developer.

Cells at 40%–60% fusion were transfected with lentivirus (Knockdown (KD) and overexpression (OE)) for 48 h. Cells at 40%–60% confluence were transfected with inhibitor BIRC5 for 48 h. Transfection efficiency was observed using fluorescence microscopy and WB. The cells (A549, A549-KD-BIRC5, A549-OE-BIRC5) were incubated in 96-well plates for 0, 1, 3, and 5 days respectively, and performed with Cell Counting Kit-8 (CCK8) assay. Cells (A549, A549-KD-BIRC5, A549-OE-BIRC5) were inoculated into Culture-Insert with scratches up to 500 μm wide. The length of scratches was observed and recorded under the microscope after culturing for 0, 24, 48, and 72 h. The area of scratched area was quantified using ImageJ software. The matrix gel was diluted at a ratio of 1:8, and 100 μL of the diluted matrix gel was added to the upper chamber of the Transwell plate, and incubated for 30 min at 37°C. Cell suspension of 200 μL (5 × 104 cells) is added to the upper chamber, and 500 μL of complete medium containing 20% FBS was added to the lower chamber. The plates were incubated for 48 h. The cells were gently wiped off the surface of the matrix gel and the transwell was fixed with 4% paraformaldehyde, and stained with 0.1% crystal violet. Six random fields were counted under the light microscope.

After trypsin digestion of cells in the logarithmic growth phase, resuspend the cells in complete medium (basal medium +10% fetal bovine serum) into cell suspension and inoculate each experimental group with 1,000 cells/well in 6-well plate culture. After 14 days of cloning culture, 1 mL of 4% paraformaldehyde was added to each well for 60 min, and the cells were washed twice with PBS. Add 1 mL of crystal violet staining solution to each well and stain the cells for 20 min. Wash the cells several times with PBS, dry them and take pictures under a light microscope.

Comprehensive analysis of tumor databases revealed that the mRNA expression level of BIRC5 was significantly higher in tumors compared to corresponding normal tissues in all relevant unpaired samples, including BLCA, BRCA, CESC, CHOL, COAD, and 18 other tumor types (Figure 1A). Similar results were observed in paired samples, such as BLCA, BRCA, CHOL, and COAD across 16 tumor types (Figure 1B). Further analysis of BIRC5 expression in various cancer and normal cell lines through the BioGPS database indicated that BIRC5 was highly expressed in multiple cancer cell lines, whereas its expression was low in normal cell lines. Notably, BIRC5 expression was elevated in stem cells and immune cells (Supplementary Table S1). The top 10 cancer and normal cell lines with the highest BIRC5 mRNA expression levels are depicted in Figures 1C, D. To assess BIRC5 expression at the protein level, we analyzed immunohistochemistry (IHC) results from the HPA database and compared them with gene expression data. The mRNA expression data from Figures 1A, B did not include lymphoma and ovarian cancer, whereas the protein expression data from the HPA database, shown in Figures 1G–K, includes lymphoma (for example, in lymphoid tissues) and other cancers (such as liver, skin, and lung). These findings suggest that while the mRNA expression analysis did not include lymphoma and ovarian cancer, the protein expression data reveal significant upregulation of BIRC5 in these and other cancers. This highlights the importance of considering both mRNA and protein expression levels in understanding the role of BIRC5 in cancer progression. These findings suggest that the upregulated expression of BIRC5 in tumors may play a role in cancer initiation and progression.

Using the TISIDB database, we observed that BIRC5 expression was significantly correlated with immune subtypes across most cancers (C1: wound healing, C2: IFN-gamma dominance, C3: inflammation, C4: lymphocyte depletion, C5: immune quiet, C6: TGF-b dominance), including ACC, BRCA, and BLCA, among 22 cancer types. Most tumors exhibited four to five immune subtypes, suggesting a strong link between BIRC5 and tumor immunity (Figure 2).

Furthermore, analysis revealed that BIRC5 expression varied significantly across molecular subtypes in 14 cancer types, including ACC, BRCA, COAD, ESCA, GBM, HNSC, KIRP, LIHC, LUSC, OV, PCPG, PRAD, UCEC, and STAD. Most of these cancers exhibited four to five distinct molecular subtypes, underscoring the strong association between BIRC5 expression and tumor classification (Figure 3). For clarity, the full names of the molecular subtype abbreviations are provided in Supplementary Table S2. These findings suggest that BIRC5 may serve as a valuable biomarker to inform personalized cancer treatment.

We conducted a survival analysis (OS, DSS, and PFI) across multiple cancer types based on BIRC5 expression. High and low BIRC5 expression groups were defined based on the [median expression level across samples]. Results were presented using forest and heat maps (Figure 4A). BIRC5 expression significantly impacted OS, DSS, and PFI in 10 cancers: ACC, KIRC, LGG, LIHC, LUAD, MESO, PAAD, SKCM, UCEC, and UVM (Figures 4B–K). For instance, in ACC, BIRC5 was associated with poor prognosis in PFI (HR = 4.66, 95% CI: 2.45–8.86, p < 0.001), OS (HR = 8.18, 95% CI: 3.72–17.99, p < 0.001), and DSS (HR = 8.30, 95% CI: 3.66–18.82, p < 0.001). Similarly, in KIRC, BIRC5 significantly affected PFI, OS, and DSS (HR = 2.78, 2.37, and 3.59, respectively, all p < 0.001). These findings suggest that BIRC5 is a risk factor that negatively influences prognosis across several cancers.

We evaluated the potential of BIRC5 to distinguish between cancerous and non-cancerous tissues across multiple cancer types using receiver operating characteristic (ROC) curves and area under the curve (AUC) values. Although ROC curves are commonly used to assess diagnostic capability, in this context, they provide insights into the broader relevance of BIRC5 in differentiating cancerous tissues and its potential role in prognosis across various cancer types. BIRC5 demonstrated a high AUC (>0.7) in 24 cancer types, including BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESAD, ESCA, ESCC, GBM, GBMLGG, KIRC, KIRP, LIHC, LUAD, LUSC, OSCC, PAAD, PCPG, PRAD, READ, SARC, STAD, and UCEC (Figure 5). In 18 of these cancers, BIRC5 exhibited particularly strong predictive power (AUC > 0.9), including in BRCA, CESC, CHOL, COAD, ESAD, ESCA, GBM, LIHC, LUAD, and UCEC. These findings suggest that BIRC5 may serve as a valuable biomarker for distinguishing cancerous tissues and potentially predicting cancer progression, supporting its broader relevance in pan-cancer analysis.

Using the STRING database and Cytoscape, we identified the top 20 proteins interacting with BIRC5 (Figure 6A). Gene Ontology (GO) enrichment analysis of these proteins revealed key biological processes, including nuclear division, organelle fission, and chromosome segregation. Cellular components involved the spindle and chromosomal regions, while molecular functions included protein serine kinase and histone kinase activity. KEGG pathway enrichment highlighted processes related to the cell cycle, oocyte meiosis, p53 signaling, and viral infections (Figures 6B, C).

To explore BIRC5’s specific role in LUAD, we identified the top 50 genes associated with BIRC5 in lung cancer datasets (Supplementary Table S3). These genes were cross-referenced with STRING-derived proteins, yielding 11 interacting proteins, including CDC20, KIF2C, and AURKB, all of which were significantly correlated with BIRC5 (Figures 6D–F). Functional enrichment analysis suggested that BIRC5 shares similar roles across various cancers, further supporting its involvement in key cancer pathways (Figures 6G–I).

Previous functional studies have identified a strong association between BIRC5 and immune responses. To further explore this, we examined the relationship between BIRC5 expression levels and the infiltration of 26 immune-related cell types. The results from the heatmap analysis of BIRC5 expression and immune cell infiltration across various cancers indicated that BIRC5 exhibits immunosuppressive effects in many tumor types (Figure 7A). Subsequently, we performed a correlation analysis between BIRC5 expression and the infiltration levels of various immune cells, including T cells, B cells, macrophages, and NK cells. The analysis revealed a general trend where high BIRC5 expression was associated with a reduced infiltration of these immune cell subtypes in most cancers (Figures 7B, C). Focusing on lung cancer as a representative tumor type, we further analyzed the correlation between BIRC5 and immune cell infiltration. The stacked bar graph showed a positive correlation between BIRC5 expression and immune cell infiltration in lung cancer (Figure 7D). To validate these findings, we conducted an additional analysis using the TIMER 2.0 database, which confirmed that BIRC5 expression was significantly positively correlated with several immune cell types, including CD8⁺ T cells (Rho = 0.159), CD4⁺ T cells (Rho = 0.404), gamma delta T cells (Rho = 0.127), follicular helper T cells (Rho = 0.130), NK cells (Rho = 0.116), B cells (Rho = 0.116), macrophage M0 (Rho = 0.250), and macrophage M1 (Rho = 0.352). Only a few immune cell types, such as Tregs (Rho = −0.082) and macrophage M2 (Rho = −0.038), showed a negative correlation with BIRC5 expression (Figures 7E, F). These results are consistent with the pan-cancer analysis, reinforcing the idea that BIRC5 may serve as an effective intervention target in tumor immunotherapy.

We first explored the genomic alterations of BIRC5 in lung cancer using the cBioPortal website and found that BIRC5 genomic changes occurred in 2.4% of cases (Figure 8A). The types of alterations in BIRC5, including amplification, gain, and diploid states, resulted in changes in gene expression, with copy number variations (CNVs) being positively associated with gene expression in some sequencing results (Figures 8B, C). Next, we investigated the role of BIRC5 in patient prognosis and disease progression across various clinical features of lung cancer. High and low expression groups were determined based on the [median expression level of BIRC5 in the cohort]. The results are presented in the form of a forest plot (Figure 8D). BIRC5 was identified as a risk factor that promotes lung cancer progression, particularly in patients with Pathologic T2 stage, N0 stage, M0 stage, and residual tumor status (R0), as well as in patients of all age groups (Figures 8E–J). To determine whether BIRC5 can serve as an effective biomarker for lung cancer prognosis, we used receiver operating characteristic (ROC) curves to evaluate its predictive performance for overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI) in lung cancer patients at 1, 3, and 5 years. The results showed that all area under the curve (AUC) values were greater than 0.5, indicating that BIRC5 is a reliable prognostic predictor at multiple time points. Further analysis of genes co-expressed with BIRC5 in lung cancer revealed eight genes significantly associated with prognosis, as determined by univariate Cox regression analysis. LASSO regression analysis identified nine genes—BIRC5, HJURP, CDK1, PLK1, CDC25C, H2AZ1, KIF23, ANLN, and CIP2A—as significant risk factors impacting the prognosis of lung cancer patients (Figures 8L–N).

We investigated genes co-expressed with BIRC5 in LUAD, identifying those positively or negatively correlated with its expression. The top 30 positively and negatively correlated genes are displayed in the heatmap (Figures 9A, B). A total of 1,233 differentially expressed genes (DEGs) were identified, including 609 upregulated and 624 downregulated genes. Protein-protein interaction (PPI) networks were constructed for both upregulated and downregulated genes, revealing hub genes (Figures 9C, D).

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the DEGs showed that BIRC5 co-expressed differential genes in LUAD were primarily involved in processes such as cilium movement, intermediate filament organization, and keratinization. Cellular components (CC) enriched in these genes included the apical plasma membrane, apical part of the cell, and cornified envelope. Molecular functions (MF) were mainly associated with peptidase inhibitor activity, endopeptidase inhibitor activity, and structural components of the skin epidermis. KEGG pathway enrichment was predominantly linked to bile secretion, cytochrome P450-mediated drug metabolism, and metabolism of xenobiotics by cytochrome P450 (Figure 9E).

Finally, Gene Set Enrichment Analysis (GSEA) was performed to explore the correlation of BIRC5-related DEGs with oncogenic and immunologic signatures, as well as gene ontology. The top five positive/negative correlations are summarized in the figures. Oncogenic signatures were primarily related to “CSR_LATE_UP.V1_UP,” “RPS14_DN.V1_DN,” “NFE2L2. V2” and “HOXA9_DN.V1_DN” (Figures 9F, G). Immunologic signatures were associated with “HOWARD_NK_CELL_INACT_MONOV_INFLUENZA_A_INDONESIA_05_2005_H5N1_AGE_18_49YO_3DY_UP,” “GSE15750_DAY6_VS_DAY10_TRAF6KO_EFF_C8_TCELL_UP,” “GSE15750_DAY6_VS_DAY10_EFF_CD8_TCELL_UP,” “GSE18893_TCONV_VS_TREG_24H_TNF_STIM_UP,” and “GSE36476_CTRL_VS_TSST_ACT_72H_MEMORY_CD4_TCELL_YOUNG_DN” (Figures 9H,I). Gene ontology was primarily enriched in “REACTOME_FORMATION_OF_THE_CORNIFIED_ENVELOPE,” “REACTOME_KERATINIZATION,” “WP_RETINOBLASTOMA_GENE_IN_CANCER,” and “REACTOME_MITOTIC_SPINDLE_CHECKPOINT” (Figures 9J, K). These results suggest a strong association of BIRC5 with oncogenic and immunologic signatures, as well as key gene ontologies, supporting its potential as a tumor biomarker and therapeutic target.

In lung adenocarcinoma (LUAD), BIRC5 protein expression was significantly higher in tumor cell lines compared to normal lung epithelial cells (Figure 10A). Gene overexpression and knockdown assays in the A549 cell line demonstrated that inhibition of BIRC5 expression markedly reduced cell proliferation, migration, and invasion (Figures 10B–J), while its overexpression enhanced these tumorigenic properties. These findings indicate that BIRC5 is critically involved in the progression of LUAD and could serve as a potential therapeutic target.