Global Advances in Hepatocellular Carcinoma Research and Therapy in 2025.

1
Introduction
Hepatocellular carcinoma (HCC) is a significant burden to the global health situation, being the sixth most prevalent malignancy and the third most frequent cause of cancer‐related mortality in the world [1]. The impact is notably pronounced in China, where the incidence of HCC was fourth, and mortality was second among all malignant tumors in 2022, and it remains an impactful burden on the health of the population [2]. There has been extensive advancement in surveillance and systematic therapies within the past decades. Though not yet with the generally optimum results, such advances have redefined the paradigm of managing HCC therapeutic approaches. It is now known that more than 60% of cases of HCC should be prevented [3]. Moreover, a combination of systematic staging, accurate classification plans, and selected multimodal therapies has been applied and led to high survival rates, especially among patients with advanced‐stage disease [3]. This article reviews the major advancements in the field of HCC through 2025 and discusses future perspectives in clinical practice.

2
Advances in Epidemiology
Hepatitis B virus (HBV) infection remains the primary etiology of HCC globally and the leading cause in China [4]. However, the prevalence of HBV infection has witnessed a sustained decline, a trend largely attributable to the systematic rollout of universal vaccination programs [5]. Likewise, the burden of hepatitis C virus‐associated HCC has declined markedly with the widespread adoption of direct‐acting antiviral therapies [6]. In contrast, metabolic dysfunction‐associated steatotic liver disease (MASLD) and metabolic dysfunction‐associated steatohepatitis (MASH) account for an expanding proportion of the oncogenic burden of HCC [7]. Projections suggest MASH‐related HCC will increase from 8% in 2022 to 11% by 2050 (a 35% growth rate), with alcohol‐related HCC similarly rising from 19% to 21% over the same timeframe [3]. In the United States, MASH and alcohol consumption have overtaken viral hepatitis to become the chief oncogenic drivers [3]. In addition, obesity and diabetes synergistically accelerate the progression from MASLD to MASH, highlighting the intricate crosstalk between metabolic pathways, lifestyle factors, and hepatocarcinogenesis [8]. Consequently, as MASLD and MASH become central drivers of HCC, prioritizing systematic screening and lifestyle intervention are essential for effective clinical management.
Guided by those changing etiological trends, the general state of HCC diagnosis and treatment is also experiencing a similar change. The classical style of diagnostic paradigm, involving the use of single forms of biomarkers like alpha‐fetoprotein (AFP) and des‐gamma‐carboxy prothrombin (DCP or PIVKA‐II), and so on, is progressively giving place to combined multi‐omics and multi‐modal frameworks incorporating metabolites, cell‐free DNA (cfDNA), and high‐quality imaging. Equally, the traditional surgical and targeted monotherapies are still being transformed into new precision approaches, such as localized precision treatment and combination targeted immunotherapy. It is important to note that this paradigm shift can be used to highlight the fact that mechanism and translational research, which are specific to certain etiologies, will be able to give completely new overviews of how HCC can be managed in the future.

3
Advances in Early Diagnosis and Screening
3.1
Conventional Diagnostic Biomarkers
The 5‐year survival rate for patients with early‐stage HCC is estimated to exceed 50%, whereas it remains below 20% for those diagnosed at an advanced stage [9]. As a result, early diagnosis is a clear‐cut issue when it comes to the prognosis of a patient. Nonetheless, an effective early detection is one of the unmet challenges of modern HCC care. At this point, serum AFP and abdominal ultrasonography as a combination have become the standard‐of‐care surveillance modality [10]. In addition to AFP, DCP, and the lens culinaris agglutinin‐reactive fraction of AFP (AFP‐L3) are established as clinically validated serum biomarkers in clinical practice [11]. These markers are frequently incorporated into multi‐parameter prediction models, including GALAD (Gender, Age, AFP‐L3, AFP, and DCP) [12], GAAD (Gender, Age, AFP, and DCP) [13], and ASAP (Age, Sex, AFP, and Protein induced by vitamin K absence or antagonist II [PIVKA‐II]) [13]. To the extent that they are based on the recurring clinical parameters and traditional methods of biomarker testing, these models demonstrate high applicability and high levels of diagnostic performance. As a result, they have been significantly incorporated in clinical HCC surveillance, being tools that cannot be done without in the modern HCC diagnosis. Another important development that might ensure optimal use of conventional biomarkers was a 2025 Phase III trial in biomarker validation that established that the GALAD score is better at detecting HCC up to 12 months prior to clinical diagnosis in cirrhotic cohorts [14], a key advance in optimizing conventional biomarker performance. Quantitatively, the model demonstrated superior discriminative power compared to AFP alone (area under the curve [AUC]: 0.78 vs. 0.66), maintaining a sensitivity of 62% and specificity of 82% at the validated threshold. Nevertheless, the overall performance of these approaches remains suboptimal. Meta‐analysis data indicate that their sensitivity and specificity are generally limited to less than 80% and 90%, respectively [13]. However, more recently, a pilot study was able to indicate that reducing the AFP‐cutoff point to 10 ng/mL may enhance the sensitivity of detecting malignancies in patients with cirrhosis, especially with respect to the increase in MASH‐related HCC [15]. Moreover, the recently introduced HES V2.0 categorization system of asking a panel of regularly laboratory‐tested biomarkers has also shown better diagnostic performance than the GALAD and ASAP systems [16, 17]. Collectively, these data underscore the need to refine existing diagnostic algorithms and re‐evaluate cutoff values to optimize early HCC screening performance.

3.2
Novel Liquid Biopsy Biomarkers and AI Integration Strategy for Early‐HCC Diagnosis
Liquid biopsy, due to its real‐time, non‐invasive, and precise forms of monitoring, is quickly establishing itself as a part of early HCC diagnosis [18]. Among other liquid biopsy targets, the most dominant ones are the so‐called cfDNA because of its early appearance in blood circulation, high temporal resolution, high signal density, and high specificity [19]. In HCC management, cfDNA serves as an emerging diagnostic biomarker capable of characterizing vital genetic and epigenetic variations, demonstrating significant clinical potential [20]. While mutation and methylation profiles are the most widely used cfDNA indicators [21], their values in the context of the early HCC diagnosis have scarcely been evaluated systematically. One of the recent studies showed that the diagnostic effectiveness of the models of the cfDNA mutations is close to that of AFP but significantly lower than that of the patterns of the cfDNA methylation. Although the level of the cfDNA concentration and integrity is high with HCC, the levels of methylation marks and fragmentation would provide better diagnostic efficiency (sensitivity > 60%, specificity > 90%) than the conventional biomarkers, such as the AFP. Notably, combining these two features failed to enhance diagnostic performance, indicating that methylation may be the more robust marker for early HCC detection. Based on these findings, researchers developed a novel pipeline termed “HCCtect” which utilizes two specific cfDNA methylation targets (OTX1 and HIST1H3G) to effectively distinguish early‐stage HCC [22]. Specifically, a methylation‐sensitive high‐resolution melting assay targeting RNF135 and LDHB demonstrated high specificity for HCC across diverse etiologies, detecting 57% of HCC cases compared to 45% for AFP alone. When combined with AFP, this dual‐marker assay achieved a sensitivity of 70% in at‐risk populations, identifying approximately 50% of very early‐stage cases that were otherwise AFP‐negative [23]. Meanwhile, cfDNA fragment signatures have emerged as refined tools for early HCC detection [24]. A recently published study developed a system to identify cfDNA fragmentation features specific to early malignancies (Multicancer Early Detection), which precisely predicted early‐stage HCC [25]. In addition, the INSPECTOR study utilized human TET enzyme‐assisted whole‐methylome sequencing to establish an early‐HCC diagnostic pattern, achieving a reported sensitivity of 100% [26]. Furthermore, the GUIDE signature, which integrates both methylation and fragmentomics features, has demonstrated outstanding discrimination power in clinical evaluations [27]. Aligning with the paradigm of precision medicine, a recent cutting‐edge study from our institution presented at the 2025 ESMO Congress (Abstract 1472P) demonstrated the powerful utility of targeted cfDNA methylation deep sequencing in the perioperative setting. Specifically, the researchers established a tumor‐naive model capable of precise preoperative risk stratification (identifying high‐risk patients with a 9.07‐fold increased recurrence risk) and a tumor‐informed model yielding exceptional accuracy for postoperative minimal residual disease (MRD) monitoring (achieving a negative predictive value of 98.6% at 1 month post‐surgery). These integrated dual models provide a robust molecular framework for tailored perioperative decision‐making and individualized adjuvant therapy. Beyond genomic cfDNA, cell‐free mitochondrial DNA also exhibits diagnostic potential [28]. A rigorously designed study leveraged unique cell‐free mitochondrial DNA fragmentomic features to develop a random forest algorithm‐based predictive model. This model demonstrated excellent capacity for distinguishing early HCC patients from healthy controls and high‐risk populations, with an AUC exceeding 0.980, sensitivity over 89.6%, and specificity over 95.00%, significantly surpassing the performance of AFP [29].
In parallel, peripheral immune cells and metabolites, which reflect the real‐time status of primary or metastatic lesions, represent critical tools for early HCC diagnosis [30, 31]. Yet, the clinical utility of such biomarkers had earlier been constrained by the computational cost of high‐dimensional data as well as the complexity of the multimodal integration architecture. As the large language models and AI advanced, other diagnostic instruments have been developed based on combining clinical, immunological, and radiological characteristics, where the diagnostic accuracy demonstrated has a high AUC of over 0.98 [32]. Also, deep‐learning techniques have been presented to create hybrid models between metabolic phenotyping and transcriptomics metadata [33]. Specifically, AI‐based radiomics transforms standard imaging into quantitative minable data, with deep learning models achieving sensitivities up to 84.8% in distinguishing HCC from benign nodules in high‐risk cirrhotic patients. Beyond diagnosis, these models predict microvascular invasion (MVI) and pathological grade preoperatively (AUC > 0.90), though lack of standardization and “black box” nature of these algorithms remain barriers to widespread clinical adoption [34]. A machine‐learning study was on serum N‐glycomics, whereby a model of eight metabolites was created, with an AUC greater than 0.95 [35]. Recently, a Phase III validation study demonstrated that the PAaM model, which includes a prognostic liver secretome signature with AFP, age, sex, ALBI, and platelet counts, allows accurate early‐HCC prediction in patients with contemporary etiologies of cirrhosis [36], supporting the clinical utility of integrating routine peripheral parameters for systematic early‐HCC screening.

3.3
Breakthrough Progress in Radiologic and Pathologic Diagnosis of HCC
The computed tomography (CT) and magnetic resonance imaging (MRI) Liver Imaging Reporting and Data System (LI‐RADS) diagnostic algorithm is an important tool for early‐HCC radiologic diagnosis [37]. Previously, the influence of key LI‐RADS features on diagnostic accuracy had not been clearly defined. This year, two significant meta‐analyses have clarified that most combinations of major features within the same LI‐RADS category have similar positive predictive values for diagnosing HCC in high‐risk patients [38]. Furthermore, the application of individual ancillary features was found to have no significant impact on the overall diagnostic performance of LI‐RADS when compared with major features alone [39]. These insights facilitate a more refined framework for operationalizing and deciphering LI‐RADS criteria in real‐world clinical settings. Meanwhile, dynamic enhancement information provided by multiphasic CT and MRI is a crucial parameter for HCC diagnosis [40]. Nevertheless, most current methods treat distinct phases as independent channels and overlook their temporal dynamics. A pilot investigation introduced a spatio‐temporal decoupling network (STD‐Net) for multiphasic liver lesion segmentation and characterization. This network achieved lower HD95, and superior classification accuracy, particularly for small or low‐contrast lesions [41]. AI is also increasingly incorporated into conventional radiological workflows for HCC diagnosis to improve precision. SALSA (System for Automatic Liver Tumor Segmentation and Detection), a newly reported AI tool, demonstrated a patient‐wise detection precision of 99.65%, outperforming both state‐of‐the‐art methods and inter‐radiologist assessments [42]. Interestingly, a multicenter study using retrospective measurements found that AI combined with radiologist evaluation on initial detection had the potential to become highly sensitive and specific, and it would reduce the workload of radiologists by 54.5%. This marks the beginning of a new paradigm of human–AI synergy, which could transform the clinical processes of the present‐day HCC management [43]. Conversely, being the main imaging modality to screen HCC, the ultrasound has also evolved within the last year as it is integrated with AI. The Model‐DCB contrast‐enhanced ultrasound (CEUS) model was developed through a multicenter study of 52 sites in China and was found to perform better than the junior CEUS radiologist and comparable with the senior CEUS and MRI radiologists in classifying focal liver lesions [44]. Specifically, for small HCC lesions (≤ 3 cm), a multicenter retrospective diagnostic study presented the ModelURC tool, based on multilayer perceptron (MLP), extreme gradient boosting (XGBoost) methods, and clinical features, which achieved an AUC of 0.934 for diagnosing small HCC lesions [45]. Therefore, combining AI with imaging scans represents a more cost‐effective and highly accurate approach for early‐HCC diagnosis and screening.
In summary, the main breakthroughs in the early diagnosis and screening of HCC over the past year have centered on enhancing diagnostic accuracy, innovating non‐invasive diagnostic technologies, and integrating AI. In the future, it will be important to undertake additional prospective and large‐scale diagnostic clinical studies on populations in the community, and particularly the high‐risk groups. These studies will contribute to the refinement of new diagnostic tools and models to better suit the clinical requirements and improve the rate of early detection of HCC significantly.

4
Basic Research Progress in HCC
4.1
Novel Insight Into the Parenchymal Oncogene Induced Carcinogenesis of HCC
Investigations into the mechanisms governing HCC initiation are critically important for elucidating the fundamental principles of its progression and guiding early diagnosis and precision therapy. While classical theories consider gene mutation and clonal expansion as key driving forces for HCC, two recent studies have identified hepatic zonation features as previously unrecognized key factors in HCC origin. Studies of Wnt/β‐Catenin signaling show that its oncogenic capacity requires not only pathway activation but also three concurrent events: cooperation with MYC to establish a high translational state, activity within hepatocytes that have escaped a Zone 3 fate, and the ability to leverage MAPK signaling to overcome spatial and differentiation limits [46]. Another study showed that the origin of HCC is closely related to hepatic metabolic zonation; Zone 3 (the pericentral region) is a high‐risk area for HCC, where the specific expression of GSTM2 and GSTM3 genes becomes a key driver of tumorigenesis by inhibiting ferroptosis. Crucially, the study established that the extent of mutant hepatocyte clonal expansion is not a definitive predictor of their tumorigenic capacity [47]. These studies challenge the long‐held notion that “the degree of clonal expansion determines carcinogenic potential,” indicating that the origin zone of hepatocytes is an additional key determinant of HCC development. PTEN inactivation is also a key mechanism in HCC development. Recent studies have shown that the atypical serine/threonine kinase RIOK1 undergoes liquid–liquid phase separation by recruiting IGF2BP1 and G3BP1 into stress granules, resulting in PTEN mRNA sequestration, reduced PTEN expression, and subsequent promotion of HCC progression. Furthermore, the molecular mechanism by which senescence regulates HCC origin was elucidated for the first time. Through lineage tracing, researchers identified a class of disease‐associated hepatocytes as key precursor cells for MASH‐associated HCC formation; senescence induces NRF2 activation in these cells, leading to the degradation of TP53 and FBP1 proteins and promoting malignant phenotype transition. These findings underscore the potential of developing potent DNA damage inhibitors as a prophylactic strategy for both MASH and HCC [48]. Dysregulation of the ubiquitin‐proteasome system is another molecular hallmark of HCC, with its role in early tumorigenesis now well‐documented. Research indicates that ADRM1‐ΔEx9 is a novel pathogenic variant that drives early HCC by directly binding to the tumor suppressor protein FBXW7 and promoting its proteasome‐mediated degradation; importantly, this degradation enhances the sensitivity of HCC cells to the PARP inhibitor Olaparib, providing a novel concept for treating early‐stage HCC [49]. Large‐cohort pan‐cancer proteomic analyses have further identified RNF5 as an E3 ubiquitin ligase specifically upregulated in HCC, which may serve as a promising therapeutic target [50]. In 2025, significant progress was made surrounding a rare pathological subtype: fibrolamellar carcinoma (FLC). Mechanistically, studies confirmed that DNAJB1‐PRKACA is the key gene leading to FLC occurrence; this fusion protein triggers SIK phosphorylation and inactivation, leading to dysregulation of the CREB transcriptional coactivator and acetyltransferase p300, causing transcriptional reprogramming and increased histone acetylation to drive malignant tumor growth. Basic and translational research in HCC is based on the notion of establishing high‐fidelity, immunocompromised animal models. In 2025, researchers proposed two effective approaches. One of them included the joint operation of several tumor driver genes and tumor suppressor genes to effectively generate 27 forms of spontaneous HCC mouse models with unaltered immune capability and unique genetic profiles. Not only did these models replicate various molecular pathological characteristics of human HCC, but they also demonstrated unique effects of the interaction between genes on the tumor growth [51]. The other approach utilized emerging extrachromosomal DNA (ecDNA) technology to construct specific, engineered ecDNA, establishing spontaneous HCC models via a cDNA‐mediated focal gene amplification strategy, providing important ideas for studying ecDNA biology and developing cancer models. This often‐overlooked characteristic of focal oncogene amplification holds significant clinical relevance and necessitates consideration in designing future murine models of ecDNA‐driven human cancers [52].

4.2
Breakthrough in Microenvironmental Ecosystem of HCC and Immunotherapy
The microenvironmental ecosystem is a new philosophy, which gives more importance to dynamic interaction, adaptive evolution, and holistic collaboration founded on the tumor microenvironment (TME). In the last 12 months, various studies devoted to the microenvironmental ecosystem have given many new materials on the systemic cognition of HCC appearance and development. To decode the ecological structure of the HCC microenvironment, researchers utilized computational pathology and spatial proteomics to reveal two stromal architecture prototypes (FR+ and FR−) that dictate contrasting prognostic outcomes [53]. Another study combined spatial transcriptomics and AI with the analysis of the comprehensive spatial distribution of immune cell infiltration into the HCC microenvironment and provided an innovative model to predict postoperative recurrence, obtaining an accuracy of 82.2%. In addition, HCC shows a profound difference in the ecosystem between recurrent and the primary version; a particular category of suppressive DCs (PD‐L1 + CD103 + DCs) explicitly promotes the advancement of recurrent HCC. As a result, liver‐targeted NRP1 inhibition is a promising niche‐guided prophylactic therapy to prevent the progression of the lesions into HCC [54]. In the early stage of HCC, a domestic team revealed for the first time a reprogramming event of tissue‐resident macrophages actively initiated by MEF2D+ precancerous cells, depicting a synergistic mutual‐promotion network between the two, providing granular insights into the primordial crosstalk between the HCC “seed” and its supportive niche. In the HCC progression stage, research from the author's department identified HMGB2 as a novel mediator of CD8+ T cell exhaustion; HMGB2 promotes HCC progression by curbing oxidative phosphorylation in CD8+ T cells while reducing the interferon‐γ response in tumor cells [55]. Furthermore, bidirectional communication between HCC cells and MDSCs mediated by ETV5 can remodel the microecological T cell landscape, hindering immune killing and driving HCC development [56]. Notch1 signaling is an important factor promoting HCC development and immune escape, and the latest research reveals sex differences in the immune response to this signal: males carrying NOTCH1‐driven tumors exhibit enhanced anti‐tumor immune responses, while females carrying NOTCH1‐driven tumors exhibit immune escape and resistance to immunotherapy caused by defects in dendritic cell (DC)‐mediated CD8+ T cell activation [57]. This paper is the first effort to show that sexual dimorphism of immunity could be mediated by sex chromosome‐bound genes but not entirely driven by sex hormones, showing how a restoration of the DC‐CD8 + T cell axis through CD40 agonist could be beneficial in treatment. In the case of HCC immunotherapy, the group of the author has worked out a novel strategy to construct artificial patches of antigens on the surface of target cells in situ, with proximity labeling to catalytically amplify tumor antigens and high‐density construction of artificial antigen patches (PATCH). This technology has proven effective in overcoming the drawbacks of high concentrations of antigens and low‐targeting specificity of conventional immunotherapy, and this method has shown high efficiency in preclinical models as well as in humanized systems. The PATCH conceptual framework defines a new paradigm for the maximization of therapeutic efficacy of such immunomodulatory plans [58]. In addition, the author's department proposed that targeting BCL9 can reshape the immunosuppressive “cold” TME into an immune‐active “hot” TME via a sophisticated dual‐action mechanism: inducing M1 polarization of macrophages, and reducing CD24 expression on the tumor cell surface, thereby sensitizing immunotherapy [59]. Another novel oncolytic virus developed by Chinese researchers is VG161, which reconstructs the immune microenvironment in a variety of ways, including enhancing infiltration and action of effector T cells and NK cells; this regimen demonstrated good outcomes in a Phase I clinical trial in patients with advanced HCC resistant to conventional systemic therapy. These results point to the need to increase the size of the eligible population of patients and to emphasize the potential of a combination of VG161 and sequential systemic regimens to maximize clinical outcomes [60]. Regarding the prediction of HCC immunotherapy efficacy, multiple novel response markers have been revealed: tumor‐infiltrating GZMK + CD8+ effector/effector memory T (Teff/Tem) cells, including progenitor exhausted T (Tpex) cells, and previously underappreciated circulating effector memory T (cTem) cells rich in HBV specificity were found in patients responding well to anti‐programmed cell death protein 1 (PD‐1) combined with lenvatinib treatment; whereas in non‐responding patients, KIR + CD8+ T cell subsets and FOXP3+ CD4+ regulatory T cells were significantly enriched. This study comprehensively delineates the distinct T‐cell dynamic landscapes characterizing combination therapy versus anti‐PD‐1 monotherapy in HCC [61].

4.3
Aberrant Metabolism as a Driving Force for Microenvironment Remodeling and HCC Deterioration
Metabolic reprogramming represents a quintessential hallmark of hepatocarcinogenesis, underscored by the liver's central role in metabolic homeostasis. With breakthroughs in metabolomics detection technologies in recent years, an increasing number of associations between novel metabolic abnormalities and HCC occurrence and development have been revealed, especially the regulatory role of tumor metabolism on the immune microenvironment. A major breakthrough in the HCC metabolism field in 2025 was revealing the impact of metabolism on Tertiary Lymphoid Structures (TLS), a key structure affecting immune response. A study confirmed that ATP‐citrate lyase (ACLY) is a core target regulating tumor metabolic‐immune interaction; targeting ACLY inhibition can upregulate B cell chemokine 13 (CXCL13), recruit B cell aggregation, increase B cell tumor infiltration, and promote TLS formation, remodeling the TME to inhibit MASH‐HCC progression. This study redefines ACLY as a critical bridge linking metabolic regulation to antitumor immunity, rather than merely a driver of proliferation. Consequently, metabolic intervention targeting this pathway emerges as a novel and potent strategy to overcome immunotherapy resistance [62]. Another study proposed molecular subtypes of immature TLS for the first time and revealed the unique characteristics of different subtypes in regulating anti‐tumor immunity and immunotherapy response; on this basis, it further elucidated a novel mechanism whereby tumor cells with high tryptophan metabolism inhibit TLS maturation by limiting the proliferation, activation, and differentiation of B/T cells and proposed a novel strategy targeting tumor tryptophan metabolism to inhibit tumor progression and improve anti‐PD‐1 efficacy. By confirming that limiting tryptophan metabolism accelerates TLS maturation and suppresses tumor growth, this research validates metabolic modulation as a viable therapeutic avenue to reinvigorate TLS‐mediated immunity in HCC [63]. The important role of bile acid metabolism in HCC was another significant breakthrough in basic metabolic research this year. It was discovered for the first time that primary bile acids induce oxidative stress in T cells, while secondary bile acids (e.g., lithocholic acid) inhibit T cell function by triggering endoplasmic reticulum stress. It was confirmed that inhibiting bile acid synthesis or supplementing ursodeoxycholic acid can significantly improve tumor immunotherapy efficacy in HCC, providing new ideas for sensitizing immunotherapy from a metabolic perspective [64]. In addition, the loss of expression of the key bile acid synthesis enzyme AKR1D1 was confirmed to be a key cause of impaired NK cell function within the HCC microenvironment; AKR1D1 can upregulate the abundance of chenodeoxycholic acid and deoxycholic acid to activate NK cells, maintaining their immune killing function, thereby inhibiting HCC pathogenesis [65]. The crosstalk between metabolism and myeloid cells in the microenvironment has been further elucidated: cross‐species longitudinal single‐cell multi‐omics analyses revealed that although adaptive changes in hepatocytes under chronic metabolic stress support short‐term cell survival and tissue homeostasis, they exert long‐term effects on the immune microenvironment, specifically activating CD9+ TREM2+ scar‐associated macrophages, laying the groundwork for HCC occurrence. This research offers a pivotal framework for comprehending tissue adaptation to stress, simplifying intricate pathological manifestations into core mechanisms. It underscores how early metabolic disruptions can initiate enduring cellular changes and shifts in homeostasis, making the liver more susceptible to cancer [66]. Notably, this study reveals that metabolic imbalances can trigger malignancy without relying on genetic mutations, a traditional pathway for cancer development. Another significant finding is that erythropoietin (EPO) produced by tumors disrupts macrophages' iron metabolism, leading to reduced intracellular iron levels. This weakens macrophages' ability to present antigens, facilitating immune evasion [67]. Recent multi‐omics and spatial analyses have also highlighted the role of tumor‐intrinsic SREBP1 signaling in creating an immunosuppressive environment by altering macrophage behavior and preventing CD8+ T‐cell infiltration [68]. Furthermore, N1‐acetylspermidine [69] and lactate [70] effluxed by HCC cells can reprogram microenvironmental macrophages, constituting a complete intercellular communication network that promotes HCC immune escape and metastasis. Finally, breakthroughs have also been made in developing novel metabolic‐based therapeutic strategies for HCC. Researchers developed a novel blocking antibody against the key lipid metabolism target CD36; through HCC animal models and humanized models, it was confirmed that this antibody effectively remodels the HCC immune microenvironment, reducing intratumoral Tregs, enhancing CD8+ T cell infiltration, and improving the cytotoxic function of CD8 + T cells, effective in both immune‐hot and cold HCC [71]. Another team, addressing the clinical dilemma of “dual metabolic‐immune disorder, lack of specific drugs, and immunotherapy tolerance” in MAFLD/MASH‐HCC, developed a “metabolic reprogramming mRNA‐lipid nanoparticle” platform (Def‐LNP@mRNATCPTP). Using a “four‐in‐one” strategy of “antioxidant‐delivery‐expression‐regulation,” a single intravenous administration achieved sustained remodeling of the liver microenvironment and significantly enhanced immunotherapy response [72]. In addition, targeting the conversion of lysophosphatidic acid to phosphatidic acid mediated by AGPAT4, which causes HCC lineage plasticity and drug resistance, the research team developed a covalent inhibitor targeting the AGPAT4 Cys228 residue, demonstrating excellent inhibitory effects in multiple preclinical models [73].

4.4
Emerging Evidence for the Correlation Between HCC Progression and Gut Microbes
A sophisticated and multifaceted interplay exists between microecological dysbiosis and the pathogenesis of HCC. As the human body's largest microecosystem, gut microbial dysbiosis exerts a profound impact on HCC progression through the “gut‐liver axis” pathway. Studies indicate that while fecal microbial diversity decreases from healthy controls to cirrhosis, it surprisingly increases from cirrhosis to early HCC. A specific 30‐microbial marker signature has achieved an AUC of 80.64% in distinguishing early HCC from non‐HCC samples in patients with HBV‐related cirrhosis [31]. In 2025, multiple novel microbes were discovered to be directly involved in HCC occurrence, development, and drug resistance. Cutibacterium mitsuokai (C. mitsuokai) can disrupt the gut barrier and translocate to the liver; after colonizing the tumor, C. mitsuokai secretes quinolinic acid to activate TIE2, subsequently activating PI3K/AKT signaling and driving HCC progression [74]. In addition, mouse model‐based research found that under HCC conditions, Klebsiella pneumoniae in the gut can translocate to tumor tissues and interact with TLR4 on HCC cells via surface PBP1B, promoting cell proliferation [75]. More importantly, a study based on tumor multifocal exome sequencing combined with 16S rRNA sequencing found heterogeneous microbial communities existing between different nodules in patients with multifocal HCC; analyzing the primary lesion (or any single lesion) alone cannot fully describe the genomic and microbiome landscape of multifocal HCC, emphasizing the necessity of multi‐point sampling [76]. Regarding HCC treatment, the latest research found that Phocaeicola vulgatus (P. vulgatus) is significantly enriched in the gut of patients unresponsive to immunotherapy. P. vulgatus inhibits the generation of indole‐3‐acetic acid (IAA) in the gut by altering tryptophan metabolism pathways, thereby weakening CD8+ T cell activity and ultimately leading to HCC immunotherapy resistance; supplementing IAA can restore CD8‐positive T cell toxicity, reverse the immunosuppressive effects caused by P. vulgatus, and restore the efficacy of immune checkpoint inhibitors (ICIs) [77]. Furthermore, valeric acid, a metabolite produced by Lactobacillus acidophilus, can inhibit HCC cell activity and improve the integrity and function of the gut barrier, thereby inhibiting HCC progression [78]. Thus, modulating gut microbial homeostasis (e.g., via probiotic supplementation or fecal microbiota transplantation) has emerged as a promising strategy for HCC prevention and therapy.

5
Advances in Treatment of HCC
5.1
Surgical Treatment Approaches for HCC
Surgery is still the main way to cure HCC at an early stage [79]. According to the long‐term research carried out by Zhongshan Hospital, Fudan University, surgery can increase the 5‐year survival rate of HCC patients at an early stage from 10% to 71% [3]. This result was confirmed by Japan's nationwide registration data between 1978 and 2015. In this study, the 5‐year overall survival (OS) rate of more than 35,000 patients who had undergone surgery for HCC was 69.1%, which further proved the great effect of modern surgical treatment on the long‐term prognosis of HCC patients [80]. Based on Korea's national registration data, the 5‐year relative survival rate of liver cancer increased from 20.6% between 2001 and 2005 to 39.4% between 2018 and 2022, reflecting sustained improvement in early detection, staging, and comprehensive treatment [81]. Overall, these nationwide data sets put the Zhongshan experience into a larger international context and suggested that hepatic resection could be the main method of surgery for suitable HCC patients. With the increasing use of minimally invasive techniques, laparoscopic liver resection (LR) has been used more often, contributing to the continuous improvement of surgical safety. A retrospective analysis demonstrates that although fluorescence‐guided laparoscopic LR involves slightly longer operative and hilar occlusion times, it significantly reduces intraoperative blood loss and postoperative complications compared to conventional laparoscopy [82]. For patients with decompensated liver function, liver transplantation (LT) provides a definitive treatment for both the malignancy and underlying end‐stage liver disease, yet the limited supply of donor organs remains a primary constraint [83]. Xenotransplantation offers a potential solution to this shortage. Dou Kefeng's team reported a milestone achievement in which a gene‐edited pig liver maintained functional survival in a human recipient for 10 days [84]. Regarding the clinical selection between LR and LT, South Korean researchers utilized multicenter big data and machine learning to develop an individualized decision model. This model predicts 3‐year OS with higher precision than conventional clinical assessments, establishing an evidence‐based framework for optimal treatment selection [85]. Furthermore, advancements in conversion therapy are broadening the population of patients eligible for surgical intervention [86]. The integration of these techniques and strategies continues to advance the surgical management of liver cancer toward a safer, more precise, and more effective paradigm.

5.2
Radiation Therapy Strategies for HCC
Historically, external beam radiation therapy (EBRT) for HCC has been largely stagnant, principally hampered by the liver's intrinsic radiosensitivity and the technical hurdles associated with conformal precision. However, the rapid evolution of targeted techniques, including stereotactic body radiation therapy (SBRT) and proton beam therapy, has redirected clinical focus toward the therapeutic value of EBRT [18]. In the curative treatment of early‐stage HCC, two studies recently published have shown that SBRT could achieve similar treatment effects compared to surgical resection and ablation [87, 88]. For intermediate‐stage HCC patients, according to the results of meta‐analysis, SBRT could achieve a similar OS rate with that of transcatheter arterial chemoembolization (TACE), but had a better local control rate, which might be used as another choice of local treatment option [89]. For advanced HCC patients, SBRT combined with immunotherapy and targeted agents might become one of the treatment options for patients with portal vein tumor thrombus, according to the results of a Phase II study [90]. Therefore, the role of radiotherapy was expanded from the palliative treatment of advanced HCC patients to the local treatment option of medium‐stage HCC patients, even to early‐stage HCC patients who were not suitable for surgery [91].

5.3
Ablation Therapy Approaches for HCC
Ablation therapy—comprising radiofrequency ablation (RFA), microwave ablation (MWA), cryoablation, and percutaneous ethanol injection—remains a definitive curative modality for very early (BCLC Stage 0) and early‐stage (BCLC Stage A) HCC, while also serving as a critical bridging strategy for patients awaiting liver transplantation [18, 92]. Furthermore, ablation techniques, particularly RFA, are well‐established and highly effective treatment options for recurrent HCC. They offer a minimally invasive yet definitive approach to achieve local tumor control, especially for small‐sized intrahepatic recurrences [93]. Although RFA is historically the most established technique, MWA has accrued substantial clinical favor, owing to its superior procedural kinetics and mitigation of the heat‐sink effect. Supporting this shift, a recent randomized controlled trial demonstrated that MWA offers superior therapeutic efficacy over single‐needle RFA, justifying its expanded implementation [94]. Despite these strengths, conventional thermal ablation is often constrained by tumor location; lesions adjacent to the liver capsule, diaphragm, or viscera pose a high risk of thermal injury to surrounding structures. To address this, nanosecond pulsed electric field (nsPEF) ablation serves as a useful non‐thermal alternative, allowing for precise targeting of tumors in anatomically challenging sites. The first prospective multicenter study on nsPEF for high‐risk HCC reported recurrence‐free survival (RFS) rates of 72.2%, 51.7%, and 43.5% at 1, 2, and 3 years, respectively, confirming its utility when thermal methods are contraindicated [95]. To address post‐ablation recurrence, switching from ultrasound to CT or MRI guidance has improved both procedural precision and safety [96]. In addition, combining AI with multiparametric MRI has outperformed conventional clinical metrics in predicting early recurrence, guiding more tailored adjuvant management [97]. These improvements in ablation techniques, imaging, and AI continue to secure ablation's role in treating HCC.

5.4
Interventional Therapy
As a primary treatment for intermediate‐to‐advanced HCC, interventional therapy has undergone steady improvements in technical refinement and combination strategies, offering more targeted options [18]. Balloon‐occluded alternative infusion of cisplatin solution and fragmented gelatin particles of transarterial chemoembolization (BOAI‐TACBOAI‐TACE) has shown clear clinical benefits in patients with intermediate‐to‐advanced HCC exceeding the “up‐to‐seven” criteria. In this cohort, it yielded an objective response rate (ORR) of 77.8% and a disease control rate (DCR) of 88.9% while preserving liver function, providing a new alternative for those beyond traditional TACE indications [98]. Another Phase II clinical trial evaluated combining ultrasound‐triggered microbubble destruction with yttrium‐90 transarterial radioembolization (⁹⁰Y‐TARE). Data indicated that this combination improved the ORR, increasing the complete response (CR) rate from 44% to 60%, which helps address the limitations of traditional radioembolization [99]. The value of postoperative adjuvant TACE (PA‐TACE) remains debated, primarily due to the challenge of accurately identifying patients at high risk of recurrence. An online predictive calculator, developed using preoperative CT imaging features and clinicopathological parameters, now enables individualized estimates of survival benefits following PA‐TACE. For the 35.4% of patients identified as suitable candidates, the 5‐year OS rate increased by 19.4%, and the median survival time increased by 22.5 months, offering an objective basis for adjuvant therapy decision‐making [100]. Combination strategies have become a key focus of modern interventional oncology research. A recent Phase III trial showed that TACE plus apatinib for unresectable HCC achieved a median progression‐free survival (PFS) of 6.1 months, compared to 3.4 months observed in the TACE monotherapy group. In addition, median OS was extended to 28.9 months, with ORR and DCR improving to 58.1% and 87.2%, respectively, with manageable toxicity [101]. Another major point of contention is that following an R0 resection, the liver is macroscopically tumor‐free; thus, administering TACE—which primarily targets visible hypervascular lesions—lacks a clear anatomical target. In contrast, hepatic arterial infusion chemotherapy (HAIC) has demonstrated more definitive advantages in high‐risk populations. Notably, a pivotal Phase III randomized controlled study established that postoperative adjuvant HAIC with FOLFOX significantly improved disease‐free survival and OS compared to active surveillance in patients with MVI‐positive HCC after curative resection [102]. This highlights the critical need to tailor adjuvant regional therapies based on specific pathological risk factors. In summary, the rapid emergence of research findings has not only expanded the technical modalities and combination strategies of interventional therapy but has also significantly improved survival outcomes through precise stratification and optimized management. These developments further solidify the essential role of interventional therapy in the comprehensive management of HCC.

5.5
Systemic Therapy for HCC
5.5.1
Perioperative Therapy for HCC
5.5.1.1
Neoadjuvant Therapy
Recent investigations into neoadjuvant strategies have focused on downstaging tumor burden and eliminating micrometastases in patients with high‐risk or borderline resectable features, utilizing combinations of ICIs, TKIs, and locoregional therapies (Table 1).
The Phase II/III CARES‐009 trial provides high‐level evidence supporting a perioperative approach for resectable HCC at intermediate or high risk of recurrence. Patients randomized to receive perioperative camrelizumab plus rivoceranib demonstrated a significantly prolonged median event‐free survival of 42.1 months compared to 19.4 months in the surgery‐alone arm (HR: 0.59), accompanied by a major pathological response (MPR) rate of 35.1% [103].
Addressing the challenges of borderline resectable HCC (BRHCC), the BRHCC‐I Phase Ib/II trial evaluated a triplet regimen comprising TACE, lenvatinib, and camrelizumab. This aggressive neoadjuvant strategy yielded promising pathological outcomes, with an MPR rate of 61.4% in the Phase II cohort and translated into 1‐year and 2‐year RFS rates that significantly exceeded those of a matched upfront surgery control group [104].
Specific attention has also been directed toward single large HCC (≥ 10 cm), a subgroup historically associated with poor surgical prognosis. The randomized Phase II NEO‐START trial investigated sequential TACE followed by camrelizumab and apatinib, reporting that the neoadjuvant arm achieved a 41.7% MPR rate and, notably, observed no recurrences at the time of reporting compared to five recurrence events in the surgery‐alone arm [105].
The CAR_Hero study explored the efficacy of HAIC combined with the bispecific PD‐1/CTLA‐4 antibody cadonilimab in patients with multinodular HCC (CNLC Ib/IIa). The combination arm demonstrated superior antitumor activity with an MPR rate of 78.6% and a significantly longer RFS compared to a propensity score‐matched direct hepatectomy cohort, suggesting that incorporating bispecific antibodies with HAIC may offer distinct advantages in complex multinodular cases [106].

5.5.1.2
Conversion Therapy
For patients with initially unresectable HCC, the paradigm has shifted toward “conversion surgery” following successful systemic or combination therapy, with recent data confirming that surgical resection in responders offers superior survival outcomes compared to non‐surgical maintenance.
The multicenter, randomized Phase III TALENTop trial addressed the critical decision‐making process for patients with locally advanced HCC who achieved partial response or stable disease after induction with atezolizumab plus bevacizumab. Interim analysis revealed that patients randomized to LR experienced a significant improvement in time‐to‐treatment failure (HR: 0.60, TTF: 11.8 vs. 20.4) and a favorable trend in OS compared with those who continued systemic therapy alone, validating the role of surgery as a consolidative treatment in responders [107].
Complementing these findings, the prospective Phase II RACB study further established the feasibility of atezolizumab plus bevacizumab in the conversion setting. The study reported a conversion resection rate of 52.0% among enrolled patients with initially unresectable disease, with a 6‐month PFS rate of 59.1%, confirming that this first‐line systemic regimen can effectively downstage tumors to facilitate curative‐intent resection [108].
Real‐world evidence from a large retrospective cohort study by Piao et al. corroborated these prospective data across various ICI‐based combination regimens. The analysis demonstrated that patients who underwent conversion surgery after meeting resectability criteria achieved a median PFS of 29.4 months versus only 11.2 months in responders who declined surgery, identifying conversion surgery as an independent prognostic factor for improved overall and PFS [109]. Collectively, these data establish the “Systemic Induction + Surgical Consolidation” paradigm. The TALENTop trial highlights the transformative value of this approach, reducing treatment failure risk (HR: 0.60) in responders.
Further validating this strategy in a real‐world setting, the multicenter retrospective TALENTRUE study provided crucial evidence supporting the “TALENT paradigm” using a triplet regimen of atezolizumab, bevacizumab, and TACE. Analyzing 107 evaluable patients, the study demonstrated a robust ORR of 60.7% (mRECIST) and a median OS of 25.8 months. Notably, 28 patients underwent R0 resection, with 15 (53.6%) of the surgical group achieving a pathological CR. The analysis also revealed that conversion rates were inversely correlated with tumor burden (42.9% for single lesions vs. 6.9% for ≥ 4 lesions), reinforcing a tiered approach where intensive multimodal therapy effectively selects patients with high but convertable burden for curative surgery [110].
As a novel conversion strategy, the Phase II UniRTX trial evaluated the combination of cadonilimab, lenvatinib, and TACE in unresectable HCC. Preliminary results showed strong antitumour activity, with an ORR of 83.3% and a DCR of 95.8%, resulting in a conversion resection rate of 54.2%. Notably, all patients who underwent R0 resection achieved MPR, and 76.9% attained complete pathological response, suggesting that bispecific antibody‐based triplet regimens can effectively induce deep pathological responses and enable surgical conversion [111].

5.5.1.3
Adjuvant Therapy
Recent clinical research in the adjuvant setting has diversified from traditional locoregional interventions to encompass targeted therapies, combination regimens, and innovative immunotherapeutic approaches aimed at reducing postoperative recurrence.
Adjuvant lenvatinib monotherapy has demonstrated meaningful clinical activity in patients with high‐risk HCC (CNLC Stage IIb/IIIa). In a prospective exploratory study, this systemic approach achieved a median RFS of 16.1 months, with more than 60% of patients remaining recurrence‐free at 1 year. These results suggest that early intervention with TKIs can effectively target residual microscopic disease in advanced surgical candidates [112].
The combination of lenvatinib and TACE has emerged as a superior strategy for patients at high risk of recurrence. The multicenter LANCE study revealed that this combined adjuvant regimen significantly extended median disease‐free survival to 19.0 months, compared to only 10.0 months in patients receiving TACE alone. This synergy between locoregional and systemic therapy appears to provide more robust protection against early relapse than single‐modality interventions [113].
In contrast to its efficacy in high‐risk populations, adjuvant TACE monotherapy does not appear to provide survival benefits for patients with early‐stage disease. A randomized Phase III trial focusing on TNM Stage I and II HCC found no significant improvement in recurrence‐free or OS when comparing adjuvant TACE to active surveillance. Consequently, routine use of TACE in the adjuvant setting for low‐risk, early‐stage patients is currently not supported by high‐level evidence [114].
Cellular immunotherapy is an emerging approach, as demonstrated by a pilot Phase II trial exploring autologous RetroNectin‐Activated Killer (RAK) cells. High‐risk patients treated with these activated T cells achieved a 1‐year RFS rate of 79% and a 3‐year RFS rate of 68%, accompanied by evidence of enhanced systemic immunity and CD8+ T‐cell expansion. These findings suggest that adoptive cell transfer can alter the postoperative microenvironment to delay recurrence [115].
Researchers are also testing sequential immune combinations, like ropeginterferon alfa‐2b with nivolumab. A phase 1 trial in HBV‐related HCC showed strong preliminary efficacy, with all participants remaining alive and recurrence‐free at a median follow‐up of over 1000 days. Alongside its oncological impact, the therapy also induced significant viral responses, including HBsAg reduction, indicating a potential benefit in both tumor control and underlying viral management [116]. Despite some early successes, adjuvant therapy for HCC still faces mixed results and negative trials that complicate clinical consensus. One such example is the phase III IMbrave050 study evaluating adjuvant atezolizumab plus bevacizumab in high‐risk HCC patients [117]. Although an initial interim analysis reported an RFS benefit [118], recently updated data showed that this early RFS benefit was not sustained with longer follow‐up, and OS remains immature. Consequently, the updated benefit‐risk profile does not currently support this combination as a routine adjuvant therapy. The inconsistent clinical outcomes observed across various adjuvant trials likely arise from three primary factors. First, the treatments tested are completely different—from single targeted drugs to complex combinations of TACE/HAIC with systemic therapy. Second, the lack of consensus on defining “high risk of recurrence” creates uneven enrollment thresholds across studies. Third, the enrolled patient populations show significant heterogeneity, with distinct underlying liver disease etiologies, baseline liver functions, and geographic demographics. These factors collectively highlight the need for unified risk criteria and precise patient stratification in future adjuvant trial designs.

5.5.2
Pharmacotherapy for Advanced HCC
5.5.2.1
First‐Line Treatment
The first‐line treatment paradigm for advanced HCC continues to evolve, shifting decisively from tyrosine kinase inhibitor (TKI) monotherapy toward sophisticated combination regimens. These strategies predominantly incorporate PD‐1/PD‐L1 inhibitors in conjunction with either anti‐angiogenic agents or CTLA‐4 inhibitors to leverage synergistic antitumor activity.
5.5.2.1.1
Targeted Therapy Combined With Immunotherapy
In 2025, several landmark Phase III trials have solidified the survival advantages of combination therapies over traditional TKI monotherapy. The dual immunotherapy combination of nivolumab and ipilimumab was evaluated against the investigator's choice of lenvatinib or sorafenib, meeting its primary endpoint with a statistically significant improvement in OS. The combination achieved a median OS of 23.7 months compared with 20.6 months in the TKI arm, supported by a superior ORR of 36% and a remarkably durable response (median: 30.4 months) [119].
Similarly, the final analysis of the global CARES‐310 trial confirmed the robust efficacy of camrelizumab combined with the VEGFR2‐TKI rivoceranib. This regimen demonstrated a median OS of 23.8 months—the longest reported in a phase III global HCC trial to date—significantly outperforming sorafenib (15.2 months; HR = 0.64; p < 0.0001). Significant improvements were also observed in median PFS (5.6 vs. 3.7 months), establishing this combination as a new standard of care, particularly in Asian populations [120].
Regional evidence further supports this paradigm, as seen in a phase III study of toripalimab plus bevacizumab conducted in a predominantly Chinese population. This anti‐PD‐1‐based combination significantly prolonged both PFS (medianL 5.8 vs. 4.0 months; HR = 0.69) and OS (median: 20.0 vs. 14.5 months; HR = 0.76) compared with sorafenib. The manageable safety profile reported in this trial has facilitated its clinical approval in China [121].
The multi‐target TKI strategy has also seen progress through the APOLLO trial, which evaluated the combination of anlotinib and penpulimab. This multicenter Phase III trial demonstrated a more than twofold increase in median PFS compared to sorafenib (6.9 vs. 2.8 months; HR = 0.52), while extending median OS to 16.5 months (vs. 13.2 months; HR = 0.69). These findings further validate the integration of TKIs with ICIs as a robust first‐line option [122]. Notably, the emergence of regimen‐specific niche populations is reshaping clinical choice: the dual‐immunotherapy (Nivo+Ipi) regimen is particularly advantageous for patients seeking long‐term durable response (mDOR 30.4 months) without the bleeding risks of anti‐angiogenics, while the Camrelizumab + Apatinib (C + A) regimen remains a preferred standard for Asian populations with HBV‐related HCC.
Innovative strategies are also moving beyond monospecific checkpoint inhibitors toward next‐generation bispecific antibodies. A 2025 real‐world study evaluated Cadonilimab, combined with TKIs in unresectable HCC. In the first‐line setting, this regimen demonstrated a median OS of 13.7 months and a median PFS of 6.7 months, showing a trend of superior survival benefits compared to a propensity score‐matched cohort receiving traditional PD‐1 inhibitors plus TKIs (median Overall Survival [mOS]: 6.7 months; median Progression‐Free Survival [mPFS]: 3.3 months). Crucially, the bispecific combination maintained a manageable safety profile (Grade ≥ 3 TRAEs: 20.5%) even in patients with significant liver dysfunction (Child‐Pugh score ≥ 8), suggesting that bispecific antibodies may represent a potent therapeutic upgrade for improved efficacy and safety in complex clinical scenarios [123].
Beyond advanced disease, the systemic combination of atezolizumab and bevacizumab is competing with the traditional role of locoregional therapy in intermediate‐stage HCC. The phase IIIb ABC‐HCC trial, which directly compared this systemic regimen against TACE, showed an improved time to failure of treatment strategy (TTFS) for the systemic arm (14.6 vs. 9.5 months; HR = 0.55). These findings indicate that earlier systemic therapy can outperform locoregional interventions in specific patient subsets [124].

5.5.2.1.2
Interventional Therapy Combined With Targeted and Immunotherapy
The combination of TACE with systemic agents aims to block the surge in angiogenic factors, such as vascular endothelial growth factor (VEGF), that typically follows embolization, thereby improving the overall immune response. The Phase III EMERALD‐1 study is the first to show a significant PFS benefit for such a combination, with patients receiving TACE plus durvalumab and bevacizumab reaching a median PFS of 15.0 months, compared with 8.2 months for TACE plus placebo (HR = 0.77; p = 0.032). The inclusion of bevacizumab proved essential for maximizing the efficacy of this triplet regimen [125].
Similarly, a Phase III trial investigated the combination of TACE, pembrolizumab, and lenvatinib in patients with non‐metastatic disease. The combination significantly improved median PFS to 14.6 months compared with 10.0 months with TACE alone (HR = 0.66). Although OS data remain immature, the observed positive trend (HR = 0.80) and the higher 24‐month OS rate (75% vs. 69%) highlight the value of this intensive approach [126].
Optimizing the delivery of these combinations is also a focus of current research, as demonstrated by a phase III trial utilizing an “on‐demand” TACE schedule combined with atezolizumab and bevacizumab. In patients with intermediate‐to‐high tumor burden, this strategy significantly improved TACE‐specific PFS to 11.3 months, compared with 7.03 months in the TACE‐only arm (HR = 0.71) [127].
Phase II data continue to support the expansion of these interventional‐systemic hybrids. TACE combined with camrelizumab and rivoceranib in unresectable HCC elicited a median PFS of 11.0 months compared with 3.1 months with TACE alone (HR = 0.34), accompanied by a substantial increase in ORR (65.0% vs. 30.0%). These results reinforce the potent synergistic potential of combining PD‐1 inhibition and anti‐angiogenic TKIs with interventional procedures [128]. The clinical significance of this synergy, as demonstrated in EMERALD‐1, lies in the use of bevacizumab to counteract post‐TACE VEGF surges. This triplet strategy is most suitable for BCLC‐B patients with preserved liver function, although its reliance on bevacizumab limits its applicability in patients with high‐risk varices.

5.5.2.2
Second‐Line Treatment
Advancements in second‐line therapy increasingly involve novel mechanisms of action, including next‐generation antibodies and biomarker‐selected targeted agents. Porustobart, a novel anti‐CTLA‐4 heavy chain‐only antibody designed to enhance regulatory T‐cell (Treg) depletion, has shown encouraging second‐line efficacy in combination with toripalimab. In a phase Ib/II study, this combination achieved an ORR of 40.0% and a median PFS of 5.7 months in patients naive to prior immunotherapy, though efficacy dropped significantly in those who had previously progressed on ICIs. Correlative analysis suggests that baseline Treg levels and T‐cell proliferation markers may serve as useful predictors of response [129].
Biomarker‐driven targeted approaches, exemplified by the selective FGFR4 inhibitor irpagratinib, represent another major second‐line strategy. In a Phase 2 study evaluating irpagratinib plus atezolizumab for FGF19‐overexpressing HCC, the combination achieved an ORR of 51.7%. Consistent efficacy was observed across both first‐line (ORR: 50.0%) and pre‐treated (ORR: 52.9%) subgroups, suggesting that molecular stratification can help overcome resistance in refractory HCC [130]. The discrepancy in outcomes between unselected immune sequencing (e.g., porustobart's 0% ORR in pre‐treated patients) and biomarker‐driven selection (e.g., irpagratinib's 52.9% ORR in FGF19+ patients) indicates that the second‐line treatment should prioritize precise molecular stratification rather than random sequencing.

6
Conclusion and Future Expectation
HCC management is currently shifting toward early intervention and the integration of novel technological approaches. This transition aims to establish a clinical framework spanning from population‐based prevention to individualized precision intervention. For prevention and screening, a global action plan led by Chinese experts has established a stratified “three‐tier defense” framework, centering on hepatitis B vaccination, antiviral therapy, and the targeted management of high‐risk populations. At the same time, combining liquid biopsies (like cfDNA) with AI radiomics is significantly improving early diagnosis capabilities. By enabling non‐invasive, high‐precision early detection during the pre‐symptomatic phase, these technologies are driving a shift from treating established disease to proactive prevention and early detection. Concurrently, basic research is being reshaped by single‐cell multi‐omics, moving toward more high‐resolution, dynamic, and translational frameworks. Researchers are currently using spatial technologies to map the TME and track how cells evolve over time, rather than just taking a static snapshot. Together, these clinical and laboratory advances help translate basic mechanistic discoveries into practical treatments that improve HCC patient care (Table 1).
6.1In summary, the management of HCC is rapidly evolving toward a multidisciplinary, precision‐based, and technology‐driven paradigm. With the ongoing accumulation of robust clinical trial data, future treatment strategies will center on “personalized, continuum‐of‐care management”—dynamically integrating local and systemic therapies at different disease stages and tailoring decisions based on tumor biology, liver functional reserve, and individual prognostic factors. Surgical approaches are expected to evolve toward ultra‐minimally invasive techniques and autonomous surgical ecosystems; AI–assisted preoperative planning and robotic surgical platforms are poised to significantly enhance the safety, precision, and recovery outcomes of complex hepatic resections. Meanwhile, innovative solutions, such as gene‐edited porcine xenotransplantation, may help alleviate the critical scarcity of donor organs. There will be closer synergy between local treatments (including stereotactic body radiotherapy, novel nonthermal ablation modalities, and refined TACE) and systemic therapies (such as ICIs combined with anti‐angiogenic agents, bispecific antibodies, and novel targeted compounds). Optimizing the timing, sequencing, and patient selection for these combinations will represent a pivotal focus of upcoming clinical research. Real‐time monitoring through liquid biopsy and AI predictive models provides a framework for evaluating therapeutic responses and predicting patient outcomes. By mapping the TME with spatial technologies, researchers can identify distinct cellular neighborhoods associated with therapeutic resistance. For example, identifying CAF‐rich populations helps select patients who might benefit from pairing FAP inhibitors with PD‐1 blockade. Using such high‐resolution spatial data, future clinical trials can shift from unselected enrollments to biomarker‐driven precision strategies, thereby improving the efficacy of combined immunotherapy in HCC. However, translating these high‐dimensional technologies into clinical practice still faces significant obstacles. The widespread use of spatial multi‐omics and AI‐driven radiomics is currently constrained by the lack of cross‐platform standardization, the “black box” nature of these algorithms, and high clinical application costs. Furthermore, the extreme inter‐ and intra‐tumoral heterogeneity of HCC means that static biomarkers often fail to reflect how resistance evolves over time. While novel targets within the metabolic machinery (e.g., ACLY, tryptophan metabolism) and the gut–liver axis have been identified, developing specific pharmacological modulators with adequate bioavailability and safety remains a significant translational challenge. Future research must therefore focus on understanding the molecular mechanisms of acquired resistance to ICI‐TKI combinations, establishing consensus guidelines for liquid biopsy, and developing cost‐effective, decentralized diagnostic tools. Bridging these gaps is essential to ensure that the promise of precision medicine becomes accessible for global HCC management, rather than being limited to academic centers. A comprehensive approach—spanning early detection to personalized therapy—holds the most realistic potential for altering the clinical course of HCC and improving patient outcomes (Table 2).

Author Contributions

Bo Hu: writing – original draft. Qiang Gao: writing – review and editing.

Funding
The authors have nothing to report.

Ethics Statement
The authors have nothing to report.

Consent
The authors have nothing to report.

Conflicts of Interest
The authors declare no conflicts of interest.