Pan-cancer analysis and breast cancer validation reveal DEAD-box helicase 23 as a crucial prognostic and immuno-biomarker Authors: Haotian Ma, Xiyidan Aimaiti, Yiyang Wang, Yongxiang Li, Jiawei Zhao, Jiayue Hou, Jiaqi Li, Chenming Guo Published: 2026 ABSTRACT Abnormal RNA metabolism is one of the core mechanisms in tumorigenesis. DDX23, a member of the family of DEAD-box protein, characterized as an ATP-dependent RNA helicase, facilitates cancer progression due to its abnormal expression. However, systematic studies examining DDX23 across the pan-cancer spectrum remain scarce. We utilized databases like TCGA and GEO to thoroughly examine the expression pattern of DDX23 in our study, its prognostic value, diagnostic efficacy, and immune-related properties across multiple cancers. We foresee possible regulatory routes by building a network of protein interactions and performing analysis of functional enrichment. Concurrently, we validate the core biological functions of DDX23 through in vitro experiments. DDX23 demonstrates markedly increased expression across a majority types of cancer, and its overexpression is strongly tied to a negative patient outcome. The level of DDX23 expression is closely linked to tumor mutational burden and the presence of infiltrating immune cells. Functional enrichment analysis indicates that DDX23 primarily participates in pathways related to RNA splicing and cell proliferation. In breast cancer (BRCA) tissue, DDX23 expression is notably higher than in normal breast tissue. In vitro experiments provide additional confirmation that DDX23 substantially boosts breast cancer cells’ ability to proliferate, invade, and migrate. In this investigation, DDX23 has been identified as a crucial biological marker for forecasting survival outcomes and immunological overview in patients with various types of cancer. It clarifies its specific expression pattern and oncogenic function in BRCA, providing crucial evidence for its clinical application in BRCA-targeted therapy and highlighting its translational potential. Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-48258-z. Introduction Cancer represents a substantial threat to human health, marked by increasing rates of occurrence and death, thereby posing a substantial challenge to global public health1. Despite clinical applications such as surgery, chemoradiotherapy, and targeted therapies improving survival outcomes for some patients, the complex pathogenic mechanisms of cancer, significant tumor heterogeneity, and treatment resistance continue to hinder the development of curative solutions2. Tumor progression is not only directly linked to molecular abnormalities within cancer cells, but also co-regulated by multiple factors within the tumor microenvironment (TME), such as immune cell infiltration and changes in the stromal structure3. In recent years, rapid advancements in bioinformatics and the deep integration of multi-omics data have provided powerful tools for identifying core regulatory genes in cancer and screening biomarkers with both diagnostic and therapeutic value. The discovery of these novel markers has injected new momentum into the development of precision oncology. DEAD-box helicase 23 (DDX23), known alternatively as PRP28 or U5-100KD, functions as a crucial ATP-dependent RNA helicases within the DEAD-box protein family4,5. This gene’s improper expression is tightly connected to dysfunction and the development of many types of human cancers. DDX23 is an essential part of the U5 snRNP complex, crucial for precursor mRNA splicing6. DDX23 affects tumor cell proliferation, invasion, migration, and drug resistance by altering various cancer-related pathways, including PI3K/Akt and Notch pathways7,8. DDX23 is localized to the 12q13.12 chromosomal locus, and its protein consists of 820 amino acids9,10. DDX23 shows elevated expression levels in various tumors, including pancreatic7, ovarian11, and lung cancers12, as well as gliomas13, and Elevated levels are commonly connected with an unfavorable prognosis in patients. Despite this, systematic pan-cancer studies on DDX23 remain scarce. Given this gap, this study leveraged existing large-scale cancer multi-omics data to conduct comprehensive bioinformatics analyses. A thorough investigation was carried out on the expression of the DDX23 gene across various types of cancer to assess its potential for diagnosis and prognosis. Functional experiments conducted both in vitro and in vivo demonstrated the significant regulatory function of DDX23 in the progression of breast cancer. Materials and methods A thorough analysis of DDX23 expression across diverse cancer datasets The UCSC database provided harmonized datasets for pan-cancer research, which included information from TCGA. These datasets comprise RNA sequencing expression profiles for 33 cancer types along with clinical data from 18,102 patients. The expression data for the DDX23 gene (ENSG00000174243.10) were collected and converted into log2-transformed TPM format for each sample, then consolidated for analysis. For data analysis and visualization, ggplot2 (3.4.4), stats (4.2.1), and car (3.1.0) were used. Subsequently, Data from GEO databases were utilized to validate the expression of DDX23 mRNA across different cancer cell lines, including GSE42568 (GPL570), GSE6791 (GPL570), GSE83889 (GPL10558), GSE45670 (GPL570), GSE29796 (GPL570), GSE115002 (GPL13497), GSE89076 (GPL16699), and GSE54129 (GPL570). To verify the differences in DDX23 protein levels among various cancer cell lines, We utilized the HPA database to acquire immunohistochemical images of 19 cancerous tissues along with their corresponding healthy tissue counterparts for our analysis14. The analysis of DDX23 diagnostic and prognostic performance The research investigated the correlation between the expression levels of DDX23 and the outcomes for patients, utilizing survival data obtained from the TCGA database. The Kaplan–Meier approach was adopted for correlation assessments15. For the intuitive visualization of our findings, we constructed forest plots and Venn diagrams using the ggplot2 package. Performed Benjamini-Hochberg (BH-FDR) multiple testing correction independently for each survival endpoint. The research utilized the ‘pROC’ software package for ROC curve analysis, emphasizing sensitivity and specificity. AUC, is crucial for evaluating diagnostic performance, and an AUC above 0.9hints that DDX23 possesses strong diagnostic capabilities, an AUC between 0.7 and 0.9 reflects moderate diagnostic capability. To further obtain reliable statistical inference results, SPSS 26.0 software was used to calculate the 95% confidence interval (95%CI) of the AUC values for different cancer types using the DeLong nonparametric test, to eliminate the impact of data distribution assumptions on the results and quantify the stability of diagnostic value; meanwhile, the statistical significance of each AUC value was verified through a null hypothesis test of ‘true AUC = 0.5’. Complete data can be found in Supplementary Table S6. Assessment of genetic variation profiles in DDX23 To investigate the genetic alterations of DDX23, we utilized the cBioPortal database16. We evaluated key genetic features including the frequency of alterations, distinct mutation types, and copy number variations (CNVs). We examined the three-dimensional (3D) structure of the DDX23 protein to thoroughly understand its genetic characteristics. We examined the distribution of CNVs among different types of cancer through utilization of the GSCA database17.Additionally, We explored the possible prognostic importance of CNVs related to DDX23 in cancer patients. A study on the relationship between DDX23 levels and immune indicators The study employed the Sangerbox3.0 platform to thoroughly explore the connection between the expression of the DDX23 gene and several important genomic elements18, Afterward, the ‘ESTIMATE’ tool was employed to quantitatively evaluate the connection between DDX23 mRNA expression and scores. Researchers conducted ss GSEA19, to further support the link between DDX23 mRNA expression and the presence of key immune cells in various cancers, several algorithms from TIMER 2.0’s ‘Immune’ module, including TIMER, EPIC, MCPCOUNTER, QUANTISEQ, and XCELL, were used20. Functional enrichment analysis of DDX23 across various cancer types We leveraged the STRING database to discover 50 proteins with experimental validation that interact with DDX23, afterward, building a PPI network diagram to visualize the interactions21. Additionally, By accessing the database of GEPIA222, we retrieved the top 100 genes that are co-expressed with DDX23 and employed a Venn diagram to determine the overlapping genes between these and the DDX23-binding proteins. Furthermore, we employed the database of TIMER 2.0 to study the association between DDX23 and the genes that coincide. enrichment analysis and result visualization were performed using the “cluster Profiler” and “ggplot2” packages, with bubble plots displaying only the top 5 pathways most highly associated with tumors within each tumor type. Additionally, Cancer SEA database data on DDX23 correlations with various functional states across cancers were visualized as heatmaps23. Based on pan-cancer gene expression and functional phenotype data from TIMER 2.0, representative cancer types were selected to analyze correlations between DDX23 expression levels and functional phenotypes. We categorized DDX23 expression levels to highly and low groups with expression levels based on a 50% threshold. The DESeq2 tool was used to examine the differences in gene expression between the two groups, and all genes along with their log2FoldChange values were included in the GSEA gene set from the MSigDB database was used in our research to emphasize functional differences DDX23 the act of expressing in the pan-cancer cohort. After 10,000 replicates of analysis, spine plots illustrated the top ten ‘response group pathways’ for each category of tumor. Collection of patient tissue samples In this study, cancer tissues and adjacent tissues of 12 BC patients who underwent surgical intervention at the Breast Surgery Department, First Affiliated Hospital of Xinjiang Medical University, during the period of January to December 2025. The patient cohort comprised representatives of various molecular subtypes, providing a more comprehensive clinical context for our findings. All participants adhered to stringent inclusion and exclusion criteria, Patients who hadn’t received radiotherapy, chemotherapy, or hormone therapy gave informed consent approved by the Ethics Committee (No. 230714-08) before surgery. After surgery, Samples were swiftly frozen for analyses. Plasmid construction and transfection To develop the vector for DDX23 knockdown (sh-DDX23), three unique small hairpin RNA (shRNA) sequences that specifically target DDX23 were designed and produced (refer to Supplementary Table S1). These sequences were subsequently cloned into the multiple cloning site of the pLKO.1-puro lentiviral vector, followed by ligation reactions, and the resulting recombinant plasmids were confirmed using Sanger sequencing. The DDX23 overexpression vector was built by adding the full-length DDX23 coding sequence into the pLVX-Puro lentiviral vector, regulated by a suitable promoter. The recombinant vector was confirmed by Sanger sequencing. HEK-293T cells at 80% confluency were transfected with 2 µg of recombinant shRNA or overexpression plasmid using Lipofectamine 3000 (Invitrogen, Shanghai, China) as per the manufacturer’s guidelines. The cells were cultured for a duration of 48 h to facilitate lentivirus production. Subsequently, the supernatant was gathered, subjected to filtration, and then ultracentrifuged at 100,000×g for 2 h at a temperature of 4 °C. Once ultrafiltration was completed, the supernatant was thrown away, and the viral pellet was reconstituted in a complete culture medium. Lentivirus infection and classification Human breast cancer (BC) cell lines, specifically HCC-1937 and MDA-MB-231, were cultured by seeding them into culture plates. Viral transduction was performed when cell confluency reached approximately 80%. Concentrated lentiviral particles were added to the cell cultures, followed by replacement with fresh complete medium at 24 h post-transduction. Cells were then maintained in culture for an additional 48 h. Utilizing the methodologies previously delineated, the cells were allocated into four distinct experimental groups: the control group with an empty vector for DDX23 knockdown, the DDX23 knockdown group (referred to as DDX23-KD), the control group for DDX23 overexpression vector, and the DDX23 overexpression group (identified as DDX23-OE). Fluorescence microscopy was used to assess gene expression 72 h post-transduction. Cells that showed the best viability were cultured for a set period before being collected for further experiments. RNA extraction and qRT-PCR RNA was extracted and analyzed using qRT-PCR. In accordance with established protocols, Trizol (Invitrogen, Shanghai, China) was used to extract total RNA, followed by analysis with the SYBR Green PCR kit (Takara, Kyoto, Japan). Amplification of the target gene was performed with the designated primers from Table S1 in a 20 µL reaction volume. All experiments were performed in triplicate (n = 3). Protein extraction and WB According to established protocols, RIPA lysis buffer with PMSF inhibitor was used to process tissues and cells for total protein extraction, the protein concentration was determined using the BC protein assay kit, with all reagents sourced from Beyotime Institute of Biotechnology in Shanghai, China. Protein samples underwent separation using a 10% SDS-PAGE gel and were subsequently transferred onto a PVDF membrane. The membrane was then blocked and incubated overnight at 4 °C with a primary antibody targeting DDX23 (ab70459, Abcam, Cambridge, UK). β-Actin served as a loading control. Secondary antibodies conjugated with HRP were applied to the membranes for an hour at room temperature, followed by visualization using the ECL chemiluminescence reagent. Signals were captured with a Bio-Rad imaging system. All experiments were performed in triplicate (n = 3). IHC staining A total of 12 paired specimens, consisting of breast cancer tissues and corresponding adjacent para-cancerous tissues, were collected from patients who received surgical intervention at the Department of Breast Surgery, the First Affiliated Hospital of Xinjiang Medical University, for subsequent analysis. Immunohistochemical staining was conducted on tissue microarray sections with a specific primary antibody targeting DDX23 (ab70459, Abcam, USA). The experimental procedure incorporated heat-mediated epitope retrieval, which was implemented using a citrate buffer solution at pH 6.0 (Invitrogen, Shanghai, China). Staining outcomes were assessed independently by two pathologists under a blinded experimental design. Each tissue sample was assigned a composite score, determined by integrating two key parameters: staining intensity and the proportion of positively stained cells. For each set of tumour and matched non-tumour tissue pairs, immunostaining signals were quantified individually, with an H-score allocated to the representative tissue core. Colony formation assay Infect BC cells with recombinant lentivirus, allow them to reach the logarithmic growth phase, count using a hemocytometer, and seed 500 cells per well into 6-well plates. Replace the medium every 3 days and continue culturing for 2 weeks until visible monoclonal colonies form. Take out the 6-well plates, wash them with PBS, and then fix them with 4% paraformaldehyde for a duration of 20 min. After that, rinse twice with PBS, apply 20 min in 0.1% crystal violet, and wash three times with PBS to remove any excess dye. Photograph under a microscope and calculate the colony formation rate. All experiments were performed in triplicate (n = 3). CCK-8 assay Cell growth was evaluated using the CCK-8 kit (Dojindo Molecular Technologies, Japan). cells were plated in 96-well plates and cultured for a period of 5 days. Afterward, CCK-8 reagent was introduced to each well and incubated at 37 °C for 1.5 h. The absorbance at 450 nm was recorded with a Tecan Infinite microplate reader, and the proliferation trends were represented through a dose-response curve. All experiments were performed in triplicate (n = 3). Transwell assay Cells belonging to both the experimental and control cohorts were seeded with the aid of a Transwell assay kit (Corning, New York, USA). To assess cellular migratory and invasive capacities, Transwell chamber assays were implemented in two configurations: one supplemented with Matrigel and the other without this matrix component. For the assay setup, the lower compartment of each Transwell system was filled with 600 µL of medium containing 10% fetal bovine serum (FBS) to act as a chemoattractant, after which the chambers were incubated for 24 h under conditions of 37 °C and 5% CO₂. Residual cells adhering to the upper surface of the chamber membrane were wiped off using a cotton swab. Cells that had migrated to the lower chamber were fixed in 4% paraformaldehyde for 30 min and then stained with 0.1% crystal violet solution (Beyotime, Shanghai, China). Finally, cell enumeration and morphological analysis were conducted under an inverted microscope (Mshot, Guangzhou, China) at a magnification of 200×. All experiments were performed in triplicate (n = 3). Wound scratch assay Cells that were in the logarithmic growth phase were plated into 6-well culture dishes at a density of 1 × 10⁵ cells per well, and then incubated for 24 h to achieve approximately 95% confluence. A straight scratch was made through the cell monolayer with a sterile 200-µL pipette tip, using a ruler as a guide. After two gentle rinses with PBS to eliminate detached cells, the culture medium supplemented with 1% FBS was added to each well. The scratch area was photographed immediately under an inverted microscope to record the 0 h baseline. The plates were then incubated for another 24 h under conditions of 37 °C and 5% CO₂, after which images of the identical regions were captured once more. All experiments were performed in triplicate (n = 3). Subcutaneous xenograft assay Approval for all animal experiments conducted was granted by the Laboratory Animal Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University. All animal experimental methods were performed in accordance with the relevant guidelines and regulations for the care and use of laboratory animals. Fifteen six-week-old BALB/c Nude mice were sourced from Beijing Vital River Laboratory Animal Technology Co., Ltd (Animal Quarantine Certificate No. 11000029312). These mice were randomly assigned into three groups, each consisting of five individuals. Cells in the logarithmic growth phase were trypsinized, washed with PBS, and resuspended to a concentration of 2 × 10⁷ cells/mL. Female nude mice, aged 6–8 weeks, were administered a 0.2 mL subcutaneous injection in the dorsal scapular area. Regularly monitor and record the body weight of the weighed mice. Starting on day 7, tumour growth was monitored every 3 days with digital callipers; the longest (a) and shortest (b) diameters were recorded and tumour volume estimated as V = ½ a b². Mice were euthanised by cervical dislocation on day 32, and tumours were excised, photographed and weighed. Statistical analysis Statistical analyses were conducted through R software (v4.4.1), visualizing the data with the “ggplot2” package. Using the DDX23 mRNA median level, the samples were allocated into HEG and LEG. For the difference in DDX23 expression levels in unpaired samples, we employed the Mann–Whitney U-test to detect the difference while using the Wilcoxon signed-rank test for paired samples. The Spearman correlation coefficient was deployed to ascertain the link between DDX23 expression and immune scores, and immune-associated genes.For in vitro experimental data, comparisons between two groups were analyzed using Student’s t-test, while comparisons among multiple groups were performed using one-way ANOVA followed by Tukey’s posthoc test for multiple comparisons. Data are presented as mean ± standard deviation (SD). p < 0.05 indicated statistical significance: *p < 0.05; **p < 0.01; ***p < 0.001; NS, Not Significant. All in vitro experiments were performed with three biological replicates (n = 3). Results Differential expression analysis of DDX23 in diverse cancer types Utilizing the TCGA database, we initially analyzed the expression levels of DDX23 in both pan-cancer and normal tissue samples. DDX23 expression was notably increased in 14 cancer types: BLCA, BRCA, CHOL, COAD, ESCA, GBM, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, READ, and STAD. In contrast, expression in KICH was lower than in the corresponding normal tissue (Fig. 1A). Fig. 1illustrates the analysis of DDX23 mRNA expression profiles across 33 distinct cancer types. (A) This examination evaluates the mRNA expression levels of DDX23 in tumor samples from TCGA in comparison to normal tissue samples. (B) A comparative analysis of DDX23 mRNA expression levels between tumor samples from TCGA and their corresponding normal tissue counterparts. Analysis of GEO datasets reveals differential expression of DDX23 in various cancer types. (C) BRCA, (D) CESC, (E) COAD, (F) ESCA, (G) LGG, (H) LUAD, (I) READ, and (J) STAD datasets. (*p < 0.05, **p < 0.01, ***p < 0.001, NS: not significant). illustrates the analysis of DDX23 mRNA expression profiles across 33 distinct cancer types. (A) This examination evaluates the mRNA expression levels of DDX23 in tumor samples from TCGA in comparison to normal tissue samples. (B) A comparative analysis of DDX23 mRNA expression levels between tumor samples from TCGA and their corresponding normal tissue counterparts. Analysis of GEO datasets reveals differential expression of DDX23 in various cancer types. (C) BRCA, (D) CESC, (E) COAD, (F) ESCA, (G) LGG, (H) LUAD, (I) READ, and (J) STAD datasets. (*p < 0.05, **p < 0.01, ***p < 0.001, NS: not significant). We investigated the mRNA expression levels of the DDX23 gene in paired samples across 23 distinct cancer types. The findings of this study indicated that DDX23 expression patterns were well-aligned with pan-cancer expression data retrieved from the TCGA database (Fig. 1B). In order to strengthen the reliability of our results, we additionally confirmed our findings by utilizing datasets obtained from the GEO database. In BRCA, COAD, CESC, ESCA, LUAD, LGG, READ, and STAD, DDX23 mRNA levels showed a statistically significant upregulation versus the corresponding normal tissue samples (Fig. 1C-J). On the whole, results indicate that DDX23 expression is in most cancers significantly upregulated than regular tissue. The levels of the DDX23 protein were evaluated through the analysis of immunohistochemical images from 19 different cancer types in the HPA dataset. Figure S1 demonstrates a significant increase in DDX23 protein levels across 18 different cancer types when compared to normal tissue samples. In renal cell carcinoma (RCC), DDX23 protein levels are significantly lower than in normal tissues, aligning with earlier findings on mRNA expression. Taken together, most malignancies exhibited a significant upregulation in DDX23 mRNA levels, and such an expression pattern was corroborated by protein-related analytical results. Diagnostic and prognostic value of DDX23 Kaplan-Meier curves showed that elevated DDX23 expression is linked to poor prognosis in most cancer types. excluding READ, UCS, and KIRC, significantly impacting survival-related metrics like OS, PFS, and DSS(Figure S2). A univariate Cox regression analysis showed that higher levels of DDX23 expression are linked to ACC, BRCA, KICH, LAML, LGG, LIHC, and LUAD. In comparison, patients with READ showed a positive impact of high DDX23 expression on OS (Fig. 2A). Hence, elevated levels of DDX23 are seen as a risk factor impacting OS in the majority of patients. Fig. 2shows the correlation between DDX23 expression and three cancer prognostic indicators. (A-C) Correlation of DDX23 expression with three cancer prognostic indicators. (D) Venn diagram illustrating the intersection of three cancer prognostic indicators. shows the correlation between DDX23 expression and three cancer prognostic indicators. (A-C) Correlation of DDX23 expression with three cancer prognostic indicators. (D) Venn diagram illustrating the intersection of three cancer prognostic indicators. Higher DDX23 expression was associated with shorter DSS in three cancer types. which included ACC, LGG, and LIHC. Conversely, higher DDX23 expression was associated with better DSS in KIRC patients (Fig. 2B). Furthermore, PFS analysis indicates that higher DDX23 expression correlates with decreased patient PFS in ACC, CESC, and LIHC, while an opposite pattern is observed in UCS (Fig. 2C). The analysis using a Venn diagram revealed that the expression levels of DDX23 were linked to OS, DSS, and PFS in patients with both ACC and LIHC, indicating that DDX23 might be a crucial molecular regulator of prognosis in these cancer types (Fig. 2D). Univariate Cox regression combined with BH-FDR correction revealed a borderline significant trend (FDR = 0.10–0.20, Supplementary Table S4) between DDX23 expression and poor prognosis in seven cancer types including LAML, BRCA, LIHC and ACC, with all HRs > 1, suggesting a consistent association. The findings indicate that elevated DDX23 expression is associated with poor prognosis across various cancers. this evidence suggests that DDX23 might serve as an potential prognostic indicator for specific types of cancer. The diagnostic performance of this gene signature was evaluated through ROC curves. We divided it into three levels according to the AUC values: AUC values between 1.0 and 0.9 indicate a high level of diagnostic accuracy, while AUC values from 0.9 to 0.7 suggest moderate diagnostic accuracy. In contrast, an AUC that falls between 0.7 and 0.5 signifies low diagnostic accuracy. Figure 3 reveals that the DDX23 gene exhibits significant diagnostic accuracy in CHOL (1.000), GBM (0.908), and SARC (0.951) (Fig. 3A). It also exhibited moderate diagnostic precision across 11 cancer types, including BRCA (0.730), CESC (0.729), COAD (0.825), ESCA (0.886), HNSC (0.834), KICH (0.767), LIHC (0.888), LUAD (0.741), LUSC (0.749), READ (0.802), and STAD (0.891) (Fig. 3B). Notably, this gene has demonstrated an AUC value of 1.0 in CHOL, indicating superior diagnostic discrimination capability. DDX23 has demonstrated high diagnostic efficacy (AUC > 0.7) across up to 14 cancer types, showing its potential as a universal cancer diagnostic marker. Fig. 3ROC curves for DDX23 expression across pan-cancer types. DDX23 expression exhibited significant diagnostic value in GBM, CHOL and SARC, while its diagnostic utility was moderate in BRCA, COAD, CESC, ESCA, HNSC, KICH, LUSC, LUAD, LIHC, READ, and STAD. ROC curves for DDX23 expression across pan-cancer types. DDX23 expression exhibited significant diagnostic value in GBM, CHOL and SARC, while its diagnostic utility was moderate in BRCA, COAD, CESC, ESCA, HNSC, KICH, LUSC, LUAD, LIHC, READ, and STAD. In conclusion, in several cancers, DDX23 is significantly related to negative outcomes and demonstrates remarkable diagnostic potential, making it a promising dual biomarker for both prognostic and diagnostic purposes. The characteristics of genetic variations in DDX23 To appraise the mutation status of DDX23 across multiple cancers, we conducted an in-depth analysis using the cBioPortal database24. Our findings disclosed that DDX23 alterations appeared in 1.7% (183/10,953) of the included tumor samples. Regarding mutation types, DDX23 genetic alterations predominantly comprised missense mutations, alongside amplification, deep deletions, truncating mutations, structural variants, and splice mutations (Fig. 4A). Further, we profiled DDX23 mutation frequencies across tumor types, revealing that DLBC (6.25%), UCEC (5.1%), and SKCM (4.07%) accounted for the top three mutation rates. In-depth analysis showed amplification as the predominant alteration in DLBC, while missense mutations constituted the core mutation type in other tumors (Fig. 4B).To comprehensively understand the DDX23 mutation landscape across protein domains in different cancer types, our study identified 155 mutation sites spanning amino acids 0 to 820. Among these, 133 were missense mutations (markedly higher than the 12 truncating mutations and 4 splicing mutations). R11S/C was the most frequent mutation site within the DDX23 domain, occurring in one SKCM patient (Fig. 4C). Figure 4D depicts the mutation status in both the three-dimensional structure and sequence of the DDX23 protein. We explored the relationship between changes in the CNV of the DDX23 gene and survival outcomes. focusing on DSS, OS, and PFS across different types of cancer (Fig. 4E). The study identified DDX23 CNV as a predictive risk factor influencing outcomes for patients with ACC, BRCA, KICH, KIRC, KIRP, LUAD, and UCEC. DDX23 CNV frequently occurs in different malignant tumors and is notably associated with unfavorable prognosis in certain cancer types. This suggests that DDX23 may contribute to cancer progression and function as an important prognostic biomarker. Fig. 4shows the DDX23 gene mutation profile in different cancers. (A) Variations in DDX23 expression observed in various tumor types. Mutation distribution and localization map for the DDX23 gene. Three-dimensional visualization of critical mutations. Effect of DDX23 copy number variation on overall cancer prognosis. shows the DDX23 gene mutation profile in different cancers. (A) Variations in DDX23 expression observed in various tumor types. Mutation distribution and localization map for the DDX23 gene. Three-dimensional visualization of critical mutations. Effect of DDX23 copy number variation on overall cancer prognosis. How DDX23 expression is connected to markers associated with immunity Recognizing the significant impact of genomic diversity on tumor prognosis and treatment, alongside the rise of genomic markers like TMB, MSI, and NEO in precision medicine, this research thoroughly evaluated the connection between DDX23 and six genomic markers in patients with various types of cancer25. The aim is to explore its potential role in regulating tumor heterogeneity. We found that DDX23 expression positively correlations with TMB in BLCA, LUAD, STAD, SARC, LGG, and READ, while negatively correlations with TMB in PRAD and THCA (Fig. 5A). It shows positive correlations with MSI in BLCA, LUAD, CESC, LUSC, ACC, and SARC, and inverse relationships with MSI in BRCA, THCA, HNSC, and DLBC (Fig. 5B). DDX23 shows negative association with NEO only in BRCA and positive association with MATH only in ACC, LUAD, and ESCA (Fig. 5C-D). Moreover, DDX23 expression exhibits significant association with HRD across seven cancer types: exhibits a positive correlation with HRD in BLCA, SARC, LUAD, LIHC, and LAML, while a negative correlation is evident in BRCA and THYM (Fig. 5E).DDX23 demonstrated notable positive correlations with LOH in LIHC, LGG, UVM, PCPG, SARC, LAML, BLCA, LUAD, and LUSC, whereas it showed negative correlations in BRCA, KIRC, CESC, and TGCT (Fig. 5F). Fig. 5presents a radar chart illustrating the distribution of tumor characteristics associated with DDX23 across various tumor types. Linkage of DDX23 expression to (A) TMB, (B) MSI, (C) NEO, (D) MATH, (E) HRD, and (F) LOH. (*p < 0.05, **p < 0.01, ***p < 0.001). presents a radar chart illustrating the distribution of tumor characteristics associated with DDX23 across various tumor types. Linkage of DDX23 expression to (A) TMB, (B) MSI, (C) NEO, (D) MATH, (E) HRD, and (F) LOH. (*p < 0.05, **p < 0.01, ***p < 0.001). The ESTIMATE algorithm was utilized to investigate the connection between DDX23 expression and immune-related scores within the TME. The research revealed an inverse relationship between DDX23 expression and Stromal, Immune, and ESTIMATE Scores in BRCA, CESC, ESCA, GBM, LUAD, LGG, LUSC, SARC, THCA, and UCEC (p < 0.05; Fig. 6A), emphasizing the relationship between DDX23 expression and indicators related to the immune system. Immune cells that penetrate tumors have a substantial impact on the TME and the advancement of cancer26. We employed the ss GSEA method to investigate the relationship between DDX23 expression and the infiltration levels of 24 different types of immune cells. Heatmap results revealed: For most cancer types (except CHOL and UCS), a statistically significant association existed between the levels of immune cell infiltration and the expression of DDX23. DDX23 demonstrated an inverse relationship with B cells, iDCs, cytotoxic a statistically significant association existed between the extent of immune cell infiltration and DDX23 expression profiles. Fig. 6DDX23 level and its relationship with immunity. (A) Correlation with tumor microenvironment score. (B) Immune infiltration correlation based on ssGSEA. (C) TIMER2.0 analysis showing association between DDX23 and CAF infiltration. (*p < 0.05, **p < 0.01, ***p < 0.001). DDX23 level and its relationship with immunity. (A) Correlation with tumor microenvironment score. (B) Immune infiltration correlation based on ssGSEA. (C) TIMER2.0 analysis showing association between DDX23 and CAF infiltration. (*p < 0.05, **p < 0.01, ***p < 0.001). lymphocytes, mast cells, neutrophils, and pDCs, while exhibiting a direct relationship with pro-tumor effector T cells (Th2), Tcm, and T helper cells (p < 0.05; Fig. 6B). Further validation of these findings was conducted using multiple algorithms including TIMER, EPIC, and MCP counter. Our investigation indicated a notable association between the expression of DDX23 in most tumors and the existence of immune cells that facilitate tumor progression, including MDSCs, Tregs, as well as pro-tumor subtypes of CD4 + T cells and neutrophils. (p < 0.05; Fig. 6C). Functional enrichment analysis of DDX23 in pan-cancer To determine if DDX23 functions independently or in conjunction with other genes in regulating biological processes during tumorigenesis and progression, we aimed to establish its potential regulatory network. We obtained 50 proteins that interact with DDX23 by querying the STRING database for the purpose of building a PPI network diagram, as illustrated in Fig. 7A. We used the GEPIA2 database to identify the top 100 genes co-expressed with DDX23, as shown in Table S2. Third, we performed a Venn diagram intersection analysis, which revealed SNRNP200 as a common gene (Fig. 7B). Subsequent analyses demonstrated a marked positive association of the expression levels of DDX23 with SNRNP200 (R = 0.71; Fig. 7C), suggestive of potential synergistic interactions between these genes in molecular functions or regulatory pathways. Following this, we implemented GO/KEGG enrichment analysis (Table S3), identifying a total of 245 enrichments: comprising 166 biological processes (BP), 53 cellular components (CC), 23 molecular functions (MF), and 3 KEGG pathways. Within cancer-associated GO-BP entries (Fig. 7D), DDX23 was significantly linked to processes including mRNA processing, RNA splicing, and mRNA splicing via the spliceosome, specifically involving transesterification reactions with bulged adenosine as the nucleophile. GO-CC analysis reveals enrichment in various complexes, including spliceosomal, U2-type spliceosomal, spliceosomal snRNP, small nuclear ribonucleoprotein, and Sm-like protein family complexes (Fig. 7E). The GO-MF analysis indicates that DDX23 may bind to snRNA, act as a transcription coregulator, and display RNA catalytic activity (Fig. 7F). KEGG analysis suggests DDX23 may function by mediating pathways including the spliceosome, RNA degradation, and the mRNA surveillance pathway (Fig. 7G)27,28. Fig. 7depicts the DDX23-related protein interaction network and functional enrichment analysis. (A) DDX23 PPI network diagram. (B) Intersection analysis of DDX23 interacting proteins in GEPIA2 and PPI databases. (C) Expression correlation scatter plot of SNRNP200 and DDX23. (D-G) Functional enrichment analysis of DDX23 interacting proteins was conducted, encompassing BP, CC, MF, and KEGG pathways. (H) Heatmap depicting the correlation between DDX23 expression and tumor characteristics across various tumors. (I) Heatmap illustrating the correlation between DDX23 expression and tumor marker pathways across different tumor types. depicts the DDX23-related protein interaction network and functional enrichment analysis. (A) DDX23 PPI network diagram. (B) Intersection analysis of DDX23 interacting proteins in GEPIA2 and PPI databases. (C) Expression correlation scatter plot of SNRNP200 and DDX23. (D-G) Functional enrichment analysis of DDX23 interacting proteins was conducted, encompassing BP, CC, MF, and KEGG pathways. (H) Heatmap depicting the correlation between DDX23 expression and tumor characteristics across various tumors. (I) Heatmap illustrating the correlation between DDX23 expression and tumor marker pathways across different tumor types. Subsequently, Our study focused on the link between DDX23 levels and the functional characteristics of cancer cells in multiple cancer types, utilizing the Cancer SEA database. The study found that DDX23 expression in BRCA was positively linked to DNA repair, proliferation, and DNA damage processes (p < 0.05). In PAAD, it was associated with angiogenesis (p < 0.05). In LAML, DDX23 correlated with cell cycle and DNA repair (p < 0.05), while in LGG, it showed positive correlations with stemness and cell cycle processes (p < 0.001; Fig. 7H). Analysis of correlations indicated that DDX23 demonstrates a negative relationship with angiogenesis in AML, while exhibiting a positive association with cell cycle progression and stem cell characteristics in HGG. In PC, it positively correlates with angiogenesis and invasion, but negatively with EMT. In the case of OV, DDX23 has a negative correlation with invasion. Conversely, in RB, it positively correlates with angiogenesis and differentiation, but negatively affects the cell cycle and DNA damage. Finally, in UM, it is negatively correlated with both apoptosis and DNA damage. These findings imply that the regulatory functions of this gene may display specific patterns that vary across different tumor types (Fig. 7I). To define the relevant pathways of the DDX23 gene in six cancers—ACC, BRCA, LGG, LIHC, LUAD, and READ—this study is based on GSEA using the Reactome pathway database. Figure 8A-F hows that genes with a positive association to DDX23 are primarily enriched in sumoylation-related pathways, DNA damage and repair (including diseases of DNA repair and DNA damage bypass), TP53-related pathways (like regulation of TP53 activity via phosphorylation), and pathways related to cell cycle regulation and aging. Pathways related to B-cell immune activation and the complement system were primarily enriched with genes that showed an inverse correlation to DDX23 expression (Figure S4). This encompasses immune regulatory interactions between lymphocytes and non-lymphocytes, B cell receptor (BCR) signaling pathways, second messenger production post-antigen-activated BCR, CD22-mediated BCR signaling regulation, the complement cascade, its initial activation, and the generation of C4 and C2 activators in the classical complement pathway. It encompasses various pathways associated with energy and material metabolism, including fatty acid and amino acid metabolism, as well as their derivatives. This indicates that a higher expression of DDX23 could be potentially linked to the reduction of B-cell immune responses and the activity of the complement system. Fig. 8This study presents the GSEA of DDX23 functional enrichment across six distinct types of cancer. In ACC, BRCA, LGG, LIHC, LUAD, and READ, the top ten pathways showed a positive correlation with DDX23 expression. This study presents the GSEA of DDX23 functional enrichment across six distinct types of cancer. In ACC, BRCA, LGG, LIHC, LUAD, and READ, the top ten pathways showed a positive correlation with DDX23 expression. Multidimensional analysis based on single-cell transcriptomic data reveals significant cancer-specific differences in single-cell heterogeneity, gene expression profiles, and functional pathway enrichment patterns across 18 cancer types (Fig. 8; Figure S3). Most cancers exhibit corresponding associations between cell clusters and characteristic genes/functional pathways, yet distinct core pathway biases emerge across different malignancies. BRCA showed a notably higher presence of immune cell subpopulations and enhanced enrichment of immune-related pathways relative to other cancers, indicating a more distinct immune microenvironment. LIHC was characterized by high enrichment of metabolic pathways, reflecting the metabolic dysregulation driving hepatocellular carcinoma. In contrast to other cancer types, both UVM and BLCA display limited diversity in cellular subpopulations, with overall low gene expression levels in each subset and no notable tendency toward subtype-specific high expression. these findings elucidate the fundamental regulatory distinctions among various cancer types at the single-cell level, guiding future research on cancer-specific mechanisms. Examination of DDX23 expression levels and their prognostic significance in BC tissues, along with the verification of the efficacy of DDX23 knockdown and overexpression in BC cells The evaluation through qRT-PCR and Western blot revealed that, in comparison to adjacent normal tissues, there is a marked overexpression of DDX23 in BC tissues (Fig. 9A-C). Additionally, an immunohistochemical assessment of samples from 12 BC patients demonstrated increased levels of DDX23 in cancerous tissues relative to neighboring non-cancerous tissues (Fig. 9D-F). Analyses of DDX23 expression via Western blot and qRT-PCR across eight BC cell lines pinpointed HCC-1937 and MDA-MB-231 cells as exhibiting the highest levels of both mRNA and protein (Fig. 9G-H). Based on the results of the above systematic screening, we selected these two TNBC cell lines to conduct subsequent functional studies, which provide an ideal cellular model for further investigating the role of DDX23 in breast cancer progression. Three independent shRNA sequences targeting DDX23 were subsequently transfected into these two lines; qRT-PCR screening showed that shDDX23-KD-2 achieved the most efficient knock-down (p < 0.001; Fig. 9I-J). Thus, shDDX23-KD-2 was selected for all subsequent loss-of-function experiments. To explore DDX23’s role in BC progression, BC cell lines with DDX23 knockdown (DDX23-KD) and overexpression (DDX23-OE) were created via lentiviral transfection, with transfection efficiency confirmed by qRT-PCR and WB (Fig. 9K-M). Fig. 9Expression analysis of DDX23 in BC tissues and validation of DDX23-KD and DDX23-OE efficiency in BC cells. (A-C) qRT-PCR and WB: DDX23 mRNA and protein levels in cancer and adjacent tissues of 12 BC patients. (D)IHC staining: Expression of DDX23 in 12 pairs of breast cancer and adjacent tissues (E)IHC staining: Representative images of DDX23 expression in BC and adjacent tissues. (F) H-score analysis of DDX23 expression in BC and adjacent tissues. (G-H) WB: DDX23 protein expression across eight breast-cancer cell lines. (I-J) qRT-PCR: The mRNA expression levels of three different KD-DDX23 primer sequences. (K-L) qRT-PCR: Verification of DDX23-KD and DDX23-OE efficiency in MDA-MB-231 and HCC-1937 cells. (M) WB: Verification of DDX23-KD and DDX23-OE efficiency inMDA-MB-231 and HCC-1937 cells. (*p < 0.05, **p < 0.01, ***p < 0.001). Expression analysis of DDX23 in BC tissues and validation of DDX23-KD and DDX23-OE efficiency in BC cells. (A-C) qRT-PCR and WB: DDX23 mRNA and protein levels in cancer and adjacent tissues of 12 BC patients. (D)IHC staining: Expression of DDX23 in 12 pairs of breast cancer and adjacent tissues (E)IHC staining: Representative images of DDX23 expression in BC and adjacent tissues. (F) H-score analysis of DDX23 expression in BC and adjacent tissues. (G-H) WB: DDX23 protein expression across eight breast-cancer cell lines. (I-J) qRT-PCR: The mRNA expression levels of three different KD-DDX23 primer sequences. (K-L) qRT-PCR: Verification of DDX23-KD and DDX23-OE efficiency in MDA-MB-231 and HCC-1937 cells. (M) WB: Verification of DDX23-KD and DDX23-OE efficiency inMDA-MB-231 and HCC-1937 cells. (*p < 0.05, **p < 0.01, ***p < 0.001). According to the CCK-8 assay results, the downregulation of DDX23 (DDX23-KD) considerably inhibited the proliferation of BC cell lines, whereas the overexpression of DDX23 (DDX23-OE) significantly stimulated cell growth (Fig. 10A-B). The colony formation assay results further validated that the knockdown of DDX23 markedly reduced the proliferation of cell lines, in comparison to DDX23 overexpression (DDX23-OE). which enhanced cell proliferation (Fig. 10C). Transwell migration experiments showed that reducing DDX23 decreased the migration of MDA-MB-231 and HCC-1937 cells, while DDX23 overexpression augmented their migration capabilities. Additionally, Transwell invasion studies demonstrated that the decrease in DDX23 levels notably diminished the invasiveness in BC cell lines, whereas increased DDX23 expression facilitated enhanced cell invasion (Fig. 10D-E). Wound-healing assays showed that reducing DDX23 significantly decreased cell migration in BC cell lines. whereas DDX23 overexpression (DDX23-OE) enhanced cell migration (Fig. 10F-I). Subcutaneous xenograft experiments in nude mice demonstrated that the knockdown of DDX23 resulted in the inhibition of proliferation in BC cell lines. while DDX23 overexpression enhanced it. Additionally, we monitored the body weight, tumor growth volume, and weight of the tumor throughout the study (Fig. 10J-M). These findings suggest that DDX23 overexpression enhances BC cell proliferation, migration, and invasion, whereas its knockdown produces contrary effects. Fig. 10Effects of DDX23-KD and DDX23-OE on proliferation, migration, and invasion of BC cells. (A and B) CCK-8 assay: Proliferation changes of BC cells after DDX23-KD and DDX23-OE. (C) Colony formation assay: Proliferation changes of cells after DDX23-KD and DDX23-OE. (D) Transwell migration assay: Migration changes of BC cell lines after DDX23-KD and DDX23-OE. (E) Transwell invasion assay: Invasion changes of BC cell lines after DDX23-KD and DDX23-OE. (F-I) Wound-healing assay: Migration changes of BC cell lines after DDX23-KD and DDX23-OE at 0 h and 24 h. (J) Subcutaneous xenograft assay: Mouse body weight curves for HCC-1937 cells following DDX23 knock-down (DDX23-KD), over-expression (DDX23-OE), and negative control (DDX23-NC). (K-M) Subcutaneous xenograft assays: Proliferative capacity of HCC-1937 cells following DDX23 knock-down (DDX23-KD), over-expression (DDX23-OE), and negative control (DDX23-NC). (*p < 0.05, **p < 0.01, ***p < 0.001). Effects of DDX23-KD and DDX23-OE on proliferation, migration, and invasion of BC cells. (A and B) CCK-8 assay: Proliferation changes of BC cells after DDX23-KD and DDX23-OE. (C) Colony formation assay: Proliferation changes of cells after DDX23-KD and DDX23-OE. (D) Transwell migration assay: Migration changes of BC cell lines after DDX23-KD and DDX23-OE. (E) Transwell invasion assay: Invasion changes of BC cell lines after DDX23-KD and DDX23-OE. (F-I) Wound-healing assay: Migration changes of BC cell lines after DDX23-KD and DDX23-OE at 0 h and 24 h. (J) Subcutaneous xenograft assay: Mouse body weight curves for HCC-1937 cells following DDX23 knock-down (DDX23-KD), over-expression (DDX23-OE), and negative control (DDX23-NC). (K-M) Subcutaneous xenograft assays: Proliferative capacity of HCC-1937 cells following DDX23 knock-down (DDX23-KD), over-expression (DDX23-OE), and negative control (DDX23-NC). (*p < 0.05, **p < 0.01, ***p < 0.001). Discussion DEAD-box RNA helicases, marked by the conserved Asp-Glu-Ala-Asp (D-E-A-D) motif, are essential in multiple RNA processing functions29. DDX23, a DEAD-box RNA helicase, plays a crucial role in ribosome assembly and pre-mRNA splicing30. Using multiple datasets, including TCGA, UALCAN, and GSEA, we reveal the molecular characteristics of DDX23 across 33 tumor types from diverse perspectives. This analysis encompasses expression patterns, diagnostic and prognostic value, and biological functions. In the current research, a significant overexpression of DDX23 was observed in the majority of cancer types and was strongly linked to unfavorable patient outcomes31. This finding was particularly evident across 10 TCGA cancer categories, including BRCA, CHOL, COAD, LIHC, and STAD. Protein levels showed consistent expression trends with mRNA expression across 18 cancer tissues, with low expression observed only in a few cancer types, such as KICH. This study corroborates previous findings that DDX23 is overexpressed in LIHC12, LGG29, and PAAD7 tissues. Earlier research demonstrated a higher expression of DDX23 in ovarian cancer (OV). However, our thorough examination of the TCGA datasets shows that the expression levels of DDX23 are, in fact, greater in the surrounding normal tissues compared to those in ovarian cancer tissues. A follow-up validation analysis utilizing the HPA database, however, indicated that the protein levels of DDX23 in ovarian cancer tissues exceed those found in nearby normal tissues. We hypothesize that the differences observed in these dataset analyses could be attributed to a limited sample size or variability among the data sources. Based on existing research evidence and the findings of this study, DDX23 is highly expressed in multiple tumor tissues. This suggests it may be a conserved pan-cancer oncogene in tumorigenesis, highlighting the need for more research into its role in tumor development. While DDX23 expression correlates with prognosis in multiple cancer patients, it appears to be strongly associated with outcomes in ACC and LIHC. Our analysis demonstrates that high DDX23 expression simultaneously impacts three survival metrics in both cancers, establishing it as a core molecular biomarker for prognostic assessment. The study conducted by Yang et al. demonstrates that DDX23 exhibits a high expression level in HCC and is associated with later stages of the disease. Our research reaffirms the notable linkage between DDX23 expression levels and OS, PFS, and DSS in patients diagnosed with LIHC. Previous studies indicate that high DDX23 expression is linked to malignancy, including proliferation and invasion, in OV, PAAD, and GBM, resulting in unfavorable patient outcomes. Experimental assays demonstrated that inhibiting DDX23 considerably reduced cell proliferation and invasion in vitro, as well as tumor growth in vivo. A recent study conducted by Lu and colleagues revealed a significant association between elevated DDX23 levels and poor prognosis in individuals diagnosed with upper urinary tract urothelial carcinoma (UTUC)32. HDAC1 and DDX23 are overexpressed in NSCLC tissues, have oncogenic roles, and are indicators of poor prognosis in NSCLC patients33. While we note these differences, we were unable to validate their prognostic value, and further clinical validation is required. In the present study, pan-cancer survival analysis showed no strict statistical significance after BH-FDR correction, a common finding in pan-cancer research. Nevertheless, these trend results were highly consistent with the oncogenic function of DDX23 confirmed by functional experiments, implying its carcinogenic role in specific tumors. Early cancer detection holds critical significance in clinical practice, enabling early disease identification and intervention to significantly improve patient prognosis34. Therefore, we utilized the AUC of the ROC curve to validate DDX23’s performance in multi-cancer diagnosis. DDX23 demonstrated exceptionally high diagnostic efficacy in CHOL, GBM, and SARC (AUC > 0.9), notably achieving an AUC of 1.0 in CHOL. The research indicated a moderate diagnostic utility, featuring an AUC ranging from 0.7 to 0.9 for 11 aggregated cancer types, and an AUC exceeding 0.7 for 14 combined cancers, Indicating its viability as a dual molecular signature for cancer diagnosis and clinical prognosis monitoring. However, given that DDX23 is not a secreted protein, its translational potential as a systemic therapeutic target requires further evaluation. Future studies should aim to create more accurate and sensitive detection techniques to inform clinical translation strategies involving DDX23. Dysregulated splicing factor expression is closely tied to tumor development and cancer progression. RNA helicases constitute a crucial subfamily within the splicing factor family35. DDX23, a key DEAD-box RNA helicase, has been linked to R-loop accumulation and genomic instability, marked by double-strand breaks (DSBs), the most severe chromosomal damage in mammalian cells36. Mutations in DDX23 and homozygous deletions have been observed in multiple cancer types, like adenoid cystic carcinoma (ACC)37. These findings indicate that deleting DDX23 might lead to genomic instability. The genetic alteration patterns observed are consistent with our findings, suggesting its viability as a therapeutic target. Furthermore, we observed that genetic alterations in DDX23 predominantly consist of missense mutations, alongside copy number variations such as amplification and deep deletion. DDX23 missense mutations have been linked to a syndrome with unusual neurodevelopmental features29. This study reveals that DDX23 CNVs correlate with patient prognosis across multiple cancers, particularly evident in KICH, KIRC, and KIRP. Future studies ought to concentrate on clarifying the molecular processes through which alterations in DDX23 affect the initiation and progression of cancer. Genomic changes like TMB, MSI, and NEO are crucial for determining ICI effectiveness38. This study explored the association of DDX23 with TMB, MSI, and NEO, identifying positive correlations in BLCA, LUAD, and SARC. Furthermore, indicators such as MATH, HRD, and LOH serve as core markers reflecting genomic heterogeneity, with their expression profiles being closely associated with patient prognosis and the effectiveness of immunotherapy39. This study further confirms that DDX23 exhibits significant correlations with these three indicators across multiple cancers, with such correlations also demonstrating cancer-specific patterns. These findings suggest that DDX23 may participate in tumor evolution by regulating genomic heterogeneity. Tumors arise within heterogeneous tissue microenvironments that promote malignant transformation, metastasis, and therapeutic resistance40. A key feature of the tumor microenvironment is the heterogeneous population of CAFs and myeloid cells41. Elevated DDX23 expression is linked to reduced antitumor immune cells, in various cancers and is negatively associated with immune Scores in most tumor microenvironments. Th1 and Th2 cells—maintaining immune homeostasis through the secretion of their respective characteristic cytokine profiles. Disruption of this balance, especially through excessive Th2 cell activation, is frequently linked to malignant tumor development and progression42. Our analysis indicates that the expression level of DDX23 exhibits a much more robust correlation with Th2 cells in comparison to Th1 cells. This implies that in tumors characterized by elevated DDX23 expression, the immune response of the host could transition from an initially anti-tumor environment to a microenvironment that fosters tumor growth. DDX23 is essential for pre-mRNA splicing. The study identifies a notable positive correlation between DDX23 and SNRNP200. SNRNP200, a crucial spliceosome complex component, is linked to tumor progression when abnormally expressed43, indicating that it and other proteins might collaboratively influence tumorigenesis via RNA splicing regulation. thorough analysis using the Cancer SEA database demonstrated a significant correlation between DDX23 expression levels and pancreatic cancer angiogenesis. This discovery reinforces previous pancreatic cancer research and reveals DDX23’s specific role in tumors. Our findings, consistent with previous research, show that DDX23 is crucial for genomic stability, RNA translation, and splicing due to its roles in RNA binding and transcriptional regulation44. Additionally, research has highlighted the importance of DDX23 in MAPK activation, cell cycle progression, EMT, and proteins associated with migration. GSEA confirmed the activation of pathways that promote tumors while also indicating a reduction in immune-related pathways within the group marked by elevated levels of DDX23 expression. This suggests that DDX23 may reshape tumor cell biology by regulating RNA splicing while simultaneously inhibiting anti-tumor immunity. We assessed DDX23 expression in paired BRCA samples with clinical tissue specimens in this study. The findings showed that malignant tissues had notably higher mRNA and protein levels than the adjacent normal tissues. Additionally, we established DDX23 knockdown and overexpression BC cell lines. After validating transfection efficiency via qRT-PCR and WB, we conducted a series of functional experiments. Our findings indicate that DDX23 overexpression enhances proliferation, migration, and invasion in BC cells. This finding aligns with DDX23’s oncogenic properties observed in pan-cancer analyses. However, it should be noted that our functional experiments were only performed in TNBC cell lines with the highest DDX23 expression. Although this selection was based on systematic screening, the generalizability of our findings to other breast cancer subtypes remains to be further verified. At the same time, due to the limitations of current time and experimental resources, this study has not yet conducted rescue experiments using shRNA-resistant DDX23 constructs. Relevant validation will be prioritized in future research. Conclusion This research represents an initial bioinformatic investigation of DDX23 within a comprehensive pan-cancer group. Our results find that DDX23 could play a significant role in tumorigenesis and advancement by influencing RNA splicing and inhibiting antitumor immune responses. To further substantiate these results, we conducted expression analyses and functional validation experiments mainly in BRCA, providing further evidence of the essential function of DDX23 in tumor progression. Notably, the pan-cancer bioinformatic results are preliminary, and the generality of DDX23 in other cancer types needs to be further verified. This study highlights DDX23’s potential as a valuable diagnostic and prognostic biomarker for breast cancer, emphasizing its promise as a target for therapy. Ethics approval and informed consent The study involving human participants (230714-07) received approval from the Ethics Committee of the First Affiliated Hospital of Xinjiang Medical University and was conducted in accordance with local regulatory standards and institutional guidelines. Prior to the start of the study, written informed consent was obtained from each participant. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5 Supplementary Material 6 Supplementary Material 6