4014 Background: Checkpoint inhibitors combined with anti-angiogenic therapy represents one of the standard first-line therapies for advanced hepatocellular carcinoma (aHCC). Nilvanstomig (ZG005) is a recombinant humanized anti-PD-1/TIGIT bispecific antibody. By blocking both pathways, it can synergistically activate T cells and enhance the anti-tumor activity of NK cells. This study evaluated ZG005 plus bevacizumab vs. sintilimab plus IBI305 (a bevacizumab biosimilar) as first-line therapy for aHCC. Methods: In this randomized, open-label, multicenter, phase 2 trial conducted in China, patients with aHCC who had not previously received systemic treatment were randomly assigned (1:1:1) to receive ZG005 10 mg/kg plus bevacizumab 15 mg/kg (Arm A); ZG005 20 mg/kg plus bevacizumab 15 mg/kg (Arm B); or sintilimab 200 mg plus IBI305 15 mg/kg (Arm C). All treatments were administered intravenously every 3 weeks until disease progression or unacceptable toxicity. Randomization was stratified by baseline AFP level ( < 400 vs. ≥400 ng/mL), macrovascular invasion or extrahepatic metastasis (presence vs. absence). The primary endpoint was IRC-assessed PFS per RECIST v1.1, with key secondary endpoints including PFS per mRECIST, ORR and DCR by both criteria, and OS. Results: As of the data cutoff (Nov 25, 2025), 95 patients were enrolled and received at least one dose of treatment (Arm A, n = 31; Arm B, n = 32; Arm C, n = 32). Baseline characteristics were well-balanced across the treatment arms. For all patients enrolled, the median age was 61 years (range, 37-75), with 85.3% of patients being male. Disease characteristics included BCLC stage B (28.4%) or C (71.6%), Child-Pugh score A5 (83.2%) or A6 (16.8%), baseline AFP ≥400 ng/mL in 45.3% of patients, and HBV positivity in 73.7%. Macrovascular invasion and/or extrahepatic metastasis were present in 71.6% (68/95) of patients. With a median follow-up of about 5 months, the IRC-assessed ORR was 32.3% in Arm A, 37.5% in Arm B, and 25.0% in Arm C per RECIST v1.1; the corresponding ORRs per mRECIST were 54.8%, 50.0%, and 34.4%. The IRC-assessed median PFS per RECIST v1.1 was not reached in Arm A or B, compared with 5.8 months in Arm C (Arm A vs. C: HR 0.40, 95% CI 0.15-1.05; Arm B vs. C: HR 0.28, 95% CI 0.10-0.79). The safety profiles were comparable across the three arms. No grade ≥3 hemorrhagic adverse events were reported in either Arm A or B. Conclusions: The combination of ZG005 and bevacizumab as first-line treatment in patients with aHCC demonstrated an early encouraging efficacy with an acceptable safety profile. Higher response rates and prolonged PFS were observed with the ZG005-based regimens. Longer follow up of the current study and future phase 3 trials are warranted to validate the efficacy and safety of ZG005 in combination with bevacizumab in aHCC. Clinical trial information: NCT06558227 .

Cholangiocarcinoma is a highly malignant tumor with an increasing incidence around the world. Discovery of novel molecular targets and effective therapies for cholangiocarcinoma are urgently needed. Palmitoylation is a reversible lipid modification mainly catalyzed by ZDHHC family palmitoyltransferases. However, its function and underlying mechanisms in cholangiocarcinoma remain poorly understood. Here, we found that protein palmitoylation levels and ZDHHC5 expression were upregulated in cholangiocarcinoma. Knockdown of ZDHHC5 inhibited the growth of cholangiocarcinoma. Mechanistically, ZDHHC5 modulates the activity of MAPK signaling pathway by regulating palmitoylation of BRAF at Cys194/195. Palmitoylation facilitates the membrane localization of BRAF and stabilizes BRAF protein. Either knockdown of ZDHHC5 or disruption of palmitoylation sites of BRAF inhibited ERK signaling. Furthermore, we found that cholangiocarcinoma cells expressing high levels of ZDHHC5 exhibited increased activities of MAPK signaling pathway and increased sensitivities to MAPK signaling pathway inhibitors. This study identifies a previously unknown ZDHHC5-BRAF-ERK axis that promotes the growth of cholangiocarcinoma and represents a potential crucial role in therapeutic.

The heterogeneous and immunosuppressive tumor microenvironment (TME) in hepatocellular carcinoma (HCC) contributes to poor immunotherapy responses. Tumor-associated macrophages (TAM) are central to the immunosuppressive TME, but how metabolic programs regulate TAM pro-tumorigenic functions remain incompletely understood. Here, we identify branched-chain amino acid transaminase 1 (BCAT1) as a metabolic checkpoint in TAMs constraining tumor progression. Compared with wild-type TAMs, BCAT1-deficient TAMs have increased intracellular crotonate, as well as enhanced histone H3 lysine 14 crotonylation, upregulated lipid metabolism genes and an immunosuppressive phenotype. In HCC mouse models, BCAT1-deficient TAMs aggravate tumor burden and suppress CD8+ T cell-mediated antitumor immunity, while myeloid-specific BCAT1 overexpression or adoptive transfer of BCAT1+ macrophages stimulates the antitumor immune response and improves anti-PD1 therapy responses. In summary, our data support a BCAT1-mediated regulation of crotonate-dependent epigenetic modulation of immunosuppressive TAMs in HCC, and indicate BCAT1+ macrophages as an adjuvant treatment for enhancing immune checkpoint blockade therapy.

Radiation-induced liver disease (RILD) poses a major clinical challenge in radiotherapy, transplantation preconditioning, or radiation accidents, yet its pathogenesis is poorly understood due to limited animal models. Here, we establish a translational pig model recapitulating human RILD pathology within 4 weeks post-40 Gy irradiation, featuring veno-occlusive disease (VOD) and centrilobular necrosis. Single-cell atlas analyses identify ferroptosis as a key driver of hepatocyte death during RILD initiation. Ferroptosis inhibition with liproxstatin-1 (Lip-1) not only prevents RILD progression but also reverses histological damage and restores liver function. Mechanistically, Lip-1 treatment restores dysregulated gene expression profiles, particularly associated with hepatocyte ferroptosis, while stimulating hepatic regeneration via coordinated proliferation of hepatocytes and endothelial cells. Our findings establish ferroptosis inhibition as a therapeutic strategy for RILD, demonstrating its dual role in cytoprotection and regeneration. This large animal model provides a robust platform to optimize radiotherapy regimens, improve transplant conditioning, and develop targeted radioprotectants.

The mechanisms underlying metabolic remodeling in metabolic dysfunction-associated steatotic liver disease (MASLD) remain unclear. Targeting the process of de novo lipogenesis (DNL) in the liver has the potential to mitigate MASLD. Here we show that interferon-related developmental regulator 1 (IFRD1) expression negatively correlates with MASLD/metabolic-associated steatohepatitis (MASH) progression in human liver tissues. In multiple mouse models, Ifrd1-/- mice exhibit an exacerbated MASLD phenotype, while hepatocyte-specific IFRD1 expression suppresses MASH progression. Mechanistically, IFRD1 promotes GLUD1's mitochondrial localization via direct interaction, stabilizing the enzyme's activity to enhance α-ketoglutarate (α-KG) production. α-KG reduces H3K36me3 level at lipogenic genes, thereby inhibiting DNL and ameliorating MASH. α-KG supplementation reverses MASH exacerbation in Ifrd1-CKO mice. Collectively, our research establishes the IFRD1-GLUD1-α-KG axis as a critical metabolic-epigenetic regulatory hub, providing novel targets for inhibiting hepatic DNL and developing therapeutic agents for MASLD/MASH.

Introduction:Hepatocellular carcinoma (HCC) harboring CTNNB1 mutations that activate the Wnt/β-catenin pathway demonstrates increased gadoxetic acid (Gd-EOB-DTPA) uptake due to overexpressed organic anion transporting polypeptide 1B3 (OATP1B3) and exhibits immune checkpoint inhibitor (ICI) resistance attributed to an immune-excluded tumor microenvironment and tumor immune barriers. This systematic review investigated Gd-EOB-DTPA-enhanced magnetic resonance imaging (MRI) for predicting immunotherapy response in HCC. Methods:A systematic search of PubMed, Web of Science, and Cochrane was conducted up to September 10, 2025. Studies linking relative enhancement ratio (RER) >0.9 on Gd-EOB-DTPA-enhanced MRI to progression-free survival (RECIST 1.1) in immunotherapy-treated patients were included. Following PRISMA 2020/SWiM guidelines, risk of bias was assessed via QUADAS-2. Meta-analysis was performed using JASP (v0.95.4) via a random-effects model with restricted maximum likelihood estimation. Heterogeneity was assessed using I 2 and τ 2. Notably, the Knapp-Hartung adjustment was applied to calculate 95% confidence interval (CI) to ensure robust inference despite the limited study number. Sources of heterogeneity and robustness were explored using subgroup analyses, meta-regression, and sensitivity analyses. Results:Five studies (n = 253; published 2021-2025) were analyzed. RER ≥0.9 was identified as a significant predictor of poor response across diverse ICI regimens, with a pooled hazard ratios (HR) of 5.79 (95% CI: 1.56-21.50; p = 0.020) and individual estimates ranging from 1.58 to 22.04. However, further analysis indicated that anti-VEGF therapy might mitigate this resistance and partially restore ICI efficacy; the association between high RER and poor survival was not statistically significant in the anti-VEGF cohort (HR = 3.39; 95% CI: 0.39-29.14; p = 0.135). Conclusions:To the best of our knowledge, this is the first systematic review evaluating the predictive utility of Gd-EOB-DTPA-enhanced MRI for HCC immunotherapy. Our findings suggest that an RER ≥0.9 serves as a potential noninvasive marker for poor treatment response. Notably, the observation that anti-VEGF combination therapy might mitigate this imaging-defined resistance is hypothesis-generating, underscoring the need for prospective studies to validate optimal strategies for patients with high-RER tumors.

Supplementary Figure 5. CHD6/SMARCA2/4-coordinated SEs-promoter looping boosts the transcription of NF-κB-related targets, related to Figure 6. A. 3C assay of the indicated genes in WT, CHD6-KO Caki-2 cells with or without DMF stimulation. “Pro” means the input-pro, while “3C” means the 3C primers targeting the indicated enhancer region. B. ChIP-qPCR of H3K27ac markers in promoters or enhancers of CCL2, ICAM1, or BCL3 gene in UOK-262 cells (n = 5). C. Immunoblotting analysis of p65 immunoprecipitates in UOK-262 cells with or without CHD6 depletion. D. ChIP-qPCR of SMARCA2/4 markers in the SEs of CCL2, ICAM1, or BCL3 gene (n = 5) in CHD6-KO UOK-262 cells with or without CHD6 restoration. E. RT-qPCR analysis of mRNA level for CCL2, ICAM1 and BCL3 genes (n = 5) in UOK-262 cells transfected with indicated siRNAs targeting cBAF, PBAF, or ncBAF. F. ChIP-qPCR of H3K27ac markers in indicated SEs of CCL2, ICAM1, or BCL3 gene (n = 5) in SMARCA2/4-KD UOK-262 cells with or without CHD6 overexpression. G. MA plot of differential H3K4me1 or H3K27ac ChIP-seq signals in UOK-262 cells treated with 12 h AU-15330 versus DMSO. Log2 fold changes of peaks are plotted on the y-axis. Significantly changed peaks (FDR< 0.05) are marked in grey. H. Gene set enrichment analysis (GSEA) revealed the inhibition of CHD6-signature by AU-15330. I. Plot of CHD6/p65 co-binding intensities in UOK-262 treated with control and AU-15330 (5 μM), respectively. P values were calculated using a two-tailed Student’s t-test (B, D-F). *p < 0.05, **p < 0.01, and ***p < 0.001. ns, no significance.

Characteristics of patients whose tumors were used for PDOs or PDXs experiments, Related to Figure 4&7.

Identification of super-enhancer-associated genes in DMF-treated ACHN cells, related to Figure 6.

Biomolecular condensates formed through liquid-liquid phase separation (LLPS) are emerging as key regulators of chromatin organization and gene expression. However, how phase separation is coupled to specific noncanonical DNA structures to confer locus-selective transcriptional regulation remains poorly understood. G-quadruplex DNA (G4 DNA) is enriched at promoters, CpG islands, and other regulatory genomic regions, suggesting a potential role in guiding chromatin regulators to defined genomic loci. KDM2B is a CpG island-binding chromatin regulator involved in transcriptional repression, yet whether and how it senses G4 DNA to establish gene-specific regulatory programs in hepatocellular carcinoma (HCC) remains unclear. Here, we identify KDM2B as a G4 DNA-responsive phase-separating chromatin regulator. KDM2B undergoes LLPS through its intrinsically disordered regions and preferentially recognizes parallel G4 DNA structures. Mechanistically, G4 DNA promotes KDM2B condensate assembly and facilitates its chromatin targeting to G4-and CpG-rich regulatory loci. This process establishes a local repressive chromatin environment and restrains inflammatory transcriptional programs. Conversely, loss of KDM2B impairs G4-dependent condensate formation and weakens chromatin-associated transcriptional repression, leading to a proinflammatory secretory phenotype characterized by increased cytokines and chemokines, including IL-6 and CXCL8. This inflammatory remodeling promotes SPP1+ macrophage accumulation, reshapes the tumor immune microenvironment, and accelerates HCC progression. Together, these findings reveal that KDM2B-G4 DNA phase-separated condensates function as an epigenetic regulatory platform that couples noncanonical genome architecture with locus-selective inflammatory gene repression, thereby restraining inflammation-associated HCC progression.

Summary of the epigenetic regulators showing the decreased sgRNA abundance in UOK-262 and ACHN cells, respectively.

BACKGROUND:Liver transplantation (LT) is a life-saving therapy for end-stage liver disease, but postoperative neurocognitive recovery and mechanisms remain unclear. Although liver-brain crosstalk is known, whether LT restores neurological dysfunction is uncertain. METHODS:Twenty-one patients undergoing LT were prospectively evaluated within 1-3 d before transplantation and 21 d postoperatively. Cognitive function was assessed using the Montreal Cognitive Assessment and the Psychometric Hepatic Encephalopathy Score, and resting-state electroencephalography (EEG) was analyzed for power spectral density, functional connectivity, and microstate dynamics. Paired plasma samples underwent nontargeted metabolomics, and candidate metabolites were further examined using in vitro inflammatory injury models, patch-clamp electrophysiology, ex vivo brain slices, and in vivo behavioral testing. RESULTS:Montreal Cognitive Assessment scores increased from 23.14 ± 2.46 to 24.67 ± 1.88 (P = 0.0002), and Psychometric Hepatic Encephalopathy Scores improved from -11.28 ± 2.42 to -6.11 ± 2.89 (P < 0.0001) after LT. EEG analyses showed increased alpha1 and beta power, enhanced functional connectivity, and altered microstate dynamics after transplantation. Plasma glial fibrillary acidic protein decreased postoperatively, suggesting reduced neuroinflammatory injury. Nontargeted metabolomics identified postoperative elevation of S-methyl-5'-thioadenosine (MTA), and experimental validation showed that MTA attenuated inflammatory neurotoxicity, reduced reactive oxygen species production, and preserved neuronal excitability and synaptic transmission. CONCLUSIONS:LT was associated with early improvement in neurocognitive function in patients with end-stage liver disease. Integrated EEG and metabolomic analyses suggest that posttransplant cognitive recovery may involve restoration of cortical network activity and systemic metabolic remodeling. MTA emerged as a candidate recovery-associated biomarker and functional metabolite that may contribute to neuroprotection by modulating neuroinflammation.

Supplementary Figure 4. CHD6 activates NF-κB signaling to potentiate FH-deficient RCC malignancy, related to Figure 5. A. Volcano plot showing differentially expressed genes in UOK-262 cells upon CHD6 knockdown. B. Unsupervised cluster analysis of differentially expressed genes in control and CHD6 knockdown UOK-262 cells. C. ATAC-seq signals showing the profiles of OCRs across the indicated peaks in UOK262 cells with versus without CHD6 knockdown. D. Heatmap exhibiting the significance of transcription factor motifs enriched in accessible loci derived from control and CHD6-KD UOK262 cells. NF-κB motif is dominantly highlighted. E. Venn diagram showing overlapping hits, defined as CHD6-signature, with CHD6 ChIP-seq peaks and changes in OCRs (ATAC-seq) and differentially expressed genes (RNA-seq). F. Western blotting assays and Co-IP analysis showing the altered CHD6-p65 interactions in UOK-262 cells with or without FH restoration. G-H. MTT (G), colony formation (H-left) assays were performed in control and p65-KD UOK-262 cells. Colony formation assays performed in p65-depleted UOK-262 cells with or without CHD6 overexpression (H-right). I. Quantification of colony formation numbers in indicated groups from (H). J. Effects of JSH-23 treatment (1 mg/kg) on UOK262-derived xenografts, as indicated (n = 6 per group, 2-way ANOVA followed by Tukey’s multiple comparisons test). Treatment started when tumors reached 50–100 mm3. K. ChIP-qPCR analysis of CHD6, Pol II-S5P and S2P in the promoter regions of the indicated genes in WT and CHD6-KO UOK-262 cells with restoration of WT CHD6. P values were calculated using 2-tailed Student’s t-test (H, K). *p < 0.05, **p < 0.01, and ***p < 0.001. ns, no significance.

Mucosal-associated invariant T cells (MAITs) are enriched in the liver and closely related to human hepatocellular carcinoma (HCC), but their role is controversial. Whether and how the plasticity of MAITs modulates HCC progression remain to be explored. Here, we revealed that CD4+ MAITs displaying Th17 features were the major source of IL-17A in human HCC. IL-17A from Th17-polarized CD4+ MAITs promoted HCC progression by enhancing lipid storage and tumor cell proliferation in a PPARα dependent manner. Additionally, we showed that both TCR-dependent and TCR-independent activation signaling induced Th17-polarized CD4+ MAIT differentiation and that strong signaling promoted their differentiation. Moreover, IL-17A production in CD4+ MAITs was promoted by glycolysis via posttranscriptional regulation, and tumor cell-derived kynurenine enhanced glycolysis and IL-17A production through the AHR pathway. These findings demonstrate that the plasticity of MAITs and the generation of CD4+ MAITs promote HCC progression via metabolic crosstalk with tumor cells.

Supplementary Figure 1. Library and cells used for in vivo epigenetic CRISPR screen, related to Figure 1. A. Schematic diagram of FH mutations and related distributions in the indicated protein domains. B. Intracellular fumarate levels were measured in a panel of indicated RCC cell lines. C. Lorenz curve showing the distribution of sgRNAs in the epigenetic-focused library. D. Workflow showing the generation of the UOK- or ACHN-clones without Cas9 for evaluating the distribution of guides that persist upon tumour formation from the tumour initiating cells (TICs), and the UOK- or ACHN-clones with Cas9 were used for further in vivo screens. E. Western blot showing superior Cas9 expressions in UOK-Cas9-Clone 5 and ACHN-Clone 4. F. Cell viability of UOK-Cas9-Clone 5 and ACHN-Cas9-Clone 4 transfected with sgRNAs targeting the essential gene CCND1. G. Representative tumor graph and growth curve of sgRNA library transduced UOK-Cas9-Clone 5 & ACHN-Cas9-Clone 4 subcutaneously injected into BABL/c nude mice. H. Percentage of maintained sgRNAs in mice injected with UOK-Clone 5 library (without Cas9) and ACHN-clone 4 library (without Cas9) cells. I. Volcano plot revealing the targets with altered sgRNA frequencies with a cutoff of p < 0.01 and log2 fold change (log2FC) > 2. J. MTT analysis showing the effects of siRNA KD of 9 CHD family genes on FH-deficient cells, respectively. The quantitative results shown are representative of 5 experiments. P values were calculated using a two-tailed unpaired t-test (F, J). *p < 0.05, **p < 0.01, and ***p < 0.001.

Background:Steatotic liver disease affects 40% of nonobese individuals, but existing screening tools inadequately detect and stage disease severity in this population because of the limited sensitivity of conventional ultrasound and the lack of dedicated prediction models. Objective:This study aimed to develop and validate an interpretable machine learning model specifically for multiclass hepatic steatosis severity prediction in nonobese individuals to support early risk stratification in this underrecognized group. Methods:Health examination data from 215,145 nonobese participants (BMI <28 kg/m²) were randomly divided into training (n=150,601, 70%) and test (n=64,544, 30%) sets. Hepatic steatosis was diagnosed and graded using the controlled attenuation parameter with established thresholds (none: <248 dB/m; mild: 248-268 dB/m; and moderate to severe: >268 dB/m). From 42 candidate variables, 14 predictors were selected using Least Absolute Shrinkage and Selection Operator regression and Recursive Feature Elimination based on Random Forest importance. Six machine learning algorithms-k-nearest neighbors, naive Bayes, multilayer perceptron, random forest, support vector machine, and Extreme Gradient Boosting (XGBoost)-were developed using 10-fold cross-validation, with hyperparameters optimized for maximal area under the receiver operating characteristic curve (ROC-AUC). Model interpretability was assessed using Shapley Additive Explanations analysis. External validation was conducted in non-Hispanic Asian participants from the National Health and Nutrition Examination Survey (n=726). Model performance was evaluated using accuracy, Cohen κ, ROC-AUC, area under the precision-recall curve, F1-score, precision, sensitivity, and specificity. Results:The final cohort included 215,145 participants, with steatosis severity classified as none (n=92,944, 43.2%), mild (n=54,121, 25.2%), and moderate to severe (n=68,080, 31.6%). Among the 6 machine learning models, XGBoost achieved the best discrimination on the test set, with an accuracy of 0.824 and a macro-average ROC-AUC of 0.941. In external validation, the model maintained strong performance (macro-average ROC-AUC=0.874). Shapley Additive Explanations analysis identified BMI, waist circumference, liver enzymes (alanine aminotransferase and aspartate aminotransferase), renal function indicators (uric acid and serum creatinine), and metabolic indices (triglycerides, continuous metabolic syndrome score, and triglyceride-glucose index) as key contributors to model predictions. The model has been implemented as an online prediction platform to facilitate clinical use. Conclusions:This interpretable XGBoost model accurately predicts controlled attenuation parameter-defined hepatic steatosis severity in nonobese individuals and demonstrates robust performance in both internal and external validation cohorts, providing a practical tool for early risk stratification in this underrecognized population.

MicroRNA (miRNA) therapy represents an attractive approach for the treatment of colorectal cancer (CRC). However, the efficacy of therapy targeting a single oncomiRNA (oncomiR) is limited, as multiple oncomiRs often act simultaneously in promoting CRC development. In addition, current anti-miRNA nucleotide (AMiN) design faces challenges, including high synthetic error rate and difficulty in synthesizing long oligonucleotides. Circular single-stranded DNA (CssDNA), an ancient form of DNA that has gained increasing attention recently, is known for its easy synthesis, high stability, low error rate, and long size potential. Here, we develop a multi-oncomiR targeting platform with low immunogenicity, high stability and long-size potential using in vivo M13 phage-generated CssDNA. This CssDNA acts as a sponge, which attracts and degrades multiple abnormally overexpressed oncomiRs in CRC. The degradation of oncomiRs promotes the expression of tumor suppressor genes (TSGs), inhibiting CRC development. Further, the degradation of oncomiRs from the sponge releases the free CssDNA for second round of action, which results in the efficacy and recyclability of CssDNA. It's visible to spread this novel, sequence customizable, multi-oncomiR targeting platform using CssDNA as a stable, low-immunogenic and recyclable AMiN in antagonizing miRNA-mediated repression of TSG expression for other tumor suppression.