Rewriting hepatic fate: emerging gene therapy strategies for liver disease

Delivery platforms for liver-directed gene therapy

The choice of delivery platform is crucial for the design of liver-directed gene therapies and significantly impacts their safety, efficacy, and durability. Among the available systems, hepatotropic AAV vectors, lentiviral vectors, and LNPs have emerged as the main options, each with distinct advantages and limitations that affect their suitability for various indications and patient populations. These established platforms, together with emerging approaches such as exosome-based systems, virus-like particles, and synthetic polymer nanocarriers, are compared in Figure 2, which summarizes their key advantages and limitations for liver targeting.

Figure 2. Comparison of different delivery platforms for liver-directed gene therapy. The figure highlights the main advantages and disadvantages of adenoviral vectors, adeno-associated virus (AAVs) vectors, lentiviral vectors, lipid nanoparticles (LNPs), exosomes, virus-like particles, and synthetic polymers/nanocarriers for hepatic gene transfer. Adenoviral vectors and AAVs are known for achieving high transduction efficiency in hepatocytes. Specific AAV serotypes like AAV8, AAV5, and AAV3B demonstrate strong liver tropism. However, they are constrained by immunogenicity, high seroprevalence of neutralizing antibodies, limited packaging capacity, and loss of episomal genomes in proliferating pediatric livers in the case of AAVs. Lentiviral vectors integrate into the host genome and can carry larger cassettes, enabling stable, long-term expression. This feature makes them suitable for ex vivo gene correction of hepatocytes or hepatocyte-like cells. Nevertheless, the risk of insertional mutagenesis and oncogenic activation currently restricts their direct in vivo liver applications. Non-viral systems, such as LNPs, exosomes, virus-like particles, and synthetic polymers/nanocarriers, have the ability to deliver mRNA, siRNA, and genome-editing components (such as CRISPR-Cas-based nucleases, base editors, and prime editors) with adaptable cargo size. They also offer the potential for repeated administration. However, they encounter challenges related to innate immune activation, reliance on LDLR-mediated uptake (for standard LNPs), cell-type specificity beyond hepatocytes, manufacturing complexity, and relatively limited clinical experience in systemic liver gene therapy.

AAV vectors have become the primary platform for in vivo liver-directed gene addition. Naturally occurring and engineered serotypes like AAV8, AAV5, and AAV3B exhibit strong tropism for hepatocytes following systemic administration^[3,24]. Once delivered, AAV vector genomes typically persist as episomal concatemers in the nucleus, allowing for long-term transgene expression in non-dividing or slowly dividing cells while minimizing integration-related genotoxicity^[24]. Clinical trials in hemophilia A and B utilizing liver-directed AAV gene therapy have shown sustained expression of factor VIII or IX over multiple years in numerous patients, resulting in significant reductions in bleeding episodes and factor concentrate usage^[27]. Similar AAV-based approaches have progressed to clinical trials for monogenic metabolic liver diseases, such as ornithine transcarbamylase deficiency and Crigler-Najjar syndrome, with initial findings indicating biochemical correction in some participants^[28,29].

However, the deployment of AAV vectors in the liver is constrained by several important factors. Many individuals harbor pre-existing neutralizing antibodies against common capsids, which can severely reduce vector transduction and preclude treatment^[30,31]. Administration of high-dose AAV can induce robust humoral and cellular immune responses against the capsid and, occasionally, the transgene product^[32]. Clinically, this often manifests as transient elevations in aminotransferases that are typically mitigated by corticosteroids, but in rare cases more serious hepatic or systemic toxicity has been observed^[33]. The risk of immune responses and toxicity must be balanced against the need to deliver sufficient vector genomes per hepatocyte to achieve therapeutic transgene expression. Furthermore, AAV’s relatively small packaging capacity, on the order of 4.7 kb, limits the size of the cDNA and regulatory elements that can be delivered, complicating therapies for genes with large coding sequences or requiring complex regulatory architectures^[32]. In pediatric patients, the non-integrating nature of AAV, which is an advantage from the perspective of genotoxicity, becomes a liability because vector genomes are diluted as hepatocytes proliferate, leading to decreased expression over time^[32]. These challenges, immunogenicity, dose-related toxicity, and limited durability, have prompted extensive efforts to engineer next-generation AAV capsids.

Lentiviral vectors are another important class of delivery vehicles. These vectors integrate into the host genome, enabling stable and long-term expression even in proliferating cells. This feature is advantageous in pediatric settings or when sustained transgene production is required across many cell divisions^[34]. In the context of liver disease, lentiviral vectors have primarily been used in ex vivo settings. Autologous hematopoietic stem cells have been transduced ex vivo to treat inherited metabolic diseases that secondarily affect the liver, illustrating the strength of lentiviral platforms in systemic disorders. Experimental approaches are exploring ex vivo gene correction of hepatocytes or hepatocyte-like cells derived from induced pluripotent stem cells, followed by transplantation into the liver. This leverages the regenerative capacity of the organ to expand corrected cells. Direct in vivo delivery of lentiviral vectors to the liver remains controversial due to concerns about insertional mutagenesis, oncogenic activation, and off-target integration in non-hepatic tissues. In particular, integration at certain genomic sites may promote hepatocarcinogenesis^[34,35]. Although modern self-inactivating lentiviral vectors with improved integration profiles are significantly safer than earlier retroviral systems^[36,37], the long-term risks in the context of chronic liver disease and potential pre-neoplastic changes require careful evaluation. For these reasons, lentiviral vectors currently have a more limited role in liver gene therapy, focusing on ex vivo applications and scenarios where the benefits of stable integration clearly outweigh the risks.

LNPs have quickly become a prominent non-viral platform for delivering nucleic acids to the liver due to their versatility^[38]. These LNPs encapsulate various types of cargo, such as mRNA, small interfering RNA (siRNA), or CRISPR-Cas components, within a lipid shell made up of ionizable lipids, cholesterol, phospholipids, and polyethylene glycol-conjugated lipids^[38]. Upon intravenous administration, LNPs are opsonized and interact with apolipoprotein E, which facilitates uptake by hepatocytes through low-density lipoprotein receptors, resulting in strong liver tropism^[39]. In the clinical setting, LNP technology has been proven effective with approved siRNA therapeutics targeting genes expressed in hepatocytes and mRNA vaccines. While mRNA vaccines are not liver-specific, they have demonstrated the scalability and safety of the LNP platform. For gene therapy purposes, LNPs offer the advantage of delivering large cargos, such as mRNA encoding genome editors and guide RNAs, without integrating into the genome^[38]. This allows for transient expression of potent genome-editing tools like base editors, which can induce permanent genetic changes while minimizing the risk of off-target effects^[40]. Early clinical trials using LNP-formulated CRISPR-Cas9 components for in vivo genome editing in the liver have produced promising results. Significant knockdown of target genes has been achieved after a single infusion, demonstrating the feasibility of this approach.

Although LNPs efficiently deliver nucleic acids to hepatocytes through apolipoprotein E-mediated uptake pathways described above, achieving precise targeting of other hepatic cell populations remains challenging^[41]. Efficient delivery to HSCs, cholangiocytes, or Kupffer cells will likely require additional engineering strategies, such as incorporating cell-specific ligands or modifying nanoparticle surface chemistry^[41]. A promising lipid candidate (CL15A6) that allows effective mRNA delivery to activated HSCs by targeting clathrin-mediated endocytosis through the platelet-derived growth factor receptor-β showed high efficacy in a mouse model undergoing liver fibrosis^[42]. Similarly, recently the incorporation of mannose-conjugated cholesterol allowed effective targeting of mRNA-dotted nanoparticles to liver sinusoidal endothelial cells and Kupffer cells following intravenous administration in mice^[43].

Despite these strengths, LNP-based delivery also has limitations. Infusion-related reactions, triggered immune responses, and transient elevations in liver enzymes can occur, particularly at higher doses^[40]. Achieving efficient editing or therapeutic expression at lower doses and increasing the transduction rates in patients lacking sufficient low-density lipoprotein receptor remains a priority, especially for indications requiring treatment in more fragile patients or in combination with other therapies^[44]. Moreover, while hepatocyte targeting via apolipoprotein E is well established, extending precise targeting to other liver cell types, such as stellate cells, cholangiocytes, or Kupffer cells, will require additional engineering, for example by incorporating specific targeting ligands on the LNP surface^[39]. Nonetheless, the modularity of LNPs, their favorable manufacturing profile, and the possibility of repeated dosing without eliciting long-lasting neutralizing antibodies make them highly attractive for many liver-directed applications.

Beyond these principal platforms, other modalities are emerging. Exosome-based delivery systems^[45,46], virus-like particles^[47], synthetic polymers/nanocarriers^[48], and hybrid approaches^[49] that combine viral and non-viral features are under investigation as potential alternatives or complements. They aim to improve cell-type specificity, reduce immunogenicity, and enable re-dosing. In parallel, advances in vector manufacturing, including suspension cell culture systems, improved purification methods, and continuous manufacturing processes, are gradually addressing scalability and cost barriers. Together, this diverse and rapidly evolving landscape of delivery platforms provides a versatile toolkit for tailoring liver gene therapies to specific disease contexts, patient characteristics, and therapeutic objectives.

Beyond safety considerations, therapeutic efficacy is a central challenge for liver-directed gene therapy. Achieving sufficient levels of hepatocyte transduction or editing is often necessary to reach the biochemical thresholds required for clinical benefit and reduce toxicities^[50]. Some metabolic disorders exhibit favorable threshold effects, where restoration of a small fraction of normal enzyme activity can improve disease manifestations. However, many conditions require higher levels of correction or sustained expression across a substantial proportion of hepatocytes. Transduction efficiency can vary considerably between individuals due to differences in liver architecture and immune systems^[24]. Factors such as fibrosis stage, vector pharmacokinetics, and host immune responses will impact gene transfer efficiencies^[19,51]. Additionally, the heterogeneous distribution of vectors within the hepatic lobule may lead to uneven gene expression^[50], potentially limiting therapeutic impact even when the total vector dose is high. Improving delivery efficiency while maintaining acceptable safety margins remains a major objective in the development of next-generation vectors and nanoparticle systems.

Therapeutic strategies: from gene addition to precision editing

Recent advances in genome editing technologies have expanded the therapeutic toolkit for liver diseases, building on earlier gene-addition approaches. This enables direct modification of endogenous loci in addition to conventional transgene delivery^[52]. This gene-addition strategy remains highly relevant and forms the basis of many ongoing trials for monogenic metabolic liver diseases. In these treatments, an AAV vector carrying a cDNA controlled by a liver-specific promoter is delivered systemically, resulting in persistent episomal expression of the therapeutic protein. Since many metabolic diseases require only modest levels of transgene expression for biochemical correction, even a small amount of expression in some hepatocytes can significantly improve clinical outcomes. For example, restoring partial activity of urea cycle enzymes can prevent hyperammonemic crises, and low-level expression of UGT1A1 can decrease unconjugated bilirubin in Crigler-Najjar syndrome^[53]. Gene addition is conceptually straightforward, typically produces a clear pharmacodynamic effect, and can be applied even when mutations are heterogeneous or distributed across large genes.

However, gene addition has limitations that become more evident as the field advances. This method does not modify the endogenous mutant locus, so it may not fully replicate the natural regulation of gene expression^[54,55]. Transgene expression depends on the promoter and regulatory elements chosen, which may not match the timing, location, or amount of endogenous gene activity. Overexpression can cause cellular stress or toxicity, while underexpression may not provide enough benefit^[54,55]. In diseases with dominant-negative or gain-of-function mechanisms, simply adding a wild-type allele may not be adequate and could, in some cases, worsen the disease. Additionally, in pediatric patients, non-integrating gene addition may require retreatment as the vector genomes are diluted with hepatocyte growth, but re-dosing is complicated by immune reactions to the vector.

These limitations have catalyzed a shift toward precision genome editing strategies that directly modify endogenous loci. CRISPR-Cas nucleases enable the targeted introduction of double-strand breaks at specific genomic sites, which are then repaired by cellular machinery through non-homologous end joining or homology-directed repair^[56]. By delivering a Cas nuclease and a guide RNA to hepatocytes, one can disrupt pathogenic genes, correct point mutations, or insert therapeutic cassettes into safe harbors or endogenous loci. In the liver, CRISPR-Cas-mediated gene disruptions has been used in preclinical models to knock out genes that promote disease, such as those involved in cholesterol synthesis or viral replication^[56]. More sophisticated strategies exploit homology-directed repair or homology-independent targeted integration to insert corrective sequences into precise genomic locations, restoring physiological regulation.

However, double-strand break-based editing raises concerns about genotoxicity, including large deletions, chromosomal rearrangements, and activation of p53 pathways^[57,58]. This has led to the development of editing modalities that do not rely on double-strand breaks. For example, base editors combine a catalytically impaired Cas protein with a deaminase enzyme, allowing for targeted conversion of single nucleotides (such as C to T or A to G) within a specific window. This enables the correction of many pathogenic point mutations or the introduction of protective variants in regulatory elements while minimizing the risk of large-scale genomic rearrangements. Prime editors further expand the range of possible edits by utilizing a reverse transcriptase fused to Cas and a specialized guide RNA (a prime editing guide RNA) that encodes the desired change^[59]. Prime editing can introduce small insertions, deletions, or arbitrary point mutations without the need for donor templates or double-strand breaks, although its efficiency in vivo is still being optimized^[59,60].

Another important factor influencing therapeutic success is the efficiency of genome editing within target hepatocytes. While preclinical models often show high editing rates under controlled experimental conditions, achieving similar efficiencies in humans is much more difficult^[61]. Editing outcomes are influenced by various factors, such as vector delivery efficiency, availability of editing components within cells, chromatin accessibility at the target site, and cellular DNA repair pathways. In many diseases, only a portion of hepatocytes may be edited in vivo, leading to questions about whether the correction levels achieved are adequate for lasting clinical benefits. Current strategies that are being investigated to overcome this limitation include improved guide RNA design, optimized editor variants with increased catalytic activity, and delivery methods that temporarily boost intracellular levels of editing machinery while minimizing prolonged exposure that could raise the risk of off-target effects^[62].

These tools are particularly well-suited for liver applications, where specific point mutations underlie many inherited diseases, and where precise tuning of gene expression through regulatory edits could modulate complex conditions^[63,64].

An additional strategic evolution involves expanding therapeutic targets beyond protein-coding sequences to include promoters, enhancers, splicing junctions, and noncoding RNA^[65]. Regulatory regions that control gene expression, such as enhancers and promoters, as well as non-coding RNAs that modulate hepatic metabolism and fibrosis are increasingly recognized as attractive intervention points. Editing or epigenetically reprogramming these elements offers a way to adjust the expression of endogenous genes in a graded fashion, potentially yielding more physiological and durable effects than overexpression from exogenous cassettes. For example, editing transcription factor binding sites in the promoter of a key metabolic enzyme could moderately upregulate its expression in response to endogenous signals, while targeting long non-coding RNAs involved in stellate cell activation might attenuate fibrogenesis without completely abolishing necessary wound-healing responses^[65].

Another conceptual shift involves cell-type specificity. To date, most liver gene therapy approaches have primarily focused on hepatocytes. However, there is now increasing attention being directed towards non-parenchymal cell populations involved in fibrosis, inflammation, and cholestatic disease^[20,66]. Emerging work is beginning to target these other cell types. For instance, therapies directed at stellate cells aim to reverse or halt fibrosis by downregulating profibrotic genes, upregulating antifibrotic mediators, or reprogramming activated stellate cells back to a quiescent phenotype^[66]. Cholangiocyte-targeted approaches are being explored for genetic cholangiopathies and cholestatic conditions, where correcting transporter function or modulating bile acid signaling in the biliary epithelium could have profound effects on disease progression^[67]. Immune cell-directed therapies, including liver-resident macrophages and (innate-like) T cells, may modulate chronic inflammation and autoimmunity or enhance anti-tumor responses in HCC^[68].

Ultimately, therapeutic strategies for liver disease are likely to become increasingly nuanced and multimodal. In some cases, such as simple loss-of-function monogenic metabolic disorders, traditional hepatocyte-directed gene addition may remain the most practical approach, especially if started early before significant tissue damage occurs. In other situations, such as dominant disorders, complex polygenic diseases, or advanced fibrosis, precision editing and cell-type-specific targeting will be necessary to achieve meaningful and long-lasting benefits. Genome editing can also be combined with temporary modulation of pathways using RNA-based therapies, small molecules, or biologics. The emerging toolkit of gene addition, knockout, knock-in, base editing, prime editing, and epigenome editing, along with the ability to select among hepatocytes and various non-parenchymal cells as targets, provides a foundation for a flexible, programmable therapeutic toolkit that can be customized to the underlying mechanism of different liver diseases.

Key challenges and safety considerations: a personal view

The translation of liver-directed gene therapy from preclinical models to clinical practice has highlighted several challenges that must be addressed to fully realize its potential. Immune responses to vectors and transgene products, dose-dependent toxicities, risks of genotoxicity and oncogenesis, durability of effect in the context of liver growth and regeneration, as well as practical issues related to manufacturing, cost, and equitable access are all central concerns.

Immunogenicity remains a major barrier to liver-directed gene therapy. As described ealier, AAV vectors can be affected by preexisting neutralizing antibodies, immune responses against capsid proteins, and limitations in vector re-administration. These immune mechanisms can reduce hepatocyte transduction efficiency and may lead to transient elevations in liver enzymes or loss of transgene expression. Consequently, strategies such as capsid engineering, transient immunomodulation, and alternative delivery platforms are being explored to mitigate immune responses and enable safer and more durable therapeutic outcomes^[95]. Cellular immune responses, especially CD8⁺ T cells targeting capsid-derived peptides presented by hepatocytes, can result in temporary or prolonged loss of transgene-expressing cells and elevated liver enzymes. This may require immunosuppressive therapy to maintain transgene expression and prevent hepatitis.

A practical limitation is the difficulty of vector re-administration. Following exposure to AAV capsids, most patients develop durable neutralizing antibodies that prevent effective redosing with the same or closely related serotypes^[52]. This constraint is particularly relevant for pediatric patients and chronic conditions in which therapeutic expression may decline over time. Current research therefore focuses on capsid engineering, transient immune modulation, and the use of alternative delivery systems that may allow repeated administration.

Similar concerns, albeit with different timing and mechanisms, apply to non-viral platforms such as LNPs. Innate immune activation and complement activation-related pseudoallergy can occur, especially at higher doses^[96,97]. Strategies to address these immune responses include engineering capsids to reduce recognition by pre-existing antibodies, temporary immunosuppression during and after vector administration, and exploring alternative platforms including LNPs or engineered exosomes that may allow for repeated dosing. However, the inability to easily re-dose many current vectors is a significant limitation, particularly for pediatric applications and chronic diseases.

Toxicity and genotoxicity are closely related concerns. As discussed, high-dose systemic administration of AAV vectors has been associated with acute hepatotoxicity, thrombocytopenia, microangiopathy, and, in rare cases, severe liver injury or multi-organ failure. The exact mechanisms remain incompletely understood but likely involve a combination of innate immune activation, complement activation, direct effects of capsid proteins, and possible off-target transduction of other tissues. Clinical experience has shown that administering very high systemic doses of AAV vectors can lead to significant toxicity^[98]. This toxicity may include acute hepatocellular injury, thrombocytopenia, complement activation, and, in rare cases, thrombotic microangiopathy. These adverse events are primarily observed in trials that require high vector genome doses to achieve effective hepatocyte transduction^[99]. This underscores the narrow therapeutic window that limits many AAV-based liver gene therapies. Therefore, improving vector potency and hepatocyte tropism to reduce the total vector dose required for therapeutic efficacy has become a primary objective of current vector engineering efforts.

Vigilant monitoring and dose titration in clinical trials are essential. In the long term, although AAV is mostly non-integrating, low-level integration events can occur. In animal models, AAV integration events have occasionally been associated with HCC, especially when vectors were administered at very high doses in neonatal animals or in the presence of underlying oncogenic stress. The extent to which these findings apply to humans is still being actively researched. Careful analysis of integration site and long-term monitoring of treated patients are crucial to evaluate any potential increased risk of malignancy.

Genome editing introduces additional safety considerations beyond those associated with gene addition. As discussed in the section on precision editing strategies, double-strand break-based approaches such as CRISPR-Cas nucleases can generate unintended genomic alterations, including large deletions or chromosomal rearrangements^[57,100]. These effects may include potential deamination at off-target sites or unintended edits in sequences with partial homology to the target. Although newer tools such as base editors and prime editors reduce reliance on double-strand breaks, careful assessment of off-target activity and long-term genomic stability remains essential. Delivering editor components through high-dose AAV or LNPs can introduce issues of immunogenicity and toxicity. To address these risks, thorough preclinical assessment of off-target editing sites using unbiased genome-wide methods is essential. Dose optimization and transient expression strategies may help minimize exposure to editing components. In addition, post-treatment genomic surveillance may be required. This becomes especially crucial when these technologies are used in younger patients with longer lifespans, as adverse effects may become more apparent over time.

Durability of therapeutic effects is another key challenge, particularly in pediatric populations. As mentioned earlier, hepatocyte proliferation during liver growth can dilute nonintegrating vector genomes such as AAV episomes, leading to a gradual loss of transgene expression^[25,52]. Clinical experience in hemophilia and metabolic diseases suggests that expression levels achieved with AAV can remain relatively stable in adults over several years but may decline more rapidly in younger patients^[85,52]. Integrating vectors or genome editing strategies that permanently modify the hepatocyte genome can overcome this limitation, but at the expense of increased concern about long-term genomic effects and oncogenic potential. This trade-off highlights the need for age-tailored strategies: for example, using transient gene addition or supportive therapies to bridge young children to an age where more durable interventions can be safely administered, or developing re-dosing protocols that utilize different serotypes or platforms to circumvent neutralizing antibodies. Observations from long-term clinical follow-up studies further emphasize the importance of sustained therapeutic durability. In several gene therapy trials, especially in hemophilia, transgene expression has been detectable for multiple years but has shown gradual decline in some patients over time^[101]. The mechanisms underlying this variability remain incompletely understood and may involve immune-mediated loss of transduced hepatocytes, epigenetic silencing of vector genomes, or hepatocyte turnover. These findings underscore the importance of developing strategies that support sustained expression, enable vector re-administration, or achieve permanent genomic correction through editing-based approaches.

Beyond biological and clinical challenges, practical and societal issues also play a significant role. Manufacturing AAV vectors and LNPs at scale with consistent quality is a technically demanding task. Production is often hindered by low yields, limited scalability, and analytical complexity. These challenges result in high production costs and uncertainties regarding long-term durability and safety^[52,102,103]. The capacity of current production infrastructure may be insufficient to meet the potential demand if gene therapies are to be widely adopted for common liver diseases, not just rare monogenic conditions. High production costs, combined with the complexity of delivery and long-term follow-up, translate into expensive therapies that raise questions about affordability and equitable access. These concerns are particularly acute for liver diseases that are highly prevalent in low- and middle-income countries, such as viral hepatitis-related cirrhosis and HCC, where healthcare resources are limited. Addressing these disparities will require not only advances in manufacturing technology to reduce costs but also novel pricing models, global partnerships, and policies that prioritize equitable distribution. In addition to biological safety concerns, several practical challenges have become apparent during clinical translation. Patient eligibility can be limited by the presence of pre-existing neutralizing antibodies against common AAV capsids, which may exclude a substantial proportion of the population from treatment. Furthermore, manufacturing clinical-grade viral vectors at large scale remains technically demanding and costly, potentially limiting broad access if these therapies are expanded beyond rare diseases. Recent work has suggested that machine learning and advances in viral immunology can enable engineering of a new generation of de-immunized capsids beyond the antigenic landscape of natural AAVs^[104]. Long-term clinical monitoring is also required to evaluate durability of expression and potential late adverse effects, adding additional logistical complexity to the implementation of gene therapy in routine clinical care.

Finally, ethical considerations are pervasive in the field of gene therapy, especially when discussing germline genome editing for inherited diseases^[105]. When it comes to treating children, especially using irreversible genome editing, there must be a delicate balance between the potential to prevent severe, life-limiting diseases and the uncertainties surrounding long-term risks^[106]. In utero interventions, which are being considered for certain severe diseases, bring up additional ethical dilemmas regarding consent, risks to the fetus and the pregnant individual, and the possibility of unintended consequences on future generations, even if germline modification is not the goal. Open and honest communication with patients and their families about the benefits and risks, strong regulatory oversight, the establishment of long-term registries to monitor outcomes, and inclusive discussions involving patient advocacy groups, ethicists, and diverse communities are all crucial to ensure that advancements in liver gene therapy are carried out responsibly.

Future directions: toward programmable and regenerative liver therapies

It is important to distinguish between technologies that have already demonstrated clinical efficacy and those that remain at an experimental or early translational stage. Liver-directed AAV gene addition therapies for hemophilia and certain systemic disorders represent the most clinically advanced applications and have reached late-phase trials or regulatory approval in some jurisdictions^[4,107]. By contrast, many genome-editing strategies, including base editing, prime editing, and epigenome editing, are still largely in preclinical development or early clinical testing. Similarly, approaches targeting non-parenchymal liver cells or complex polygenic diseases remain largely conceptual. Recognizing these differences in maturity is critical when evaluating the near-term clinical potential of emerging gene therapy platforms.

Looking ahead, the field of liver-directed gene therapy is evolving from a collection of disease-specific products to a more integrated ecosystem of programmable platforms that can be adapted across indications and combined with regenerative and immunomodulatory strategies. Several converging research directions are likely to shape this future. One central theme is the development of next-generation vectors and delivery systems that improve specificity, potency, and safety. Advances in protein engineering, high-throughput screening, and machine learning are being leveraged to create AAV capsids with enhanced hepatocyte tropism, reduced binding to pre-existing antibodies, and minimized off-target transduction of other tissues such as the dorsal root ganglia or myocardium^[108,109]. Directed evolution approaches, performed in vivo in animal models or using humanized liver chimeric mice, can identify capsids that efficiently transduce human hepatocytes. These strategies can yield capsids that efficiently transduce human hepatocytes at lower doses, thereby reducing toxicity and manufacturing demands. Parallel efforts aim to engineer LNPs with tailored properties, including optimized ionizable lipids, surface modifications with targeting ligands, and controlled release characteristics. These improvements may enable preferential delivery to specific liver cell types such as HSCs or cholangiocytes. Such precision delivery systems could form the backbone of modular therapies, in which the therapeutic cargo (for example, a base editor or regulatory RNA) can be exchanged while the delivery platform remains largely unchanged.

A second theme involves the refinement of genome editing to achieve safer, more versatile, and more predictable outcomes. The trajectory of editing tools is moving away from double-strand break-dependent systems toward approaches that minimize DNA damage. Next-generation base editors with reduced off-target deamination and improved specificity, as well as prime editors with enhanced efficiency, are actively being optimized for in vivo use^[110]. RNA editing platforms, particularly antisense RNA-guided adenosine deaminase acting on RNA (ADAR) which operate at the transcript level and thus do not permanently alter the genome, provide another layer of flexibility, particularly for conditions where reversible modulation of gene expression or function is advantageous^[111]. Additionally, epigenome editing tools that recruit transcriptional activators or repressors to specific loci without changing the underlying DNA sequence offer a way to titrate gene expression up or down, which may be especially useful in polygenic or multifactorial liver diseases where subtle modulation of multiple pathways may be preferable to complete ablation or overexpression of a single target^[112].

A third area of active development involves harnessing and directing the regenerative capacity of the liver. Gene therapy can intersect with regenerative medicine in several ways. In situ reprogramming strategies aim to convert one cell type into another within the liver, for example by reprogramming activated stellate cells or myofibroblasts into hepatocyte-like cells, thereby simultaneously reducing fibrosis and increasing functional parenchyma^[113]. Gene therapies can also be used to enhance the engraftment and proliferation of transplanted gene-corrected cells, such as hepatocytes or liver organoids derived from patient-specific induced pluripotent stem cells^[114]. In this scenario, patients might receive a combination of ex vivo gene-corrected cell transplants and in vivo gene therapies that modulate the microenvironment to favor engraftment, survival, and appropriate integration of these cells into the liver architecture. Partial reprogramming approaches that transiently activate developmental transcriptional programs or transcription factors without fully reverting cells to a pluripotent state are being explored as a means to stimulate regeneration while minimizing oncogenic risk, and gene therapy tools could provide temporal and spatial control over such interventions^[115].

Combination therapies are likely to become increasingly important, particularly for complex or advanced liver diseases. For example, in patients with HCC arising in cirrhotic livers, gene therapies may directly target tumor cells, example through oncolytic viruses or constructs delivering pro-apoptotic genes. They may be combined with systemic immunotherapies, such as checkpoint inhibitors, as well as anti-fibrotic gene therapies directed at HSCs. In chronic viral hepatitis, gene editing designed to disable viral genomes or host factors essential for viral replication could complement antiviral drugs and immunomodulatory treatments, potentially leading to deeper and more durable cures. In metabolic diseases, gene addition or editing might be paired with small molecules that modulate metabolic pathways or with diet-based interventions, allowing lower doses of gene therapy and reducing toxicity. Rational design of such multimodal regimens will require systems-level understanding of liver pathophysiology and careful clinical trial design to assess synergistic effects and manage overlapping toxicities.

Data-driven personalization will also shape future liver gene therapies. High-resolution multi-omics profiling of liver tissue and blood, including genomics, transcriptomics, proteomics, and metabolomics, can provide detailed snapshots of disease state, treatment response, and emerging resistance mechanisms^[116,117]. These tools can be combined with advanced imaging and non-invasive biomarkers such as circulating cell-free DNA and extracellular vesicles. Together, they may enable the dynamic adjustment of gene therapy strategies over time. For example, baseline genomic data might identify patients at higher risk for off-target editing or vector integration at specific hotspots, guiding vector choice or the intensity of post-treatment surveillance^[117]. Longitudinal monitoring could detect early signs of clonal expansion or pre-neoplastic changes, prompting interventions or closer follow-up. Artificial intelligence-driven analysis of large datasets from gene therapy registries and trials could uncover patterns that inform dosing, vector selection, and identification of subgroups that derive particular benefit or experience specific adverse events^[118,119].

Finally, the future of liver gene therapy will be shaped by how effectively the field addresses ethical, regulatory, and societal considerations^[120]. Ensuring that advances do not exacerbate existing health disparities will require concerted efforts to make therapies affordable and accessible, including in regions with high burdens of liver disease. Engagement with diverse patient communities and stakeholders is fundamental to building trust, understanding preferences and concerns, and co-designing interventions and follow-up protocols. Regulatory frameworks will need to balance the urgency of addressing severe, otherwise untreatable diseases with the caution warranted by irreversible interventions with long-term unknowns. Innovative trial designs in clinical research, such as adaptive trials and platform trials integrated in Integrated Research Platforms, may be particularly well suited to evaluate modular gene therapy across multiple diseases^[121]. Long-term registries and international collaborations will be essential to capture late-emerging outcomes and to refine best practices over time. Consequently, the future of liver-directed gene therapy probably lies in more precise, programmable, and context-specific interventions that combine gene addition, editing, and reprogramming with pharmacologic, cellular, and immunologic therapies. Achieving this vision will require not only technological advances but also careful clinical translation, thorough safety monitoring, and a dedication to fair implementation.