In 2023, the term metabolic dysfunction-associated steatotic liver disease (MASLD) replaced nonalcoholic fatty liver (NAFL) disease to emphasize the central role of metabolic dysfunction in disease pathogenesis and to encompass a broader spectrum of liver diseases [1]. Similarly, nonalcoholic steatohepatitis was renamed metabolic dysfunction-associated steatohepatitis (MASH) to highlight the influence of metabolic factors [2]. This updated nomenclature provides a precise, non-stigmatizing framework that emphasizes the link between fatty liver diseases and metabolic risk factors such as obesity, diabetes, and insulin resistance. The revised framework centers metabolic dysfunction as the primary driver of disease progression, influencing both clinical diagnostic criteria and research priorities. MASLD reflects the heterogeneity of metabolic dysfunction, with genetic variants (e.g., PNPLA3, TM6SF2, HSD17B13) and variations in insulin resistance and lipid dysregulation contributing to disease progression [3]. Recent research has underscored a strong causal relationship between MASLD and metabolic syndrome, type 2 diabetes, and insulin resistance, which act not only as risk factors but also as key drivers of progression. These comorbidities increase the risk of cardiovascular disease, chronic kidney disease, and various extrahepatic malignancies, including colorectal, breast, and pancreatic cancers, thereby positioning MASLD as a systemic metabolic disorder [4]. MASLD is a highly heterogeneous condition, with differences in clinical presentation, progression, and therapeutic responses influenced by genetic factors (e.g., PNPLA3, TM6SF2, HSD17B13) as well as environmental and lifestyle factors, necessitating personalized management strategies [5].

Despite extensive research efforts, no definitive pharmacological therapy exists for MASLD or its progressive form, MASH, underscoring the need for novel therapeutic approaches. Previous studies have shown that differentially expressed non-coding RNAs (ncRNAs) play crucial roles in MASLD pathogenesis by regulating hepatic gluconeogenesis, lipogenesis, insulin resistance, oxidative stress, metabolic inflammation, tissue regeneration, and fibrogenesis [6]. Recognized as potential biomarkers, these ncRNAs offer promising opportunities for diagnosis and disease staging, and emerging strategies such as RNA interference and overexpression techniques show promise, although further investigation is needed [7].

This systematic review consolidates all currently available data on human MASLD, focusing on identifying potential biomarkers through the analysis of differential expression patterns of long non-coding RNAs (lncRNAs) and circular RNAs (circRNAs) across the MASLD disease spectrum. Additionally, it evaluates the diagnostic accuracy of these ncRNAs in differentiating healthy individuals from those with nonalcoholic fatty liver NAFL, MASH, or fibrosis. By elucidating the roles and mechanisms of lncRNAs, circRNAs, and microRNA (miRNAs), this study systematically assesses their therapeutic potential in MASLD/MASH.

Since this study is a systematic review and does not involve human or human-origin materials, neither institutional review board approval nor informed consent was required.

This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement available at https://www.prisma-statement.org/.

Studies were selected based on predefined inclusion and exclusion criteria. Eligible studies included those employing cell or animal models of MASLD/MASH, analyzing RNA expression profiles in MASLD/MASH patients, conducting controlled experiments with clearly defined methodologies, or utilizing liver tissue, serum, plasma, or blood samples. Furthermore, only studies published between 2010 and 2023 were included to ensure that recent advancements in ncRNA research were considered. Excluded studies comprised duplicates, systematic reviews, meta-analyses, case reports, comments, letters, and editorials, as well as non-English publications and studies lacking relevant outcomes or appropriate negative controls. Additionally, studies lacking proper controls, having small sample sizes (<10), or not including statistical validation (P<0.05) were excluded to ensure methodological rigor and reliable conclusions.

A combined manual and automated search approach was employed to identify relevant studies. We systematically searched multiple databases, including PubMed, Cochrane Library, Scopus, Embase, and Web of Science, using comprehensive search terms from 2010 to 2023.

The search terms used in PubMed, Cochrane Library, Scopus, Embase, and Web of Science are presented in Suppl. 1.

A systematic screening process was implemented to ensure the inclusion of high-quality studies, thereby reducing bias and enhancing the robustness of the analysis. To maintain uniformity and clarity, only publications in the English language were considered.

Data extraction was independently performed by 3 investigators (Q.A.N., C.X., and M.A.K), who screened the titles and abstracts of the selected studies. Discrepancies were resolved through discussion, and when disagreements persisted, a fourth investigator (D.Y.) mediated the resolution to maintain objectivity.

Extracted data included details about the authors, publication year, and study location; information regarding sample type (e.g., liver tissue, serum) and sample size; the RNA variants studied; dosage and mode of administration; expression patterns (upregulation or downregulation); and study outcomes, such as fat accumulation, inflammation, and fold-change data when available. Any inconsistencies or missing data were addressed through consensus among the investigators, ensuring a comprehensive and accurate synthesis of the therapeutic potential of ncRNAs in MASLD/MASH.

A comprehensive database search identified 2,180 unique titles and abstracts. The study selection process adhered to PRISMA guidelines for transparency and reproducibility, as illustrated by the PRISMA flowchart (Figure 1), which details the identification, screening, eligibility, and inclusion stages.

After removing duplicates (n=352) and excluding 919 articles that did not meet the inclusion criteria, 315 articles remained for full-text review. Of these, 181 were excluded for reasons such as being review articles (n=8), lacking a MASLD or MASH model (n=16), not investigating miRNA-based therapies (n=78), or failing to use quantitative real-time polymerase chain reaction (qRT-PCR) for RNA validation (n=8). Additionally, 35 studies involved non-human subjects, 21 examined multiple miRNAs, 8 were irrelevant to the objectives, and 7 lacked therapeutic effect assessments or outcome reporting. Consequently, 134 articles were included in the systematic review, comprising 96 studies on miRNAs, 38 studies on 44 distinct lncRNAs, and 11 studies on circRNAs. The included studies displayed significant methodological diversity: 38 (28.6%) used cell-based models, 51 (38.3%) employed animal models, and 44 (33.1%) incorporated both. Most studies validated ncRNA expression using qRT-PCR, RNA sequencing, or microarray analysis, while a subset explored the functional roles of ncRNAs in lipid metabolism, inflammation, and fibrosis progression, highlighting their critical involvement in MASLD/MASH pathogenesis. Publication trends indicate an increasing recognition of ncRNAs as potential biomarkers and therapeutic targets for MASLD/MASH.

However, notable gaps remain. The majority of studies relied on preclinical models, with only about 26% utilizing human-derived samples (liver tissue, serum, or plasma). Moreover, the limited sample sizes in human studies may restrict the generalizability of these findings. These limitations underscore the need for large-scale, multicenter clinical studies to validate the clinical utility of ncRNAs in diagnosing and treating MASLD/MASH.

The included studies were evaluated for relevance, methodological rigor, and validity, ensuring that appropriate models (cell, animal, or human samples) and validated techniques (e.g., qRT-PCR) were employed. This rigorous assessment enhanced the review's reliability. Data synthesis involved thematic analysis and, where applicable, statistical pooling using RevMan 5.4. Studies were categorized by ncRNA type (miRNA, lncRNA, circRNA) and their roles in lipid metabolism, inflammation, and fibrosis, highlighting consistent findings. Statistical pooling provided a quantitative overview of ncRNA expression patterns and their association with MASLD/MASH progression.

A review of 38 studies on lncRNAs in MASLD identified 44 distinct lncRNAs exhibiting differential expression patterns across various sample sources (liver tissue, serum/plasma, peripheral blood mononuclear cells [PBMCs]) and employing diverse methodologies (qRT-PCR, RNA-seq, microarray) (Suppl. 2). These lncRNAs regulate key hepatic processes including lipid and glucose metabolism, oxidative stress, and immune regulation and play pivotal roles in MASLD pathogenesis and progression [8–14]. Many lncRNAs, such as uc.372, uc.333, CCAT1, lnc-SPARCL1-1:2, NEAT1, PVT1, and MALAT1, influence MASLD progression through miRNAs sponging. Most studies focused on liver tissue with small sample sizes, while a few examined serum/plasma and PBMCs, emphasizing the need for a comprehensive approach that links liver and serum/plasma levels to enhance diagnostic and therapeutic potential.

Several lncRNAs emerged as promising biomarkers. Alshehri et al. [14] found elevated lncARSR levels in both serum and liver samples from MASLD patients, suggesting its potential as a novel biomarker. Additionally, 2 studies reported increased NEAT1 levels in MASLD patient serum and PBMCs, with NEAT1 in PBMCs showing strong diagnostic accuracy (area under the receiver operating characteristic curve [AUC]), 0.822; sensitivity, 86.47%; specificity, 82.03%) [8,14]. Serum levels of lncPRYP4-3 and RP11-128N14.5 were significantly elevated in MASLD compared to healthy controls, consistently across NAFL and MASH subtypes [15,16]. In contrast, lncPRYP4-4 was upregulated in NAFL, but not in MASH [15]. Guo et al. [13] and Lee et al. [17] demonstrated that serum PVT1 (n=112) effectively distinguished MASLD patients (AUC, 0.895; sensitivity, 84.0%, specificity; 84.6%), while HCG18 showed high diagnostic accuracy (AUC, 0.934; sensitivity, 82.8%; specificity, 89.1%) [18].

The lncRNA MEG3 exhibited inconsistent expression patterns in MASLD studies. Some studies reported increased expression in MASLD and MASH [17,19], while others found decreased expression correlated with MASLD severity [20]. These findings suggest a compensatory role for MEG3 in MASLD pathogenesis. MEG3 is also involved in regulating hepatic endothelial senescence and insulin resistance in obesity [17].

MASH, a severe form of MASLD characterized by hepatic necroinflammation and rapid fibrosis progression, requires early detection to manage complications. Seven studies identified differentially expressed lncRNAs in MASH serum or liver samples (Suppl. 3), highlighting their potential as biomarkers for MASH diagnosis and prognosis. Park et al. found that plasma LeXis was independently associated with MASH, with an AUC of 0.743, sensitivity of 54.3%, and specificity of 100% [21]. Similarly, lnc-SPARCL1-1:2, which is upregulated in MASLD, simple steatosis, and MASH, predicted MASH with robust diagnostic capabilities (AUC, 0.870 for MASH vs. healthy controls; AUC, 0.790 for MASH vs. simple steatosis, AUC; 0.974 for MASH vs. other MASLD) [22]. These studies are limited by small cohorts and a lack of external validation, underscoring the need for larger studies. Additional lncRNAs, such as LncPRYP4-3 and RP11-128N14.5, were upregulated in MASH serum with fold-changes of 4.7 and 2.33, respectively, distinguishing MASH from early-stage MASLD or healthy controls [23]. In contrast, LncPRYP4-4 showed no significant difference between MASH and healthy controls, suggesting its relevance to earlier MASLD stages [24]. MEG3, involved in hepatic endothelial senescence and insulin resistance, exhibited inconsistent expression patterns in MASH: some studies reported upregulation (fold-change, 1.9) [25], while others observed downregulation (fold-changes, 0.48 [26] and 0.48), indicating a context-dependent role in MASH pathogenesis.

Comparing lncRNAs between MASH and earlier MASLD stages is crucial for improving diagnostics and therapies. Suppl. 3 (MASH vs. MASLD) highlights lncRNAs and circRNAs that distinguish these stages. LeXis displayed differential expression between MASH and MASLD, supporting its role as a stage-specific biomarker [21]. Similarly, MALAT1 and lnc-SPARCL1-1:2 showed distinct expression profiles in MASH, aligning with their roles in lipogenesis, fibrosis, and extracellular matrix remodeling [22,27]. In contrast, MEG3 and LncPRYP4-3 exhibited opposite expression patterns, suggesting complex regulatory functions [19,26]. These findings highlight lncRNAs as potential biomarkers for MASH pathogenesis and diagnosis. Larger, multicenter studies are needed for validation. Future research should explore lncRNA mechanisms in MASH, particularly in inflammation and fibrosis pathways such as those involving lnc-SPARCL1-1:2 to improve diagnostics and personalize treatment strategies.

We extracted data on lncRNA expression in MASLD and MASH, including sample sources, fold-change values, statistical significance, and associated biological pathways. Figure 2A highlights the central role of liver tissue in MASLD pathogenesis, with serum-derived lncRNAs representing potential non-invasive biomarkers. The limited data on PBMC-derived lncRNAs underscore the need for further exploration. LncRNAs were categorized according to their roles in lipid metabolism, inflammation, fibrosis, and oxidative stress. For instance, lncRNAs such as HOTAIR and MALAT1 regulate lipid metabolism by interacting with key transcription factors involved in lipid biosynthesis and catabolism, while also modulating extracellular matrix components that influence fibrosis progression in MASLD. These molecular interactions may serve as promising therapeutic targets to modulate disease progression and mitigate fibrosis. Lipid metabolism emerged as the predominant category, reflecting its key role in MASLD progression. Inflammation and fibrosis are also significant, underscoring their involvement in the transition to MASH, while oxidative stress, though less frequently observed, remains an integral component of disease pathology (Figure 2B). LncRNAs such as NEAT1 and lncARSR show robust upregulation in larger cohorts, indicating their reliability as diagnostic markers or therapeutic targets. Downregulated lncRNAs with modest fold-changes may represent protective mechanisms (Figure 2C). Liver-derived lncRNAs (e.g., MEG3, MALAT1) are significantly associated with lipid metabolism and fibrosis pathways, while serum-derived lncRNAs, such as lncHCG18, reinforce their potential as disease indicators (Figure 2D). Most differentially expressed lncRNAs are upregulated, suggesting they drive metabolic dysregulation and fibrotic remodeling, whereas downregulated lncRNAs may function as inhibitory elements, guiding future therapeutic strategies (Figure 2E).

CircRNAs, which are ncRNAs with closed-loop structures, play a pivotal role in the pathogenesis of MASLD and MASH. This review identified 2 circRNAs that were differentially expressed in MASLD and 9 in MASH (Suppl. 4). In MASLD, circRNA_0046367 and circRNA_0001805 were significantly downregulated. Specifically, circRNA_0046367 (fold-change, 0.4; P<0.01) is associated with the circRNA_0046367/miR-34a/PPARα axis, thereby influencing lipid metabolism, while circRNA_0001805 (fold-change, 0.63; P<0.05) modulates the miR-106a-5p/miR-320a pathways that regulate ABCA1 and CPT1, which are critical in lipid and energy metabolism [12,28]. In MASH, all 9 circRNAs exhibited downregulation. Notably, mitochondrial circRNA SCAR showed the most significant decrease (fold-change, 0.16; P<0.001). This ncRNA plays a crucial role in regulating mitochondrial ROS production and mitigating oxidative stress; its reduced expression has been linked to impaired mitochondrial function, increased oxidative stress, and inflammatory signaling that contribute to MASH pathogenesis [29]. Other mitochondrial circRNAs, including hsa_circ_0089763, hsa_circ_0008882, and hsa_circ_0032777, showed fold-change values ranging from 0.28 to 0.74. As illustrated in Figure 3A, most circRNAs originate from liver or mitochondrial samples, underscoring their involvement in hepatic and mitochondrial pathways. Figure 3B categorizes circRNAs by functional roles, with mitochondrial function being the predominant category, followed by lipid metabolism and oxidative stress. Figure 3C demonstrates a correlation between fold-change and sample size, confirming consistent downregulation. Finally, Figure 3D and 3E display statistically significant associations and fold-change patterns, reinforcing the potential of circRNAs as biomarkers for differentiating MASLD and MASH from healthy states and for tracking disease progression.

MiRNAs are crucial regulators of gene expression in MASLD and MASH, affecting disease onset, progression, and severity. Multiple studies have reported altered miRNA expression patterns that highlight their diagnostic, prognostic, and therapeutic potential (Suppl. 5). In MASLD, miR-130a-3p was downregulated in serum (fold-change, ≤1.3; P<0.05) and influenced TGFBR1 and TGFBR2 through the transforming growth factor (TGF)-β/SMAD pathway [30]. Conversely, miR-130 was upregulated in liver tissue (fold-change, ≤2; P<0.05), thereby activating the IGFBP2/AKT pathway [22]. These bidirectional expression patterns suggest that miRNA regulation is context-dependent. As shown in Figure 4A, the distribution of miRNAs across sample sources reveals diverse expression patterns in serum and liver tissue. Upregulated miRNAs, such as miR-34a, consistently appear across liver, serum, and HepG2 models, where they modulate PPARα and SIRT1, both key regulators of lipid metabolism. Additionally, Let-7b-5p was increased in serum (fold-change, ≥2; P<0.05) and influenced TGF-β signaling [12,13]. In contrast, downregulated miRNAs such as miR-26a (P<0.05) in liver tissue have been linked to pro-inflammatory interleukin (IL)-6–IL-17 signaling [12]. The functional roles of these miRNAs are further illustrated in Figure 4B, where their involvement in lipid metabolism, inflammation, and fibrosis is categorized, emphasizing their role in MASLD pathogenesis. Reduced expression of specific miRNAs may exacerbate inflammatory responses and metabolic disturbances.

In MASH, miR-223-3p was downregulated in both serum and liver (P<0.05), leading to the suppression of E2F1, a transcription factor involved in cell proliferation and fibrosis. This suggests that miR-223-3p may have a protective, anti-fibrotic role in disease progression [31]. In contrast, inflammatory miRNAs such as miR-155 were consistently upregulated in liver and blood samples (P<0.05), correlating with enhanced inflammatory signaling and metabolic dysregulation [32]. Figure 4C illustrates the fold-change distribution, reflecting the variability observed among miRNAs and sample types. Altered miRNA profiles thus influence metabolic pathways central to MASLD/MASH progression. For example, miR-451, which is downregulated in serum (P<0.05), reduces Akt signaling and promotes autophagy, a process important for cellular homeostasis and lipid turnover [33]. Conversely, miR-21, upregulated in MASLD liver tissue, is linked to WNT/β-catenin and other oncogenic pathways (e.g., the HBP1-p53-Srebp1c axis), underscoring its role in dyslipidemia, steatosis, and potential tumorigenic transformation in MASLD [34,35]. The statistical significance of these findings is demonstrated in Figure 4D, which features a heatmap of p-values across sample sources, underscoring the robustness of the observed trends. Across ncRNA classes, lncRNAs were predominantly upregulated in MASLD and MASH, whereas circRNAs were consistently downregulated. In MASLD, most lncRNAs (e.g., lncHCG18, lncPVT1, lncARSR, NEAT1, MALAT1, MEG3) were upregulated relative to healthy controls [36,37]. This pattern is further supported by Figure 4E, which compares fold-changes by expression direction, emphasizing the loss of protective regulatory mechanisms in MASLD.

In MASH, lncRNA expression generally increased, whereas only one lncRNA (lnc-SPARCL1-1:2) was downregulated [22]. This trend suggests that lncRNAs may continue to influence disease progression from MASLD to MASH. CircRNAs in MASH were also downregulated, possibly due to mitochondrial dysfunction or impaired regulation in advanced stages. MiRNA expression changes were variable. Several miRNAs linked to lipid accumulation, inflammation, and insulin resistance (e.g., miR-34a, miR-9, miR-21, miR-122, miR-103) were upregulated in both MASLD and MASH, while others (e.g., miR-130a-3p, miR-223-3p, miR-146 family) were downregulated, indicating a loss of protective regulatory mechanisms.

A detailed analysis of miRNA expression in MASH (Suppl. 6) reveals significant dysregulation, with miRNAs playing critical roles in oxidative stress, inflammation, lipid metabolism, and fibrogenesis. Upregulated miRNAs, such as miR-34a and miR-155, promote disease progression by driving oxidative damage, inflammation, and fibrosis; miR-34a increases ROS production and lipid peroxidation, while miR-155 activates inflammasomes and fibrotic remodeling. Additionally, miR-9 disrupts gluconeogenesis and lipid homeostasis by targeting SIRT1 and Onecut2. Conversely, downregulated miRNAs such as miR-223 and miR-451 appear to serve protective roles by suppressing inflammation and promoting autophagy, respectively, which are crucial for maintaining metabolic balance. The loss of these protective miRNAs may further facilitate disease progression, suggesting that restoring their expression could offer a therapeutic strategy to mitigate inflammation and metabolic imbalance in MASH. The interplay between upregulated and downregulated miRNAs in MASH underscores their complex roles in molecular pathways. For example, while miR-21 upregulation drives lipogenesis and fibrosis via WNT/β-catenin signaling, miR-144 downregulation enhances TLR2-mediated inflammation. These findings reinforce the role of miRNAs as key drivers of MASH pathology and potential therapeutic targets.

As shown in Figure 5A, most miRNA studies utilize liver samples, underscoring the liver's central role in disease progression. Serum-derived miRNAs also show promise as non-invasive biomarkers, while miRNAs from cell-based models such as HepG2 cells and primary hepatocytes reflect diverse experimental approaches. Figure 5B categorizes miRNAs by their mechanistic pathways, highlighting their impact on lipid metabolism, oxidative stress, inflammation, and fibrosis in MASLD/MASH. For example, miR-34a and miR-21 are associated with dysregulated lipid metabolism and fibrosis, whereas miR-223 is involved in anti-inflammatory and anti-fibrotic processes. Figure 5C links fold-change with sample size, emphasizing the robustness of the miRNA findings. Highly upregulated miRNAs like miR-34a and miR-7, supported by larger datasets, further reinforce their significance as biomarkers and therapeutic targets. Figure 5D presents a heatmap of p-values that demonstrates the statistical reliability of the associations, with miRNAs such as miR-223 and miR-34a consistently significant in both liver and serum, indicating their systemic and hepatic relevance. Finally, Figure 5E compares miRNAs by expression direction, with upregulated miRNAs (e.g., miR-34a, miR-155) promoting lipid accumulation, oxidative stress, and inflammation, while downregulated miRNAs (e.g., miR-223, miR-451) are linked to protective mechanisms such as anti-inflammation and autophagy-mediated lipid clearance.

A systematic review of 134 high-quality studies revealed differential expression patterns of ncRNAs across various sample sources and their functional implications in MASLD/MASH. This comprehensive analysis highlights the potential of ncRNAs as diagnostic biomarkers and therapeutic targets, offering novel avenues for clinical intervention.

Despite the promising potential of ncRNAs as biomarkers and therapeutic targets in MASLD and MASH, several challenges remain. Variability in ncRNA expression profiles across studies affected by differences in sample sources (e.g., liver tissue versus serum), detection methods (e.g., RNA-seq versus qRT-PCR), and normalization strategies hinders reproducibility and comparability. Most studies rely on preclinical models, and human-based research is limited by small sample sizes and low diversity, which restricts the generalizability of the findings. Furthermore, mechanistic studies often focus on isolated ncRNA-mRNA interactions, neglecting broader networks such as competing endogenous RNA crosstalk and epigenetic regulation. Clinical translation is further impeded by issues related to ncRNA stability, tissue-specific delivery, and a lack of in vivo validation of therapeutic efficacy. Additionally, the absence of longitudinal studies linking ncRNA dynamics to histopathological progression or imaging biomarkers limits their prognostic value. Addressing these challenges will require large-scale, multicenter studies, standardization of methodologies, and integrative multi-omics approaches to facilitate the clinical application of ncRNAs in MASLD/MASH.

The findings of this review elucidate the complex roles of lncRNAs, circRNAs, and miRNAs in MASLD and MASH, highlighting their potential as diagnostic biomarkers and therapeutic targets. The differential expression patterns and functional associations of these ncRNAs underscore their involvement in key metabolic, inflammatory, and fibrotic pathways. Addressing current limitations through comprehensive, integrative studies will pave the way for the clinical application of ncRNA-based diagnostics and therapeutics, ultimately enhancing the management and treatment of MASLD and MASH. While ncRNA-based diagnostics show considerable promise, their clinical implementation requires further validation, standardization, and regulatory approval.