Elevated TARDBP expression correlates with an unfavorable prognosis, immune evasion, and poor immune efficacy in hepatocellular carcinoma

Introduction

Hepatocellular carcinoma (HCC) is a major cause of cancer-related mortality worldwide (1). Despite improvements in surveillance, many patients are diagnosed at advanced stages. The advent of immune checkpoint inhibitors (ICIs) has improved the therapeutic landscape (2-4), with clinical trials confirming their safety and tolerability (5). However, objective response rates remain modest—typically under 20% for monotherapy and rarely exceeding 30% for combination regimen. There is, therefore, a pressing need to identify biomarkers that can reliably predict therapeutic efficacy and stratify patients.

The liver exhibits inherent immunological tolerance. In HCC, this facilitates the accumulation of exhausted or dysfunctional immune cells, suppressing anti-tumor immunity (6,7). While biomarkers such as programmed death-ligand 1 (PD-L1), tumor mutational burden (TMB), and specific genomic alterations have been associated with outcomes (8), and pathways like vascular endothelial growth factor (VEGF), transforming growth factor beta (TGF-β), and Wnt/β-catenin are linked to resistance(9), novel markers are required to improve clinical benefit.

Transactivation DNA-binding protein (TARDBP) is a DNA- and RNA-binding protein integral to transcriptional and post-transcriptional regulation, including splicing and messenger RNA (mRNA) transport (10-13). Elevated TARDBP is a poor prognostic indicator in melanoma and HCC (14,15) and promotes progression in triple-negative breast cancer (TNBC) through alternative splicing (16,17). It also modulates metabolism and the cell cycle to support tumor growth (18,19). However, the specific mechanisms linking TARDBP to immune infiltration and clinical outcomes in HCC remain to be elucidated. We present this article in accordance with the MDAR and TRIPOD reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0712/rc).

Methods

Patient characteristics

Gene expression data and clinical information for 374 HCC cases and 50 normal tissues were retrieved from The Cancer Genome Atlas (TCGA). Clinicopathological characteristics were stratified by TARDBP expression levels (Table S1).

Data collection and processing

Gene expression data for liver hepatocellular carcinoma (LIHC) were obtained in fragments per kilobase of exon model per million mapped fragments (FPKM) format from the TCGA portal and subsequently converted to transcripts per kilobase million (TPM) for downstream analysis. The immune landscape of the TCGA-LIHC cohort was further characterized using the six pan-cancer immune subtypes (C1–C6) previously defined by Thorsson et al. (20). In our study, the HCC samples were distributed across C1 to C5, representing wound healing (C1), IFN-γ dominant (C2), inflammatory (C3), lymphocyte depleted (C4), and TGF-β dominant (C5) .The study included 374 HCC cases. RNA-sequencing data from the University of California, Santa Cruz Xena Functional Genomics Explorer (UCSC Xena) were also incorporated for pan-cancer analysis. To validate TARDBP protein expression differences between tumor and normal tissues, immunohistochemical images were acquired from the Human Protein Atlas (HPA). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Cell culture and siRNA transfection and drug treatment

HuH7 human HCC cells (HuH7 Product Number: CL0166) were cultured in RPMI-1640 medium containing 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin at 37 ℃ in a humidified 5% CO₂ incubator. HepG2 cells were grown in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin under identical conditions. At 60–70% confluence, the cells were transfected with three distinct siRNAs targeting TARDBP, PD-L1, V-domain immunoglobulin suppressor of T-cell activation (VISTA), and B7 homolog 3 (B7-H3) using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol. Final siRNA concentrations were 200 nM for HuH7 and 300 nM for HepG2. Total RNA was extracted 24 hours after transfection for quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) analysis, and proteins were collected at 48 hours for immunoblotting. Among the three siRNA constructs, si-1 yielded the most efficient knockdown and was consequently chosen for all subsequent functional assays. Oxaliplatin (OXA) was administered at final concentrations of 0, 2, 4, 10, 15, and 20 µM. To induce PD-L1 expression, cells were exposed to OXA for 72 h before lysis for protein analysis.

qRT-PCR

Total RNA was isolated using the EZ-press RNA Purification Kit (EZBioscience, Cat. B0004DP, Nanjing, China). Then, 500 ng of RNA was reverse-transcribed into cDNA with PrimeScript RT Master Mix (Perfect Real Time) (Takara, Cat. RR036A, Kusatsu, Shiga, Japan). QRT-PCR was performed using TB Green Premix Ex Taq II (Tli RNaseH Plus) (Takara, Cat. RR820, Kusatsu, Shiga, Japan), with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) serving as the internal control. Relative mRNA expression levels were calculated via the 2^–ΔΔCt method (21). All primer sequences are listed in Table S2.

Western blotting

Cells were lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer containing protease and phosphatase inhibitors. Equal protein amounts (30–70 µg) were separated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Burlington, MA, USA). After blocking with 5% BSA, the membranes were incubated overnight at 4 ℃ with primary antibodies against TARDBP, PD-L1, phospho-p65 (Ser536), total p65, and GAPDH. Following incubation with HRP-conjugated secondary antibodies, protein bands were visualized using enhanced chemiluminescence (ECL).

Survival analysis

Survival distributions were estimated using the Kaplan-Meier method and compared via the log-rank test, with patients stratified by median TARDBP expression. Univariate and multivariate Cox proportional hazards regression analyses were conducted to identify independent prognostic factors, and results were visualized using forest plots (R package ggplot2).

DNA methylation analysis

We interrogated TARDBP promoter methylation and its correlation with gene expression and patient survival using the UALCAN and MethSurv platforms (22,23).

Differentially expressed gene (DEG) analysis

Patients were stratified into high- and low-TARDBP expression groups using the median value as the cutoff. DEGs were identified via the DESeq2 package (24), applying significance thresholds of adjusted P<0.05 and |log₂FC| >1. Additionally, Spearman correlation coefficients were calculated to assess the relationship between TARDBP and the top 10 DEGs.

Functional enrichment analysis

Functional enrichment was assessed via Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses (adjusted P<0.05), with results visualized using the GOplot package (25). Additionally, gene set enrichment analysis (GSEA) was performed using clusterProfiler (26,27) to identify significant pathways [adjusted P<0.05, false discovery rate (FDR) <0.25].

Protein-protein interaction (PPI) network analysis

A PPI network was constructed from DEGs using STRING (http://string-db.org/, confidence score >0.7) and visualized in Cytoscape (version 3.5.1) (28). The CytoHubba plugin identified the top 10 hub genes (29).

Immune cell infiltration analysis

We assessed the relative enrichment of immune cells through single-sample gene set enrichment analysis (ssGSEA) implemented in the GSVA R package. A panel of 24 immune cell types quantified the infiltration levels (30). Spearman’s correlation analysis evaluated relationships between TARDBP expression and immune cell populations, while the Wilcoxon rank-sum test assessed group differences.

Multiplex immunofluorescence (mIFC) staining

Formalin-fixed, paraffin-embedded tissue sections (4 µm) underwent sequential staining with a tyramide signal amplification (TSA) 7-color kit (abs50015-100T, Absinbio, Shanghai, China). Sections were incubated with anti-TARDBP (80002-1-RR, Proteintech, Wuhan, China), treated with HRP-conjugated secondary antibody (abs50015-02, Absinbio, Shanghai, China), and developed with TSA 520 for 10 minutes, with Tris buffer washes between each step. We repeated this sequence for anti-PD-1 (WD52, Wondfo Biotech, Guangzhou, China), anti-CD3 (WB53), and anti-CD45RO (WB56), followed by DAPI (abs47047616, Absinbio) counterstaining. After coverslipping, slides were scanned on a Pannoramic MIDI II scanner (3DHISTECH, Budapest, Hungary) and analyzed using Indica HALO software (Indica Labs., Albuquerque, NM, USA).

Single-cell analysis of TARDBP expression

The single-cell RNA-seq dataset GSE149614 was processed using the Seurat v3.0 pipeline for dimensionality reduction and clustering (31). Batch effects were corrected via the ComBat algorithm (32), and clusters were visualized using Uniform Manifold Approximation and Projection (UMAP) (33). TARDBP expression across distinct cell subtypes was illustrated using violin plots generated with the ggplot2 package (34). Subsequently, subtype-specific signature genes were identified, and functional enrichment analysis was performed using clusterProfiler (27).

Immunotherapy prediction analysis

To evaluate the potential of TARDBP as an immunotherapy biomarker, we assessed its Spearman correlation with TMB, microsatellite instability (MSI), and immune checkpoint genes across pan-cancer datasets. For HCC-specific analysis, immunophenoscores (IPS) were retrieved from The Cancer Immunome Atlas (TCIA) (35). Additionally, associations with T-cell exhaustion markers were examined via Pearson correlation. Finally, the predictive value of TARDBP was validated in an independent cohort of 65 melanoma patients treated with nivolumab (GSE91061) (36).

Construction and validation of the nomogram

A prognostic nomogram for overall survival (OS) was constructed based on independent predictors identified via multivariate Cox regression. The performance of the nomogram was evaluated using the concordance index (C-index) and calibration plots generated with the rms package (version 5.1.4). Additionally, predictive accuracy was assessed via time-dependent receiver operating characteristic (ROC) curves using the timeROC package.

Drug sensitivity analysis

Gene expression and corresponding drug sensitivity data were retrieved from the CellMiner database. Pearson correlation analysis was subsequently performed, focusing exclusively on FDA-validated compounds, to evaluate the association between TARDBP expression and therapeutic response.

Statistical analysis

Data from all experiments were independently repeated at least three times and are presented as the mean ± standard error of the mean (SEM). For two-group comparisons, two-tailed Student’s t-tests were employed, while one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test was applied for multiple groups. In the discovery cohort, TARDBP expression differences were assessed using the Wilcoxon rank-sum test (for unpaired samples) and the paired t-test (for matched pairs). Clinical associations were evaluated using the Wilcoxon rank-sum test and logistic regression. All statistical tests were two-sided, and a threshold of P<0.05 was defined as statistically significant. All bioinformatics and statistical analyses were performed using R software (version 4.1.0). Data visualization was predominantly conducted using the “ggplot2” and “pheatmap” R packages.

Results

Elevated expression of TARDBP in HCC

Initial pan-cancer analysis revealed significant TARDBP upregulation in 11 tumor types—notably including LIHC, breast invasive carcinoma (BRCA), and lung adenocarcinoma (LUAD)—while observing downregulation in four others, such as kidney renal clear cell carcinoma (KIRC) and thyroid carcinoma (THCA) (Figure 1A). Focusing on the TCGA-LIHC cohort, TARDBP expression was markedly elevated in both paired and unpaired HCC tumor samples relative to normal tissues (Figure 1B,1C). ROC analysis confirmed its strong diagnostic potential, yielding an area under the curve (AUC) of 0.867 [95% confidence interval (CI): 0.819–0.915] (Figure 1D). Immunohistochemical (IHC) staining was performed using archived tissue images from the Human Protein Atlas (HPA) database (antibody CAB003703). In normal liver tissues, TARDBP staining was weak to moderate and predominantly localized in the cytoplasm of hepatocytes, with occasional faint nuclear signal. In contrast, HCC tissues exhibited markedly intensified staining, with strong diffuse positivity in both the cytoplasm and nucleus of malignant cells (Figure 1E). Quantitative analysis of IHC staining intensity confirmed that TARDBP protein expression was significantly higher in HCC tissues than in normal liver tissues (P<0.001, Wilcoxon rank-sum test; Figure 1F).

Figure 1 Expression levels of TARDBP in different types of tumors and hepatocellular carcinoma. Expression of TARDBP. (A) TARDBP expression across tumor types compared to normal tissues from TCGA and GTEx. (B) TARDBP levels in HCC versus non-matched normal liver tissues (TCGA + GTEx). (C) TARDBP expression in matched tumor-normal pairs from TCGA. (D) ROC curve analysis for distinguishing HCC from normal liver tissues in TCGA. (E) Representative immunohistochemistry images depict TARDBP in normal liver and HCC (https://www.proteinatlas.org/) (magnification: ×200). (F) Quantification of TARDBP immunohistochemical staining intensity in normal liver and HCC tissues. Data were obtained from the Human Protein Atlas database. Each dot represents an individual tissue sample. Statistical significance was determined by the Wilcoxon rank-sum test. ns, not significant (P≥0.05); *, P<0.05; **, P<0.01; ***, P<0.001. AUC, area under the curve; CI, confidence interval; FPR, false positive rate; GTEx, Genotype-Tissue Expression; HCC, hepatocellular carcinoma; ROC, receiver operating characteristic; TARDBP, transactivation DNA-binding protein; TCGA, The Cancer Genome Atlas; TPM, transcripts per kilobase million; TPR, true positive rate.

We next assessed the clinical relevance of TARDBP expression in HCC patients. Overexpression of TARDBP significantly correlated with multiple adverse clinicopathological parameters, including advanced T stage, higher pathological stage, older age, elevated alpha-fetoprotein (AFP) levels, and higher histological grade (Figure S1; Table S3). These robust associations strongly suggest that TARDBP plays a contributing role in HCC progression.

Prognostic value of TARDBP in HCC

To determine the prognostic significance of TARDBP, we stratified patients according to clinical characteristics. High TARDBP expression consistently predicted poorer OS across all subgroups, including patients aged ≤60 or >60 years, those with T1–T2 or T3–T4 stages, N0 status, and stage III–IV disease (Figure 2). These results establish TARDBP expression as a negative prognostic biomarker in HCC.

Figure 2 Prognostic values of TARDBP expression in patients with hepatocellular carcinoma evaluated by the Kaplan-Meier method in different subgroups. Overall survival is stratified by high versus low TARDBP expression within the indicated clinical subgroups. AFP, alpha-fetoprotein; CI, confidence interval; HR, hazard ratio; N, node; T, tumor; TARDBP, transactivation DNA-binding protein.

TARDBP expression and methylation

To investigate potential mechanisms of TARDBP dysregulation, we analyzed promoter DNA methylation. UALCAN analysis indicated significantly lower TARDBP promoter methylation in HCC than in normal liver tissues (Figure 3A). Two CpG sites (cg22693994 and cg00330490) exhibited hypomethylation in tumors (Figure 3B). High methylation at cg22693994 correlated with shorter OS, whereas elevated methylation at cg00330490 was associated with improved prognosis (Figure 3C-3H). These data suggest that promoter hypomethylation contributes to TARDBP overexpression and adverse outcomes in HCC.

Figure 3 DNA methylation of TARDBP and its prognostic relevance in HCC. (A) TARDBP promoter methylation levels from UALCAN. (B) Correlation between TARDBP mRNA expression and its promoter methylation. (C-H) Kaplan-Meier survival curves for selected TARDBP CpG sites. HCC, hepatocellular carcinoma; HR, hazard ratio; LIHC, liver hepatocellular carcinoma; LR, likelihood ratio; TARDBP, transactivation DNA-binding protein; TCGA, The Cancer Genome Atlas.

PPI network analysis

Comparing high- and low-TARDBP expression groups revealed 4,022 DEGs, comprising 2,505 upregulated and 1,517 downregulated genes (Figure 4A). Spearman correlation analysis demonstrated strong associations between TARDBP expression and the most upregulated genes, such as CEACAM7, MAGEA4, LGALS14, TEX15, HS3ST4, HMGA2, WIF1, MAGEA10, NR0B1, and ZPLD1 (Figure 4B).

Figure 4 TARDBP-related DEGs and functional enrichment (GO/KEGG) in HCC. (A) Volcano plot of DEGs, with downregulated genes in blue, upregulated genes in red, and the top 10 upregulated genes labeled. (B) Heatmap showing correlations between TARDBP and the top 10 upregulated DEGs. (C) GO enrichment analysis. (D) KEGG pathway analysis. ***, P<0.001. BP, biological process; CC, cellular component; DEGs, differentially expressed genes; ECM, extracellular matrix; GO, Gene Ontology; HCC, hepatocellular carcinoma; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function; TARDBP, transactivation DNA-binding protein; TPM, transcripts per kilobase million.

PPI network constructed from 578 DEGs identified the top 10 hub genes as INS, KRT19, MUC1, MUC5AC, CHGA, GAD2, SST, KRT20, AGR2, and CT45A1 (Figure S2, Table S4). These results reveal distinct molecular profiles between high- and low-TARDBP groups, supporting the idea that TARDBP modulates key oncogenic networks in HCC.

Functional enrichment analysis

GO and KEGG analyses of DEGs showed significant enrichment in immune-related processes, including humoral immune response, immunoglobulin complex formation, and antigen binding (Figure 4C). KEGG pathway analysis highlighted neuroactive ligand-receptor interaction, PPAR signaling, and chemical carcinogenesis (Figure 4D). GSEA comparing high- and low-TARDBP groups indicated that multiple immune-related pathways were enriched in the low-expression group (Figure S3). Collectively, these findings suggest that high TARDBP expression correlates with diminished immune activation and a suppressed immunophenotype in HCC.

Correlation between TARDBP expression and immune infiltration

To evaluate the relationship between TARDBP and the tumor immune microenvironment, we applied ssGSEA. TARDBP expression exhibited strong positive correlations with Th2 cells and T helper cells, and negative correlations with Th17 cells, dendritic cells (DCs), and natural killer (NK) cells (Figure 5A). The high-TARDBP group displayed greater enrichment of Th2, NK CD56bright, and helper T cells, but lower DC infiltration compared to the low-TARDBP group (Figure 5B-5I). mIFC further validated these observations. In tumor regions with low TARDBP expression, CD3 and CD45RO were abundant, while PD-1 expression remained low. Conversely, TARDBP-high regions showed diminished CD3 and CD45RO expression alongside elevated PD-1 (Figure 5J). These data suggest that high TARDBP expression is associated with an immunosuppressive tumor microenvironment.

Figure 5 Association between TARDBP expression and immune infiltration in HCC. (A) Spearman correlations between TARDBP and the relative abundance of 24 immune cell types, with dot size proportional to the absolute correlation coefficient |r|. (B-E) Infiltration scores for DCs, NK cells, T helper cells, and Th2 cells compared between high- and low-TARDBP expression groups. (F-I) Scatter plots of TARDBP expression against enrichment scores for the corresponding immune cells. (J) mIFC images from four patients show TARDBP (green), CD3 (red), CD45RO (purple), and PD-1 (pink) (magnification: 400×). ns, not significant (P≥0.05); *, P<0.05; **, P<0.01; ***, P<0.001. aDC, activated dendritic cell; DC, dendritic cell; HCC, hepatocellular carcinoma; iDC, immature dendritic cell; mIFC, multiplex immunofluorescence staining; NK, natural killer; pDC, plasmacytoid dendritic cell; TARDBP, transactivation DNA-binding protein.

The relationship between the expression of TARDBP and the TMB, MSI, T-cell exhaustion-related genes, and immune checkpoint genes

We next assessed the potential of TARDBP as a predictor of immunotherapy efficacy by analyzing its correlations with TMB, MSI, and immune checkpoint gene expression. TARDBP expression was positively correlated with TMB in bladder urothelial carcinoma (BLCA), colon adenocarcinoma (COAD), lower grade glioma (LGG), LUAD, and testicular germ cell tumors (TGCT) (Figure 6A), and with MSI in BLCA, BRCA, glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal papillary cell carcinoma (KIRP), LGG, LIHC, sarcoma (SARC), stomach adenocarcinoma (STAD), and TGCT (Figure 6B). In HCC specifically, TARDBP expression positively correlated with most T-cell exhaustion-related genes, except PLCG2 and CEACAM1 (Figure 6C). Moreover, immune checkpoint genes including CD86, CD44, CD80, CTLA4, CD27, and CD276 were significantly downregulated in the high-TARDBP group (Figure 6D-6I). These results imply that TARDBP upregulation is linked to immune exhaustion and an immunosuppressive phenotype.

Figure 6 Genomic and immunologic correlates of TARDBP. (A) Correlations between TARDBP expression and TMB across various cancers. (B) Correlations between TARDBP expression and MSI across cancers. (C) Correlations between TARDBP and T-cell exhaustion-related genes. (D-I) Correlations between TARDBP and immune checkpoint genes. *, P<0.05; **, P<0.01; ***, P<0.001. MSI, microsatellite instability; TARDBP, transactivation DNA-binding protein; TMB, tumour mutational burden.

TARDBP knockdown attenuates oxaliplatin-induced NF-κB activation and PD-L1 expression

To investigate the role of TARDBP in oxaliplatin-treated HuH7 cells, we performed siRNA-mediated TARDBP knockdown. qRT-PCR and western blotting confirmed efficient silencing of TARDBP (Figure 7A,7B). TARDBP knockdown reduced the mRNA and protein levels of PD-L1, VISTA, and B7-H3 (Figure 7A,7B). PD-L1 protein levels increasing in a dose-dependent manner (Figure 7C). TARDBP knockdown attenuated oxaliplatin-induced PD-L1 and phosphorylation of p65 (Figure 7D).

Figure 7 Knockdown TARDBP attenuates oxaliplatin-induced NF-κB activation and checkpoint expression in HuH7 cells. HuH7 cells were transfected with si-NC, or siRNAs targeting TARDBP and/or treated with oxaliplatin. (A) The relative mRNA expression of TARDBP (left) and PD-L1, VISTA and B7-H3 (right) was analyzed by qRT-PCR 24 h later. (B) The protein expression of TARDBP and PD-L1 after TARDBP knockdown was analyzed by western blot 72 h later. (C) HuH7 cells were treated with increasing concentrations of oxaliplatin (0–20 µM) for 72 h, and the protein expression of PD-L1 was analyzed by western blot. (D) HuH7 cells with or without TARDBP knockdown were treated with oxaliplatin for 72 h, and the protein expression of PD-L1 and p-p65 was analyzed by western blot. **, P<0.01; ****, P<0.0001. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NC, negative control; NF-κB, nuclear factor-κB; oxa, oxaliplatin; p-p65, phospho-p65; PD-L1, programmed death-ligand 1; qRT-PCR, quantitative real-time reverse transcription polymerase chain reaction; TARDBP, transactivation DNA-binding protein; VISTA, V-domain immunoglobulin suppressor of T-cell activation.

Single-cell analysis of the expression of TARDBP in HCC

To gain further insights into the major cell types expressing TARDBP in the cancer microenvironment, we utilized the GSE149614 dataset to analyze TARDBP at the single-cell level. After quality control, we divided all cells into 22 subgroups and identified the highly expressed genes in each group using a bubble plot (Figure 8A). We then displayed the distribution of each group using UMAP (Figure 8B) and divided all the cells into eight subgroups based on their highly expressed genes (Figure 8C). We visualized the expression of TARDBP in each subgroup using a violin diagram and noted that TARDBP exhibited high expression levels in cycling, endothelial, and tumor cells (Figure 8D). Enrichment analysis of all subgroups showed that the cycling and endothelial cell groups mainly participated in mitotic nuclear division and epidemiological cell migration (Figure 8E). Our findings suggested TARDBP is mainly expressed in tumor cells and regulates the biological behavior of tumors.

Figure 8 Single-cell characterization of TARDBP expression in tumor and stromal compartments (GSE149614). (A) Bubble plot displaying marker gene expression across 22 cell clusters. (B,C) UMAP visualization of cell clustering (resolution =0.5). (D) Violin plots illustrate TARDBP expression across major cell subsets. (E) Functional enrichment analysis of highly variable genes in each cell subset. TARDBP, transactivation DNA-binding protein; UMAP, Uniform Manifold Approximation and Projection.

TARDBP and the efficiency of immunotherapy

We established an IPS score based on the expression of TARDBP, which is currently the most promising marker of ICB response. The findings demonstrated a significant decrease in the IPS score among individuals with high TARDBP expression as opposed to those with low TARDBP expression (Figure 9A-9D). Subsequently, we studied the relationship between TARDBP expression and the prognosis of immunotherapy within a melanoma immunotherapy cohort, GSE91061. Our study revealed a significant association between low TARDBP expression and improved patient OS (Figure 9E) and these patients were more likely to respond to immunotherapy than patients with high TARDBP expression (Figure 9F). We further divided 406 patients with liver cancer from the TCGA into 5 immune subgroups (C1–5) (20) and analyzed the expression of TARDBP in each subgroup. The results indicated a concentration of patients with low TARDBP expression in C3, while high TARDBP expression patients were concentrated in C1 and C2 (Figure 9G). These findings suggested that the expression of TARDBP might affect inflammatory reactions and immune cell invasion, thus serving as an effective marker to predict the efficiency of immunotherapy.

Figure 9 Association between TARDBP expression and IPS and response to immunotherapy. (A) IPS in patients negative for both CTLA4 and PD-1. (B) IPS in CTLA4-negative/PD-1-positive patients. (C) IPS in CTLA4-positive/PD-1-negative patients. (D) IPS in CTLA4-positive/PD-1-positive patients. (E) Kaplan-Meier curves for overall survival in the GSE91061 cohort, stratified by low versus high TARDBP expression. (F) TARDBP expression levels compared between responders and non-responders to immunotherapy (box plot). (G) TARDBP expression across five distinct TCGA-LIHC immune subtypes. FPKM, fragments per kilobase of exon model per million mapped fragments; IPS, immunophenoscores; LIHC, liver hepatocellular carcinoma; PD-1, programmed cell death protein 1; TARDBP, transactivation DNA-binding protein; TCGA, The Cancer Genome Atlas.

Construction and validation of a nomogram

We conducted univariate and multivariate analyses to determine the prognostic factors for HCC. Our univariate analysis showed that T staging (T3, T4) and TARDBP expression were independent risk factors, and multivariate analysis confirmed that TARDBP expression was an independent predictive index [adjusted hazard ratio (HR) =1.610, 95% CI: 1.125–2.304, P=0.009] (Table S5). A forest plot showed the results of multivariate analysis (Figure 10A).

Figure 10 Prognostic modeling integrating TARDBP. (A) Forest plot from multivariate Cox regression analysis for overall survival. (B) Nomogram predicting 1-, 3-, and 5-year overall survival probabilities in HCC. (C) Calibration curves for the predicted 1-, 3-, and 5-year overall survival. AFP, alpha-fetoprotein; CI, confidence interval; HCC, hepatocellular carcinoma; HR, hazard ratio; N, node; T, tumor; TARDBP, transactivation DNA-binding protein.

To predict the outcome of patients diagnosed with HCC, we used age, sex, tumor-node-metastasis (TNM) staging, AFP, and TARDBP expression to establish a nomogram (Figure 10B). The nomogram exhibited a bootstrap corrected C-index of 0.652 (SE =0.039), suggesting a moderate predictive accuracy for the OS of patients with HCC. Furthermore, calibration curves (Figure 10C) were utilized to evaluate the predictive efficacy of the nomogram, confirming its suitability to predict the prognosis of patients with HCC.

Drug sensitivity evaluation

In the CellMiner database, TARDBP expression was matched to various drugs. Our analysis showed that certain drugs, such as CHELERYTHRINE, NELARABINE, PX-316, and ribavirin were more sensitive under conditions of low TARDBP expression, while LY-3021414, PQR-620, and AZD-3147 were more sensitive under conditions of high TARDBP expression (Figure S4). The results suggested a robust correlation between TARDBP expression and drug sensitivity in HCC, highlighting its possibility as a promising biomarker for drug sensitivity.

Discussion

HCC is characterized by a generally poor prognosis, with relapse rates approaching 70% within five years following surgical intervention (37). While current prognostic parameters—such as AFP levels, surgical margin status, tumor staging, and vascular invasion—are utilized (38-40), their inherent limitations often prevent routine clinical application in therapeutic decision-making. Consequently, the urgent need to identify novel, robust biomarkers is paramount for refining prognostic stratification and guiding personalized treatment strategies.

TARDBP has been extensively studied across multiple cancer type (41-44). Specifically, in breast cancer, Guo et al. reported that the loss of TARDBP reduces the abundance of CD44 variants, thereby inhibiting stemness in breast cancer stem cells (BCSCs). Mechanistically, the splicing factor SRSF3—which interacts with TARDBP—was shown to act upstream to maintain CD44 variant isoforms and BCSC stemness (16). In lung cancer, Chen et al. proposed that TARDBP modulates miR-423-3p to promote cell migration, while also upregulating miR-500a-3p and binding its mature sequence. Reduced miR-500a-3p expression correlated with poor patient outcomes, suggesting a tumor-suppressive role for TARDBP via this microRNA (42). In HCC, TARDBP has been shown to suppress apoptosis and enhance proliferation through several mechanisms. Recently, TARDBP has also been reported to upregulate gap junction protein beta 2 (GJB2), thereby promoting tumor progression and immune evasion in HCC (45), further supporting the oncogenic role of TARDBP in liver cancer. Earlier work demonstrated that TARDBP interacts directly with ABHD2, stimulating free fatty acid production and ROS generation via fatty acid oxidation (46). TARDBP also inhibits glycogen synthase kinase 3 beta translation, suppressing Wnt/β-catenin pathway activation and promoting hepatoma cell proliferation (47). Elevated TARDBP expression further modulates glycolysis through the TARDBP/miR-520/PFKP axis (14). Since hepatocellular carcinogenesis is closely linked to liver fibrosis, Yong et al. observed that TDP-43 accumulation in mice induces mitochondrial damage and fibrosis progression, suggesting that targeting mitochondrial TARDBP could enhance anti-fibrotic therapies (48). Our data confirmed higher TARDBP expression in tumor tissues compared to adjacent non-tumor tissues, with elevated expression associated with increased promoter methylation. Consistent with prior studies, high TARDBP expression predicted significantly worse OS.

Although ICB benefits some patients, most “immune-cold” solid tumors remain unresponsive. Overcoming the diverse immune evasion mechanisms in these tumors represents a major challenge for converting “cold” tumors into “hot” ones (49). Since ICIs depend on T cell infiltration, a central goal of tumor immunotherapy is to transform non-inflamed, immunosuppressive tumors into T cell-inflamed phenotypes (50). Notably, recent evidence from a machine learning-based analysis of tumor immune infiltrates has validated the predictive value of immune profiling for atezolizumab plus bevacizumab outcomes in HCC patients (51). Consistent with literature reporting that oxaliplatin stimulates PD-L1 expression (52), we similarly observed a dose-dependent increase in PD-L1 following oxaliplatin treatment. Our results further showed that oxaliplatin enhanced p65 phosphorylation p-p65 concurrent with increased PD-L1 protein abundance. Crucially, TARDBP knockdown attenuated oxaliplatin-induced p-p65 and PD-L1 expression. This indicates that TARDBP acts as a pivotal regulator in the oxaliplatin-mediated NF-κB/PD-L1 signaling pathway, presenting it as a promising therapeutic target for sensitizing HCC to chemotherapy and ICB.

Due to the substantial amount of missing data regarding background liver fibrosis and cirrhosis status in the TCGA-LIHC dataset, we were unable to stratify our clinicopathologic analysis based on fibrosis stages. Given that HCC arising in cirrhotic or advanced fibrotic livers possesses distinct biological and prognostic features, future studies utilizing independent clinical cohorts with comprehensive histopathological records are warranted to validate our findings within specific fibrosis contexts.

Conclusions

Our comprehensive integrated analysis—spanning clinical, molecular, and functional datasets—unequivocally demonstrates that elevated TARDBP expression in HCC is associated with immune suppression, poorer patient survival, and diminished benefit from immunotherapy. These findings collectively establish TARDBP as a promising biomarker with significant translational potential, simultaneously identifying it as a novel therapeutic target for effectively reprogramming the immune tumor microenvironment in HCC.

Acknowledgments

We express our gratitude to Dr. Jianming Zeng and his bioinformatics team at the University of Macau, especially Biotrainee, for generously sharing their expertise and code. We would like to acknowledge the use of the biorstudio high-performance computing cluster (https://biorstudio.cloud) at Biotrainee and the support from Shanghai HS Biotech Co., Ltd for conducting the research presented in this paper.

Footnote

Reporting Checklist: The authors have completed the MDAR and TRIPOD reporting checklists. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0712/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0712/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0712/prf

Funding: This work was supported by National Natural Science Foundation of China (No. 82102971), Guangdong Basic and Applied Basic Research Foundation, China (Nos. 2023A1515220229 and 2020A1515111074), Guangzhou Key Medical Discipline Construction Project Fund (No. 2025-2027), Science and Technology Projects in Guangzhou (No. 2025A04J3971), Guangzhou Health Science and Technology Project (No. 20251A010075) and Plan on Enhancing Scientific Research in Guangzhou Medical University.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0712/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Chakraborty E, Sarkar D. Emerging Therapies for Hepatocellular Carcinoma (HCC). Cancers (Basel) 2022;14:2798. [Crossref] [PubMed]
3. Sangro B, Sarobe P, Hervás-Stubbs S, et al. Advances in immunotherapy for hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol 2021;18:525-43. [Crossref] [PubMed]
4. Dai S, Chen Y, Cai W, et al. Combination immunotherapy in hepatocellular carcinoma: synergies among immune checkpoints, TKIs, and chemotherapy. J Hematol Oncol 2025;18:85. [Crossref] [PubMed]
5. Liu Z, Liu X, Liang J, et al. Immunotherapy for Hepatocellular Carcinoma: Current Status and Future Prospects. Front Immunol 2021;12:765101. [Crossref] [PubMed]
6. Pinter M, Jain RK, Duda DG. The Current Landscape of Immune Checkpoint Blockade in Hepatocellular Carcinoma: A Review. JAMA Oncol 2021;7:113-23. [Crossref] [PubMed]
7. Seyhan D, Allaire M, Fu Y, et al. Immune microenvironment in hepatocellular carcinoma: from pathogenesis to immunotherapy. Cell Mol Immunol 2025;22:1132-58. [Crossref] [PubMed]
8. Bruni D, Angell HK, Galon J. The immune contexture and Immunoscore in cancer prognosis and therapeutic efficacy. Nat Rev Cancer 2020;20:662-80. [Crossref] [PubMed]
9. Fridman WH, Zitvogel L, Sautès-Fridman C, et al. The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol 2017;14:717-34. [Crossref] [PubMed]
10. Klim JR, Pintacuda G, Nash LA, et al. Connecting TDP-43 Pathology with Neuropathy. Trends Neurosci 2021;44:424-40. [Crossref] [PubMed]
11. Liu B, Chen X, Wang J, et al. TDP-43: unveiling the hidden key to cellular fate decisions. Cell Commun Signal 2025;23:453. [Crossref] [PubMed]
12. Hawley ZCE, Campos-Melo D, Strong MJ. Evidence of A Negative Feedback Network Between TDP-43 and miRNAs Dependent on TDP-43 Nuclear Localization. J Mol Biol 2020;432:166695. [Crossref] [PubMed]
13. Chendrimada TP, Gregory RI, Kumaraswamy E, et al. TRBP recruits the Dicer complex to Ago2 for microRNA processing and gene silencing. Nature 2005;436:740-4. [Crossref] [PubMed]
14. Park YY, Kim SB, Han HD, et al. Tat-activating regulatory DNA-binding protein regulates glycolysis in hepatocellular carcinoma by regulating the platelet isoform of phosphofructokinase through microRNA 520. Hepatology 2013;58:182-91. [Crossref] [PubMed]
15. Zeng Q, Cao K, Liu R, et al. Identification of TDP-43 as an oncogene in melanoma and its function during melanoma pathogenesis. Cancer Biol Ther 2017;18:8-15. [Crossref] [PubMed]
16. Guo L, Ke H, Zhang H, et al. TDP43 promotes stemness of breast cancer stem cells through CD44 variant splicing isoforms. Cell Death Dis 2022;13:428. [Crossref] [PubMed]
17. Ke H, Zhao L, Zhang H, et al. Loss of TDP43 inhibits progression of triple-negative breast cancer in coordination with SRSF3. Proc Natl Acad Sci U S A 2018;115:E3426-35. [Crossref] [PubMed]
18. Chiang PM, Ling J, Jeong YH, et al. Deletion of TDP-43 down-regulates Tbc1d1, a gene linked to obesity, and alters body fat metabolism. Proc Natl Acad Sci U S A 2010;107:16320-4. [Crossref] [PubMed]
19. Tauffenberger A, Vaccaro A, Aulas A, et al. Glucose delays age-dependent proteotoxicity. Aging Cell 2012;11:856-66. [Crossref] [PubMed]
20. Thorsson V, Gibbs DL, Brown SD, et al. The Immune Landscape of Cancer. Immunity 2018;48:812-830.e14. [Crossref] [PubMed]
21. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 2001;25:402-8. [Crossref] [PubMed]
22. Chandrashekar DS, Bashel B, Balasubramanya SAH, et al. UALCAN: A Portal for Facilitating Tumor Subgroup Gene Expression and Survival Analyses. Neoplasia 2017;19:649-58. [Crossref] [PubMed]
23. Modhukur V, Iljasenko T, Metsalu T, et al. MethSurv: a web tool to perform multivariable survival analysis using DNA methylation data. Epigenomics 2018;10:277-88. [Crossref] [PubMed]
24. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. [Crossref] [PubMed]
25. Walter W, Sánchez-Cabo F, Ricote M. GOplot: an R package for visually combining expression data with functional analysis. Bioinformatics 2015;31:2912-4. [Crossref] [PubMed]
26. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50. [Crossref] [PubMed]
27. Yu G, Wang LG, Han Y, et al. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 2012;16:284-7. [Crossref] [PubMed]
28. Szklarczyk D, Gable AL, Lyon D, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 2019;47:D607-13. [Crossref] [PubMed]
29. Chin CH, Chen SH, Wu HH, et al. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Syst Biol 2014;8:S11. [Crossref] [PubMed]
30. Bindea G, Mlecnik B, Tosolini M, et al. Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer. Immunity 2013;39:782-95. [Crossref] [PubMed]
31. Stuart T, Butler A, Hoffman P, et al. Comprehensive Integration of Single-Cell Data. Cell 2019;177:1888-1902.e21. [Crossref] [PubMed]
32. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 2007;8:118-27. [Crossref] [PubMed]
33. Becht E, McInnes L, Healy J, et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nat Biotechnol 2018; Epub ahead of print. [Crossref] [PubMed]
34. Ito K, Murphy D. Application of ggplot2 to Pharmacometric Graphics. CPT Pharmacometrics Syst Pharmacol 2013;2:e79. [Crossref] [PubMed]
35. Charoentong P, Finotello F, Angelova M, et al. Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade. Cell Rep 2017;18:248-62. [Crossref] [PubMed]
36. Riaz N, Havel JJ, Makarov V, et al. Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab. Cell 2017;171:934-949.e16. [Crossref] [PubMed]
37. Villanueva A, Hoshida Y, Battiston C, et al. Combining clinical, pathology, and gene expression data to predict recurrence of hepatocellular carcinoma. Gastroenterology 2011;140:1501-12.e2. [Crossref] [PubMed]
38. Llovet JM, Kelley RK, Villanueva A, et al. Hepatocellular carcinoma. Nat Rev Dis Primers 2021;7:6. [Crossref] [PubMed]
39. El Jabbour T, Lagana SM, Lee H. Update on hepatocellular carcinoma: Pathologists' review. World J Gastroenterol 2019;25:1653-65. [Crossref] [PubMed]
40. Cerrito L, Ainora ME, Mosoni C, et al. Prognostic Role of Molecular and Imaging Biomarkers for Predicting Advanced Hepatocellular Carcinoma Treatment Efficacy. Cancers (Basel) 2022;14:4647. [Crossref] [PubMed]
41. Kim PY, Tan O, Liu B, et al. High TDP43 expression is required for TRIM16-induced inhibition of cancer cell growth and correlated with good prognosis of neuroblastoma and breast cancer patients. Cancer Lett 2016;374:315-23. [Crossref] [PubMed]
42. Chen X, Fan Z, McGee W, et al. TDP-43 regulates cancer-associated microRNAs. Protein Cell 2018;9:848-66. [Crossref] [PubMed]
43. Zhang H, Lin J, Yahaya BH. Comprehensive analysis of co-expressed genes with TDP-43: prognostic and therapeutic potential in lung adenocarcinoma. J Cancer Res Clin Oncol 2024;150:44. [Crossref] [PubMed]
44. Giannini M, Bayona-Feliu A, Sproviero D, et al. TDP-43 mutations link Amyotrophic Lateral Sclerosis with R-loop homeostasis and R loop-mediated DNA damage. PLoS Genet 2020;16:e1009260. [Crossref] [PubMed]
45. Zhu Y, Hu M, Xu R, et al. TARDBP upregulates GJB2 to promote tumor progression in hepatocellular carcinoma. Biochem Biophys Res Commun 2026;796:153123. [Crossref] [PubMed]
46. Liu BW, Wang XY, Cao JL, et al. TDP-43 upregulates lipid metabolism modulator ABHD2 to suppress apoptosis in hepatocellular carcinoma. Commun Biol 2022;5:816. [Crossref] [PubMed]
47. Guo F, Wang H, Jiang M, et al. TDP-43 induces EMT and promotes hepatocellular carcinoma metastasis via activating Wnt/β-catenin signaling pathway. Am J Cancer Res 2020;10:3285-301. [PubMed]
48. Yong H, Shan S, Wang S, et al. Activation of mitophagy by rapamycin eliminated the accumulation of TDP-43 on mitochondrial and promoted the resolution of carbon tetrachloride-induced liver fibrosis in mice. Toxicology 2022;471:153176. [Crossref] [PubMed]
49. Zhang J, Huang D, Saw PE, et al. Turning cold tumors hot: from molecular mechanisms to clinical applications. Trends Immunol 2022;43:523-45. [Crossref] [PubMed]
50. Wei C, Ma Y, Wang F, et al. Igniting Hope for Tumor Immunotherapy: Promoting the "Hot and Cold" Tumor Transition. Clin Med Insights Oncol 2022;16:11795549221120708. [Crossref] [PubMed]
51. Scheiner B, Lombardi P, D'Alessio A, et al. Preliminary qualification of a machine learning-based assessment of the tumor immune infiltrate as a predictor of outcome in patients with hepatocellular carcinoma treated with atezolizumab plus bevacizumab. J Immunother Cancer 2025;13:e010975. [Crossref] [PubMed]
52. Liang XJ, Guo X, Zhu PH, et al. An Enhanced Cancer Immunogenic Cell Death Strategy: Fibrate-Oxaliplatin(IV) Conjugates with Modulating Tumor Immune Microenvironment. J Med Chem 2025;68:26328-47. [Crossref] [PubMed]