Comparative efficacy of different diet modalities on metabolic profiles and liver functions in metabolic dysfunction-associated fatty liver disease: a network meta-analysis Authors: Jiahui Zou, Fei Tian, Shuqing Cui, Jing Hou, Zhenhua Li, Yaochen Wei, Haozhe Cui, Yunpeng Li, Shuanghao Zhou, Zechao He, Xiangming Ma Published: 2025 ABSTRACT AbstractBackgroundMetabolic dysfunction-associated fatty liver disease (MAFLD) is a global public health challenge that is primarily managed through lifestyle interventions, particularly diet. This network meta-analysis aimed to thoroughly evaluate the effectiveness of four different dietary patterns: Mediterranean diet (MD), intermittent fasting (IF), Dietary Approaches to Stop Hypertension (DASH), and low-fat diet (LF) in improving liver function and metabolic profiles in adults diagnosed with MAFLD.MethodsIn accordance with the PRISMA guidelines, we conducted a search in PubMed, Embase, Web of Science, the Cochrane Library, ClinicalTrials.gov, and WHO-ICTRP for randomized controlled trials (RCTs) up to Oct. 19, 2025, and registered the study on PROSPERO (CRD42024609524). The Cochrane Risk of Bias Tool 2.0 was used to evaluate study quality, and the surface under the cumulative ranking curve (SUCRA) was used to rank the intervention effects.ResultsIF emerged as the most effective intervention for weight loss (WMD = -1.96 kg, 95% CI -3.01,-0.92) and reducing waist circumference (WMD = -1.73 cm, 95% CI -2.89,-0.57). Sensitivity analysis indicated that intensive IF regimens might drive stronger improvements than time-restricted feeding. The MD ranked highest for improving hepatic pathology, showing statistically significant reductions in intrahepatic lipid (IHL) (WMD = -1.68%, 95% CI -3.21, -0.16), and triglycerides (WMD = -22.54 mg/dL, 95% CI -42.79, -2.29).ConclusionsIF and MD demonstrate distinct therapeutic profiles for MAFLD. IF serves as a robust strategy for weight management, particularly with more intensive regimens. Conversely, the MD demonstrates potential as a favorable strategy for targeting hepatic steatosis and lipid metabolism disorders. Therefore, clinicians are advised to adopt a personalized strategy, tailoring dietary prescriptions to the patient's primary metabolic characteristics, pending validation from high-quality, long-term RCTs.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-025-04535-1.Keywords: Metabolic dysfunction-associated fatty liver disease, Diet, RCTs, Network meta-analysis, Liver function Introduction Metabolic dysfunction-associated fatty liver disease (MAFLD) is common worldwide [1]. It is estimated to affect around 38% of the global population, and its prevalence is increasing rapidly [2–4]. It is strongly associated with obesity, dyslipidemia, type 2 diabetes mellitus, and metabolic syndrome [5–7]. MAFLD can develop into metabolic dysfunction-associated steatohepatitis (MASH), cirrhosis, and hepatocellular carcinoma, making it a serious public health threat [8–10]. Currently, there are no approved pharmacological treatments for MAFLD, and lifestyle interventions, particularly dietary interventions, have become the mainstay of disease management [11, 12]. Diet plays a crucial role in regulating hepatic metabolism and energy homeostasis. A balanced diet may reduce hepatic fat accumulation, improve insulin resistance, and alleviate hepatic inflammation [13, 14]. The Mediterranean diet (MD) is a dietary pattern that has shown benefits for liver health [15, 16]. It is characterized by high consumption of fruits, vegetables, whole grains, nuts, and olive oil, and moderate intake of fish and poultry. The high content of antioxidants, monounsaturated fatty acid, and dietary fiber content of MD is thought to reduce hepatic fat content and improve metabolic indices [17]. By comparison, the low-fat diet (LF) restricts the intake of saturated and trans fats, thereby reducing total calorie intake and hepatic lipid synthesis [18, 19]. The LF diet is based on the National Health and Medical Research Council and American Heart Association Dietary Recommendations [20, 21]. Target macronutrient energy contributions for the LF diet were 50% from carbohydrate, 30% from fat (with < 10% of energy as saturated fat), and 20% from protein. Intermittent fasting (IF), which involves alternating fasting and feeding periods, may improve insulin sensitivity and reduce hepatic oxidative stress [22]. The Dietary Approaches to Stop Hypertension (DASH) diet, abundant in fruits, vegetables, low-fat dairy, and whole grains with reduced sodium, has been shown to significantly enhance cardiovascular and metabolic health [23]. Although the aforementioned dietary patterns are supported by evidence, comprehensive analyses of these patterns in the context of MAFLD are lacking. Notably, network meta-analysis offers a distinct advantage: it enables the simultaneous evaluation of the effects of multiple interventions. This allows for a more systematic ranking of different dietary patterns based on their impact on hepatic metabolic indices in patients with MAFLD. Currently, the MD, LF, and DASH Diet are widely recognized for their metabolic benefits; Concurrently, IF has garnered increasing attention as an emerging strategy focusing on eating timing rather than nutrient composition [24–26]. However, existing evidence remains fragmented: most prior studies have restricted their scope to pairwise comparisons of a single diet against a control group or focused primarily on liver enzyme changes, with limited comprehensive assessment of Liver Stiffness Measurement (LSM) and Intrahepatic Lipid (IHL) content [27, 28]. Crucially, the scarcity of direct comparisons between IF and dietary patterns such as MD and DASH creates uncertainty in clinical decision-making. Clinical guidelines often recommend 'healthy diets' broadly, but patients often struggle to choose between the structural restriction of IF and the compositional changes of MD, DASH, or LF. To bridge this gap, we conducted a network meta-analysis to simultaneously compare the efficacy of these four dietary patterns. This study constructed a comprehensive comparative network to systematically evaluate the relative efficacy of IF against MD, DASH, and LF. By rigorously incorporating quantitative imaging endpoints alongside traditional enzymatic markers, we aim to provide a more solid evidence base for precision interventions in MAFLD. Materials and methods Search strategy This network meta-analysis followed the PRISMA guidelines and was registered with PROSPERO (CRD42024609524). Databases such as PubMed, Embase, Web of Science, Cochrane Library, ClinicalTrials.gov, and WHO-ICTRP were searched to identify randomized controlled trials related to dietary management in patients with MAFLD. No language restrictions were imposed in this study’s inclusion criteria. The initial literature search was performed up to July 15, 2025. To ensure the inclusion of the most recent evidence, the search was updated during the revision process, extending the coverage to October 19, 2025. The keywords included: Metabolic dysfunction-associated fatty liver disease, MAFLD, diet, mediterranean diet, intermittent fasting, low-fat diet, DASH diet, randomized controlled trial. The detailed search strategy is provided (Supplementary 1). Eligibility criteria and study selection Two independent reviewers assessed the titles, abstracts, and full texts of the studies to establish their eligibility according to the PICO (Population, Intervention, Comparison, Outcome) framework's inclusion and exclusion criteria (Supplementary 2). A third reviewer synthesized these results and resolved all discrepancies with the first two reviewers through consensus. The included participants were adults (≥ 18 years) with MAFLD diagnosed using imaging modalities (ultrasound, controlled attenuation parameter [CAP], magnetic resonance spectroscopy [MRS], and magnetic resonance imaging [MRI]) or histology. Four forms of dietary interventions were involved, with alternative dietary regimens, placebo, or usual care as control groups. The primary outcomes included at least one liver enzyme-related index. Only randomized controlled trials were included. The exclusion criteria were studies involving animal models or participants with a history of relevant medication use. The screening procedure adhered to the PRISMA checklist for systematic reviews and meta-analyses (Supplementary 3). Data extraction and quality assessment Following the guidelines of the Cochrane Handbook for Systematic Reviews of Interventions, we created a uniform data collection form that included author names, year of publication, study location, type of intervention, intervention duration, and various recorded clinical parameters. Two independent researchers entered the data into a spreadsheet and cross-checked them to identify potential errors. The primary outcomes were hepatic enzyme levels. The secondary endpoints were lipid profiles, blood glucose levels and body composition indicators (Supplementary 2). For multi-arm trials, independent comparisons were included individually during network construction to ensure no loss of direct evidence. For duplicate intervention groups, effect sizes were pooled using a random-effects model to avoid bias introduced by redundant data. The agreement between authors was represented by Cohen’s kappa value. The quality of all included trials was independently evaluated by two authors, and any issues were resolved through discussions. If there were disagreements, a third author was brought in for assessment. The quality of all included trials was evaluated using the Cochrane Risk of Bias Tool version 2.0 (RoB 2.0). We assessed the risk of bias across the following five domains: (1) bias arising from the randomization process; (2) bias due to deviations from intended interventions; (3) bias due to missing outcome data; (4) bias in measurement of the outcome; and (5) bias in selection of the reported result. The overall risk of bias for each study was categorized as 'low risk', 'some concerns', or 'high risk'. We also used the Confidence in Network Meta-Analysis (CINeMA) framework to assess the certainty of the evidence in six key areas: within-study bias, reporting bias, indirectness, imprecision, heterogeneity, and inconsistency. Statistical analysis Network meta-analysis of the included RCTs was performed using StataMP 17. Analyses under the frequentist framework were implemented using the Stata command 'mvmeta'. For the included studies, the weighted mean difference (WMD) was adopted as the effect size metric. Under the random-effects model, the effects of each intervention were compared with those of the control/non-intervention group, and the 95% confidence interval (CI) for the WMD was calculated. StataMP 17 was used to visualize the network geometry and adjusted network plot: nodes in the plot represent individual interventions, with node size positively correlated with sample size; connecting lines denote direct comparisons between interventions, and line thickness corresponds to the number of studies conducting direct comparisons. Publication bias was evaluated through visual inspection of the adjusted funnel plots and quantified using Egger’s test. Inconsistency was evaluated by calculating and comparing the random-effects standard deviations of the consistency and inconsistency models to determine whether there was evidence of inconsistency between interventions. The node-splitting method was employed to assess local inconsistency within the network, which validated the congruence between direct and indirect evidence for specific split nodes through comparative analysis. Surface under the cumulative ranking curve (SUCRA) probabilities were computed using StataMP 17. SUCRA is a numerical metric that summarizes the ranking distribution of interventions across various outcome measures. A higher SUCRA value indicates a superior relative effectiveness of the intervention, and the interpretation of rankings must be considered alongside the statistical significance (95% CI) of pairwise comparisons. The transitivity assumption of the network was assessed by comparing the baseline characteristics of the included studies. To assess the potential violation of the transitivity assumption due to the heterogeneity of Intermittent Fasting regimens, we performed sensitivity analyses. We systematically excluded studies utilizing specific regimens (5:2 fasting, Alternate-Day Fasting [ADF], and Time-Restricted Feeding [TRF]) to evaluate the robustness of the estimates and identify potential drivers of heterogeneity. Results Characteristics of included studies A total of 25 relevant full-text articles were systematically selected from 4,977 records identified in PubMed, Embase, Web of Science, Cochrane Library, ClinicalTrials.gov, and WHO-ICTRP databases. Following the removal of 1,248 duplicate publications from identical studies and 3,074 items that failed to meet the inclusion criteria, all 25 selected articles were published between 2017 and 2025 (Fig. 1) [29–53]. The research comprised 13 studies from Asia, 8 from Europe, 1 from the Americas, and 3 from Australia, totaling 1,816 participants. All the subjects in these studies were adults diagnosed with MAFLD. The mean duration of the intervention was 20.08 weeks, with the treatment intervention period ranging from 12 weeks to 2 years (Supplementary 4). Among the included trials, the most common intervention group was IF, followed by the MD, the LFD, and the DASH diet. Twenty-four trials were evaluated based on imaging modalities, including ultrasound, FibroScan, MRI, and transient elastography. One of the trials was based on histological findings. The results showed that the Kappa value for initial screening was 0.79, for full-text screening was 0.85, and for data extraction was 0.78. Flow chart of the research selection process Methodological quality and confidence in evidence According to the RoB 2.0 assessment, 11 studies were rated as low risk in the randomization process domain. For deviations from intended interventions, only 5 studies presented a low risk, while 20 studies raised some concerns or high risk. Regarding missing outcome data, all trials were assessed as low risk; however, for bias in measurement of the outcome, 12 studies raised some concerns. Finally, in the selection of the reported result domain, 4 studies were classified as having some concerns. Overall, only 1 study was considered to have a low risk of bias, while 24 were rated as having some concerns or high risk (Fig. 2). Sensitivity analyses for the IF node demonstrated the robustness of the primary findings while highlighting regimen-specific differences. For ALT, the pooled effect of IF vs Control remained favorable across all exclusion scenarios (WMD ranging from −3.79 to −10.39). Notably, excluding TRF studies resulted in a substantial increase in efficacy (WMD = −10.39, 95% CI −15.43, −5.35), suggesting that intensive regimens like 5:2 and ADF drive the strongest improvements. Conversely, excluding 5:2 fasting attenuated the effect size and statistical significance for ALT (WMD = −3.79) and GGT (WMD = 1.12), indicating that the 5:2 regimen is a key contributor to the observed benefits (Supplementary 5.1). We rigorously evaluated both global and local inconsistency to validate the network model. The global consistency assessment showed that all clinical outcomes had P-values greater than 0.05; specifically, the P-value was 0.69 for ALT and 0.59 for AST, indicating no statistically significant global inconsistency (Supplementary 5.2). We performed a node-splitting analysis to assess local inconsistency within closed loops. The results showed no significant discrepancies between direct and indirect evidence for any outcome. For example, the difference coefficients for ALT comparisons were non-significant (P > 0.05), and similar consistency was observed for lipid and glycemic markers (Supplementary 5.3). Since no significant inconsistency was identified that would alter the treatment effect estimates, the consistency model was deemed robust and applied for all analyses. Risk of bias of randomized clinical trials To validate the transitivity assumption, we evaluated key effect modifiers across the network. Despite variations in diagnostic modalities and dietary protocols, the study populations exhibited a consistent metabolic phenotype. Participants were broadly middle-aged and overweight with confirmed MAFLD. The balanced distribution of these clinical and demographic characteristics across comparisons supports the transitivity required for the network meta-analysis (Supplementary 5.4). Comparison-adjusted funnel plots for most efficacy outcomes showed generally symmetric distributions around the midline, indicating a low risk of publication bias. Some scatter was observed, likely attributable to variable sample sizes and study quality; however, overall symmetry was preserved (Supplementary 6.1). Egger’s test indicated no significant publication bias for most outcomes. Caution is warranted due to potential biases, such as fasting insulin (P < 0.05). Further research incorporating more literature is needed to validate these conclusions (Supplementary 6.2). The Confidence in Network Meta-Analysis (CINeMA) framework was used to assess the quality of the evidence. The results showed that for most intervention comparisons, there was a low risk of reporting bias. Major concerns were mostly observed in imprecision, while heterogeneity and inconsistency were mostly rated as no or some concerns. The overall confidence rating for the evidence was low. This suggests that the conclusions should be interpreted with caution (Supplementary 7). Summary of liver injury marker and liver stiffness assessment with quantitative imaging modalities improvements In the present study, the results of liver enzyme-related indicators and liver stiffness assessment indicators were the primary outcomes. Alanine Aminotransferase (ALT), Aspartate Aminotransferase (AST), Gamma-Glutamyl Transferase (GGT), and Alkaline Phosphatase (ALP) were used as representative liver injury markers. Twenty-one studies reported results for ALT [29, 30, 32, 33, 35, 37–52]. For ALT, there were 827 patients in four treatment interventions and another 613 patients in the control group (Fig. 3). The most efficient intervention for lowering ALT levels was MD, with the highest SUCRA (68.10%) (WMD = −6.56 IU/L, 95% CI −11.59, −1.52), followed by IF (61.40%) (WMD = −5.95 IU/L, 95% CI −9.96, −1.94) (Fig. 4). In contrast, no significant reductions were observed for DASH (WMD = −6.41 IU/L, 95% CI −15.25, 2.43) or LF (WMD = −4.33 IU/L, 95% CI −16.32, 7.65). Nineteen studies reported results for AST [29, 30, 32, 33, 35, 37–39, 41–46, 48–52]. This includes a total of 738 patients in 4 treatment interventions and 577 patients in the control group (Fig. 3). Compared to the control group, IF was found to significantly reduce AST (WMD = −3.33 IU/L; 95% CI −6.09, −0.58). No significant differences were observed between the other pairwise comparisons. Fifteen studies reported results for GGT [29, 30, 35, 37, 39–42, 44, 46–50, 52]. For GGT, there were 551 patients in total, 4 treatment interventions, and another 341 patients in the control group (Fig. 3). The SUCRA analysis revealed that MD (WMD = −6.02 IU/L, 95% CI −15.73, 3.70) had the highest probability of reducing GGT (SUCRA: 62.7%) (Fig. 5). However, these effects did not reach statistical significance relative to the control group (Fig. 4). Three studies reported results for ALP [33, 35, 37]. In contrast, MD (65.3%) was the only intervention to show a statistically significant reduction (WMD = −3.46 IU/L, 95% CI −5.48, −1.45). Network maps of primary outcomes. The size of the nodes was proportional to the number of participants included in the trial, and the thickness of lines between the interventions relates to the number of studies for that comparison. The numbers on the connecting lines indicate the exact count of randomized controlled trials for each direct comparison Forest plots of network meta-analysis results for primary outcomes. The plots display the Weighted Mean Difference (WMD) and 95% Confidence Intervals (CIs) for each dietary intervention compared to the Control group. The dashed vertical line at 0 represents the line of no effect. Results to the left of the line indicate a reduction in the outcome levels (favoring the intervention), while results to the right indicate an increase SUCRA plots showing the comparative probability of efficacy for each dietary intervention on liver injury markers and stiffness. Interventions with curves closer to the upper-left corner cover a larger surface area (higher SUCRA value), indicating a higher probability of being the superior treatment. Conversely, curves closer to the lower-right indicate lower efficacy Five studies reported the results of IHL [35, 37, 43, 44, 46]. The MD (62.5%) was the only intervention to show a statistically significant reduction (WMD = −1.68%, 95% CI −3.21, −0.16) (Fig. 5). Thirteen studies reported the results of the LSM (Fig. 3) [31, 34, 39–46, 50–52]. In the network meta-analysis, no significant differences were observed between pairwise interventions (Fig. 4). Summary of lipid and glucose metabolism improvements Additionally, metabolic indicators, such as blood lipid and glucose levels, can be used as secondary outcomes to observe the effects of different dietary interventions on MAFLD. To evaluate lipid metabolic profiles, we selected triglycerides (TG), total cholesterol (TC), high-density lipoprotein (HDL), and low-density lipoprotein (LDL) levels as representative markers. The network meta-analysis showed that MD (79.3%) (WMD = −22.54 mg/dL, 95% CI −42.79, −2.29) was the only dietary intervention that significantly lowered TG concentrations compared with the control. For TC and HDL, no significant differences were observed among the dietary patterns. Notably, MD uniquely exerted a significant lowering effect on LDL levels (73.7%) (WMD = −3.41 mg/dL, 95% CI −4.67, −2.14) (Supplementary 9). Although IF and MD achieved the highest SUCRA rankings for glycemic parameters including FBG, HbA1c, and insulin, no statistically significant differences were observed in pairwise comparisons between any dietary patterns or against the control group (Supplementary 9). Summary of improvements in body composition indicators We also conducted network meta-analyses to evaluate the body composition indicators of the diet. The SUCRA analysis showed that the IF (WMD = −1.96 kg, 95% CI −3.01, −0.92) had the highest probability (69.2%) of reducing weight. Network meta-analysis revealed that both IF (WMD = −0.67 kg/m2, 95% CI −1.07, −0.28) and MD (WMD = −0.76 kg/m2, 95% CI −1.27, −0.24) significantly reduced BMI. The analysis indicated that IF was the unique intervention associated with a significant decrease in waist circumference (WC) (69.8%) (WMD = −1.73 cm, 95% CI −2.89, −0.57). For Fat mass, although the MD achieved a SUCRA ranking of 63.1%, the reduction was not statistically significant (WMD = −0.18 kg, 95% CI −1.11, 0.74) (Supplementary 9). Discussion Principal findings The relative advantages and disadvantages of different dietary patterns, as well as the priority of clinical selection, remain unresolved. This network meta-analysis systematically evaluated and compared the comprehensive effects of four dietary patterns (IF, MD, DASH, and LF) on liver, blood lipid, and blood glucose indices in patients with MAFLD. A total of 1,816 participants from 25 randomized controlled trials were included in this study. Interpretation of results and discussion on mechanisms IF emerged as a top-performing intervention for improving body weight and waist circumference. This metabolic benefit aligns with prior research suggesting that cyclic energy restriction implies favorable lipolysis [54–56]. Furthermore, IF demonstrated statistically significant efficacy in improving liver enzyme profiles, specifically lowering ALT (WMD = −5.95 IU/L; 95% CI −9.96, −1.94) and AST (WMD = −3.33 IU/L; 95% CI −6.09, −0.58). Our findings suggest that IF may be a viable strategy for metabolic management. This finding aligns with recent experimental conclusions, providing further corroboration of the potential benefits of IF on metabolic dysfunction in patients with MAFLD [57]. MD may offer distinct advantages in improving hepatic pathological features and controlling lipid metabolism. First, MD appears to be a promising regimen for lipid modulation. In our analysis, it had the highest probability of increasing HDL (SUCRA: 82.6%) and lowering TG levels (SUCRA: 79.3%). Second, MD demonstrated notable potential in improving MAFLD parameters. It was associated with a significant reduction in ALT, a key biomarker of hepatocellular injury (WMD = −6.56 IU/L, 95% CI −11.59, −1.52). Additionally, MD significantly decreased IHL (WMD = −1.68%, 95% CI −3.21, −0.16). The MD diet components are rich in antioxidants and unsaturated fatty acids, which have been theoretically linked to reduced hepatic steatosis and inflammation [58–60]. In contrast, the DASH and LF diets exhibited limited efficacy in improving core MAFLD indicators. DASH and LF appear to be less focused on core MAFLD pathologies compared to IF and MD, such as and hepatic steatosis. This finding is consistent with previous conclusions that low-fat diets have limited efficacy in MAFLD [61]. Although pooling various IF regimens is statistically justified by the consistent direction of liver enzyme improvement, significant heterogeneity exists regarding the magnitude of efficacy, a finding that aligns with recent studies [62, 63]. Our analysis indicates an 'intensity-response' relationship: excluding TRF trials substantially increased the estimated efficacy of IF in reducing ALT, surpassing even the MD Diet. Conversely, excluding the 5:2 regimen notably weakened the results, particularly for GGT. The observed differential metabolic effects associated with fasting intensity corroborate the results of a recent meta-analysis conducted by Pascual et al. (2023) [64]. Consequently, while TRF serves as a valid initial intervention, regimens involving greater caloric deficits appear necessary for the more effective reversal of hepatic steatosis and enzyme elevation. Clinicians should consider this hierarchy of efficacy when tailoring dietary prescriptions. Consistencies and discrepancies with existing studies To our knowledge, this is the first network meta-analysis to simultaneously compare these four dietary patterns in patients with MAFLD. Prior evidence indicates that the MD effectively reduces hepatic fat content and transaminase levels. Evidence from randomized controlled trials and meta-analyses consistently demonstrates the hepatic benefits of the MD. Ryan et al. found a remarkable 39% reduction in hepatic steatosis, while Haigh et al. systematically confirmed significant reductions in liver enzymes like ALT across multiple studies [65, 66]. While more research is needed, current evidence supports the utility of the MD in managing liver health. Similarly, Lange et al. conducted a systematic review focusing on IF confirmed its potential to ameliorate hepatic steatosis and improve liver enzyme profiles [67]. In a meta-analysis, Khalafi et al. analyzed over 1,000 participants and found consistent improvements: IF was found to decrease AST and reduce ALT [68]. Our network meta-analysis results corroborate these findings, specifically demonstrating a significant reduction in ALT and AST. The distinct contribution of the current study is the provision of clearer evidence, derived from both direct and indirect comparisons. Limitations This study has several strengths. First, we included 25 randomized controlled trials involving 1,816 participants, providing the largest body of evidence to date for indirect comparisons in this field. Second, we employed the rigorous Cochrane Risk of Bias tool and CINeMA framework for evidence assessment. Funnel plots and Egger's test were used to assess for publication bias. Symmetry was observed for most outcomes. Several limitations of this study warrant mention. First, our analysis was restricted to four mainstream dietary patterns, which may not fully encompass the therapeutic potential of all nutritional interventions for MAFLD; thus, future large-scale studies are warranted to broaden this scope. Second, the mean intervention duration of 20.08 weeks is a significant limitation. This relatively short timeframe may explain why no dietary pattern demonstrated a significant improvement in LSM, as it may be insufficient to reverse established hepatic stiffness or fibrosis. Furthermore, the inherent challenges of blinding participants and personnel in dietary interventions introduce the potential for subjective bias. Although we controlled for confounding factors during the analysis, and 5 of the 25 included studies were assessed as high risk of bias, this remains an intrinsic limitation of the evidence. We acknowledge the internal heterogeneity within the dietary intervention nodes, particularly Intermittent Fasting. Despite variations in fasting windows, these protocols share a robust mechanistic core: they consistently induce periodic energy restriction that homogeneously improves hepatic insulin sensitivity and promotes lipolysis, providing a valid basis for pooling these interventions. Finally, the overall confidence in the evidence assessed using the CINeMA framework was low for several outcomes. The low confidence of the network meta-analysis evidence necessitates caution in generalizing our conclusions and limits their immediate clinical application. Future research should prioritize optimized study designs to enhance the quality of evidence in this field. Clinical implications and future perspectives This study provides evidence-based support for dietary management of MAFLD. We identified distinct clinical applications for the IF and MD. The clinical value of IF is highlighted by its notable capacity for metabolic modulation. IF was significantly effective in promoting weight loss (WMD = −1.96 kg, 95% CI −3.01, −0.92) and reducing WC (WMD = −1.73 cm, 95% CI −2.89, −0.57). While a 1.96 kg weight loss may not by itself meet the threshold required to reverse hepatic steatosis, IF serves as an effective initial intervention [68, 69]. Consequently, it is preferentially recommended for patients requiring initiation of weight management. However, clinical implementation must carefully consider patient adherence. Conversely, the clinical significance of the MD stems from its direct improvements in hepatic pathology and cardiovascular risk factors. The MD was the dietary pattern observed to significantly reduce both ALT (WMD = −6.56 IU/L, 95% CI −11.59, −1.52) and IHL (WMD = −1.68%, 95% CI −3.21, −0.16). A reduction in ALT of 6.56 IU/L is indicative of alleviated hepatocellular injury, while an absolute reduction in IHL of 1.68% may help address the core pathology of hepatic fat accumulation, consistent with previous observation [66, 70]. Furthermore, the benefits of the MD in improving the lipid profile are of profound significance in reducing the risk of cardiovascular mortality in patients with MAFLD [71]. Therefore, the MD could be considered for patients characterized by dyslipidemia. In clinical practice, dietary plans should be tailored based on patient characteristics, such as the presence of obesity or hyperlipidemia, to optimize overall therapeutic results. Future studies with extended follow-up periods are warranted to assess the long-term efficacy of these interventions in delaying hepatic fibrosis progression. Conclusions This network meta-analysis suggests that both IF and the MD offer viable therapeutic strategies for managing MAFLD, though their clinical strengths diverge. IF appears to be an effective option for body composition improvements, making it preferentially suitable for patients with MAFLD characterized by obesity. Sensitivity analyses suggest that regimen intensity plays a critical role in efficacy. Conversely, the MD demonstrated the highest probability of improving core hepatic markers and lipid profiles, suggesting it may be considered for patients presenting primarily with hepatic injury and lipid metabolism disorders. However, these findings should be interpreted with caution due to the low overall certainty of the current evidence and the relatively short duration of included trials. Future research should focus on high-quality, large-scale RCTs with extended follow-up. Supplementary Information Supplementary Material 1. Authors’ contributions J-H Z、F T、S-Q C、J H、Z-H L、Y-C W、H-Z C、Y-P L、S–H Z、Z–C H and X-M M designed and implemented the study. J-H Z、F T and S-Q C provided per study study results and performed statistical analyses. J-H Z and F T drafted the manuscript. X-M M critically revised the manuscript for important intellectual content and approved the final paper. Funding The author(s) declare that financial support was received for the research and/or publication of this article.This study was supported by the 2023 Talent Project of Tangshan City (B202302016). Data availability The study's original contributions are available in the article and supplementary material, and further questions can be addressed to the corresponding author. Declarations Ethics approval and consent to participate Not applicable. Consent for Publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jiahui Zou and Fei Tian contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The study's original contributions are available in the article and supplementary material, and further questions can be addressed to the corresponding author.