Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide, with an evolving etiological shift towards metabolic dysfunction-associated fatty liver disease (MAFLD)^[1-4]. Its treatment landscape has transformed significantly from localized therapies to systemic options. The advent of tyrosine kinase inhibitors (TKIs), such as sorafenib and lenvatinib, marked the first breakthrough in targeted therapy for advanced disease^[5-7]. Subsequently, a profound paradigm shift was ushered in by immune checkpoint inhibitors (ICIs). While initial monotherapy results were modest, combination strategies - such as atezolizumab plus bevacizumab (“T+A”) and durvalumab plus tremelimumab (“STRIDE”) - have significantly improved patient outcomes and established new first-line standards [Table 1]^[8-14]. Despite this progress, a substantial clinical bottleneck persists.

Currently, only 20%-30% of patients achieve durable objective responses to ICI-based regimens, leaving the vast majority to face primary or secondary resistance, which severely limits overall survival gains^[10,15,16]. The mechanisms driving this resistance are primarily rooted in the tumor microenvironment (TME). Primary resistance is often associated with an immune-“cold” TME, characterized by a lack of cytotoxic T-cell infiltration. This is frequently driven by constitutive Wnt/β-catenin signaling that impairs chemokine-mediated immune cell recruitment^[17-20]. Conversely, secondary resistance develops in initially responsive tumors through adaptive immune evasion, such as the compensatory upregulation of alternative inhibitory checkpoints [e.g., lymphocyte-activation gene 3 (LAG-3), T-cell immunoglobulin and mucin-domain containing-3 (TIM-3), T-cell immunoreceptor with Ig and ITIM domains (TIGIT)] leading to T-cell exhaustion, and the expansion of immunosuppressive cells [e.g., regulatory T-cells (Tregs) and myeloid-derived suppressor cells (MDSCs)]^[21-24]. This results in the dynamic remodeling of the TME from an “inflamed” to an “excluded” state^[25,26]. The multifaceted nature of these resistance mechanisms is illustrated in Figure 1 and further detailed in Table 2^[27-34].

While landmark trials such as CheckMate-040^[13] and KEYNOTE-224^[8] established the role of ICI monotherapy in HCC, the subsequent failure of KEYNOTE-240^[35] highlighted the persistent challenge of primary resistance. Even with current front-line combinations (e.g., IMbrave150^[10] and HIMALAYA)^[12], the objective response rate (ORR) has plateaued at approximately 30% - a clinical ceiling increasingly attributed to the gut-liver axis. The most compelling evidence for this link comes from the “antibiotic penalty”. Recent multicenter cohorts reveal that patients receiving broad-spectrum antibiotics shortly before or after initiating ICI therapy experience significantly worse clinical outcomes [ORR, progression-free survival (PFS), and overall survival (OS)]. This disruption of microbial diversity underscores that an intact, diverse gut microbiota is a prerequisite for optimal systemic anti-tumor immune responses^[36].

The clinical challenge is profound, as the overarching efficacy of immunotherapy is fundamentally constrained by these resistance patterns^[37]. However, the traditional tumor-centric perspective cannot fully explain the observed inter-patient heterogeneity in treatment response, pointing to the involvement of systemic host factors^[38,39]. Recent evidence highlights the gut microbiota, operating through the “gut-microbiota-liver axis”, as a key systemic regulator of hepatic anti-tumor immunity and a decisive factor in ICI efficacy^[40,27]. To overcome the resistance observed in major trials such as IMbrave150, future efforts must focus on translating microbiome research into clinical practice. This includes developing microbiome-based companion diagnostics for patient stratification and integrating microbial “preconditioning” [e.g., fecal microbiota transplantation (FMT) or next-generation probiotics] with established ICI regimens. Such strategies, currently being explored in early-phase trials [e.g., Fecal Microbiota Transplant combined with atezolizumab/bevacizumab in patients with hepatocellular carcinoma (FAB-HCC)^[41]], aim to transform the gut microbiota from a driver of resistance into a therapeutic ally, potentially breaking the current efficacy bottleneck in HCC immunotherapy. This review systematically explores the mechanisms by which the gut microbiota influences immunotherapy resistance in HCC and evaluates the translational potential of microbiota-targeting strategies to overcome this challenge.

The gut microbiota is a vast and dynamic microbial community residing within the human intestinal tract, primarily composed of bacteria, archaea, fungi, and viruses. The collective sum of their genomes is formally defined as the “microbiome”. The functional repertoire encoded by this microbiome vastly exceeds that of the human genome, earning it the designation of the human “second genome”^[42]. Beyond its fundamental roles in nutrient digestion and metabolic homeostasis, the gut microbiota plays a central role in educating host immunity and maintaining intestinal barrier integrity. In the context of HCC, the gut-microbiota-liver axis serves as a crucial bidirectional communication hub. Its unique functionality is rooted in distinctive anatomical connectivity and intricate physiological cross-talk.

The portal venous system constitutes the anatomical cornerstone of this axis. Venous blood from the intestines drains directly into the liver via the portal vein. This vascular hardwiring allows bioactive products derived from the gut microbiota - such as microbiota-associated molecular patterns (MAMPs), metabolic byproducts, and trace amounts of translocated bacteria - to bypass the systemic circulation and undergo efficient “first-pass” exposure within the hepatic sinusoids [Figure 2]^[43]. Consequently, the liver physiologically becomes the “first-line” organ for sensing and processing gut-derived signals. The liver is rich in innate immune cells, particularly resident macrophages known as Kupffer cells. These cells, equipped with an abundance of pattern recognition receptors (PRRs) [e.g., Toll-like receptor 4 (TLR4)], continuously monitor these gut-derived signals, thereby coupling the dynamic state of the gut microbiota in real-time with the liver’s immune milieu^[44].

Intestinal barrier integrity acts as the critical gatekeeper maintaining this homeostasis. An intact epithelial barrier effectively restricts the pathological translocation of bacteria and their structural components. Conversely, barrier dysfunction (often termed “leaky gut”) precipitates the influx of highly immunogenic substances, such as lipopolysaccharide (LPS), into the portal circulation. This endotoxemia triggers persistent hepatic inflammatory responses and fosters a profoundly immunosuppressive microenvironment conducive to tumorigenesis^[45-47]. Beyond MAMPs, enterohepatic bile acid circulation establishes another sophisticated bidirectional regulatory loop. Primary bile acids synthesized by the liver enter the intestine, where they are biotransformed by the microbiota into secondary bile acids with distinct bioactivities. These secondary metabolites can activate critical nuclear receptors [e.g., farnesoid X receptor (FXR)]^[48] and membrane-bound receptors [e.g., Takeda G protein-coupled receptor 5 (TGR5)]^[49] within hepatocytes to modulate both metabolic pathways and innate immunity. Concurrently, these bile acids exert antimicrobial effects, forming a feedback loop that shapes the composition and spatial organization of the gut microbiota itself, culminating in a dynamically balanced regulatory system^[50,51]. The complexity of this axis is further amplified by systemic network extensions, including lymphatic trafficking and neural connections between the gut and liver, which collectively constitute a multi-layered bidirectional communication network beyond the portal vein^[52,53].

In summary, the gut-microbiota-liver axis is an integrated biological system encompassing anatomical hardwiring, continuous material exchange, and sophisticated signal transduction. It empowers the gut microbiota to exert systemic, remote control over the hepatic immune microenvironment, thereby providing a robust physiological rationale for targeting the microbiota to overcome immunotherapy resistance in HCC.

The efficacy of immunotherapy for HCC exhibits significant individual variability, rooted not only in the tumor itself but also intimately linked to the systemic immune state of the host. In recent years, the gut microbiota has been established as a key systemic modulator and a core factor driving differential responses to ICI therapy, demonstrating dual potential as both a predictive biomarker and a therapeutic target.

Through metagenomic, 16S rRNA sequencing, and metabolomic analyses of HCC patients undergoing ICI treatment, studies consistently reveal systematic differences in the gut microbiota between responders and non-responders in terms of species composition, ecological networks^[27], and functional output [Table 3]^[54-58]. For instance, early clinical observations from specific cohorts suggest that responders’ intestines are often enriched with beneficial genera such as Akkermansia, Lachnospiraceae, and Faecalibacterium prausnitzii, whose metabolic output is associated with enhanced anti-tumor immunity^[28]. However, it is critical to interpret these associations with caution. Microbiome studies in oncology are notoriously heterogeneous, and the identification of universal predictive taxa has proven challenging. Microbial signatures predictive of ICI response in one cohort often fail to reproduce in independent cohorts across different geographic regions or hospital centers^[59]. Conversely, non-responders frequently exhibit a dysbiotic pattern characterized by genera such as Ruminococcus and Enterococcaceae, often accompanied by the formation of an immunosuppressive microenvironment^[60]. These differential profiles hold promise as potential biomarkers, but their broad clinical utility demands validation in large-scale, multi-center prospective trials.

While ICI-based combination therapies have significantly improved the treatment landscape for advanced HCC, primary and acquired resistance mediated by an immune-“cold” tumor microenvironment remains a central bottleneck. Traditional localized sensitization strategies are often limited by profound tumor heterogeneity; thus, targeting the gut microbiota offers a novel “outside-in” systemic intervention paradigm. The feasibility of this strategy is firmly rooted in the unique pathophysiological wiring of the gut-microbiota-liver axis in HCC. Against the backdrop of chronic liver disease, impaired intestinal barrier function drives the continuous translocation of microbial products (e.g., lipopolysaccharide, LPS), which act directly on the liver via the portal vein. By engaging PRRs (e.g., TLR4) on hepatic innate immune cells (e.g., Kupffer cells), these translocated products trigger chronic local inflammation and actively sculpt an inhibitory immune niche^[61,62]. Consequently, intervening at the level of the gut microbiota effectively intercepts erroneous immune signaling at its source, achieving a “remote reprogramming” of the TME.

In-depth mechanistic studies elucidate that the gut microbiota precisely regulates therapeutic resistance through a complex, multi-pathway network^[63]. Specific microbial taxa can directly orchestrate a resistant microenvironment. For example, specific pathobionts such as Fusobacterium nucleatum and Escherichia coli (E. coli) have been shown to upregulate host multidrug resistance efflux pumps - most notably ATP binding cassette subfamily B member 1 (ABCB1) (P-glycoprotein) and ATP binding cassette subfamily C member 1 (ABCC1) - on the surface of malignant cells, thereby actively pumping out chemotherapeutic agents and targeted drugs. Concurrently, these microbial signals can activate the Janus kinase 1/protein kinase B/signal transducer and activator of transcription 3 (JAK1/AKT/STAT3) signaling cascade to inhibit tumor cell apoptosis, or induce protective autophagy via the TLR4/myeloid differentiation primary response 88 (MyD88) axis, collectively dampening the efficacy of systemic therapies^[64-66]. Furthermore, preclinical studies demonstrate that FMT from clinically resistant patients is sufficient to transfer the ICI-resistant phenotype to recipient mice^[67]. Simultaneously, specific microbiota-derived metabolites, such as indole-3-acetic acid (IAA) and butyrate, have been shown under certain contexts to directly impair anti-tumor immunity by enhancing the suppressive function of regulatory T cells or by exacerbating the pre-cancerous inflammatory environment via interleukin-35 (IL-35)^[68,69]. Conversely, beneficial dominant bacteria and their metabolites can play crucial sensitizing roles. For instance, specific Clostridium species can reverse the Wnt/β-catenin signal-dominated “immune desert” state by modulating secondary bile acid metabolism, promoting cytotoxic T cell infiltration^[70]. Moreover, the tryptophan metabolite indole-3-aldehyde, derived from Bifidobacterium, acts as a competitive ligand for the aryl hydrocarbon receptor (AhR), antagonizing the potent immunosuppressive signals of tumor-derived kynurenine, restoring cluster of differentiation 8 positive (CD8⁺) T-cell function, and downregulating inhibitory checkpoints such as TIM-3^[71,72] [Figure 3].

Despite these compelling mechanistic insights, gut microbiome research is characterized by significant heterogeneity, with seemingly contradictory associations reported for certain taxa across different studies, underscoring the profound complexity of microbe-host interactions. For example, while the genus Veillonella was found to be enriched in responders in a study by Lee et al.^[54], it is frequently associated with a pro-inflammatory state in contexts such as inflammatory bowel disease. These discrepancies may stem from several key factors: First, population and etiological heterogeneity - driven by differences in genetic backgrounds, underlying liver disease etiology [e.g., hepatitis B virus (HBV), hepatitis C virus (HCV), metabolic dysfunction-associated steatotic liver disease (MASLD)], dietary habits, and medication history among different cohorts collectively shape unique microbiota profiles^[73]. Second, technical and methodological inconsistencies - ranging from sample collection and DNA extraction protocols to sequencing platforms (16S rRNA vs. shotgun metagenomics) and bioinformatics pipelines - drastically reduce cross-study reproducibility. Third, functional divergence at the strain level is critical; different strains within the same genus can harbor vastly different genomic and metabolic capacities. Metagenomic analyses increasingly suggest that strain-specific features, rather than species-level abundance, are the true determinants of biological function^[58]. Additionally, methodological differences in research may also lead to inconsistent results^[74].

Therefore, future predictive modeling must transcend reliance on single bacterial species and adopt a multi-dimensional, integrative strategy. This necessitates the combined analysis of specific strains with clear functional annotations, microbiota functional outputs (e.g., metabolomic profiles), and the ecological network structure of the microbial community. Combinatorial biomarkers - such as a structural ratio of “high Akkermansia + high Faecalibacterium + low Ruminococcus”, the secondary-to-primary bile acid ratio, and the abundance of microbial butyrate synthesis genes - hold immense potential for accurately predicting ICI efficacy. Recently, machine learning frameworks have been deployed to navigate this complexity. For instance, one study utilized a random forest algorithm to integrate gut microbiota features (such as the relative abundances of Akkermansia and Ruminococcus) with clinical variables, constructing a robust predictive model that significantly outperformed any single clinical indicator in distinguishing treatment responses^[75].

In summary, the gut microbiota represents a critical node for understanding and therapeutically intervening in the HCC immunotherapy response. Its composition and functional output directly dictate treatment outcomes by modulating intestinal barrier integrity, systemic inflammation, and the TME landscape. While current preclinical models and early clinical cohorts provide a robust conceptual foundation for developing microbiome-based companion diagnostics and precision intervention strategies, these approaches are still in their infancy. Moving forward, the field must transition from associative small-cohort studies to standardized, adequately powered, multi-center randomized clinical trials to critically evaluate their true clinical utility. Future translational research must dissect the microbe-host interaction network at a granular level, propelling HCC immunotherapy into a new era of systemic remodeling and precise regulation.

The composition of the gut microbiota, at both the species and strain levels, is not a passive bystander but an active, preconfigured determinant that shapes hepatic immune surveillance. A healthy, diverse, and stable microbial community supports systemic immune homeostasis by maintaining intestinal barrier integrity and generating beneficial metabolites. Specific beneficial microbiota fortify host defenses through both direct and indirect mechanisms^[76].

Akkermansia muciniphila (A. muciniphila) represents a prototypical species in this context. Its outer membrane protein, Amuc_1100, can specifically interact with Toll-like receptor 2 (TLR2) on intestinal epithelial cells. This interaction triggers downstream signaling cascades that promote the synthesis and secretion of mucin 2 (MUC2) by goblet cells, thereby thickening and fortifying the protective mucus layer^[77]. Furthermore, short-chain fatty acids (SCFAs, e.g., acetate) produced by A. muciniphila metabolism can activate G protein-coupled receptors (e.g., GPR43) on epithelial cells, subsequently upregulating the expression of tight junction proteins and directly enhancing epithelial barrier function^[78]. Human evidence robustly supports these preclinical findings: a randomized controlled trial demonstrated that supplementation with pasteurized A. muciniphila significantly reduced plasma levels of gut permeability markers (e.g., LPS-binding protein) and systemically improved host insulin sensitivity and inflammatory status. This provides crucial support for its translational value in remodeling the HCC immune microenvironment^[79].

Beyond A. muciniphila, classical probiotics such as Bifidobacterium and Lactobacillus also reinforce the barrier through distinct functional pathways. For instance, Bifidobacterium longum subsp. infantis produces specific glycosyl hydrolases that degrade dietary fibers to generate functional oligosaccharides. These molecules serve as essential precursors to promote mucin synthesis^[80]. Certain Lactobacillus strains, via the binding of their cell wall component lipoteichoic acid to TLR2 on dendritic cells, can strongly induce the production of IL-10 and transforming growth factor beta (TGF-β). This cytokine milieu, in turn, promotes Treg differentiation. These cells can secrete tissue repair factors, such as keratinocyte growth factor, directly stimulating epithelial cell proliferation and tight junction protein synthesis, thereby achieving an immune-mediated barrier repair^[81].

Other beneficial bacteria, such as Limosilactobacillus reuteri (L. reuteri), can induce the expression of heat shock proteins in intestinal epithelial cells, enhancing cellular resistance to oxidative stress and maintaining epithelial integrity. A pilot study in cirrhotic patients found that L. reuteri supplementation successfully reduced serum endotoxin levels and improved liver function scores, suggesting its therapeutic potential to alleviate liver disease via the gut-microbiota-liver axis^[82]. In a study of patients with unresectable HCC, fecal samples from immunotherapy responders showed significant enrichment of Lachnospiraceae, Lachnoclostridium, and Veillonella. Notably, the abundance of Lachnoclostridium was highly positively correlated with systemic levels of secondary bile acids, such as ursodeoxycholic acid and ursocholic acid. Furthermore, the coexistence of Lachnoclostridium enrichment and Prevotella 9 depletion emerged as a significant predictor of superior overall survival^[54]. Coprococcus, particularly Coprococcus comes, is another consistently observed critical genus in responders. This bacterium is closely associated with dietary fiber digestion and the robust production of butyrate; its abundance strictly correlates with a favorable treatment response. Microbial interaction network analyses have revealed denser positive correlations among species enriched in responders, indicating a more synergistic, resilient, and stable community structure^[27].

Conversely, the enrichment of pathogenic bacteria driven by gut dysbiosis serves as a major risk factor promoting HCC progression and ICI resistance. The Fap2 protein of Fusobacterium nucleatum has been shown to specifically bind and activate the inhibitory receptor TIGIT on natural killer (NK) cells and T cells, directly suppressing the cytotoxic function of NK cells and CD8⁺ T cells, thereby fostering a profoundly immunosuppressive niche at the tumor site^[83]. A recent study combining metagenomics and liver tissue bacterial culture revealed a significant enrichment of Klebsiella pneumoniae in both the intestines and livers of HCC patients and their corresponding transplanted mouse models. Mechanistic models demonstrated that monocolonization with this specific bacterium was sufficient to drive HCC development by disrupting the gut barrier and inducing severe systemic and local hepatic inflammation. The core mechanism involves the direct interaction of the bacterial surface protein penicillin-binding protein 1B (PBP1B) with the TLR4 receptor on hepatocytes, thereby hyperactivating downstream oncogenic signaling pathways^[84]. This crucial finding not only elucidates the pathogenic role of specific gut pathobionts in HCC but also provides novel targets for developing precision therapies targeting the “pathogen-host” interaction interface. Similarly, Ruminococcus was consistently found enriched in non-responders. In studies related to chronic hepatitis B progression, Ruminococcus enrichment was associated with delayed viral clearance and the maintenance of an immune-tolerant state. It may indirectly foster an inhibitory hepatic immune environment by encoding bile salt hydrolase (BSH), which severely alters the host’s bile acid metabolic balance^[55]. Additionally, Enterococcaceae exhibited a clear trend of enrichment during HCC progression, with its relative abundance significantly increasing from early to advanced disease stages. This dynamic change was heavily correlated with elevated levels of the gut injury marker regenerating islet-derived protein 3 alpha (REG3α) and the microbial translocation marker soluble cluster of differentiation 14 (sCD14), underscoring its potential role in disrupting intestinal barrier integrity and driving relentless systemic inflammation^[56].

In summary, the homeostasis of intestinal barrier function serves as a crucial and actionable target within the HCC immune landscape [Figure 4]. Future microbiota-targeted intervention strategies must delicately balance “reinforcing the beneficial” and “eliminating the pathogenic”. This dual strategy entails supplementing with beneficial consortia that possess barrier-repair functions, while simultaneously employing targeted eradication of specific pro-carcinogenic pathobionts or utilizing precise receptor (e.g., TLR4) signal modulation. Through such a comprehensive approach, it becomes possible to drastically reduce the antigenic and inflammatory burden on the liver at its source, laying the essential foundation for reversing the “cold” tumor microenvironment and maximizing the efficacy of immunotherapy.

Strategies for modulating the gut microbiota are rapidly advancing from conceptual frameworks toward early-phase clinical exploration, forming a diverse intervention armamentarium that ranges from whole-community transplantation to precision rational design, and from broad-spectrum to fine-tuned approaches. These translational strategies primarily encompass FMT, next-generation probiotics and synbiotics, and synthetic biology-driven engineered bacteria. They collectively aim to therapeutically reshape the “gut-microbiota-liver axis”, reverse established immunosuppression, and thereby sensitize tumors or overcome resistance to immunotherapy in HCC. However, it is crucial to note that the translational readiness of these modalities varies significantly, spanning from experimental proof-of-concept models to ongoing early-phase clinical trials [Figure 11].

FMT, serving as the most direct whole-microbiota intervention [Figure 11A], has successfully entered early clinical exploration. A prospective single-arm study demonstrated that among 13 patients with advanced gastrointestinal cancers (including 4 with HCC) who were refractory to nivolumab, the administration of FMT combined with continued nivolumab therapy enabled previously resistant HCC patients to achieve a partial response after receiving an FMT from a responder donor. Crucially, this clinical benefit was tightly associated with the successful engraftment of specific immune-activating bacterial strains^[139]. Furthermore, interim analysis from a Phase II pilot study (FAB-HCC) evaluating patients resistant to the first-line “T+A” regimen reported highly encouraging results: following a single donor FMT and the continuation of the original systemic therapy, an ORR of 33% and a disease control rate of 83% were observed among six evaluable patients, preliminarily validating the safety and feasibility of this approach^[140].

In-depth longitudinal analyses reveal that successful FMT is a highly dynamic process of microbial community reconstruction. In clinical responders, beneficial donor strains stably colonize and expand, while the relative abundance of potentially deleterious pathobionts correspondingly decreases^[141]. This microbial shift translates into the detectable clonal expansion of tumor antigen-specific T cells and an improved TCR repertoire diversity in the host’s peripheral blood, subsequently leading to enhanced infiltration and cytotoxic function of CD8⁺ T cells within the tumor microenvironment^[142,143]. Expanding beyond systemic ICIs, the microbiota’s influence also dictates responses to locoregional interventions. For instance, in the context of radiotherapy, studies corroborate that gut dysbiosis can suppress tumor antigen presentation and effector T-cell function via the impairment of the cGAS-STING-type I interferon pathway. Conversely, microbiota-mediated activation of this pathway can synergistically enhance the anti-tumor efficacy of PD-1 blockade combined with radiation^[144].

Despite these highly promising prospects, establishing FMT as a routine, standardized adjuvant therapy for HCC still faces multiple hurdles, including the strict standardization of donor screening, the mitigation of long-term safety risks (e.g., pathogen transfer), and the navigation of complex regulatory pathways^[145,146]. Driven by the sheer complexity and undefined nature of FMT, intervention strategies are increasingly evolving towards precision medicine using next-generation probiotics and synbiotics [Figure 11B] based on specific, highly characterized functional strains. This paradigm shift is grounded in the recognition that the strain - rather than the species - is the core functional unit. Different strains within the same species can exhibit profound differences in their genomic architectures, metabolomic profiles, and immunomodulatory capacities. For example, not all Akkermansia muciniphila strains effectively fortify the intestinal barrier; their therapeutic efficacy strictly depends on the expression and structural conformation of the outer membrane protein Amuc_1100, which is highly strain-specific^[147]. Similarly, the capability of Bifidobacterium to sensitize tumors to immunotherapy exhibits strict strain dependency.

Clinical and preclinical studies have firmly validated the potential of strain-specific interventions. One notable trial found that in patients receiving dual checkpoint blockade (nivolumab plus ipilimumab), concurrent supplementation with the live biotherapeutic Clostridium butyricum strain MIYAIRI 588 (CBM588) significantly extended median progression-free survival (12.7 months vs. 2.5 months)^[29]. Additionally, Lactobacillus rhamnosus Probio-M9 has been shown to accelerate post-antibiotic microbiota reconstitution and subsequently enhance anti-PD-1 efficacy^[148]. In the context of altering the pre-tumorigenic background, Lactobacillus acidophilus exerts potent preventive effects in NASH-HCC models through the production of its metabolite valeric acid. This SCFA activates GPR41/43 receptors and inhibits the Ras homolog family member A (RhoA) pathway [a canonical member of the Ras homolog-guanosine triphosphatase (Rho-GTPase) family], thereby suppressing inflammation-driven tumor progression^[149]. Furthermore, a rationally constructed synthetic consortium comprising 11 defined bacterial strains was shown to robustly induce the generation of interferon gamma positive (IFN-γ⁺) CD8⁺ T cells in the intestine, enhancing host anti-infection immunity and synergistically improving the efficacy of ICIs^[150].

However, the continuous administration of exogenous probiotics may occasionally delay the autochthonous recovery of the host’s indigenous microbiota^[151]. Therefore, synbiotics - rational combinations of specific probiotics with targeted prebiotics - are strategically designed to selectively enhance the in vivo colonization and metabolic output of the target strains. For instance, a synbiotic combination of bilberry anthocyanins and citrus pectin significantly increased fecal butyrate levels, synergistically enhancing the anti-tumor effect of PD-L1 blockade^[152]. Modulating the underlying hepatic background is also a critical pre-conditioning strategy. In patients with NAFLD, meta-analyses of multiple randomized controlled trials indicate that interventions with probiotics, prebiotics, or synbiotics can significantly ameliorate liver steatosis, fibrosis markers, and elevated liver enzymes^[153]. For example, the daily consumption of probiotic yogurt containing Lactobacillus acidophilus La5 and Bifidobacterium lactis Bb12 for eight weeks significantly reduced serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), total cholesterol, and low-density lipoprotein (LDL)-cholesterol levels^[154]. Similarly, treatment with a tailored probiotic mixture of 6 strains for 12 weeks in obese patients with NAFLD [confirmed by magnetic resonance imaging-proton density fat fraction (MRI-PDFF)] significantly reduced intrahepatic fat content. This clinical improvement was accompanied by a marked expansion of beneficial endogenous taxa, such as Agathobacter and Blautia^[155].

A more cutting-edge frontier involves leveraging synthetic biology to engineer bacteria [Figure 11C], developing them as programmable “living therapeutics” capable of actively homing to tumors, sensing the hostile tumor microenvironment, and intelligently releasing therapeutic payloads. For example, researchers have successfully constructed engineered E. coli capable of specifically targeting and colonizing the hypoxic cores of HCC tumors. Drug-loaded nanoparticles anchored to their bacterial surface enable in situ antigen expression and the targeted co-delivery of chemotherapeutic agents and genetic payloads, demonstrating potent anti-tumor and anti-metastatic effects in murine models^[156]. Another breakthrough study designed an intelligent, self-lysing delivery system based on attenuated Salmonella. This engineered bacterium efficiently invades malignant cells and lyses intracellularly to release therapeutic protein drugs, effectively inhibiting primary HCC and lung metastases^[157]. Additionally, engineered E. coli designed to specifically sense the acidic tumor microenvironment and locally induce the expression of anti-PD-1 single-chain variable fragments have successfully activated robust anti-tumor immunity and suppressed post-surgical recurrence in HCC models^[158].

Within this translational landscape, postbiotics [Figure 11D] - formally defined as preparations of inanimate microorganisms and/or their components that confer a health benefit on the host - have emerged as a highly promising intervention modality. Their distinct advantages include superior safety profiles, enhanced structural stability, and highly standardized manufacturing processes. Clinical studies have demonstrated that combined supplementation with heat-inactivated preparations of specific Limosilactobacillus fermentum, Limosilactobacillus reuteri, and Lactiplantibacillus plantarum for 60 days significantly improved liver function and modulated uric acid levels in patients with MAFLD, while concurrently optimizing their baseline gut microbiota architecture^[159]. Basic research further corroborates this, showing that postbiotic compounds extracted from Lactiplantibacillus plantarum exert profound protective effects in preclinical acute liver injury models^[160]. Remarkably, even pasteurized Akkermansia muciniphila retains the potent biological activity of its outer membrane protein Amuc_1100, fully maintaining its function in preserving the critical intestinal barrier in experimental models^[79].

In summary, the seamless clinical translation of microbiota-targeted therapies must still overcome significant challenges related to long-term safety, standardized pharmaceutical-grade production, and individualized efficacy prediction. Integrating deep multi-omics data - such as high-resolution metagenomics and metabolomics - to construct AI-driven predictive models represents a paramount future perspective. This technological leap will eventually enable clinicians to precisely match individual patients with their optimal bacterial strains or synthetic formulations, firmly driving microbiota-targeted interventions away from the era of “broad-spectrum” empirical supplementation and into a transformative new epoch of precision microbial modulation.

A primary mechanism by which the gut microbiota dictates the efficacy of HCC immunotherapy is through its functional output: bioactive metabolites. Consequently, “metabolite therapy” - a strategy that directly targets these signaling molecules with defined biological activities - offers a highly controllable and translationally viable avenue for overcoming immune resistance. This therapeutic strategy primarily encompasses two distinct directions: first, the exogenous supplementation of beneficial immunomodulatory metabolites; and second, the precise pharmacological blockade of deleterious metabolic pathways. Both approaches aim to actively reshape the host systemic immune tone and remodel the TME [Figure 12].

The direct systemic or targeted supplementation of beneficial metabolites is a widely investigated strategy. Its core translational challenge lies in traversing the “double-edged sword” nature of metabolite function and achieving efficient, spatiotemporally targeted delivery. Butyrate, a prominent SCFA, serves as a prototypical example. Systemically high concentrations of butyrate may paradoxically promote regulatory T-cell (Treg) differentiation by inhibiting HDACs, which can inadvertently suppress anti-tumor immunity^[161]. However, robust research has also revealed its potent capacity for localized immune activation: butyrate can stimulate HCC cells to upregulate the expression of the critical chemokine CXCL11 through epigenetic chromatin remodeling, effectively recruiting NK cells to infiltrate the tumor bed. Preclinical models confirm that oral butyrate supplementation significantly inhibits tumor growth, an effect that is strictly dependent on the presence of functional NK cells^[85]. These nuanced findings suggest that localized administration (e.g., via specialized enemas) or the development of liver-targeting/TME-responsive nanodelivery systems may maximize the immunostimulatory effects of butyrate while firmly mitigating its systemic immunosuppressive risks. Furthermore, exogenous acetate supplementation has demonstrated significant therapeutic potential; it attenuates the production of the pro-inflammatory cytokine interleukin 17A (IL-17A) by type 3 innate lymphoid cells (ILC3s) in the liver through HDAC inhibition mechanisms, exhibiting potent synergistic anti-tumor effects when combined with PD-1 inhibitors in preclinical models^[162].

Within the realm of bile acid metabolism, the direct supplementation of UDCA - a hydrophilic secondary bile acid already FDA-approved for specific liver diseases with a well-established safety profile - is considered a highly translatable strategy. Accumulating evidence suggests that while specific hydrophobic secondary bile acids (e.g., lithocholic acid^[163]) possess profound immunosuppressive properties, UDCA actively counteracts these effects. Direct UDCA supplementation aims to rapidly improve the hepatic immune microenvironment by beneficially skewing the composition of the circulating bile acid pool, and preliminary clinical explorations combining it with ICIs are currently underway^[164].

Compared to direct supplementation, the pharmacological blockade of deleterious metabolic pathways represents a highly targeted intervention. Among these, antagonizing the tryptophan-kynurenine-AhR axis remains a major research focus. Within the TME, the microbially modulated enzymes IDO1 and tryptophan 2,3-dioxygenase (TDO2) metabolize tryptophan into kynurenine. Kynurenine acts as an endogenous AhR ligand, causing its persistent activation, which subsequently drives Treg differentiation, induces T-cell exhaustion, and polarizes macrophages toward a suppressive M2 phenotype. This axis is a fundamental driver in the formation of an immune-“cold” microenvironment^[164,165]. Therefore, developing IDO1/TDO2 inhibitors or direct AhR antagonists aims to sever this harmful signaling cascade upstream. Although early clinical trials of IDO1 inhibitors have yielded complex and sometimes underwhelming results in certain solid tumors, this strategy remains potentially highly valuable in specific HCC subgroups characterized by intense IDO1 overexpression^[165].

An even more innovative frontier is “precision microbial enzyme inhibition”. This paradigm involves utilizing small-molecule drugs to specifically inhibit key bacterial enzymes responsible for catalyzing harmful metabolites, entirely bypassing the need to exert broad-spectrum bactericidal pressure. For instance, the enzyme cascades encoded by the bile acid-inducible (bai) operon, expressed by specific anaerobic taxa, are responsible for converting primary bile acids into highly immunosuppressive secondary bile acids (e.g., lithocholic acid and deoxycholic acid), which actively promote HCC progression^[166]. Designing highly specific, non-lethal inhibitors targeting these bacterial enzymes holds immense promise for reducing the generation of harmful secondary bile acids directly at their source. Crucially, this approach preserves the overall structural integrity and diversity of the gut microbiota while profoundly improving the hepatic immune landscape.

In summary, metabolite therapy - by directly and rationally manipulating the functional output of the gut microbiota - provides a multidimensional framework for precise intervention to overcome HCC immunotherapy resistance. Nevertheless, this strategy must navigate substantial challenges on its path toward routine clinical translation. For supplementation strategies, issues regarding concentration gradients, context-dependent immunological outcomes, delivery efficiency, and profound patient heterogeneity must be resolved. For blockade strategies, ensuring absolute target specificity, overcoming metabolic pathway redundancy, and accumulating robust long-term safety data are imperative. Moving forward, integrating advanced nanotechnology, synthetic biology, and multi-omics-guided personalized diagnostics will propel metabolite therapy from conceptual blueprints to clinical reality, firmly establishing it as a fundamental pillar within the comprehensive HCC treatment paradigm.

Beyond FMT, precision probiotics/synbiotics, engineered live biotherapeutics, and metabolite therapies, interventions targeting the gut microbiota are rapidly expanding into highly diversified and precision-guided dimensions. Although many remain in the conceptual or preclinical stages, these strategies either macroscopically modulate the host-microbiota interaction landscape or employ novel biologics for targeted microbial editing and delivery. Together, they constitute a robust, multi-layered, and complementary experimental armamentarium.

Dietary modulation represents the most accessible and fundamental intervention. Its role extends far beyond basic nutritional provision, serving as a core modality to systemically sculpt the structure and functional output of the gut microbiota. A large-scale prospective cohort study involving 63,275 Singaporean Chinese individuals (with an average follow-up of 17.6 years) demonstrated that long-term adherence to a healthy, plant-based dietary pattern was associated with a significantly reduced risk of HCC^[167]. However, given the observational nature of this study, this association is inevitably subject to confounding variables, as these cohorts plausibly adhere to overall healthier lifestyles (e.g., smoking cessation, reduced alcohol consumption). Therefore, while it is strongly hypothesized that shaping a specific, protective microbial-metabolic environment through such diets contributes to these cancer-preventive benefits, the definitive mechanistic link remains speculative and strictly warrants validation through controlled clinical trials. Beyond chronic dietary patterns, chrono-nutritional interventions also exhibit profound translational potential. Compelling preclinical studies indicate that short-term starvation cycles or fasting-mimicking diets can radically remodel the tumor immune microenvironment. Mechanistically, this involves regulating the metabolic reprogramming of TAMs [via the fructose-1,6-bisphosphatase (FBP1)/AKT/Ras-related protein Rab-27A (Rab27a) signaling axis] to drastically reduce the secretion of immunosuppressive PD-L1-positive exosomes, thereby significantly enhancing the efficacy of anti-PD-1/PD-L1 therapy in multiple murine models of HCC^[168]. This provides critical proof-of-concept for deploying specific fasting regimens as “metabolic adjuvants” to immunotherapy. Ultimately, “precision nutrition” paradigms - leveraging specific functional components like defined dietary fibers or polyphenols - are expected to enable the targeted enrichment and functional activation of beneficial commensal consortia.

For the highly targeted elimination of specific pathobionts, bacteriophage (phage) therapy exhibits irreplaceable advantages. Unlike broad-spectrum antibiotics, which inflict devastating collateral damage on the commensal ecosystem, phages can specifically lyse target bacterial strains. This achieves the precise clearance of pathogens while maximally preserving the integrity of the beneficial gut microbiota. For instance, customized phage “cocktails” could be rationally designed against key bacteria implicated in HCC progression and immunosuppression, such as Fusobacterium nucleatum or Klebsiella pneumoniae. This strategy not only eradicates the pathogenic source but also holds immense promise for decisively reversing the bacteria-driven immunosuppressive TME through the precise, surgical editing of the microbiota composition^[83,84]. Although phage therapy for HCC remains strictly in the preclinical arena, its unparalleled specificity offers a highly attractive translational approach for overcoming pathogen-driven resistance phenotypes.

Furthermore, bacterial outer membrane vesicles (OMVs) are rapidly gaining traction as naturally derived, nanoscale immunomodulatory carriers. OMVs intrinsically harbor various highly immunogenic components from their parent bacteria, endowing them with potent built-in adjuvant properties. An innovative preclinical study leveraged synthetic biology to construct dual-targeting engineered OMVs [OMV-CD47/glypican-3 (GPC3)] that simultaneously display an anti-CD47 nanobody and an anti-GPC3 single-chain variable fragment on their surface. These highly specialized vesicles can precisely recognize HCC cells. Mechanistically, they dually enhance macrophage-mediated tumor phagocytosis by blocking the CD47 “don’t eat me” signal, while simultaneously exploiting their natural immunostimulatory payload to activate dendritic cells and radically remodel the TME. This dual-action strategy effectively inhibited tumor growth and metastasis in an orthotopic HCC model^[169].

It is imperative to emphasize that the strategic timing and synergistic sequencing of these interventions will dictate their ultimate translational success. Microbiota-targeted therapies will rarely be administered as isolated monotherapies. Instead, the precise temporal integration of these interventions with established standard-of-care modalities - including surgery, transarterial chemoembolization (TACE), radiotherapy, and targeted systemic therapies - will profoundly shape the final therapeutic trajectory. For example, as a strong translational hypothesis, the peri-procedural adjuvant use of mucoprotective probiotics during TACE could strategically mitigate treatment-induced intestinal barrier damage and systemic endotoxemia, thereby cultivating a significantly more favorable, immune-permissive baseline for subsequent systemic ICI therapy^[33]. Looking ahead, the formulation of personalized intervention blueprints - rooted deeply in the patient’s baseline microbiota profile (enterotype) and metabolomic biomarkers - represents the ultimate translational zenith, acknowledging that this is a future paradigm rather than an immediate clinical reality. Guided by AI-driven multi-omics analyses, individual patients could theoretically be seamlessly matched with their most optimal probiotic consortium, synthetic community, FMT donor, or chrono-nutritional regimen. This milestone will finally propel microbiota-targeted oncology from an empirical, “one-size-fits-all” approach into the definitive era of true precision medicine.

The crucial role of the gut microbiota in orchestrating the response and resistance to immunotherapy for HCC not only provides new therapeutic targets but also poses novel demands and opportunities for clinical laboratory practice. Advances in this field clearly dictate that future laboratory medicine must evolve beyond the traditional paradigm focused primarily on single-pathogen identification or static biochemical marker analysis. It must actively shift towards a novel microbiome diagnostics system that integrates multi-omics data, emphasizes dynamic longitudinal monitoring, and focuses on actionable functional interpretation. This transformation aims to more precisely serve efficacy prediction, treatment monitoring, and the guidance of personalized microbiota-targeted interventions in cancer immunotherapy.

Regarding the scope of laboratory samples, the traditional concept of the tumor “liquid biopsy” urgently warrants expansion. Currently, tumor-derived markers such as circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs) dominate the primary focus. However, gut microbiome research highlights the equal strategic importance of a systemic host “microbiome liquid biopsy”. This paradigm shift demands that clinical laboratories fundamentally elevate the clinical value of fecal samples. Feces must no longer be viewed merely as crude specimens for routine occult blood or parasite screening; rather, they are the core biological materials for accurately assessing the functional status of the “gut-microbiota-liver axis”. In the future, establishing standardized workflows for fecal metagenomic and metabolomic testing, alongside the development of rapid molecular assays targeting key predictive bacterial taxa (e.g., Akkermansia) or functional gene clusters, should become integral components of companion diagnostics for immunotherapy. Concurrently, the analytical scope of peripheral blood samples requires simultaneous expansion. Beyond traditional tumor markers, laboratories must incorporate systemic indicators reflecting microbiota-host interactions, such as microbial translocation markers (e.g., LPS-binding protein, soluble CD14) and key microbiota metabolite profiles (e.g., SCFAs, specific secondary bile acids, and the tryptophan/kynurenine ratio). These indicators provide indirect yet vital real-time information regarding the systemic immunological tone of the tumor microenvironment. Integrating these microbiota-derived markers with traditional tumor molecular markers to construct a multi-dimensional biomarker landscape will significantly enhance the predictive capacity for treatment response and resistance risk.

This profound change directly necessitates a rapid capability upgrade in laboratory technologies. Comprehensive microbiome analysis relies heavily on high-throughput sequencing and complex bioinformatics. Therefore, laboratory professionals must master the basic principles, operational standards, and quality control essentials of next-generation sequencing (NGS) technologies. They must also acquire foundational bioinformatics interpretation skills to translate massive datasets - such as microbial diversity metrics, compositional structures, and functional pathway predictions - into clinical significance. Simultaneously, to promote widespread clinical adoption, developing rapid, cost-effective testing methods suitable for routine clinical laboratories - independent of massive sequencing platforms - is crucial. Examples include specific quantitative polymerase chain reaction (qPCR) assays for particular beneficial or pathobiontic strains, or automated chemiluminescence immunoassays for key immunomodulatory metabolites. A more profound paradigm shift addresses the inherent limitations of isolated, single-analyte test results. Future laboratory diagnostics will inextricably rely on advanced data integration and artificial intelligence (AI) models. The role of the laboratory department will inevitably evolve from merely outputting discrete data points to utilizing machine learning algorithms to integrate multi-omics datasets - encompassing the patient’s microbial features, metabolite profiles, and clinical parameters. This integration will generate highly intelligible reports, such as an “Immunotherapy Response Index” or “Microbiota Modulation Recommendations”, directly supporting personalized clinical decision-making.

This technological shift consequently drives a fundamental transformation in the mode of report interpretation and clinical communication. The interpretation of a microbiome test report transcends the complexity of routine laboratory panels. It requires laboratory physicians to pivot from simple binary judgments of “normal vs. Abnormal” towards a comprehensive, biologically grounded interpretation of microbiota composition, functional potential, and clinical relevance. Reports should strive to elucidate the biological significance behind the data, for example: “Reduced gene abundance in the butyrate synthesis pathway suggests potentially impaired immunomodulatory function of the gut microbiota, which is associated with a heightened risk of T-cell functional suppression.” Furthermore, because the gut microbiota is highly dynamic, laboratory practice must advocate for and implement a longitudinal monitoring concept. Analyzing serial samples collected before, during, and after treatment allows for the dynamic assessment of the biological effects of microbiota-targeted interventions (e.g., probiotics, FMT) and provides a quantitative basis for real-time regimen adjustments. This dynamic landscape demands that laboratory physicians proactively strengthen communication and collaboration with multidisciplinary tumor boards (MTBs) - including oncology, hepatology, and clinical nutrition departments - positioning the laboratory as the indispensable bridge connecting complex microbiome data to bedside clinical practice.

Finally, the cornerstone for the robust, sustainable development of this new field lies in rigorous quality control, standardization, and strict ethical norms. The clinical translation of microbiome testing currently faces immense standardization challenges. Strict industry consensus and standardized operating procedures (SOPs) are urgently needed for every pre-analytical and analytical step - from sample collection, stabilization, and DNA extraction to sequencing library preparation and bioinformatics pipelines - to guarantee the reproducibility and comparability of results across different institutions. Strengthening the architecture of internal quality control (IQC) and external quality assessment/proficiency testing (EQA/PT) programs is paramount. Additionally, because microbiome data inherently contains highly sensitive genomic and personal health information, laboratory departments must establish impregnable data security and privacy protection frameworks, strictly adhering to ethical guidelines while vigorously promoting technological advancement.

In conclusion, gut microbiome research has unequivocally opened new therapeutic avenues for HCC immunotherapy and holds the immense potential to guide laboratory medicine into a transformative new era characterized by a greater systemic perspective, longitudinal dynamism, and unprecedented precision. For the laboratory profession, this represents not merely a technological innovation, but a fundamental future upgrade in clinical mindset and disciplinary positioning. Clinical laboratories are poised to transform from traditional service departments providing discrete diagnostic data into core clinical hubs that integrate multi-dimensional biological intelligence to generate personalized diagnostic and therapeutic decision support. By proactively embracing this profound capability reconstruction, laboratory medicine will play an increasingly critical and irreplaceable role in overcoming cancer therapy resistance and realizing the ultimate goal of truly personalized precision medicine.

The gut microbiota has emerged as a promising predictive biomarker for patient stratification in ICI therapy. Early clinical observations suggest that responders’ intestines are often enriched with beneficial taxa such as Akkermansia, Lachnospiraceae, and Faecalibacterium prausnitzii, whose metabolic output is strongly associated with enhanced systemic anti-tumor immunity. Conversely, non-responders frequently exhibit a dysbiotic pattern. However, microbiome studies in oncology are notoriously heterogeneous. Single microbial signatures predictive of ICI response in one cohort often fail to reproduce across different geographic regions or hospital centers due to underlying population and methodological differences. Therefore, future companion diagnostics must shift towards a multi-dimensional, integrative strategy. Combining multi-omics data (e.g., metagenomics, metabolomics, immunomics) and leveraging advanced artificial intelligence, such as random forest algorithms, can generate highly robust predictive models. Combinatorial biomarkers - such as a structural ratio of “high Akkermansia + high Faecalibacterium + low Ruminococcus,” systemic secondary-to-primary bile acid ratios, and the abundance of microbial butyrate synthesis genes - hold immense potential for the early stratification of patients’ therapeutic response and resistance risk.

When considering real-world clinical implications, the profound impact of concurrent medications - particularly antibiotics - on established immunotherapies cannot be overlooked. The current treatment landscape for advanced HCC has been unequivocally revolutionized by several pivotal registration trials, notably IMbrave150 (atezolizumab plus bevacizumab)^[10], HIMALAYA (tremelimumab plus durvalumab)^[12], as well as foundational studies (e.g., CheckMate-040^[13], KEYNOTE-224^[8], and KEYNOTE-240^[35]) assessing monotherapy with PD-1 inhibitors. While these landmark trials successfully established ICIs as the standard of care, it is a critical limitation of the current oncology literature that these original study designs did not incorporate prospective gut microbiome profiling as primary or secondary endpoints. Consequently, direct microbiome sub-analyses from these specific pivotal cohorts remain sparse.

However, emerging real-world evidence bridging these established regimens with microbiome-disrupting factors provides crucial clinical insights. Recent retrospective cohorts evaluating HCC patients treated with the IMbrave150 “T+A” regimen or anti-PD-1 monotherapies have consistently demonstrated that prior or concurrent exposure to broad-spectrum antibiotics - which violently disrupt the delicate microbial interaction network required for optimal immune priming - significantly diminishes ORRs and strictly correlates with poorer overall survival in HCC patients^[170]. This strongly implies that the profound clinical efficacy observed in these major trials is inherently dependent on an intact, eubiotic baseline gut microbiome. Therefore, critically analyzing these standard-of-care regimens through the lens of microbial ecology underscores an urgent translational need: rigorous antibiotic stewardship must be implemented to purposefully preserve the patient’s baseline microbial immune tone, and future iterations of global, large-scale HCC clinical trials must prospectively integrate microbiome stratification to fully optimize personalized immunotherapeutic strategies.

Beyond pharmacological interventions, dietary modulation represents a safe, highly accessible, and foundational strategy to fundamentally remodel the intestinal microenvironment. Diet profoundly shapes both microbiota composition and functional output. For instance, diets rich in specific fermentable fibers and prebiotics actively promote the expansion of SCFA-producing taxa (such as Lachnospiraceae and Ruminococcaceae). The resulting increase in systemic butyrate and other beneficial metabolites can finely modulate mucosal T-cell responses, strictly preserve intestinal epithelial barrier integrity, and favorably condition the hepatic immune microenvironment. Integrating personalized dietary guidance into the standard oncological care of HCC patients may serve as a potent, synergistic “preconditioning” strategy to enhance subsequent ICI regimens.

Direct microbiota-targeted interventions, including next-generation probiotics and synbiotics, are rapidly emerging as powerful tools to reverse immunotherapy resistance. Basic research confirms that defined specific strains hold immense translational value. For example, postbiotic compounds extracted from Lactiplantibacillus plantarum exhibit potent protective effects in liver injury models, and even heat-inactivated Akkermansia muciniphila retains the robust activity of its outer membrane protein (Amuc_1100) to maintain intestinal barrier integrity. Moving forward, intervention strategies must definitively evolve from a “one-size-fits-all”, broad-spectrum supplementation approach towards true precision medicine. By comprehensively evaluating a patient’s unique enterotype and metabolic profile, clinicians could theoretically supplement specifically defined immune-sensitizing strains, employ customized phages to target pathogenic bacteria, or utilize engineered bacteria strictly designed to deliver specific immunomodulatory molecules directly to the gut.

In addition to specific dietary components, metabolic interventions have shown immense potential. Preclinical studies indicate that short-term starvation cycles can profoundly reshape the gut microbiota architecture, reduce systemic inflammation, and synergistically potentiate the anti-tumor efficacy of ICIs^[171].

While early correlative small-cohort studies provide intriguing evidence, these microbiota-targeted strategies are still in their clinical infancy. Strategies integrating microbial preconditioning (e.g., FMT or rationally designed consortia) with established ICI regimens are currently being aggressively explored in ongoing clinical trials, such as the FAB-HCC trial^[41]. The ultimate success of these interventions hinges entirely on establishing a rigorous clinical validation loop. The field must rapidly transition to standardized, adequately powered, multi-center phase II/III randomized controlled trials to critically evaluate the optimal combination, optimal timing, and true clinical utility of these novel therapies. Furthermore, extensive collaborative efforts with regulatory agencies are urgently needed to establish clear evaluation and approval pathways, ensuring the long-term safety and standardized, pharmaceutical-grade production of FMT and live biotherapeutic products. Ultimately, standardizing the entire translational process - from precise sample collection to AI-driven data analysis - will propel HCC immunotherapy into a transformative new era of systemic remodeling and precise regulation.