siRNA therapeutics for metabolic dysfunction-associated Steatohepatitis: Targets, delivery, clinical evidence, and translational challenges

Abstract

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most prevalent chronic liver disease worldwide. Its progressive inflammatory form, metabolic dysfunction-associated steatohepatitis (MASH), is driven by interacting genetic, metabolic, inflammatory, and fibrogenic processes that complicate therapeutic development. Small interfering RNA (siRNA) is an emerging modality for sequence-specific suppression of genetically supported targets and intracellular proteins that are difficult to modulate with conventional drugs. Chemical modification and liver-directed delivery have improved siRNA stability and hepatic exposure. However, most MASH-directed programs remain preclinical or in early clinical development, and human data have not established consistent histological benefit. Major limitations include delivery to nonparenchymal cells, inefficient endosomal escape, off-target effects, immunogenicity, and uncertain long-term safety. This review critically evaluates target selection, delivery, clinical and preclinical evidence, unsuccessful programs, competing nucleic acid modalities, and combination strategies. The clinical role of siRNA in MASH remains to be established.

Highlights

• •SiRNA targets genetically supported drivers of MASH with sequence specificity.
• •Clinical evidence for siRNA in MASH remains limited and histology is unproven.
• •Delivery, endosomal escape, and long-term safety remain translational barriers.
• •Precision selection and rational combinations may define the role of siRNA in MASH.

1. Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly termed non-alcoholic fatty liver disease (NAFLD), is the most prevalent chronic liver disease worldwide. Its inflammatory form, metabolic dysfunction-associated steatohepatitis (MASH; formerly termed non-alcoholic steatohepatitis, NASH), may progress to fibrosis, cirrhosis, hepatic decompensation, and hepatocellular carcinoma (HCC) [1]. Modeling studies using the historical NAFLD terminology projected increasing liver-related complications and mortality between 2015 and 2030 [2], while prevalence estimates based on the former definition indicated a substantial but geographically variable disease burden [3]. In 2023, a multisociety consensus recommended MASLD and MASH to emphasize metabolic dysfunction and reduce stigmatizing terminology [4]. These terms are used throughout this review; NAFLD and NASH are retained only when reporting the original terminology of historical studies, clinical trials, or publication titles.

MASH is histologically defined by steatosis, hepatocellular ballooning, and lobular inflammation, with or without fibrosis [5]. Disease progression is heterogeneous and nonlinear, but clinically significant fibrosis is closely associated with adverse liver-related outcomes [6]. MASH also occurs within a systemic metabolic context and is associated with cardiovascular disease, chronic kidney disease, HCC, and extrahepatic malignancies [7]. This biological and clinical heterogeneity complicates therapeutic development. The relative contributions of genetic susceptibility, lipid dysregulation, inflammation, and fibrogenesis vary among patients and across disease stages.

Consequently, an intervention effective for one pathogenic driver or disease stage may not benefit all patients. Systemic metabolic therapies may not adequately address established fibrosis, whereas anti-inflammatory and antifibrotic strategies must balance efficacy against effects on immune regulation and tissue repair.

Small interfering RNA (siRNA) enables sequence-specific suppression of disease-associated transcripts and may be particularly suitable for genetically supported targets and intracellular proteins that are difficult to modulate with small molecules or antibodies. The liver is accessible to oligonucleotide therapeutics through platforms such as N-acetylgalactosamine conjugates and lipid nanoparticles. However, hepatocyte uptake does not ensure delivery to hepatic stellate cells or macrophages, and transcript knockdown does not establish histological or clinical benefit. The consequences of sustained gene silencing also require long-term evaluation in a chronic disease. This review critically evaluates the biological rationale and prioritization of siRNA targets, delivery strategies, clinical and preclinical evidence, and translational limitations in MASH. It distinguishes clinically tested programs from preclinical concepts, examines lessons from unsuccessful programs, and considers the potential role of siRNA in precision and combination-treatment strategies.

2. Biological rationale for siRNA target selection in MASH

MASH results from interacting genetic, metabolic, inflammatory, and fibrogenic processes rather than a fixed sequence of insults. The “multiple parallel hits” model captures this heterogeneity but does not imply that simultaneous inhibition of several pathways will necessarily improve clinical outcomes [8]. For siRNA development, its principal value lies in identifying disease drivers that are causal, accessible, and safely suppressible. The major systemic and hepatic processes involved in MASH pathogenesis are summarized in Fig. 1.

Fig. 1.

Pathogenesis of MASH. Created with BioRender.com (https://BioRender.com/zqtrah9).

2.1. Lipid metabolism, lipotoxicity, and genetic susceptibility

Excess fatty-acid influx and hepatic lipid synthesis can exceed the metabolic capacity of hepatocytes, inducing lipotoxicity, endoplasmic reticulum stress, mitochondrial dysfunction, and cell death [8,9]. Regulators of lipid synthesis or uptake, including SREBP-1c and CD36, have been investigated as potential siRNA targets [[10], [11], [12], [13]]. However, the evidence remains predominantly cellular or preclinical. Because these proteins also contribute to physiological lipid and glucose homeostasis, sustained suppression may induce metabolic compensation or impair normal hepatic function.

Human genetic evidence can provide a stronger basis for target selection, but the direction of the genetic effect is critical. The PNPLA3-I148 M variant promotes pathogenic PNPLA3 accumulation on hepatic lipid droplets, supporting reduction of its expression as a genotype-directed strategy [14]. Protective loss-of-function variants in HSD17B13 support therapeutic silencing, and an HSD17B13-directed RNAi candidate has entered clinical evaluation [15,16]. By contrast, reduced TM6SF2 function impairs lipid secretion and increases hepatic fat, indicating that indiscriminate TM6SF2 silencing could reproduce rather than correct the disease phenotype [17]. Genetic association alone therefore does not establish whether a target should be suppressed. Clinically advanced THRβ agonist and GLP-1 receptor agonist therapies also provide efficacy benchmarks against which metabolic siRNA programs should be evaluated [[18], [19], [20], [21]].

2.2. Inflammation and fibrogenesis

Lipotoxic hepatocyte injury promotes the release of damage-associated molecular patterns, which activate resident and recruited immune cells and sustain inflammatory signaling [22,23]. HMGB1 and CCR2 have been evaluated as preclinical siRNA targets, whereas TLR4 provides mechanistic support for hepatocyte inflammatory-fibrogenic signaling through the Jagged1/Notch pathway [[24], [25], [26]]. Prolonged suppression of innate immune pathways could nevertheless impair host defense, resolution of injury, or tissue repair. The unsuccessful Phase III development of the CCR2/CCR5 inhibitor cenicriviroc in the AURORA trial (NCT03028740) further indicates that pathway involvement in MASH does not predict therapeutic efficacy [27].

Recent studies have identified additional intracellular regulatory nodes, but their suitability for siRNA differs according to the required direction of modulation and the accessible target cell. Hepatocyte-specific CEBPA expression decreases during MASH progression, and experimental CEBPA loss aggravates fibrosis [28]. Direct CEBPA silencing would therefore be directionally inappropriate; activation of CEBPA or suppression of its upstream repressors or profibrotic downstream mediators may be more rational. EFHD2 has been identified as a macrophage- and monocyte-associated regulator of hepatic interferon-γ signaling, and its genetic deletion attenuated experimental steatohepatitis and fibrosis [29]. However, therapeutic siRNA activity and selective macrophage delivery have not been established. NLRP3 sensing has also been implicated in neutrophil extracellular trap-mediated activation of hepatic stellate cells and monocytes in alcohol-accelerated MASH fibrosis [30]. This finding supports pathway relevance but does not constitute evidence for an NLRP3-directed siRNA therapy in MASH. Together, these examples illustrate that discovery of an intracellular disease node must be followed by evaluation of modulation direction, target-cell accessibility, therapeutic window, and direct RNA-silencing evidence.

Persistent injury activates hepatic stellate cells (HSCs) and promotes extracellular-matrix production through TGF-β/SMAD and related profibrotic pathways [22,31]. Proposed siRNA targets include HSP47, TAZ, TGF-β1, Jagged1, IL-11, and PDGFR-β, but most are supported primarily by cell or animal studies [26,[32], [33], [34], [35], [36]]. BMS-986263, an HSP47-targeting siRNA formulation, entered Phase II evaluation in patients with MASH-related compensated cirrhosis but was terminated because of insufficient short-term efficacy [37]. This outcome shows that biological plausibility and progression to clinical testing do not establish antifibrotic efficacy.

Fibrotic targets also present delivery and safety constraints. GalNAc conjugates predominantly enter hepatocytes and are unsuitable for direct HSC targeting without an alternative ligand or delivery system [32]. Cell selectivity observed in animal models may not be reproduced in fibrotic human liver. Sustained or systemic suppression of pleiotropic pathways such as TGF-β could also disrupt immune regulation, wound healing, tumor surveillance, and tissue homeostasis [31]. These concerns favor cell-selective delivery and controlled partial suppression.

2.3. Systemic metabolic drivers and the limits of liver-directed silencing

Insulin resistance increases adipose-tissue lipolysis and hepatic fatty-acid exposure, thereby amplifying lipotoxicity and MASH progression [38,39]. Gut dysbiosis and impaired intestinal barrier function may further alter bile-acid signaling and increase hepatic exposure to microbial products [40]. Because these drivers involve obesity, adipose-tissue dysfunction, diet, and extrahepatic insulin resistance, a single liver-directed siRNA is unlikely to correct the broader systemic phenotype.

Systemic metabolic interventions may therefore be more appropriate for these disease components. Liver-directed siRNA may instead have a complementary role against genetically supported or intrahepatic targets that remain active despite systemic treatment. Such programs must demonstrate incremental benefit over established therapies while preserving tissue selectivity and long-term metabolic safety.

2.4. Prioritizing siRNA targets for MASH

Biological involvement in MASH is insufficient to establish a viable siRNA target. Candidate targets should have human genetic, functional, or causal evidence that defines both their disease relevance and the appropriate direction of therapeutic modulation. Their expression and disease-stage relevance should correspond to a hepatic cell population accessible to a sufficiently selective delivery system. Target engagement should be measurable and linked to histological or clinical benefit rather than inferred from transcript knockdown, aminotransferase reduction, or decreased liver fat alone.

Prioritization should also consider physiological target functions, the therapeutic window, homeostatic compensation, and the reversibility of adverse biological effects. A viable siRNA strategy should provide superior or complementary value relative to approved or clinically advanced therapies. Human evidence should therefore be weighted more strongly than isolated animal efficacy, and histological or clinical outcomes more strongly than pharmacodynamic biomarkers. Targets that do not meet these criteria should remain classified as mechanistic or preclinical concepts rather than therapeutically validated candidates.

3. Current MASH treatment Landscape and therapeutic limitations

MASH management now includes lifestyle intervention, metabolic surgery, and disease-specific pharmacotherapy. However, treatment responses remain heterogeneous, and available options do not address every disease stage or patient subgroup. Their efficacy and limitations provide the clinical benchmarks against which siRNA therapeutics should be evaluated. Representative approved, investigational, and unsuccessful pharmacological strategies for MASH are summarized in Table 1.

Table 1.

Selected pharmacological benchmarks for siRNA development in MASH.

Therapy	Status and key evidence	Main limitation	Relevance to siRNA
Resmetirom [18,43]	Accelerated approval; histological benefit	Incomplete response; cirrhosis excluded	Liver-directed efficacy benchmark
Semaglutide [44,45]	Accelerated approval; histological and metabolic benefit	Gastrointestinal adverse effects; outcomes pending	Strong comparator for systemic metabolic therapy
Pioglitazone [[46], [47], [48]]	Off-label; improves steatohepatitis and insulin sensitivity	Weight gain, edema, and fracture risk	Selectivity and tolerability benchmark
Pegozafermin [19]	Investigational; Phase II activity and Phase III development	Long-term benefit unconfirmed	Metabolic and combination comparator
Obeticholic acid [[49], [50], [51], [52]]	Fibrosis signal; not approved	Pruritus, increased LDL-C, unfavorable benefit–risk balance	Surrogate improvement may be insufficient
Simtuzumab [53]	Phase IIb unsuccessful	No meaningful antifibrotic efficacy	Pathway relevance does not establish validation
Selonsertib [54]	Phase III unsuccessful	No significant fibrosis improvement	Highlights preclinical translation risk
Cenicriviroc [27]	Phase III unsuccessful	Primary histological endpoint not met	Supports caution with inflammatory targets

3.1. Lifestyle intervention and metabolic surgery

Lifestyle modification remains the foundation of MASLD and MASH management. In a prospective study with paired liver biopsies, weight loss of at least 5% was associated with improvement in hepatic steatosis, whereas weight loss of at least 10% produced the greatest histological benefit, including improvement in steatohepatitis and fibrosis [41]. Achieving and sustaining this degree of weight reduction nevertheless remains difficult, particularly for patients with type 2 diabetes, advanced disease, or multiple metabolic comorbidities.

Metabolic surgery can produce greater and more durable weight loss in appropriately selected individuals with severe obesity. A meta-analysis of 21 studies involving 2374 patients found improvement in steatosis and steatohepatitis in most patients, whereas fibrosis improved or resolved in approximately 30% [42]. The 30% estimate applies specifically to fibrosis, rather than to all three histological outcomes collectively. Surgical eligibility, perioperative risk, patient preference, and uncertainty in advanced cirrhosis limit its broader application.

3.2. Approved and advanced pharmacological therapies

The pharmacological management of MASH changed with the approvals of resmetirom and semaglutide for adults with noncirrhotic disease and moderate-to-advanced fibrosis. Resmetirom is an oral, liver-directed, thyroid hormone receptor-β-selective agonist that reduces hepatic lipid accumulation and improves atherogenic lipid profiles. In the Phase III MAESTRO-NASH trial, both the 80-mg and 100-mg doses produced higher rates of MASH resolution without worsening of fibrosis and fibrosis improvement without worsening of disease activity than placebo. MASH resolution was reported as NASH resolution in the original trial [43]. The US Food and Drug Administration granted resmetirom accelerated approval in 2024 for adults with noncirrhotic MASH and F2–F3 fibrosis, in conjunction with diet and exercise [18]. However, histological response was incomplete, cirrhosis is excluded from the indication, and long-term clinical benefit requires confirmation [18].

Semaglutide addresses systemic metabolic drivers through glucagon-like peptide-1 receptor agonism. In the planned week-72 interim analysis of the Phase III ESSENCE trial, MASH resolution without worsening of fibrosis occurred in 62.9% of semaglutide-treated participants and 34.3% of placebo recipients. Fibrosis improvement without worsening of MASH occurred in 36.8% and 22.4%, respectively [44]. Semaglutide received accelerated FDA approval in August 2025 for adults with noncirrhotic MASH and moderate-to-advanced fibrosis [45]. It also provides weight-loss and cardiometabolic benefits, although gastrointestinal adverse events were more frequent than with placebo. The ongoing ESSENCE trial is assessing whether histological improvement translates into fewer liver-related clinical events [44].

Pioglitazone can improve steatohepatitis and insulin sensitivity in selected patients with prediabetes or type 2 diabetes, but weight gain, edema, fracture risk, and long-term tolerability concerns limit broader use [[46], [47], [48]]. The FGF21 analogue pegozafermin improved histological endpoints in Phase II [19] and is now being evaluated in Phase III trials (NCT06318169; NCT06419374). Together, these findings show that single-target metabolic and liver-directed therapies can produce histological benefit. Earlier program failures should therefore be interpreted according to target validity, safety, delivery, disease stage, and trial design.

3.3. Lessons from unsuccessful development programs

Several development programs show why biological plausibility and early histological signals should not be equated with clinical validation. Obeticholic acid, a farnesoid X receptor agonist, produced an antifibrotic signal but was associated with pruritus and adverse lipid changes and was not approved for MASH [[49], [50], [51], [52]].

Other fibrogenic and stress-response targets also failed in late-stage development. Simtuzumab, an antibody against lysyl oxidase-like 2, did not improve fibrosis or liver-related outcomes in patients with bridging fibrosis or compensated cirrhosis caused by MASH [53]. Selonsertib, an apoptosis signal-regulating kinase 1 inhibitor, failed to improve fibrosis in the Phase III STELLAR-3 and STELLAR-4 trials [54]. In the Phase III AURORA trial, fibrosis improvement without worsening of MASH occurred in 22.3% of cenicriviroc-treated participants and 25.5% of placebo recipients [27].

These outcomes do not establish that all single-target, anti-inflammatory, or antifibrotic strategies are ineffective. Possible explanations include inadequate causal validation, pathway redundancy, insufficient exposure in the relevant cell population, advanced and less reversible disease, placebo-associated histological improvement, and biopsy variability. The same distinction applies to siRNA: efficient gene silencing cannot compensate for an unsuitable target, inaccessible target cell, inappropriate disease stage, or inadequate therapeutic window.

3.4. Defining the potential added value of siRNA

The approvals of resmetirom and semaglutide raise the evidentiary threshold for siRNA therapeutics: target knockdown must translate into benefit beyond approved treatments [18,[43], [44], [45]]. siRNA may provide added value by suppressing genetically or causally supported intracellular hepatic targets that are not readily addressed by current drugs, particularly PNPLA3 and HSD17B13 [[14], [15], [16],55]. Its likely role is therefore in biomarker-defined populations or in patients with a persistent hepatic driver despite metabolic treatment. Clinical adoption will require incremental histological or clinical benefit, acceptable long-term safety, a manageable treatment burden, and reasonable cost. Molecular selectivity and durable knockdown alone are insufficient.

4. siRNA as a therapeutic modality

The therapeutic value of siRNA depends on its mechanism of transcript suppression, chemical design, duration of activity, and suitability relative to other nucleic acid modalities.

4.1. Mechanism of RNA interference and target scope

Therapeutic siRNAs are chemically synthesized double-stranded RNAs that act in the cytoplasm through the Argonaute 2-containing RNA-induced silencing complex. Following cellular uptake and endosomal escape, the passenger strand is removed, whereas the guide strand directs sequence-specific cleavage of complementary messenger RNA, thereby reducing target-protein expression [[56], [57], [58], [59]].

This mechanism is suitable for targets whose reduced expression is expected to be therapeutic. It cannot restore the function of a deficient protein, and its applicability depends on the required direction and magnitude of modulation, target-protein turnover, and whether partial suppression provides an adequate therapeutic window.

4.2. Chemical design, potency, and durability

Unmodified siRNAs are susceptible to nuclease degradation, rapid clearance, inefficient cellular uptake, and innate immune recognition. Clinically optimized molecules commonly incorporate patterned 2′-O-methyl and 2′-fluoro ribose modifications together with limited terminal phosphorothioate linkages. The position and combination of these modifications must balance nuclease resistance, Argonaute loading, potency, and immunological tolerability [59,60].

Sequence design is equally important. Partial complementarity between the guide-strand seed region and unintended transcripts can produce microRNA-like off-target repression, whereas sequence- or structure-dependent recognition by innate immune receptors may induce inflammatory responses. Bioinformatic screening, transcriptome-wide assessment, nucleotide modification, and rigorous purification can reduce, but not eliminate, these risks [61,62].

Persistent intracellular guide-strand activity can sustain target suppression for weeks or months and may reduce dosing frequency. The same durability delays pharmacological reversal if adverse effects occur and may permit homeostatic compensation during prolonged treatment. Dose selection must therefore account for the magnitude, duration, and recovery of target suppression rather than potency alone.

4.3. Comparison with other nucleic acid modalities

Modality selection should begin with the biological requirements of the target rather than a preference for a particular platform [56,57,59]. Antisense oligonucleotides (ASOs) are single-stranded molecules that induce RNase H1-mediated RNA degradation or alter RNA processing through steric mechanisms. Whereas siRNA acts predominantly in the cytoplasm through Argonaute 2, ASOs can act in both nuclear and cytoplasmic compartments. Both modalities can be conjugated to N-acetylgalactosamine for hepatocyte-directed delivery and can produce prolonged but reversible effects. The development of the PNPLA3-targeting ASO AZD2693 shows that ASOs are a direct alternative to siRNA for genetically defined MASH targets [63].

miRNA-based interventions use mimics to restore miRNA activity or antisense inhibitors to block endogenous miRNAs. Their ability to regulate multiple partially complementary transcripts may be useful for interconnected pathways but complicates dose selection, mechanistic attribution, and prediction of unintended effects. Experience with the liposomal miR-34a mimic MRX34 outside MASH also illustrates the potential for immune-mediated toxicity and delivery-related limitations [57,64].

CRISPR-based approaches permanently modify genomic DNA and may be considered for strongly validated causal targets. In vivo liver editing has shown clinical feasibility outside MASH, but its irreversible effects differ fundamentally from the pharmacologically reversible activity of siRNA and ASOs [65]. Unintended genomic alterations, immune responses, and irreversible on-target consequences impose a particularly high benefit-risk threshold in a heterogeneous chronic disease.

siRNA is therefore most appropriate when therapeutic benefit requires partial and reversible suppression of a causally supported transcript, the relevant cell population is accessible, and durable activity offers an advantage over conventional drugs or alternative nucleic acid modalities. The principal differences among these approaches are summarized in Table 2.

Table 2.

Comparison of nucleic acid therapeutic modalities relevant to MASH.

Modality	Core Mechanism	Maturity and main limitation	Potential role in MASH
siRNA [56,59,66]	AGO2-mediated cytoplasmic mRNA cleavage	Established class outside MASH; off-target repression and nonparenchymal delivery remain limiting	Partial, reversible suppression of causally supported intracellular targets
ASO [63]	RNase H1-mediated degradation or steric modulation	Established class; sequence- and chemistry-dependent toxicity	Alternative to siRNA, including PNPLA3 suppression
miRNA modulation [57,64]	Mimicry or inhibition of endogenous miRNA networks	Limited clinical maturity; broad network and immune effects	Exploratory modulation of interacting disease pathways
CRISPR-based editing [65]	Permanent genomic modification	Early clinical proof of concept outside MASH; irreversible and potentially off-target	Exceptionally well-validated causal genetic targets

5. Delivery strategies for siRNA therapeutics in MASH

Delivery determines whether an siRNA reaches the disease-relevant liver cell and becomes pharmacologically active (see Table 3). An effective platform must protect the RNA, enter the intended cell population, release sufficient guide strand into the cytosol, and support RNA-induced silencing complex (RISC) loading. These requirements differ among hepatocytes, hepatic stellate cells (HSCs), liver sinusoidal endothelial cells (LSECs), Kupffer cells, and recruited macrophages [59,75,76]. Representative siRNA targets and their corresponding delivery strategies in MASH are summarized in Fig. 2.

Table 3.

Clinical evidence and development status of siRNA programs relevant to MASH.

Target/Candidate	Developer/Sponsor	Platform	Phase and status	Reported finding	Main limitation
PNPLA3	Arrowhead Pharmaceuticals/Janssen	GalNAc-siRNA, SC	Phase I, Completed (NCT04844450)	Up to 46% liver-fat reduction in I148 M homozygotes [55,67]	Direct target knockdown and histological benefit not established
ARO-PNPLA3/JNJ-75220795
PNPLA3	Alnylam Pharmaceuticals/Regeneron Pharmaceuticals	GalNAc-siRNA, SC	Phase I, completed; Phase I/IIa and IIa, recruiting (NCT06024408; NCT05648214; NCT07527910)	No results posted	Target engagement and efficacy unproven
ALN-PNP
PNPLA3	Amgen	Hepatocyte-directed siRNA, SC	Phase I, completed (NCT04857606)	No results posted	Clinical activity and further development uncertain
AMG 609
PNPLA3	Eli Lilly and Company	Hepatocyte-directed siRNA, SC	Phase I, active, not recruiting (NCT05395481)	No results posted	Target engagement and efficacy unproven
LY3849891
HSD17B13	Arrowhead Pharmaceuticals/GlaxoSmithKline	GalNAc-siRNA, SC	Phase I/IIa, completed; Phase IIb, active, not recruiting (NCT05583344)	Hepatic mRNA reduced by up to 93.4%; mean ALT reduced by 42.3% [15]	Histological efficacy not established
Gatuzosiran/GSK4532990
HSD17B13	Alnylam Pharmaceuticals	GalNAc-siRNA, SC	Phase I, published; Phase II, recruiting (NCT04565717; NCT05519475)	Median hepatic mRNA reduction of 78% at 400 mg [68]	No histological benefit demonstrated; Phase I protocol ended for business reasons [69]
Rapirosiran/ALN-HSD
MTARC1	Novo Nordisk	Hepatocyte-directed siRNA, SC	Phase I, completed and recruiting studies (NCT05599945; NCT06891365)	No results posted; target relationship reported by sponsor [70]	Target engagement and efficacy unproven
NNC0581-0001/NN6581
HSP47	Nitto Denko/Bristol Myers Squibb	Retinoid-conjugated LNP-siRNA, IV	Phase II, terminated (NCT04267393)	No efficacy demonstrated in MASH; separate human data derived from post-HCV fibrosis [71]	Terminated for lack of short-term efficacy; reason for failure remains unresolved [37]
BMS-986263/ND-L02-s0201

Abbreviations: SC, subcutaneous; IV, intravenous; LNP, lipid nanoparticle; ALT, alanine aminotransferase.

Fig. 2.

Representative siRNA Targets and Delivery Strategies in MASH. Created with BioRender.com (https://BioRender.com/exklu2q).

5.1. Hepatocyte delivery with GalNAc conjugates and lipid nanoparticles

N-acetylgalactosamine (GalNAc) conjugation is the most clinically mature strategy for hepatocyte-directed siRNA delivery. Multivalent GalNAc ligands bind the asialoglycoprotein receptor (ASGPR), which is predominantly expressed on hepatocytes, and promote receptor-mediated internalization. This platform supports subcutaneous administration, prolonged activity, and relatively infrequent dosing. Its validation in RNAi therapeutics outside MASH provides a translational foundation for hepatocyte-expressed targets such as PNPLA3 and HSD17B13 [59].

GalNAc conjugates do not provide comparable functional delivery to HSCs, Kupffer cells, or other nonparenchymal populations. Moreover, efficient receptor-mediated internalization does not establish productive cytosolic exposure.

Lipid nanoparticles (LNPs) protect siRNA from degradation and can promote hepatic uptake and endosomal membrane destabilization. Their intrahepatic distribution depends on lipid composition, particle size, surface properties, and the adsorbed protein corona and may include hepatocytes and cells of the reticuloendothelial system [76,77]. LNP-mediated delivery, including the combined delivery of siRNA and small-molecule cargo, has shown activity in preclinical MASLD models, but comparable efficacy has not been established in humans [78].

GalNAc conjugates provide relatively well-defined hepatocyte selectivity and a clinically established subcutaneous route. LNPs offer greater formulation flexibility but have more complex biodistribution, generally require intravenous administration, and introduce carrier-dependent safety considerations. Platform selection should therefore follow the cellular location of the target and the exposure required for effective suppression.

Outside MASH, the APOC3-directed GalNAc-siRNA plozasiran has produced substantial triglyceride lowering and received FDA approval for familial chylomicronemia syndrome [66,79]. This experience supports the feasibility of durable hepatocyte-targeted RNAi but does not provide evidence of MASH resolution or fibrosis improvement.

5.2. Delivery to hepatic stellate cells and liver macrophages

Activated HSCs are major producers of extracellular matrix, whereas Kupffer cells and recruited monocyte-derived macrophages regulate inflammation, hepatocyte injury, HSC activation, and fibrosis resolution. These cells lack the functional ASGPR-mediated uptake that supports GalNAc delivery to hepatocytes, requiring alternative ligands or nanoparticle properties [32].

Experimental HSC-directed systems include hyaluronic acid carriers targeting CD44 and nanoparticles modified with peptides recognizing platelet-derived growth factor receptor-β (PDGFR-β) [80,81]. Retinol-based strategies exploit the physiological role of HSCs in retinoid storage [36], while ligand-tethered LNPs have shown preferential HSC delivery and antifibrotic activity in animal models [82].

These findings require cautious interpretation. CD44 and PDGFR-β are not exclusive to HSCs, and receptor expression or accessibility may change during HSC activation. Ligand binding in vitro or increased whole-liver accumulation therefore does not establish selective cytosolic delivery. Convincing evidence requires cell-resolved biodistribution, target knockdown in isolated HSCs, and pharmacodynamic effects that cannot be attributed to uptake by hepatocytes or macrophages.

The phagocytic activity of liver macrophages can increase nanoparticle uptake but does not distinguish delivery to resident Kupffer cells from recruited macrophages or other phagocytic populations. Macrophage-directed siRNA LNPs have reduced inflammatory features in preclinical MASH models [24], and liver-directed CCR2 siRNA has reduced monocyte recruitment and fibrosis in experimental cirrhosis [25]. Neither finding alone proves selective macrophage delivery.

Most HSC- and macrophage-directed platforms remain supported by rodent data. Differences in receptor expression, macrophage composition, fibrotic architecture, nanoparticle clearance, and ligand accessibility limit direct extrapolation to human MASH. Translation therefore requires cell-resolved target engagement in human tissue or clinically relevant models.

5.3. Endosomal escape and intracellular bioavailability

Cellular uptake and productive cytosolic delivery are distinct. After receptor- or nanoparticle-mediated endocytosis, most internalized siRNA remains within endosomal compartments and is recycled, retained, or directed toward lysosomal degradation. Imaging studies of LNP delivery indicate that only a small fraction reaches the cytosol [83]. However, this fraction varies by carrier, cell type, dose, and measurement method; estimates such as “less than 1%” should not be generalized across GalNAc conjugates and LNPs.

Ionizable lipids become more protonated in the acidic endosomal environment and interact with anionic membrane lipids, promoting membrane destabilization and cytosolic release. The corresponding escape mechanisms of GalNAc-siRNA conjugates are less completely defined. For both platforms, total cellular uptake can substantially overestimate pharmacologically active intracellular exposure [59,83].

Increasing membrane-disruptive activity may improve escape but can also damage cellular membranes, disturb lysosomal function, activate inflammatory pathways, or cause nonspecific tissue injury. Productive delivery therefore requires controlled release rather than maximal membrane disruption.

Delivery studies should distinguish total liver accumulation and target-cell uptake from cytosolic or RISC-loaded guide-strand exposure and functional target knockdown. The latter measurements provide stronger evidence of pharmacologically productive delivery. Fluorescent labeling can be misleading if the label dissociates from the siRNA or remains trapped in endosomes. Cell-resolved measurements of guide-strand loading, target mRNA reduction, and downstream protein suppression are therefore preferable.

In vitro transfection confirms that a transcript is susceptible to siRNA-mediated suppression but does not validate a delivery platform because it bypasses the vascular, tissue, and intracellular barriers encountered in vivo [76].

5.4. Disease-related anatomical barriers

The hepatic microenvironment changes during MASH progression. In healthy liver, fenestrated LSECs facilitate exchange between sinusoidal blood and the space of Disse. Chronic metabolic injury promotes LSEC defenestration, basement membrane formation, sinusoidal capillarization, and extracellular-matrix deposition. These changes increase tissue stiffness and restrict access to hepatocytes and HSCs.

Experimental restoration of sinusoidal fenestrae improved subsequent delivery of an antifibrotic nanoassembly, supporting vascular accessibility as a determinant of treatment exposure [84]. Delivery established in healthy animals therefore cannot be assumed to persist in advanced MASH.

Particle size, surface charge, lipid composition, shape, and protein adsorption influence passage through the hepatic microvasculature. Smaller particles may improve tissue penetration but can alter cargo capacity, stability, clearance, and cellular uptake. Cationic surfaces may promote membrane interaction while increasing protein adsorption, macrophage sequestration, complement activation, and toxicity. Surface shielding can prolong circulation but impair cellular entry or endosomal escape [76,77,82].

No single size threshold, including “less than 70 nm,” can be considered universally optimal. The appropriate physicochemical profile depends on the carrier, target cell, administration route, fibrosis stage, and required intracellular exposure. Delivery should therefore be evaluated across relevant disease stages, including advanced fibrosis, rather than inferred from healthy-liver biodistribution alone.

5.5. Delivery-related safety and immunogenicity

Delivery-related toxicity is platform-specific. LNP-associated risks include formulation-dependent inflammation, complement activation, hepatic accumulation, and infusion-related reactions [77,85]. PEGylated lipids may interact with pre-existing or treatment-induced anti-PEG antibodies, potentially altering pharmacokinetics or increasing hypersensitivity during repeated administration [86].

GalNAc conjugates avoid several liabilities associated with intravenously administered nanoparticles but remain predominantly hepatocyte-directed and do not solve delivery to nonparenchymal cells [59]. Platform selection should balance target-cell accessibility, administration route, formulation complexity, and carrier-specific tolerability.

6. Translational evidence for siRNA therapeutics in MASH

Human evidence for siRNA therapeutics in MASH remains limited and is concentrated in genetically supported hepatocyte targets, particularly PNPLA3 and HSD17B13. Most inflammatory, fibrogenic, and cell-death targets remain preclinical. Published target engagement, biomarker changes, liver-fat reduction, trial registration, and histological benefit represent different levels of evidence and should not be treated as equivalent.

6.1. Clinically evaluated programs

ARO-PNPLA3, also known as JNJ-75220795, provides the most developed genotype-defined human evidence among PNPLA3-directed programs. In the completed Phase I study, the 400-mg dose reduced liver fat by up to 46% from baseline in PNPLA3-I148 M homozygotes, corresponding to an approximately 39% placebo-corrected reduction. No comparable response was observed in heterozygous participants [55,67]. The published analysis reported no serious or severe adverse events or treatment discontinuations. These findings support genotype-associated liver-fat reduction but do not establish direct hepatic PNPLA3 knockdown, MASH resolution, fibrosis improvement, or clinical benefit.

Other PNPLA3-directed candidates are supported primarily by trial registration. An initial Phase I study of ALN-PNP was completed in nine participants without posted results (NCT06024408). A broader Phase I/IIa study remains recruiting (NCT05648214), as does a Phase IIa study evaluating ALN-PNP with or without tirzepatide in homozygous PNPLA3-related MASLD (NCT07527910). The Phase I study of AMG 609 has been completed, whereas the Phase I study of LY3849891 is active but not recruiting; no results have been posted for either study (NCT04857606 and NCT05395481). These registrations establish clinical development activity, not target engagement or efficacy.

HSD17B13-directed programs have produced evidence of target suppression but not histological efficacy. Gatuzosiran, formerly ARO-HSD and now designated GSK4532990, reduced hepatic HSD17B13 mRNA and protein, with hepatic mRNA reduction reaching 93.4% and mean ALT reduction reaching 42.3% at 200 mg [15]. These findings establish target silencing and biochemical activity but not MASH resolution or fibrosis improvement. The Phase IIb HORIZON trial is registered under the historical NASH terminology and is active but no longer recruiting; no results have been posted (NCT05583344).

Rapirosiran, also known as ALN-HSD, produced dose-dependent HSD17B13 suppression, with a median hepatic mRNA reduction of 78% at 6 months in the 400-mg group [68]. The Phase I study was not designed or powered to establish histological efficacy, and the available biopsy data were insufficient to demonstrate MASH resolution or fibrosis improvement. Its protocol was terminated for business reasons and changes in development strategy, with the registry reporting that no safety issue had been identified (NCT04565717) [69]. This termination should not be classified as an efficacy or safety failure. A separate genotype-enriched Phase II study in participants with MASH and F2–F3 fibrosis remains recruiting (NCT05519475).

BMS-986263 is a retinoid-conjugated LNP containing an siRNA against HSP47 and was designed to target activated hepatic stellate cells. The Phase II study reported by Lawitz et al. enrolled patients with advanced hepatic fibrosis after sustained virological response to hepatitis C treatment and therefore does not establish efficacy in MASH [71]. A separate Phase II study in MASH-related compensated cirrhosis was terminated for lack of efficacy during its short-term acute phase (NCT04267393) [37]. The registry does not determine whether this outcome resulted from inadequate HSC delivery, insufficient target suppression, treatment duration, advanced disease, endpoint sensitivity, or failure of HSP47 inhibition. The mechanistic cause of non-efficacy is therefore unknown.

MTARC1, also termed MARC1 in sponsor documents, is being evaluated through NNC0581-0001/NN6581 [70]. A first-in-human, single-dose Phase I study in healthy participants and participants with hepatic steatosis has been completed without posted results (NCT05599945). A multiple-dose Phase I study in participants with hepatic steatosis and suspected steatohepatitis is recruiting (NCT06891365). Publicly available information does not establish MTARC1 knockdown or effects on liver fat, histology, or clinical outcomes.

6.2. Selected preclinical targets by pathogenic domain

Preclinical candidates were included only when direct therapeutic siRNA experiments were reported. Representative preclinical siRNA candidates and the evidence supporting them are summarized in Table 4. Genetic associations, knockout models, ASOs, viral vectors, miRNAs, and nonspecific pharmacological inhibition were not classified as therapeutic siRNA evidence. These studies establish experimental feasibility but not human target engagement, histological efficacy, or repeat-dose safety.

Table 4.

Selected preclinical siRNA targets relevant to MASH.

Target/intended cell	Model and siRNA approach	Preclinical interpretation
HMGB1/hepatic macrophages [24]	Mannose-modified LNP in murine MASH	Reduced inflammatory and steatotic features; repeat-dose macrophage targeting remains unvalidated
TAZ/hepatocytes [33]	GalNAc–siTAZ in hepatocyte-humanized MASH mice	Reduced liver injury and fibrosis; limited by short duration and a single hepatocyte donor
Jagged1/hepatocytes [26]	GalNAc–siJag1 in diet-induced murine MASH	Reduced Notch signaling, HSC activation, and fibrosis; long-term Notch-related safety is unknown
TGF-β1/non-cell-selective [34]	TGF-β1 siRNA in a high-fat diet plus CCl4 rat model	Reduced collagen expression and fibrosis; non-MASH model and pleiotropic pathway effects limit translation
IL-11 or IL-11RA1/activated HSCs [35]	HSC-targeted nanoparticles in two murine MASH models	Reduced steatosis, inflammation, and fibrosis; human HSC delivery and repeat-dose safety remain unproven
Inducible GPX4 isoform/hepatocytes [72]	In vitro siRNA supported by dietary genetic mouse models	Reduced ferroptotic injury; therapeutic siRNA delivery and efficacy were not demonstrated in vivo
Fas/non-cell-selective [73]	Fas siRNA in a western diet plus cigarette-smoke mouse model	Reduced apoptosis, inflammation, and liver injury; unconventional model and no human validation
Rubicon/hepatocytes [74]	Hepatocyte-targeted siRNA nanoliposome in Western diet-fed mice	Enhanced autophagy and reduced steatosis; repeat-dose safety and human efficacy unknown

6.3. Distinguishing target validation from therapeutic validation

Target validation and therapeutic validation are distinct. Human genetic associations, altered expression in diseased liver, and experimental perturbation may support target validation but do not establish that a therapeutic siRNA will benefit patients. Therapeutic validation requires a defined product to reach the intended cell population, achieve quantitative suppression at a tolerable exposure, and improve histological or clinical outcomes [59].

Hepatic mRNA reduction, ALT decline, MRI-PDFF reduction, and whole-liver accumulation are separate forms of intermediate evidence. None alone establishes cell-specific delivery, MASH resolution, fibrosis improvement, or long-term safety. Evidence should progress from human or causal target support through cell-specific delivery, target engagement, biological response, histological benefit, clinical outcomes, and long-term safety.

6.4. Why some programs advance while others fail

PNPLA3 and HSD17B13 are supported by human genetic evidence, are expressed in hepatocytes accessible to GalNAc-mediated delivery, and permit quantitative pharmacodynamic assessment. PNPLA3 inhibition has produced genotype-associated liver-fat reduction, while HSD17B13 programs have achieved hepatic target suppression and ALT reduction. Neither target has yet demonstrated histological benefit. The business-related termination of the rapirosiran Phase I protocol does not constitute evidence against HSD17B13 biology.

The HSP47 program presents a different problem. Activated HSCs are more difficult to reach than hepatocytes, and the terminated study enrolled participants with compensated cirrhosis, in whom fibrosis may be less reversible. The registry establishes lack of short-term efficacy but does not identify the responsible biological or technical mechanism. Inadequate delivery, insufficient knockdown, treatment duration, disease stage, endpoint sensitivity, and failure of the target itself remain possibilities rather than confirmed explanations.

Clinical advancement therefore requires alignment among causal target evidence, target-cell accessibility, measurable target engagement, disease stage, and downstream benefit. A negative trial cannot be attributed to target biology without separating these factors.

7. Translational barriers and development challenges

Clinical translation of siRNA therapeutics for MASH is limited by differences between experimental models and human disease, the safety implications of sustained gene silencing, and the practical demands of manufacturing and clinical evaluation.

7.1. Limited predictive value of preclinical models

No single experimental model reproduces the metabolic dysfunction, hepatocellular injury, inflammation, fibrosis, cardiovascular risk, and clinical heterogeneity of human MASH. Diet-induced models can reproduce obesity and insulin resistance but often develop fibrosis slowly or inconsistently. Nutrient-deficient diets and models incorporating carbon tetrachloride accelerate inflammation and fibrosis but do not fully reflect the metabolic context or natural history of human disease. Genetically engineered models provide mechanistic information but may overemphasize individual pathways [87].

The timing of treatment also affects translation. Many preclinical studies administer siRNA before or shortly after disease induction, whereas clinical trials typically enroll patients with established F2–F3 fibrosis or cirrhosis. An intervention that prevents early steatosis or inflammatory signaling may not reverse established extracellular matrix remodeling. Species differences in target expression, immune responses, liver-cell composition, and nanoparticle disposition may further alter pharmacological activity and delivery efficiency.

Preclinical studies should therefore use complementary models selected according to the biological target and intended disease stage. Whenever feasible, treatment should begin after the relevant phenotype has developed. Evaluation should include target engagement, cell-specific biodistribution, histological response, recovery after treatment withdrawal, and systemic metabolic effects. Human primary cells, multicellular liver models, organoids, and precision-cut liver slices may help identify species-specific effects, although none reproduces the complete systemic environment of MASH [87].

7.2. Long-term safety and homeostatic compensation

MASH may require treatment for years, making the safety of sustained transcript suppression more relevant than short-term tolerability alone. Candidate targets may contribute to normal lipid handling, stress responses, immune surveillance, extracellular matrix turnover, or tissue repair. Prolonged or excessive silencing could therefore disrupt physiological homeostasis even when short-term treatment improves disease-associated biomarkers [59,88].

Human genetic evidence, including naturally occurring loss-of-function variants, can strengthen target selection but does not eliminate this risk. Lifelong partial reduction caused by a germline variant is not equivalent to pharmacological silencing initiated in adulthood. The magnitude, timing, and cellular distribution of suppression may also differ. Dose-ranging studies should determine whether partial target reduction is sufficient and whether a therapeutic window exists between disease modification and disruption of normal function.

Safety assessment must also reflect the intended population. Patients with MASH frequently have obesity, type 2 diabetes, cardiovascular or kidney disease, and polypharmacy. These factors may influence oligonucleotide disposition, hepatic uptake, compensatory responses, and susceptibility to adverse effects. Long-term studies should assess metabolic adaptation, liver and kidney function, immune effects, drug–drug interactions, and the consequences of treatment interruption [88].

7.3. Off-target effects and immunogenicity

Sequence-dependent off-target effects should be distinguished from the intended consequences of target suppression. Partial complementarity between the guide-strand seed region and unintended transcripts can produce miRNA-like repression and has contributed to hepatotoxicity during preclinical siRNA selection. Candidate optimization should combine sequence-based prediction, transcriptome-wide profiling, testing in human-relevant cells, and chemical strategies that reduce seed-mediated repression without compromising on-target potency [59,61].

Synthetic siRNAs may also activate innate immune receptors in a sequence- and chemistry-dependent manner [62]. Appropriate nucleoside modification and rigorous purification can reduce immune recognition, but clinical assessment remains necessary [60]. Monitoring should be tailored to the sequence, chemical design, and expected tissue exposure and should include hepatic and renal function, inflammatory or immune abnormalities, the effects of organ impairment, and interactions with concomitant medications [88]. Long-term observation is required because short-term tolerability cannot establish the safety of repeated gene suppression in patients with metabolic comorbidities.

7.4. Durability, repeat dosing, and reversibility

Plasma pharmacokinetics do not directly indicate the duration of the pharmacodynamic response. Circulating siRNA concentrations may decline rapidly, whereas intracellular retention and RISC-mediated activity can sustain target suppression for weeks or months [59,88]. This durability may reduce dosing frequency but also limits the ability to reverse biological activity promptly if adverse effects occur.

Dosing intervals established for approved siRNA products in other diseases cannot be directly extrapolated to MASH [59]. The duration of silencing depends on the sequence, chemical design, dose, target turnover, delivery platform, and target cell. Claims that MASH-directed siRNAs will permit biannual or annual administration are therefore not yet supported.

Clinical studies should characterize the onset, magnitude, duration, and recovery of target suppression after each dose. Repeat-dose evaluation should assess cumulative hepatic and extrahepatic toxicity, potential changes in delivery efficiency as fibrosis progresses, and whether target expression recovers predictably after treatment cessation [59,88]. Multi-year repeat-dose safety data in patients with MASH remain limited.

7.5. Manufacturing, cost, and scalability

Large-scale siRNA production requires control of oligonucleotide identity, purity, duplex composition, conjugation efficiency, residual impurities, potency, and stability. GalNAc conjugates have a relatively defined composition. Nanoparticle systems introduce additional variables, including particle size, encapsulation efficiency, lipid composition, surface properties, and release behavior. These attributes can affect biodistribution and biological activity and must remain consistent across manufacturing lots [56,59,89].

Multiplex formulations create further challenges because the stability, potency, and exposure of each component must be controlled. Specialized analytical methods, storage requirements, parenteral administration, and manufacturing complexity may also increase treatment costs. Because MASLD and MASH affect large populations, scalability and cost-effectiveness will be more consequential than in the rare-disease indications in which several siRNA products were initially developed.

7.6. Regulatory and clinical trial challenges

Clinical development in MASH is complicated by slow disease progression and reliance on histological surrogate endpoints. Histological assessment is based on established features of steatohepatitis and fibrosis [5,90]. FDA draft guidance issued under the historical NASH terminology identifies two endpoints for trials supporting accelerated approval: resolution of steatohepatitis without worsening of fibrosis and improvement in fibrosis without worsening of steatohepatitis [90].

Liver biopsy is invasive and subject to sampling error and reader variability. These limitations can affect eligibility, fibrosis-stage classification, estimates of treatment effect, and statistical power [91]. Non-invasive biomarkers and imaging may support patient selection, dose evaluation, and pharmacodynamic assessment, but changes in liver fat, aminotransferases, or circulating fibrosis markers do not independently establish histological or clinical benefit. Trials should therefore distinguish target engagement from therapeutic efficacy, standardize biopsy assessment, specify the handling of missing or unevaluable samples, and determine whether histological responses are associated with durable clinical benefit.

8. Clinical positioning and future directions

8.1. Patient selection and precision therapy

Patient selection for siRNA trials should be guided by target biology rather than by a diagnosis of MASH alone. PNPLA3 has substantial human genetic and mechanistic support, while naturally occurring loss-of-function variants in HSD17B13 are associated with protection from chronic liver disease [14,16]. Genotype alone, however, is insufficient because not every variant carrier develops active steatohepatitis or progressive fibrosis, and the targeted pathway may not remain a dominant disease driver at every stage. A proposed framework for the precision development of siRNA therapeutics in MASH is presented in Fig. 3.

Fig. 3.

Proposed Framework for Precision Development of siRNA Therapeutics in MASH.

Created with BioRender.com (https://BioRender.com/lq8u7tg).

Created with BioRender.com (https://BioRender.com/uxoon7a).

Eligibility and stratification should integrate genotype with histological activity, fibrosis stage, target expression where measurable, target-cell accessibility, and a pharmacodynamic marker of target engagement. Metabolic, inflammatory, and fibrotic phenotypes may further identify patients in whom a liver-specific driver remains therapeutically relevant.

Disease stage is both a biological and pharmacological variable. Strategies targeting lipid accumulation or early hepatocellular injury may be more effective before advanced fibrosis develops. In established cirrhosis, limited tissue reversibility and altered hepatic architecture may reduce both biological response and delivery efficiency. Precision treatment should therefore account for mechanism, disease stage, and delivery feasibility rather than relying on genotype alone.

8.2. Combination strategies and demonstration of incremental benefit

Combination development should pair therapies with complementary mechanisms rather than assume that combining active agents will produce synergy. Resmetirom acts through liver-directed thyroid hormone receptor-β signaling, whereas semaglutide targets systemic metabolic dysfunction. An siRNA could provide additional benefit by suppressing a genetically supported hepatic target that remains active despite these treatments [18,[43], [44], [45]]. The recruiting Phase IIa study of ALN-PNP with or without tirzepatide (NCT07527910) provides an early test of this approach, but no results are available.

Combination trials should generally follow adequate characterization of the siRNA dose, target engagement, safety, and preliminary monotherapy activity. A randomized add-on design comparing optimized background therapy plus siRNA with the same background therapy plus placebo would isolate the incremental contribution of RNA silencing.

Evaluation should extend beyond target engagement to incremental effects on MASH resolution, fibrosis improvement, cardiometabolic outcomes, treatment burden, and overlapping toxicities. A biomarker change alone would not justify combination use without clinically meaningful benefit and an acceptable benefit–risk profile.

8.3. Multiplex and multi-target RNA silencing

The biological heterogeneity of MASH does not in itself justify simultaneous silencing of multiple targets. Multiplex approaches require evidence that more than one independently validated pathway must be inhibited and that the same effect cannot be achieved with a simpler regimen.

Possible strategies include co-administration of separate siRNA products, co-formulation of multiple siRNAs in one delivery system, multifunctional RNA constructs, and combination of a single siRNA with a small molecule or peptide-based therapy. These approaches differ in pharmacokinetics, dose optimization, manufacturing, and regulatory requirements.

Additional components may increase the off-target burden, complicate attribution of adverse events, and obscure the contribution of individual targets. Differences in intracellular potency and duration may also prevent a fixed formulation from maintaining an appropriate degree of suppression for every target. Multiplex development therefore requires evidence that each component contributes to efficacy, that the dose ratio is justified, and that combined toxicity remains acceptable [56,58,59].

8.4. Evidence needed for clinical translation

Clinical translation requires more than hepatic mRNA reduction. Quantitative target engagement in the intended liver-cell population must be followed by reproducible histological benefit and, ultimately, improvement in clinical outcomes. Changes in liver fat, aminotransferases, or circulating biomarkers may support pharmacological activity but should not be interpreted as substitutes for histological or clinical benefit [90].

Because MASH is chronic, efficacy must be accompanied by evidence of repeat-dose safety, reversibility, and an acceptable therapeutic window. Comparative and add-on trials should determine whether siRNA provides meaningful benefit beyond approved therapies, while manufacturing scalability, cost, and accessibility will influence its eventual clinical viability.

9. Conclusions

siRNA enables suppression of intracellular and genetically supported targets that may be difficult to modulate with conventional drugs. Its clearest potential in MASH lies in targets with strong causal evidence, a defined target cell, feasible delivery, a measurable marker of target engagement, and an adequate therapeutic window. However, clinical evidence remains limited. Experience with the HSP47 program illustrates that delivery and mechanistic plausibility do not necessarily translate into histological efficacy.

The clinical value of siRNA will depend on whether target engagement produces durable histological and clinical benefit without unacceptable long-term toxicity. Its role may therefore lie in genotype- or mechanism-directed treatment of selected patients, either alone or in rational combinations with established liver-directed or systemic metabolic therapies, rather than as a universal replacement for existing treatment.

CRediT authorship contribution statement

Yanan Guo: Writing – original draft, Resources, Methodology, Investigation. Qunyan Yao: Writing – review & editing, Supervision.

Consent for publication

Not applicable.

Ethics approval and consent to participate

Not applicable.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation and revision of this work, the authors used Google Gemini and OpenAI ChatGPT to assist with English-language refinement and the organization of responses to peer-review comments. After using these tools, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Funding source

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

No data was used for the research described in the article.