Introduction
With the ongoing transformation of modern lifestyles and dietary habits, metabolic dysfunction-associated fatty liver disease (MAFLD),1,2 also widely referred to as nonalcoholic fatty liver disease (NAFLD),3 has become the most prevalent chronic liver disorder worldwide. It is now recognized as a major contributor to progressive liver injury, including cirrhosis and hepatocellular carcinoma, and affects the health of hundreds of millions of individuals globally. MAFLD affects approximately 30% of the global population, with rising prevalence in China and worldwide.1,2 This growing burden underscores the urgent need to identify novel immunometabolic drivers of disease progression.
MDSCs are a heterogeneous population of immunosuppressive myeloid cells that represent precursors of dendritic cells (DCs), macrophages (Mφ), and/or granulocytes. These cells have emerged as important regulators in a wide range of pathological conditions, including cancer, chronic infection, autoimmune disorders, and chronic metabolic disease.4 By inhibiting both innate and adaptive immune responses, MDSCs are increasingly recognized as pivotal modulators of inflammatory microenvironments and tissue remodeling.
The identification of MDSCs has provided a new perspective on immune regulation during liver disease progression. However, their precise role in MAFLD remains incompletely understood. This review therefore aims to summarize the current understanding of the biological functions and mechanistic roles of MDSCs in MAFLD, with particular emphasis on how MDSC-mediated immune regulation contributes to disease initiation and progression. In addition, we discuss the potential of MDSCs as therapeutic targets and highlight how interventions directed at MDSC-dependent immunoregulatory pathways may offer new opportunities for precision treatment of MAFLD.
Despite accumulating evidence from murine models, several critical gaps remain: (i) whether MDSC subsets identified in mice are phenotypically conserved in human MAFLD; (ii) the lack of validated circulating MDSC biomarkers for disease staging; and (iii) the safety concerns surrounding MDSC-targeted therapies in metabolic diseases. This review critically appraises current evidence, explicitly addresses these translational hurdles, and highlights subset-specific functions that may inform precision therapeutics.
Overview of MAFLD
MAFLD is closely associated with obesity, type 2 diabetes mellitus, hypertension, and related metabolic disorders. Its disease spectrum encompasses metabolic dysfunction-associated fatty liver (MAFL) and metabolic dysfunction-associated steatohepatitis (MASH), as well as progressive hepatic fibrosis and cirrhosis.5,6 The pathogenesis of MAFLD has not yet been fully elucidated, and the currently prevailing framework is the “multiple-hit hypothesis”.7 The initial insult is centered on hepatic lipid metabolic imbalance, characterized by increased lipogenesis, elevated free fatty acids, and insulin resistance, all of which promote hepatic lipid accumulation. Subsequent hits, including oxidative stress, mitochondrial dysfunction, and activation of inflammatory pathways, act synergistically to induce hepatocellular injury, persistent inflammation, and fibrogenesis, ultimately driving progression toward cirrhosis or hepatocellular carcinoma. Genetic susceptibility and environmental influences also play important roles in MAFLD pathogenesis.8 Taken together, MAFLD arises through a multidimensional and multilayered pathogenic process, and clarification of these mechanisms is essential not only for a deeper understanding of disease biology but also for the development of diversified targeted therapeutic strategies.
Immune regulation also plays a pivotal role in MAFLD progression, with multiple immune cell populations contributing to disease progression. T cells promote hepatic stellate cell (HSC) activation through secretion of cytokines such as interleukin-4 (IL-4), interleukin-13 (IL-13), and interleukin-17 (IL-17), thereby enhancing liver fibrogenesis, while simultaneously contributing to hepatocellular injury, metabolic dysregulation, and insulin resistance.9,10 Mφ trigger inflammatory responses and liver injury through the release of proinflammatory mediators. Lipotoxic stimulation restricts the self-renewal of Kupffer cells (KCs), while infiltrating bone marrow-derived monocytes differentiate into proinflammatory monocyte-derived macrophages, together exacerbating hepatic injury and fibrosis.11,12 DCs infiltrate MASH livers, and their abundance is positively correlated with disease severity. By driving proinflammatory T-cell phenotypic reprogramming, DCs mediate an inflammation–fibrosis cascade that aggravates hepatic pathological injury.13 MDSCs, as a heterogeneous population with potent immunosuppressive properties, also participate in MAFLD progression by regulating lipid metabolic disturbances, inflammatory responses, and fibrotic remodeling. Emerging evidence suggests bidirectional crosstalk between MDSCs and KCs: MDSC-derived interleukin-10 (IL-10) and transforming growth factor-β (TGF-β) may suppress KC self-renewal,14 while activated KCs produce CCL2 that can recruit myeloid cells including MDSCs,15 potentially establishing a feed-forward loop that perpetuates immune dysregulation.
Biological characteristics of MDSCs
Definition and classification of MDSCs
MDSCs are a highly heterogeneous population composed of myeloid progenitors and immature myeloid cells (IMCs), and they are characterized by potent immunosuppressive activity. Under physiological conditions, IMCs generated in the bone marrow rapidly differentiate into mature granulocytes, Mφ, or DCs, and therefore MDSCs are present only at low levels in healthy individuals. However, under pathological conditions such as cancer, trauma, infectious diseases, and certain autoimmune or chronic inflammatory disorders, the differentiation of IMCs into mature myeloid cells is blocked, resulting in the abnormal expansion of the MDSC compartment.4 According to their developmental and phenotypic characteristics, MDSCs are broadly classified into two major subsets: granulocytic MDSCs (G-MDSCs) and monocytic MDSCs (M-MDSCs). It should be noted, however, that the phenotypic definition of these populations differs between humans and mice.4,16,17 This species-specific difference should be taken into account when interpreting experimental findings and translating them into clinical settings.
Mechanisms of MDSC-mediated immunosuppression
In chronic inflammatory diseases, MDSCs exert profound immunosuppressive effects on both innate and adaptive immune responses, thereby contributing to disease progression. Their suppressive activity is primarily mediated through high expression of arginase-1 (ARG1), inducible nitric oxide synthase (iNOS), and reactive oxygen species (ROS), all of which impair T-cell function (Fig. 1).4 Among these subsets, G-MDSCs preferentially rely on ROS and ARG1 to mediate immunosuppression and do so largely independently of iNOS, whereas M-MDSCs mainly depend on nitric oxide (NO) as well as immunosuppressive cytokines such as IL-10 and TGF-β (Fig. 1).4,18
MDSCs suppress immune responses through multiple coordinated mechanisms. T helper 2 (Th2)-associated cytokines activate the Janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway, thereby upregulating ARG1 expression and causing depletion of L-arginine, which in turn inhibits expression of the CD3ζ chain and blocks synthesis of cyclin D3 and cyclin-dependent kinase 4, ultimately impairing T-cell antigen recognition and proliferation.18,19 Upon stimulation with T helper 1 (Th1) cytokines, iNOS competes with ARG1 for the common substrate L-arginine and catalyzes the production of NO. NO disrupts T-cell proliferation and differentiation by nitrating the T-cell receptor (TCR), blocking interleukin-2 (IL-2) receptor signaling, and inhibiting the Janus kinase 3–signal transducer and activator of transcription 5 (JAK3-STAT5) pathway.4,14 ROS, in turn, directly suppress T-cell function by damaging the CD3ζ chain via H2O2 and by mediating ONOO−-dependent nitration of the TCR; notably, inhibition of ROS production in MDSCs can almost completely reverse their immunosuppressive activity.4,18,20 In addition, MDSCs secrete TGF-β, which downregulates NKG2D receptor and perforin expression on natural killer (NK) cells, suppresses interferon-γ (IFN-γ) production, and interferes with the antigen-presenting capacity of DCs through oxidized lipids.21,22 Together, these synergistic pathways constitute the highly effective immunosuppressive network orchestrated by MDSCs.
Key roles of MDSCs in MAFLD progression
MDSCs regulate immune responses through the secretion of cytokines such as IL-10 and TGF-β, as well as immunomodulatory mediators including ROS, ARG1, and prostaglandin E2 (PGE2). Through dynamic interactions with other immune cell populations, MDSCs promote hepatic lipid accumulation and lipid metabolic dysregulation, while simultaneously sustaining chronic inflammatory responses, accelerating fibrogenesis, and ultimately driving the progression of MAFLD.
MDSCs promote hepatic lipid accumulation and lipid metabolic disturbance
MDSCs contribute to hepatic steatosis through multiple cooperative mechanisms that integrate immune regulation with metabolic remodeling. First, MDSCs secrete immunosuppressive cytokines, including IL-10 and TGF-β. IL-10 negatively regulates M1 polarization of Mφ, reduces inflammatory cell infiltration within the liver, and thereby attenuates immune responses triggered by lipid accumulation.14,23 In addition, IL-10 enhances the expansion and suppressive function of regulatory T cells (Tregs), further dampening effector T-cell activity and reducing immune-mediated clearance of hepatic lipid deposits. TGF-β directly promotes hepatic lipid storage by upregulating key lipogenic enzymes, including fatty acid synthase and acetyl-CoA carboxylase. In a choline-deficient amino acid-defined (CDAA) diet-induced MASH model, hepatocyte-specific deletion of Tgfbr2 (Tgfbr2ΔHEP) significantly reduced serum alanine aminotransferase (ALT) and triglyceride levels, lowered the MAFLD activity score, and attenuated steatosis, inflammation, and hepatocellular ballooning, indicating that hepatocyte-intrinsic TGF-β signaling promotes liver injury, steatosis, and inflammation in CDAA-fed mice and thereby promotes disease progression. Moreover, TGF-β binds to receptors on T cells and suppresses their proliferation and cytotoxic activity. Second, MDSCs alter the hepatic metabolic state by modulating lipid metabolism-related signaling pathways. Compared with mice fed a normal diet, high-fat diet (HFD)-fed mice show a significant increase in hepatic MDSC abundance, accompanied by marked activation of the STAT3 signaling pathway and elevated levels of phosphorylated STAT3.24 Conversely, CCR5-deficient G-MDSCs display reduced STAT3 activity in the hepatic microenvironment, which suppresses the expression of lipogenic genes such as acetyl-CoA carboxylase 1 and stearoyl-CoA desaturase 1, thereby reducing de novo lipogenesis.24 In a bone-marrow chimera model with G-MDSC-specific CCR5 deficiency (n = 6 per group), hepatic triglyceride content, hepatocyte phospho-STAT3 levels, and Acetyl-CoA carboxylase 1 (Acc1)/Stearoyl-CoA desaturase 1 (Scd1) mRNA expression were all significantly reduced compared with controls (all P < 0.05); conversely, global CCR5 knockout aggravated steatosis, with increased liver weight/body weight ratio, serum aspartate aminotransferase (AST) and alkaline phosphatase (ALP), and hepatic triglycerides (all P < 0.05, n = 6). These findings suggest that MDSCs influence hepatic lipid accumulation and metabolism, at least in part, through regulation of STAT3-dependent metabolic programs. Finally, hepatic lipid accumulation itself can promote the expansion and activation of MDSCs. Lipotoxic molecules such as ceramides released in steatotic livers stimulate MDSC proliferation and functional activation. Indeed, MDSC activation has been reported to be closely associated with hepatic fatty acid accumulation, arachidonic acid metabolic dysregulation, and ceramide overload, all of which may contribute to MAFLD progression.25,26 Consistent with this, HFD-fed mice exhibit enhanced differentiation of myeloid progenitors into MDSCs together with stronger immunosuppressive activity compared with mice fed a normal diet.27
MDSCs promote persistent hepatic inflammation
MDSCs also play a major immunoregulatory role in hepatic inflammation. They secrete immunosuppressive cytokines,28 including IL-10, TGF-β, and IL-6, which suppress local immune responses and thereby facilitate the persistence of chronic inflammation. As previously discussed, MDSC-derived IL-10 and TGF-β establish an immunosuppressive microenvironment that perpetuates inflammation. In the inflammatory context, these cytokines additionally suppress DC activation and weaken immune-mediated clearance of inflamed hepatic tissue. In parallel, TGF-β, together with MDSC-derived ROS, ARG1, and PGE2, suppresses T-cell activity, while PD-L1/PD-L2-PD-1 interactions further drive T-cell dysfunction.15 In a polyunsaturated fatty acid diet-induced mouse model, MDSC-derived IL-6 promoted hepatic macrophage activation and proinflammatory mediator release through activation of the JAK-STAT3 pathway, thereby contributing to sustained and chronic hepatic inflammation. IL-6 acts synergistically with ROS and ARG1 produced by MDSCs to establish an immunosuppressive microenvironment that perpetuates chronic liver inflammation. In HFD-induced obese mice, factors associated with MDSCs are markedly altered in hepatic tissue, with significant upregulation of both transcript and protein levels of monocyte chemoattractant protein-1 (MCP-1), S100A8, and S100A9.29 Among these, MCP-1 may mediate the migration of MDSCs from the peripheral circulation into the liver, thereby facilitating the formation of a chronic inflammatory niche. Quantitatively, circulating CD11b+Gr-1+ MDSCs in HFD-fed mice reached 22.20% ± 2.99% of leukocytes versus 7.00% ± 0.48% in lean controls (an approximately threefold increase, P < 0.05), while hepatic S100A8 and S100A9 transcripts were elevated 2.5- and 2.3-fold (both P < 0.01) and their protein levels 3.1- and 1.6-fold (both P < 0.001), respectively.
Consistent with these experimental observations, MDSC abundance has been reported to increase in livers affected by chronic liver disease, where these cells accumulate in inflammatory regions and correlate positively with the severity of hepatic inflammation and the stage of fibrosis. In a cohort of 98 patients with chronic liver disease, hepatic CD33+/HLA-DR− MDSC abundance correlated positively with both histological inflammation grade (Spearman r = 0.610, P < 0.001) and fibrosis stage (r = 0.687, P < 0.001), and was significantly elevated in hepatitis, cirrhosis, and hepatocellular carcinoma tissues versus normal liver (all P < 0.001). Although these findings suggest a potential link in human liver disease, they warrant validation in MAFLD-specific cohorts. In methionine-choline-deficient (MCD) diet-induced mouse models, MDSCs are mainly localized within inflammatory cell clusters and produce abundant ROS, directly participating in the pathological progression of hepatic inflammation.15 Taken together, these findings indicate that MDSCs regulate immune responses through multiple interconnected mechanisms, establish an immunosuppressive microenvironment, prevent efficient clearance of hepatic inflammation, and may further aggravate liver injury, thereby sustaining and promoting chronic inflammatory progression in MAFLD. Critically, the bidirectional crosstalk between MDSCs and KCs represents a key axis in this process: MDSCs suppress KC self-renewal through TGF-β-dependent mechanisms, while lipotoxic hepatocytes release ceramides that activate MDSCs, creating a self-perpetuating cycle of immune dysregulation.
MDSCs promote the progression of liver fibrosis
During the progression of MAFLD, expansion of MDSCs is closely associated with hepatic fibrogenesis. By exerting potent immunosuppressive effects, MDSCs may weaken hepatic repair mechanisms and thereby accelerate the transition from MAFLD to MASH, cirrhosis, and even hepatocellular carcinoma.30 The profibrotic role of MDSCs is primarily reflected in their ability to promote activation of hepatic stellate cells (HSCs) through the secretion of immunosuppressive and regulatory cytokines, thereby accelerating fibrotic remodeling. In carbon tetrachloride (CCl4)-induced murine liver injury models, MDSCs recruited to the injured liver exhibit increased expression of multiple cytokines, including IL-10, VEGF, and TGF-β.14,23 HSCs are non-parenchymal cells located in the space of Disse and normally remain in a quiescent state; following liver injury, however, they proliferate and become activated. MDSCs can directly induce HSC activation through secretion of TGF-β, driving their transdifferentiation into myofibroblasts, the key effector cells of liver fibrosis that synthesize and secrete collagen. In parallel, IL-10 and TGF-β derived from MDSCs may act synergistically to further enhance HSC activation and collagen production. In both high-fat-high-carbohydrate (HFHC) diet-induced and CCl4-induced mouse models of liver fibrosis, MDSCs upregulate PPP1R15A through TGF-β signaling, which in turn promotes expression of ARG1 and S100A8/9 while suppressing T-cell proliferation.23 In addition, MDSCs generate ROS, thereby inducing oxidative stress, whereas S100A8/9 proteins can further promote inflammation and fibrogenesis through ROS-dependent pathways.
Notably, MDSCs may also exert a negative feedback loop after hepatic injury by suppressing excessive inflammation and thereby partially attenuating fibrotic progression.31,32 Infusion of bone marrow cells containing MDSCs into patients or mice with liver fibrosis has been shown to inhibit fibrogenesis. In particular, the CD11b+Gr-1+/high F4/80− subpopulation promotes increased IL-10 expression by activated HSCs, reduces hepatic collagen production, and consequently alleviates liver fibrosis. Quantitatively, in CCl4-induced fibrotic mice (8–12 per group), infusion of CD11b+Gr1+ bone-marrow cells significantly reduced α-SMA and COL1A1 mRNA in isolated HSCs within 24 h (α-SMA: P < 0.05; COL1A1: P < 0.01), expanded hepatic regulatory T cells, and reduced F4/80+ macrophage infiltration (both P < 0.05); these anti-fibrotic effects were abolished when IL-10-deficient bone-marrow cells were used, establishing IL-10 dependence.33 Similarly, accumulation of monocytic MDSCs (M-MDSCs) in the livers of bile duct-ligated mice can suppress fibrosis progression.34 However, this effect is not uniformly beneficial. The immunosuppressive activity of MDSCs can also reshape the hepatic immune microenvironment, rendering hepatocytes more susceptible to injury and potentially accelerating fibrotic development.17 Thus, the role of MDSCs in liver fibrosis appears to be highly context-dependent, with both protective and pathogenic consequences. These seemingly contradictory findings reflect the context-dependent functional plasticity of MDSCs, which is systematically analyzed in the following section.
Context-dependent dual roles of MDSCs in MAFLD
While the preceding sections emphasize the pathogenic contributions of MDSCs to steatosis, inflammation, and fibrosis, accumulating evidence indicates that their functions are not unidirectionally deleterious. Rather, MDSCs exhibit context-dependent plasticity, with both protective and pathogenic consequences shaped by subset composition, disease stage, and microenvironmental cues (Fig. 1). A critical appraisal of this dualistic nature is essential for designing rational therapeutic strategies that preserve beneficial immune regulation while targeting pathogenic MDSC activity. The multifaceted and context-dependent contributions of MDSCs to MAFLD progression are summarized in Table 1.14,15,23,24,29,32,33
Table 1Mechanism-centered roles of MDSCs in MAFLD progression
| Process | Core Mechanism | Key Molecules | Major Consequence | Quantitative Evidence | Evidence Level* |
|---|
| Hepatic steatosis | Suppression of immune clearance; promotion of lipogenesis | IL-10, TGF-β, STAT3, ACC1, SCD1 | Lipid accumulation | G-MDSC-specific CCR5 deficiency (bone-marrow chimera, n = 6/group): hepatic triglycerides, hepatocyte phospho-STAT3, and Acc1/Scd1 mRNA all significantly reduced vs controls (all P < 0.05)24; conversely, global CCR5-KO on HFD increased liver weight/body weight ratio, serum AST and ALP, and hepatic triglycerides (all P < 0.05, n = 6)24 | Mouse: 14,23,24; Human: limited |
| Chronic inflammation | Establishment of immunosuppressive niche | IL-6, IL-10, TGF-β, ARG1, ROS, PD-L1/PD-1 | Persistent inflammation | HFD mice: circulating CD11b+Gr-1+ MDSCs 22.20% ± 2.99% vs 7.00% ± 0.48% in lean controls (∼3-fold, P < 0.05); hepatic S100A8/S100A9 mRNA increased 2.5- and 2.3-fold (both P < 0.01).29 Human cohort (n = 98): hepatic CD33+/HLA-DR− MDSC density correlated with inflammation grade (Spearman r = 0.610, P < 0.001)32 | Mouse: 15,29; Human: 32 (correlation) |
| Fibrogenesis | Activation of HSCs; promotion of collagen synthesis | TGF-β, IL-10, VEGF, S100A8/9, ROS | ECM deposition, fibrosis | Human cohort (n = 98): hepatic MDSC density correlated with fibrosis stage (Spearman r = 0.687, P < 0.001) and was elevated in hepatitis, cirrhosis and HCC vs normal liver (all P < 0.001).32 Mouse CCl4 model (n = 8–12/group): CD11b+Gr1+ BMC infusion reduced HSC α-SMA and COL1A1 mRNA within 24 h (α-SMA: P < 0.05; COL1A1: P < 0.01) in an IL-10-dependent manner, supporting the opposing anti-fibrotic arm33 | Mouse: 14,23,33; Human: 32 |
Subset-specific functional plasticity
Granulocytic MDSCs (G-MDSCs) predominantly exert proinflammatory and profibrotic effects through ROS and ARG1, whereas monocytic MDSCs (M-MDSCs) may adopt an anti-fibrotic phenotype under certain conditions. In bile duct-ligated mice, hepatic M-MDSC accumulation suppresses fibrosis progression,34 whereas in CCl4-induced injury, CD11b+Gr-1+/high cells promote IL-10-mediated HSC quiescence.33 Conversely, G-MDSCs appear consistently associated with STAT3-driven lipogenesis and chronic inflammation.24 This subset dichotomy suggests that therapeutic strategies must distinguish between G-MDSC and M-MDSC modulation to avoid inadvertently abrogating protective immune regulation. Furthermore, the differential reliance on immunosuppressive mechanisms—G-MDSCs primarily utilizing ROS/ARG1 and M-MDSCs employing NO, IL-10, and TGF-β—implies that targeted interventions may yield distinct functional outcomes depending on which subset predominates in a given disease context.
Stage-dependent functional shifts
During the simple steatosis stage of MAFLD, MDSC-mediated immunosuppression may attenuate lipotoxicity-driven inflammation by limiting effector T-cell responses and promoting regulatory T-cell expansion.14 However, as disease progresses to metabolic dysfunction-associated steatohepatitis (MASH) and liver cirrhosis, sustained MDSC accumulation fosters an immunosuppressive niche that prevents efficient resolution of inflammation, promotes HSC activation, and potentially facilitates immune evasion during hepatocellular transformation.30 Thus, the net effect of MDSCs transitions from transient tissue protection in early injury to chronic pathogenic remodeling in advanced disease. This stage-dependent paradigm has important therapeutic implications: interventions aimed at MDSC depletion in early-stage MAFLD might inadvertently remove transiently protective immune regulation, whereas in advanced fibrosis, selective suppression of pathogenic G-MDSCs while preserving M-MDSC-mediated repair mechanisms may represent a more nuanced approach.
Microenvironmental determinants
The functional polarity of MDSCs is further dictated by local microenvironmental cues. Hypoxia-inducible factor-1α (HIF-1α) stabilization in steatotic livers promotes MDSC recruitment and enhances their suppressive capacity through upregulation of glycolytic metabolism and PD-L1 expression. TGF-β concentration gradients appear critical: low-level TGF-β sustains MDSC suppressive programming through SMAD-dependent transcriptional activation, whereas high concentrations in fibrotic foci drive myofibroblast differentiation and collagen deposition via non-canonical pathways.23 Additionally, ceramide-enriched lipotoxic environments stimulate MDSC expansion but may alter their cytokine profile toward enhanced IL-6 and reduced IL-10 production,25,26 tipping the balance toward pathogenicity. Understanding these contextual switches is essential for designing stage-selective MDSC interventions. Notably, the reciprocal interactions between MDSCs and hepatocytes—wherein lipotoxic hepatocytes activate MDSCs through ceramide release, and MDSCs in turn suppress hepatocyte-autonomous immune surveillance—exemplify the complex feedback loops that govern MDSC function in the MAFLD microenvironment.
Critical appraisal: The concept of MDSC plasticity in MAFLD is predominantly derived from murine models utilizing dietary (MCD, HFD) or toxic (CCl4) injury paradigms. Whether the protective functions observed in these experimental systems translate to human MAFLD remains uncertain. No prospective human studies have directly assessed MDSC subset switching across MAFLD stages, and the limited clinical data (e.g., Ref 32) remain correlative rather than mechanistic. Human MDSCs are phenotypically defined by CD14−/lowCD15+ or HLA-DR−CD33+ markers rather than the murine CD11b+Gr-1+ system, and cross-species functional conservation has not been rigorously validated. Future studies employing humanized models and prospective clinical cohorts are urgently needed to define the therapeutic window within which MDSC modulation might be beneficial versus harmful.
Quantitative interpretation of evidence presented in Table 1. The murine lipogenesis data (G-MDSC-specific CCR5 deficiency, n = 6 per group) demonstrate effect sizes consistent with biological relevance—hepatic triglyceride reduction, attenuated phospho-STAT3, and downregulation of Acc1/Scd1 mRNA (all P < 0.05)—yet the small sample size and absence of confidence intervals preclude robust effect-size estimation. The human correlation data (n = 98) yield r values of 0.610–0.687, reflecting moderate-to-strong monotonic associations between hepatic MDSC density and histological severity. These coefficients should not be interpreted as the proportion of outcome variance explained in a regression framework, as they derive from nonparametric Spearman correlation. Clinically, an r = 0.687 for fibrosis stage suggests that MDSC density cannot reliably discriminate individual patients across fibrosis stages without substantial overlap, reinforcing the need for multiparameter models rather than single-cell metrics.
Evidence-level and clinical-stage assessment of therapeutic strategies are shown in Table 2.24,29,35–38 All five strategies remain at the preclinical or early translational phase in MAFLD. MitoTEMPO and anti-αMβ2/STAT3 inhibition are supported by quantitative mechanistic evidence, yet neither has entered human MAFLD trials. CCR5 antagonism (maraviroc) has the most advanced clinical footprint, but the Hepmarc and MAVMET trials were feasibility/phase II studies in HIV-coinfected populations with negative efficacy signals for hepatic endpoints; this evidence level is insufficient to support off-label use in immunocompetent MAFLD. Leptin-receptor blockade and CXCR4/CXCL12 disruption have promising murine efficacy data but lack dose-response characterization, pharmacokinetic profiling, and toxicity assessment. Collectively, no MDSC-targeted strategy currently meets the threshold for clinical recommendation; the certainty of evidence for all strategies is low to very low per GRADE criteria. All should be regarded as experimental interventions warranting controlled phase I/II mechanistic trials in MAFLD-specific populations.
Table 2MDSC-targeted therapeutic strategies in MAFLD
| Target/Strategy | Mechanism | Effect on MDSCs | Quantitative Evidence | Evidence Level | Clinical Stage |
|---|
| MitoTEMPO | Mitochondrial ROS scavenging | Down MDSC expansion, down proinflammatory mediators | HFD mice (n = 5–8/group): hepatic S100A8 and S100A9 mRNA reduced 5.9- and 5.2-fold (P < 0.001), protein 1.9- and 1.4-fold (P < 0.01), and MCP-1 reduced (P < 0.001).29 Circulating MDSC frequency fell from 22.20% ± 2.99% to 12.52% ± 1.22%, and the decrease was significant (P < 0.05), indicating reduced circulating MDSC frequency29 | Mouse (HFD model) | Preclinical |
| CCR5 inhibition | Blockade of CCR5 signaling | Down G-MDSC anti-inflammatory program, down STAT3 | Mouse, G-MDSC-specific CCR5 deficiency (chimera, n = 6/group): hepatic triglycerides, hepatocyte phospho-STAT3, and Acc1/Scd1 mRNA all reduced (P < 0.05).24 Human: the 96-week Hepmarc feasibility trial (n = 53; add-on maraviroc n = 23, HIV+MAFLD) found no significant between-group differences in LSM, CAP, or ELF change, with 95% CIs containing zero35; the 48-week MAVMET RCT (n = 90) likewise found no MRI-PDFF liver-fat reduction with maraviroc | Mouse24; Human (feasibility RCT)35 | Emerging clinical interest* |
| Leptin receptor blockade | Antagonism of leptin signaling | Down MDSC abundance | HFD-fed mice: soluble leptin receptor (leptin antagonist) significantly reduced MDSC abundance vs controls (P < 0.05; 3 treated mice and 14 controls pooled from two experiments); conversely, anti-Gr-1-mediated MDSC depletion aggravated insulin resistance and adipose inflammation.36,37 Effect size not reported in primary source (absolute counts and percentage change unavailable) | Mouse | Preclinical |
| CXCR4/CXCL12 axis | Disruption of hepatic recruitment | Down MDSC liver infiltration | Mouse MASH model: adoptive transfer of exogenous MDSCs significantly reduced serum ALT and AST vs controls (P < 0.05)40; absolute values and effect size not reported in primary source. Quantitative reduction of MDSC infiltration after pharmacological CXCR4 blockade in MAFLD has not been reported; in a chronic CCl4 model the CXCR4 antagonist AMD3100 aggravated fibrosis with increased intrahepatic neutrophils, underscoring context-dependence | Mouse | Preclinical |
| Anti-alphaMbeta2 / STAT3i | Blockade of CCN1/alphaMbeta2/STAT3 axis | Down MDSC expansion & function | In vitro human system (n = 6): the STAT3 inhibitor Stattic completely abrogated CCN1+GM-CSF-induced MDSC expansion, and αMβ2-neutralizing antibody significantly reduced expansion (Stattic: P < 0.01; αMβ2 antibody: P < 0.05); CCN1-expanded MDSCs suppressed T-cell proliferation via iNOS, reversed by the iNOS inhibitor 1400W (P < 0.05)38 | Human in vitro38 | Preclinical |
Targeted therapeutic strategies
Current preclinical and emerging clinical strategies directed at MDSCs or their upstream signaling pathways are outlined in Table 2. At present, MDSC-directed therapies in MAFLD remain largely preclinical. However, several targets have progressed toward clinical evaluation in related metabolic diseases. Notably, CCR5 antagonists, originally developed for HIV therapy, have entered early-phase trials for metabolic dysfunction-associated steatohepatitis (MASH), although their specific effects on MDSC reprogramming in human liver remain uncharacterized. The following sections summarize preclinical findings and explicitly note translational gaps that must be addressed prior to clinical deployment.
Mitochondrial antioxidant MitoTEMPO
Studies based on high-fat diet (HFD)-induced MAFLD mouse models have shown that the mitochondrial antioxidant MitoTEMPO significantly attenuates hepatic steatosis and inflammatory responses.29 Following MitoTEMPO treatment, the number of MDSCs in both the liver and peripheral blood was markedly reduced, accompanied by a clear decrease in the degree of hepatic steatosis. In addition, MitoTEMPO suppressed the expression of MDSC-associated proinflammatory mediators, including MCP-1, S100A8, and S100A9. Quantitatively (n = 5–8 mice per group), MitoTEMPO reduced hepatic S100A8 and S100A9 mRNA by 5.9- and 5.2-fold (P < 0.001), their protein levels by 1.9- and 1.4-fold (P < 0.01), and MCP-1 mRNA and protein (P < 0.001); notably, the reduction in circulating MDSC frequency (from 22.20% ± 2.99% to 12.52% ± 1.22%) reached statistical significance (P < 0.05), indicating that MitoTEMPO reduced circulating MDSC frequency and its associated proinflammatory mediators. Clinical Translation Challenges: No clinical trials have evaluated MitoTEMPO in MAFLD to date. Its systemic antioxidant effects may disrupt physiological ROS signaling required for normal immune responses, warranting careful dose-escalation studies. Furthermore, the pharmacokinetics of mitochondria-targeted antioxidants in human liver tissue remain poorly characterized, and off-target effects on non-MDSC immune populations have not been assessed.
Targeting the chemokine receptor CCR5
Evidence indicates that the chemokine receptor CCR5 plays an important role in the pathological progression of MAFLD.24 This receptor is expressed not only in hepatocytes but also in G-MDSCs. In the MAFLD microenvironment, CCR5 signaling sustains the immunosuppressive secretory program of G-MDSCs (IL-10, TGF-β, and PD-L1); blockade of CCR5 in G-MDSCs attenuates this program while markedly reducing STAT3 phosphorylation, lipogenic gene expression, and proinflammatory mediator production in adjacent hepatocytes, thereby decreasing hepatic lipid deposition and inflammatory infiltration in mice. In the supporting bone-marrow chimera study (n = 6 per group), G-MDSC-specific CCR5 deficiency significantly reduced hepatic triglyceride content, hepatocyte phospho-STAT3 levels, and Acc1/Scd1 mRNA expression versus controls (all P < 0.05). However, clinical translation remains unproven: in the 96-week Hepmarc randomized feasibility trial (n = 53; 23 receiving add-on maraviroc) in people with HIV and MAFLD, adherence reached 92% and no serious adverse reactions occurred, yet there were no significant between-group differences in liver stiffness, controlled attenuation parameter, or enhanced liver fibrosis score changes, with 95% confidence intervals for ALT and liver stiffness trends containing zero; consistently, the 48-week MAVMET randomized trial (n = 90) found no MRI-PDFF liver-fat reduction with maraviroc versus control.35 These observations suggest that CCR5 inhibition may represent a promising therapeutic strategy for MAFLD, although efficacy in immunocompetent MAFLD patients has yet to be demonstrated.
Repurposing Potential and Species-Specific Barriers: While CCR5 inhibition reduces G-MDSC-mediated STAT3 activation in murine models, human G-MDSCs are defined by CD15+CD33+ rather than Ly6G+ markers, and whether CCR5 blockade produces equivalent phenotypic effects in human MAFLD requires validation. The optimal timing of CCR5-targeted intervention during disease progression also remains undefined. Notably, CCR5 antagonists such as maraviroc have entered early-phase trials for MASH,36 although their specific effects on MDSC reprogramming in human liver remain uncharacterized.
Blockade of the leptin receptor
Obesity-related metabolic inflammation promotes the expansion and activation of MDSCs by upregulating proinflammatory mediators such as IL-6, TNF-α, and PGE2.36 Under obese conditions, MDSCs infiltrate the peripheral circulation, liver, and adipose tissue.37,40 Leptin, an adipose-derived hormone, plays a central role in regulating energy balance, appetite, and multiple metabolic processes. One study37 demonstrated that leptin promotes MDSC accumulation, whereas MDSCs in turn suppress leptin levels through a negative-feedback mechanism. Animal experiments further demonstrated that administration of a soluble leptin receptor, which antagonizes leptin signaling, significantly reduces MDSC abundance in HFD-fed mice (P < 0.05 versus controls); however, the primary study did not report absolute MDSC counts, percentage changes, or per-group sample sizes, precluding effect-size estimation. This incomplete quantitative reporting is explicitly acknowledged as an evidence gap in the Critical Appraisal section. Conversely, depletion of MDSCs by anti-Gr-1 antibodies aggravates insulin resistance and adipose tissue inflammation, indicating that MDSCs play a critical role in obesity-associated metabolic dysregulation.40 Detailed quantitative effect sizes for leptin-receptor blockade in MAFLD await reporting in future studies; the available evidence confirms statistical significance (P < 0.05) without enabling precise magnitude estimation.36
Systemic Safety Constraints and Metabolic Interconnectivity: Leptin resistance in advanced obesity complicates soluble receptor strategies; clinical translation must account for the disconnect between circulating leptin levels and central leptin sensitivity. Moreover, systemic leptin antagonism may disrupt neuroendocrine regulation of energy homeostasis, raising safety concerns that have limited clinical development. Human MDSC responses to leptin blockade have not been evaluated.
Targeting the CXCR4/CXCL12 axis
Studies have demonstrated that administration of exogenous MDSCs significantly lowers serum ALT and AST levels in mouse models of MASH (P < 0.05 versus controls), suggesting a potential effect on biochemical markers of hepatic injury41; however, absolute enzyme values, per-group sample sizes, and effect sizes were not reported in the primary study, precluding quantitative synthesis. The signaling axis formed by CXCR4 and its ligand CXCL12 is a key pathway mediating hepatic recruitment of MDSCs. In the MASH microenvironment, CXCL12 expression is upregulated, attracting MDSCs into the liver and thereby aggravating inflammation and fibrosis. This axis also participates in remodeling of the hepatic immune microenvironment by regulating both the quantity of recruited MDSCs and their immunosuppressive activity. These findings suggest that therapeutic intervention targeting the CXCR4/CXCL12 axis may offer a new direction for MASH treatment. However, quantitative reductions in MDSC hepatic infiltration following pharmacological CXCR4 blockade have not been reported in MAFLD models, and in a chronic CCl4 model the CXCR4 antagonist AMD3100 paradoxically aggravated fibrosis with increased intrahepatic neutrophil accumulation, underscoring the context-dependence of this axis.
Chronic Dosing Paradox and Hematopoietic Trade-offs: CXCR4 antagonists (e.g., plerixafor) are approved for stem cell mobilization but have not been evaluated in MAFLD. Given the essential role of CXCR4 in hematopoietic stem cell homing, long-term blockade may impair bone marrow function, necessitating intermittent rather than continuous dosing strategies. Whether CXCR4/CXCL12 disruption selectively reduces pathogenic MDSC recruitment without compromising protective immune cell trafficking remains unknown.
Anti-αMβ2 antibodies or STAT3 inhibitors
Recent studies have identified cysteine-rich protein 61 (Cyr61/CCN1) as a liver injury-associated molecule that is highly expressed in a liver-specific manner during MAFLD progression and positively correlates with the degree of hepatic MDSC infiltration.38 CCN1 binds to the integrin αMβ2 receptor on the surface of MDSCs, activates a STAT3 phosphorylation cascade, and thereby drives both MDSC expansion and enhancement of their immunosuppressive function. Experimental evidence indicates that blockade of integrin αMβ2 or pharmacological inhibition of the STAT3 pathway can significantly reverse the CCN1-mediated expansion of MDSCs: in an in vitro human system (n = 6), the STAT3 inhibitor Stattic completely abrogated CCN1+GM-CSF-induced MDSC expansion and αMβ2-neutralizing antibody significantly reduced expansion (Stattic: P < 0.01; αMβ2 antibody: P < 0.05), while CCN1-expanded MDSCs suppressed T-cell proliferation via iNOS, an effect reversed by the iNOS inhibitor 1400W (P < 0.05). Notably, this evidence derives from human in vitro experiments rather than murine models, strengthening its translational relevance. These findings confirm that the CCN1/αMβ2/STAT3 signaling axis represents a critical molecular pathway underlying immune microenvironmental dysregulation in MAFLD (Fig. 1).
Druggability Hurdles and Cell-Specific Delivery Imperatives: STAT3 inhibitors face significant safety hurdles given the pleiotropic roles of STAT3 in normal physiology, including liver regeneration and immune surveillance. The therapeutic window for MDSC-specific STAT3 inhibition may be narrow, requiring targeted delivery systems to minimize off-target effects.
Emerging potential of MDSCs as noninvasive biomarkers
Beyond therapeutic targeting, circulating MDSCs and their secreted mediators (e.g., S100A8/A9, arginase activity) may serve as noninvasive biomarkers for MAFLD staging.42 In patients with chronic liver disease, hepatic MDSC abundance correlates with inflammation grade and fibrosis stage (in a cohort of 98 patients, Spearman r = 0.610 and r = 0.687, respectively; both P < 0.001).32 However, standardized immunophenotyping protocols for human MDSCs remain inconsistent, ranging from CD14+/CD15+ gating to Lin-HLA-DR−CD33+ definitions, limiting cross-study comparability. Future research should integrate single-cell transcriptomics and spatial profiling to define subset-specific signatures that complement existing noninvasive scores such as NFS and FIB-4.43,44 The development of MDSC-derived exosomal cargo assays may further enable liquid biopsy approaches for monitoring disease progression and treatment response in MAFLD patients.
Clinical-grade quantitative appraisal of MDSC biomarker potential. The correlation coefficients reported in the human cohort (r = 0.610–0.687) reflect moderate-to-strong monotonic associations; however, their clinical interpretability is constrained by several statistical and methodological factors. First, the sample size (n = 98) provides adequate power (>0.99) to detect correlations of this magnitude at α = 0.05, yet subgroup stratification by MAFLD etiology, BMI, or glycemic status was not performed, precluding assessment of effect modification. Second, the cohort was predominantly HBV-related, introducing substantial spectrum bias that limits generalizability to metabolically defined MAFLD. Third, immunohistochemical CD33+/HLA-DR− phenotyping lacks the precision of multiparameter flow cytometry and was not validated against interobserver agreement metrics (e.g., κ statistic). Fourth, no multivariable models adjusting for age, sex, insulin resistance, or concomitant therapies were reported, leaving residual confounding unquantified. Fifth, and most critically, receiver-operating characteristic (ROC) analysis, optimal cutoff determination, sensitivity/specificity profiles, and likelihood ratios were absent, meaning the diagnostic and prognostic discrimination of hepatic MDSC density remains uncharacterized. From a clinical decision-making perspective, these data provide proof-of-concept support only: they justify prospective biomarker-development studies but do not yet warrant incorporation into MAFLD screening algorithms, biopsy-avoidance strategies, or treatment-response monitoring.
Critical appraisal and limitations
In this section, we critically appraise the strengths and limitations of the current MDSC-MAFLD evidence base, explicitly discuss how these methodological constraints limit the clinical applicability of existing findings, and delineate actionable implications for MAFLD screening, staging, and therapeutic development.
Strengths and limitations of animal models. Murine studies offer decisive experimental strengths: (i) genetically homogeneous backgrounds and controlled dietary interventions (MCD, HFD, HFHC) enable mechanistic isolation of MDSC-driven pathways; (ii) loss-of-function validation via CCR5-deficient bone-marrow chimeras, anti-Gr-1 depletion, and hepatocyte-specific Tgfbr2 knockout supports causal inference that cannot be obtained from human observational data; (iii) access to liver tissue permits spatial and single-cell profiling of MDSC-hepatocyte-stellate cell crosstalk. However, these advantages are offset by critical translational limitations. Murine MDSCs are defined by CD11b+Gr-1+ (Ly6G/Ly6C) markers, whereas human equivalents are identified by CD14−/lowCD15+ or HLA-DR−CD33+ phenotypes; cross-species functional conservation has not been rigorously validated. Dietary induction spans weeks to months, whereas human MAFLD develops over decades with superimposed metabolic syndrome, aging, and polypharmacy. Moreover, murine Kupffer cells constitute ∼20–30% of hepatic leukocytes versus ∼50% in humans, and the gut-liver axis differs markedly in microbial composition and bile-acid metabolism, potentially altering MDSC recruitment kinetics and microenvironmental programming. Consequently, murine data excel at hypothesis generation and preclinical target prioritization but cannot reliably predict human pharmacokinetics, dosing windows, or adverse-event profiles.
Strengths and limitations of human studies. Human evidence provides indispensable ecological validity and direct relevance to clinical phenotypes. The n = 98 cohort32 established that hepatic MDSC density increases across the spectrum of chronic liver disease and correlates with histological severity, offering a biologically plausible rationale for biomarker development. Nevertheless, the limitations are equally consequential. The study was cross-sectional, precluding any inference about temporal precedence or disease trajectory. The cohort was predominantly HBV-related rather than MAFLD-specific, meaning the observed MDSC associations may reflect viral rather than metabolic immunopathology. No multivariable adjustment for age, BMI, diabetes, dyslipidemia, or pharmacotherapy was reported, leaving unmeasured confounding unresolved. Immunohistochemistry-based phenotyping (CD33+/HLA-DR−) lacks standardization against flow-cytometric reference methods, and no interobserver reliability data were provided. Finally, no ROC analysis, diagnostic cutoffs, or calibration metrics were generated, meaning the discriminative performance of MDSC density for MAFLD staging versus noninvasive scores (NFS, FIB-4) remains unknown. These constraints relegate the human data to hypothesis-generating status; they do not currently support clinical deployment as diagnostic, prognostic, or predictive biomarkers.
Study-level appraisal shaping clinical inference. The anti-fibrotic benefit of CD11b+Gr1+ bone-marrow cells was demonstrated in a CCl4 model (8–12 mice per group) with IL-10 dependence formally established,33 but acute toxin-induced fibrosis differs fundamentally from chronic metabolic fibrosis in its inflammatory kinetics, collagen cross-linking, and reversibility, severely limiting extrapolation to MAFLD. For MitoTEMPO, the highly significant suppression of S100A8/S100A9 (P < 0.001) was accompanied by a significant reduction in MDSC frequency (P < 0.05), indicating effects on both MDSC abundance and associated inflammatory mediators.29 Most importantly for clinical practice, the only randomized controlled evidence for an MDSC-relevant target—the Hepmarc (n = 53) and MAVMET (n = 90) trials of the CCR5 antagonist maraviroc—was conducted in HIV-coinfected, virologically suppressed populations, was not powered for efficacy, and did not measure MDSC reprogramming; their negative hepatic-endpoint results therefore cannot be extrapolated to immunocompetent MAFLD patients, and CCR5 blockade should presently be regarded as mechanistically promising but clinically unproven.35 The murine evidence for G-MDSC-specific CCR5 targeting rests on a single bone-marrow chimera experiment with only six mice per group, without reported randomization or blinded outcome assessment, and the divergent phenotype of global CCR5 knockout (exacerbated steatosis and liver injury) further cautions against compartment-agnostic clinical extrapolation.24 Finally, the evidence for leptin-receptor blockade and CXCR4/CXCL12-axis intervention is weakened by incomplete quantitative reporting in the primary studies, which precludes cross-strategy comparison and materially lowers confidence in these two targets.36,41
Implications for clinical practice and trial design. The current evidence base generates five actionable, albeit provisional, implications for clinicians and translational researchers:
Screening and case-finding: Circulating MDSCs should not be adopted as standalone noninvasive screening biomarkers for MAFLD. Until standardized multiparameter flow-cytometric protocols (e.g., Lin−HLA-DR−CD33+ or CD14−/lowCD15+ gating with viability dyes) are validated in prospective, metabolically homogeneous cohorts with appropriately large sample sizes (sufficient for biomarker derivation and external validation), their use in clinical practice is premature.
Disease staging and biopsy decision-making: Hepatic MDSC density (assessed by immunohistochemistry or in situ hybridization) shows moderate-to-strong correlation with inflammation and fibrosis severity, yet it requires invasive tissue sampling. Its additive value to existing noninvasive scores (NFS, FIB-4, ELF, or vibration-controlled transient elastography) has not been tested. Clinicians should continue to rely on established staging algorithms; MDSC-based assessment remains investigational.
Noninvasive surrogate development: Serum S100A8/A9 heterodimers and plasma arginase activity represent promising circulating surrogates for hepatic MDSC burden, but normative ranges, intra-individual biological variation, and confounding by systemic inflammation (e.g., infection, autoimmune flare) remain undefined. These analytes should be incorporated into prospective biomarker panels rather than deployed as single-analyte tests.
Therapeutic target prioritization: MDSC-directed strategies (mitochondrial antioxidants, CCR5 antagonists, leptin-receptor blockade, CXCR4/CXCL12 disruption, STAT3 inhibitors) remain exclusively preclinical in MAFLD. Given the dual protective and pathogenic roles of MDSCs, depletion strategies carry theoretical risks of uncontrolled infection, impaired immune surveillance, and loss of tissue-repair function, particularly in cirrhosis. Any first-in-human trial must include rigorous safety monitoring for infectious complications and hepatic decompensation.
Clinical trial endpoint design: Future interventional trials should prioritize functional MDSC endpoints (arginase activity, ROS generation, T-cell suppression assays) over simple enumeration, given the dissociation between MDSC frequency and suppressive function observed with MitoTEMPO. Moreover, subset-specific endpoints (G-MDSC vs. M-MDSC) should be prespecified, as these populations exert divergent effects on lipogenesis and fibrogenesis.
Future studies must prioritize: (i) humanized liver-on-chip or organoid models that recapitulate human metabolic comorbidities; (ii) prospective MAFLD-specific cohorts with serial biosampling, standardized immunophenotyping, and multivariable statistical modeling; (iii) randomized mechanistic trials that incorporate MDSC functional endpoints alongside conventional histological and elastographic outcomes; and (iv) rigorous safety profiling of MDSC modulation in cirrhotic and immunocompromised populations to advance MDSC biology from bench to bedside.
Conclusions
MDSCs are increasingly recognized as central immune regulators in the pathogenesis of MAFLD, linking metabolic dysfunction, chronic inflammation, and fibrotic remodeling into an integrated disease network. Rather than acting as bystanders, these cells actively shape the hepatic microenvironment through immunosuppressive cytokines, metabolic mediators, and redox-dependent mechanisms. By modulating macrophage polarization, T-cell activity, dendritic-cell function, hepatic stellate-cell activation, and lipid metabolic signaling, MDSCs participate in multiple stages of disease progression, from steatosis and inflammatory persistence to fibrogenesis and, potentially, transition toward advanced liver injury. Importantly, the available evidence suggests that the effects of MDSCs are not unidirectional; under certain conditions, they may attenuate excessive inflammation and transiently limit tissue damage, whereas in other contexts they promote immune evasion, chronic injury, and fibrosis. This functional plasticity highlights the complexity of MDSC biology in MAFLD and indicates that their pathological significance is highly dependent on subset composition, disease stage, and local microenvironmental cues.
From a translational perspective, MDSCs provide a potential framework for mechanism-based intervention in MAFLD. Experimental studies suggest that mitochondrial antioxidants, chemokine receptor targeting, blockade of leptin signaling, modulation of the CXCR4/CXCL12 axis, and inhibition of the CCN1/αMβ2/STAT3 pathway may influence MDSC abundance or function and, in preclinical models, modify disease severity. However, current evidence remains largely preclinical, and several critical issues remain unresolved, including the precise distinction between protective and pathogenic MDSC subsets, the reproducibility of animal-model findings in humans, and the risk that excessive suppression of MDSCs may disrupt immune homeostasis. Future research should therefore integrate immunophenotyping, single-cell and spatial approaches, and clinically relevant validation to define MDSC subset-specific functions with greater precision. Such efforts may, in theory, enable the development of selective therapeutic strategies that not only suppress pathogenic inflammation and fibrosis but also preserve essential immune balance, provided that the molecular distinctions between protective and pathogenic MDSC subsets can be reliably defined and pharmacologically targeted. Ultimately, this would advance precision treatment for MAFLD.