Higher adherence to the EAT–lancet diet is associated with reduced risk of metabolic dysfunction–associated steatotic liver disease: a 5-year middle eastern cohort study Authors: Farzam Kamrani, Mobina Imannezhad, Elahe Etesami, Nazanin Zamanian, Jalaledin Mirzay Razaz, Sayed Hossein Davoodi, Reza Homayounfar, Ali Nikparast Published: 2026 ABSTRACT BackgroundMetabolic dysfunction–associated steatotic liver disease (MASLD) is a rapidly increasing metabolic disorder worldwide. Sustainable dietary patterns such as the EAT–Lancet Diet (ELD), designed to promote both human and planetary health, may help prevent MASLD; however, prospective evidence from Middle Eastern populations remains limited.MethodsThis prospective cohort study included 5,058 Iranian adults aged 20–70 years from the MMRT study who were free of MASLD at baseline. Dietary intake was assessed using a validated 125‑item food frequency questionnaire, and adherence to the ELD was quantified using the Planetary Health Diet Index (PHDI). Incident MASLD over a 5‑year follow‑up was defined using hepatic steatosis (CAP ≥ 238 dB/m) in the presence of metabolic dysfunction. Multivariable logistic regression, restricted cubic splines, and mediation analyses were applied.ResultsDuring follow‑up, 562 participants developed MASLD, yielding a 5-year cumulative incidence of 11.1%. In fully adjusted models, individuals in the third and fourth quartiles of PHDI had 39% (OR 0.61; 95% CI 0.45–0.81) and 34% (OR 0.66; 95% CI 0.50–0.88) lower odds of MASLD, respectively, compared with the lowest quartile (P‑trend < 0.001). Each one‑standard‑deviation increase in PHDI was associated with a 16% lower MASLD risk (OR 0.84; 95% CI 0.76–0.94). The inverse relationship was largely linear, with a suggestive plateau at higher PHDI scores (P for non-linearity = 0.08). Mediation analyses indicated substantial contributions from dietary fiber (71.4%), HOMA‑IR (69.3%), CAP (53.8%), and CRP (44.0%).ConclusionsGreater adherence to a PHDI was associated with lower MASLD risk in Iranian adults, potentially through improvements in insulin resistance, inflammation, hepatic steatosis, and dietary fiber intake.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13098-026-02198-w. Introduction Metabolic dysfunction–associated steatotic liver disease (MASLD) is the term introduced in 2023 to replace what was previously known as nonalcoholic fatty liver disease (NAFLD) [1]. This condition refers to fatty liver disease that is linked to metabolic dysfunction. The rebranding aims to characterize the disease by its nature (a metabolic dysfunction) rather than by what it is not (non-alcoholic). Current estimates indicate that approximately 38% of all adults are affected by MASLD. Projections suggest that by 2040, the prevalence of MASLD among adults could rise to over 55% [2]. In a recent cohort study conducted within the Iranian population, 33.7% were classified as MASLD patients [3]. MASLD is a metabolic dysfunction where diet and lifestyle interventions play a crucial role in prevention. The Mediterranean diet has gained attention for its moderate improvements in liver function, evidenced by reductions in liver enzymes [4]. Other dietary practices show varying benefits: the very low-calorie ketogenic diet (VLCKD) is highly effective for weight loss, while the dietary approaches to stop hypertension (DASH) excels at lowering blood lipid levels, though its effects on liver enzymes are inconsistent. Additionally, a low free sugar diet (LFSD) has shown improved glucose metabolism [5]. While these dietary approaches offer promise for managing MASLD, discussions are now expanding to include planetary health alongside human health. Sustainability has recently gained significant attention, particularly in the context of climate change and its impacts. The EAT-Lancet Diet (ELD), developed by the EAT-Lancet Commission in 2019, is a mostly plant-based dietary framework designed to optimize human health while promoting environmental sustainability [6]. The ELD emphasizes a variety of whole grains, fruits, vegetables, legumes, nuts, and sources of unsaturated fats. It recommends moderate consumption of dairy and poultry, limits red and processed meats, and minimizes added sugars and refined grains. The overarching goal of this dietary approach is to provide essential nutrients within planetary boundaries, thereby mitigating the risk of non-communicable diseases and reducing greenhouse gas emissions, land use, and biodiversity loss associated with food production [7]. Because the EAT–Lancet diet is predominantly plant‑based, it has been hypothesized to confer protection against metabolic diseases. For instance, a large-scale meta-analysis involving 2.21 million participants found that adherence to ELD was associated with a 22% reduced risk of diabetes, a 16% reduced risk of cardiovascular disease mortality, a 17% reduced risk of all-cause mortality [8]. In terms of MASLD prevention, several studies have highlighted the ELD diet as a pattern that promotes both human and environmental health. For example, Zhang et al. conducted an analysis involving more than 170,000 participants, primarily from the UK, which demonstrated that high adherence to the ELD diet reduced the risk of MASLD by 23% to 32%, depending on the ELD scoring method used [9]. In another study by Wu et al., the UK population was examined, revealing that those in the highest tertile of ELD adherence experienced a 33% reduction in the risk of MASLD, after a median follow-up period of 13.3 years [10]. However, the existing evidence is primarily derived from Western or East Asian cohorts, creating a significant gap in our understanding of populations with distinct genetic susceptibilities and unique dietary customs [9, 10]. While higher-income countries have demonstrated benefits from adherence to the ELD as a sustainable dietary pattern, this diet may be unaffordable for billions. In low-income countries, it could exacerbate deficiencies in essential nutrients such as vitamin A, B12, calcium, iron, and zinc, owing to limited access to animal-source foods [11]. This concern is further underscored by individual-level evidence suggesting that the ELD may not be appropriate for low-income populations. For instance, a study by Karim Dehnavi et al. revealed that adherence to the diet varies by socioeconomic status (SES). Although higher adherence can lead to overall cost reductions, this advantage does not apply to low-income individuals, highlighting the presence of economic barriers [12]. Therefore, conducting a large longitudinal study focused on a Middle Eastern population could yield valuable insights into health outcomes associated with sustainable diets in this region. This prospective cohort study aimed to evaluate the impact of adherence to the ELD on the risk of developing MASLD over a five-year follow-up period. The study contributes to the evidence regarding the effectiveness of a sustainable diet within the Middle Eastern population and its health outcomes. Methods Study design and population This investigation employed a prospective cohort design, drawing upon data collected through the Monitoring of Metabolic Diseases Risk Factors in Tehran (MMRT) initiative, which examines determinants of non-communicable conditions in Tehran’s adult population [13]. Recruitment utilized convenience sampling methods targeting individuals between 20 and 70 years of age attending a comprehensive health facility. During the recruitment window spanning January 2017 through December 2018, 8,760 potential participants were contacted, with 7,836 agreeing to enroll, yielding an 89.4% acceptance rate. Written informed consent was obtained from all enrollees, who received anonymized identification codes ensuring data confidentiality. Initial data acquisition involved structured in-person interviews, physical examinations, and biochemical analyses capturing information on demographics, lifestyle patterns, health history, nutritional habits, and body measurements. Five years following baseline (January 2022–December 2023), the original cohort was re-contacted for follow-up evaluation employing measurement protocols matching those used initially. The follow-up assessment window was fixed, meaning the time from baseline to follow-up was nearly identical for all participants. Retention remained robust at 95%, with 7,600 participants completing the second assessment wave. The analytical cohort was defined by excluding individuals presenting with baseline MASLD (n = 1,660), cancer (n = 30, due to potential disease-related alterations in dietary intake, body composition, and inflammation), renal failure (n = 50, due to its impact on dietary behaviors and metabolic state), gallstone (n = 114, as it can directly confound hepatic assessment), rheumatoid arthritis (n = 255, due to systemic inflammatory state and potential dietary modifications), and thyroid pathologies (n = 431, owing to their direct effects on hepatic metabolism). Importantly, identical diagnostic procedures and criteria—namely FibroScan with CAP ≥ 238 dB/m in conjunction with the same metabolic dysfunction criteria—were uniformly applied for both baseline exclusion and incident case identification during follow-up. After applying these criteria, 5,058 participants with complete longitudinal data comprised the final study sample. Ethical oversight was provided by the National Nutrition and Food Technology Research Institute’s ethics committee (approval code: IR.SBMU.NNFTRI.REC.1404.003), with all procedures conducted in accordance with Helsinki Declaration guidelines. Assessment of EAT–lancet diet adherence Nutritional assessment utilized a validated 125-item Food Frequency Questionnaire (FFQ) derived from the Willett framework and modified to capture Iranian eating patterns [14]. Respondents indicated consumption frequencies for each food across the preceding year using daily, weekly, monthly, or annual categories. Portion sizes followed standards established by the U.S. Department of Agriculture (USDA) food database, with household measurements transformed into gram units where applicable. Nutritional composition data were obtained predominantly from the USDA Food Composition Table, with the Iranian Food Composition Table providing supplementary information for region-specific foods. Adherence to the ELD was quantified using an index adapted from the Planetary Health Diet Index (PHDI), originally developed and validated by Cacau et al. [6]. The index is based on the reference diet proposed by the EAT-Lancet Commission, which defines recommended intake ranges for major food groups within a 2,500 kcal/day dietary pattern. To allow comparability across individuals with different total energy intakes, all food group recommendations were operationalized using an energy-density approach. Specifically, for each participant, we first calculated the energy contribution (in kcal) from each food group. We then expressed this contribution as a percentage of the individual’s total daily energy intake (% of total kcal). The resulting energy percentages were directly compared against the proportional intake ranges defined by the 2,500 kcal EAT-Lancet reference diet (e.g., the reference diet recommends that whole grains contribute approximately 32% of total energy, which corresponds to 811 kcal out of 2,500 kcal). This method renders the scoring independent of absolute caloric intake, ensuring that participants with varying energy needs are evaluated on a comparable basis. Food items from the FFQ were aggregated into predefined food groups according to the EAT–Lancet classification; the specific FFQ items contributing to each PHDI component are detailed in Supplementary Table S1. The adherence score comprised 16 components categorized into four scoring domains: 1. Adequacy components (higher intake reflects greater adherence): nuts and peanuts, legumes, fruits, vegetables, and whole grains. 2. Optimum components (moderate intake reflects greater adherence): eggs, fish and seafood, tubers and potatoes, dairy products (excluding dairy fats), and vegetable oils. 3. Ratio components: ratio of dark green vegetables to total vegetables and ratio of red and orange vegetables to total vegetables. 4. Moderation components (lower intake reflects greater adherence): red meat (beef, lamb, pork), poultry and substitutes, animal fats, and added sugars. Each component was scored proportionally according to predefined cut-offs derived from the EAT–Lancet reference diet energy contributions. Depending on the component, a maximum of 5 or 10 points was assigned. The total adherence score ranged from 0 to 150 points, with higher scores indicating greater adherence to the EAT–Lancet dietary pattern (Supplementary Table S2) [6]. Assessment of Metabolic Dysfunction–Associated Steatotic Liver Disease (MASLD) Hepatic steatosis was evaluated through transient elastography (FibroScan™, Echosens, Paris, France), employing the XL probe for participants with elevated body weight. The Controlled Attenuation Parameter (CAP), expressed in decibels per meter (dB/m), served to quantify liver fat accumulation. CAP readings of ≥ 238 dB/m were considered indicative of hepatic steatosis [2, 15]. MASLD diagnosis required documented hepatic steatosis alongside one or more metabolic abnormalities: increased BMI (≥ 25 kg/m²) or waist circumference (men > 94 cm, women > 80 cm), elevated blood pressure (≥ 130/85 mmHg or use of antihypertensive agents), raised triglycerides (≥ 150 mg/dL or lipid-modifying medication), reduced HDL-C (men < 40 mg/dL, women < 50 mg/dL or relevant therapy), dysglycemia with fasting glucose 100–125 mg/dL, HbA1c between 5.7 and 6.4%, postprandial glucose ≥ 140 mg/dL at 2 h, or established diabetes mellitus with or without pharmacological management [1]. Assessment of covariates Sociodemographic characteristics, lifestyle practices, and clinical backgrounds were collected via a structured, interviewer-led questionnaire. Captured variables encompassed age, sex, educational attainment, tobacco use, pharmaceutical consumption, and existing health conditions. Economic status was determined through a wealth score index (WSI) constructed using Multiple Correspondence Analysis (MCA), incorporating indicators such as household appliance ownership, electronic device possession, vehicle access, television type, and international travel history [16]. Physical activities were quantified using the Modifiable Activity Questionnaire (MAQ), which computes weekly metabolic equivalent task (MET) hours. This instrument has demonstrated validity in Iranian cohorts [17]. Certified technicians obtained weight, height, and waist circumference measurements following standardized procedures. Body mass index (BMI) was derived by dividing weight (kg) by the square of height (m²). Body composition was assessed via dual-energy X-ray absorptiometry (DXA; Lunar iDXA, GE Healthcare, USA), and fat mass index (FMI) was calculated by dividing total fat mass by height squared [18]. Blood pressure readings were obtained via automated sphygmomanometer following a five-minute seated rest period. Analytical values represented the mean of two consecutive measurements. Venous blood specimens were drawn after an 8–12 h overnight fasting period. Serum concentrations of fasting blood glucose (FBG), triglycerides (TG), total cholesterol (TC), and high-density lipoprotein cholesterol (HDL-C) were determined through enzymatic colorimetric techniques using automated instrumentation. All biochemical analyses utilized the AutoAnalyzer platform (Selectra E, Vitalab, Holliston, the Netherlands) with Pars Azmoon reagent kits. Low-density lipoprotein cholesterol (LDL-C) concentrations were calculated via the Friedewald Equation [19]. Insulin quantification employed electrochemiluminescence immunoassay methodology, with insulin resistance indexed through HOMA-IR, calculated as (fasting blood glucose [mg/dL] × fasting insulin [µU/mL]) / 405. Hepatic enzyme activities, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT), were determined using conventional enzymatic methods. Statistical analysis Participant characteristics were summarized across PHDI quartiles using descriptive statistics, presenting continuous data as means ± standard deviations and categorical data as frequencies and proportions. Between-quartile comparisons employed linear regression for continuous outcomes and chi-square (χ²) testing for categorical outcomes. The relationship between PHDI and MASLD likelihood was examined through multivariable logistic regression modeling. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for each PHDI quartile relative to the lowest quartile as reference. Model 1 controlled for age and sex. Model 2 incorporated additional covariates: physical activity, wealth score index, education level, smoking, and fat mass index. Model 3 further accounted for hypertension, and type 2 diabetes mellitus, while the final model added total energy intake. Linear trend testing utilized median quartile values as continuous predictors. Given that the follow-up window was fixed for all participants, logistic regression was deemed an appropriate analytical approach; the introduction of person-time was not required as there was no variation in follow-up duration. Stratified analyses examined associations within subgroups defined by age, sex, BMI, smoking habits, WSI (above vs. below median), PA (above vs. below median), T2DM, and hypertension status. Potential non-linear relationships between PHDI and MASLD odds were explored using random-effects restricted cubic splines with four knots positioned at the 5th, 35th, 65th, and 95th PHDI percentiles [20]. Mediation analyses were performed using the “mediation” package in R to assess whether CAP, HOMA-IR, C-reactive protein, and dietary fiber mediated the relationship between PHDI and MASLD. The temporal sequence for mediation was supported by the prospective design: the exposure (PHDI) was derived from baseline dietary intake, the mediators were measured at baseline, and the outcome (incident MASLD) was assessed prospectively five years later. The analysis estimated direct, indirect, and total effects with 95% confidence intervals, employing bootstrapped standard errors based on 10,000 replications. Additionally, the proportion of the total effect mediated by each of these factors was calculated to quantify their individual contributions to the observed association. A complete-case analysis approach was employed, wherein only participants with complete data for all variables included in the final models were retained in the analytical sample. Given the large number of statistical tests performed, particularly in the component-level, subgroup, and sensitivity analyses, the results of these exploratory secondary analyses should be interpreted with caution and regarded as hypothesis-generating. No formal correction for multiple testing was applied, as such corrections can be overly conservative in nutritional epidemiology, particularly when tests are not independent; however, our primary conclusions rest solely on the pre-specified dose–response analyses of PHDI and incident MASLD. Statistical computations were performed using Stata version 17 (StataCorp, College Station, TX, USA) and R software version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was defined as a two‑tailed P value < 0.05. Results Baseline characteristics A total of 5,058 participants were included in the present analysis (Fig. 1). The mean age of the study population was 41.95 ± 9.90 years, and 52.7% were men. The mean PHDI score was 50.78 ± 5.67, ranging from 43.66 ± 2.47 in the lowest quartile (Q1) to 58.16 ± 4.01 in the highest quartile (Q4) (P-trend < 0.001). Fig. 1Flow chart of the study Flow chart of the study Across increasing quartiles of PHDI (Table 1), participants tended to be younger (P-trend < 0.001) and reported higher physical activity levels (P-trend < 0.001). The proportion of men varied significantly across quartiles (P-trend < 0.001), while no significant trends were detected in smoking status or education level (both P-trend > 0.05). Table 1Baseline characteristics of participants across quartiles of planetary health diet index.aCharacteristicsTotal population (n = 5058)Q1 (n = 1085)Q2 (n = 1245)Q3 (n = 1418)Q4 (n = 1310)P-trend bAge41.95 ± 9.9042.89 ± 9.9542.30 ± 10.3741.66 ± 9.5141.14 ± 9.75 < 0.001 Men, n (%)2665 (52.7%)527 (48.6%)728 (58.5%)781 (55.1%)629 (48.0%) < 0.001 Smoker, n (%)599 (11.9%)123 (11.3%)158 (12.7%)160 (11.3%)158 (12.1%)0.657Education, n (%)0.537Low928 (18.3%)209 (19.3%)241 (19.4%)263 (18.5%)215 (16.4%)Middle2984 (59.0%)627 (57.8%)726 (58.3%)839 (59.2%)792 (60.5%)High1146 (22.7%)249 (22.9%)278 (22.3%)316 (22.3%)303 (23.1%)WSI71.41 ± 14.2869.99 ± 14.0671.24 ± 14.7172.68 ± 14.1471.37 ± 14.09 0.004 Physical activity (MET/H/week)24.01 ± 2.7523.70 ± 2.6923.87 ± 2.7124.26 ± 2.8024.11 ± 2.76 < 0.001 Weight (kg)84.90 ± 13.4786.48 ± 14.0886.47 ± 13.5683.00 ± 12.9484.14 ± 13.09 < 0.001 Height (cm)166.74 ± 6.14166.70 ± 6.14167.07 ± 6.24166.81 ± 6.02166.38 ± 6.160.093Waist circumference (cm)84.82 ± 14.5387.84 ± 15.2684.98 ± 14.7682.80 ± 13.7484.35 ± 14.12 < 0.001 Body mass index (kg/m2)27.31 ± 4.5127.60 ± 4.5927.74 ± 4.5326.79 ± 4.3927.21 ± 4.47 < 0.001 Fat mass index (kg/m2)7.28 ± 3.228.07 ± 3.417.39 ± 3.266.69 ± 2.927.18 ± 3.17 < 0.001 SBP (mmHg)118.33 ± 7.85118.64 ± 7.99118.80 ± 7.99117.87 ± 7.45118.11 ± 7.99 0.010 DBP (mmHg)72.64 ± 9.2272.98 ± 9.4872.58 ± 9.1772.39 ± 9.1472.67 ± 9.150.375Fasting blood sugar (mg/dl)97.92 ± 14.6999.05 ± 14.8897.89 ± 14.6696.76 ± 14.6498.27 ± 14.520.087Fasting Insulin (µU/mL)8.42 ± 2.739.23 ± 2.938.58 ± 2.837.90 ± 2.488.17 ± 2.54 < 0.001 HOMA-IR1.29 ± 0.541.47 ± 0.581.34 ± 0.541.15 ± 0.471.23 ± 0.50 < 0.001 Triglyceride (mg/dl)138.31 ± 16.69141.00 ± 17.28138.85 ± 17.19136.03 ± 15.37138.05 ± 16.73 < 0.001 Total cholesterol (mg/dl)170.52 ± 26.73172.74 ± 28.71172.34 ± 26.43168.15 ± 24.99169.52 ± 26.89 < 0.001 LDL-C (mg/dl)95.00 ± 15.2496.63 ± 15.2595.77 ± 15.3193.71 ± 14.7894.30 ± 15.52 < 0.001 HDL-C (mg/dl)44.45 ± 7.1043.52 ± 7.1144.05 ± 7.2745.12 ± 6.9844.88 ± 6.96 < 0.001 ALT (U/L)26.86 ± 4.6627.44 ± 4.8427.59 ± 4.6626.31 ± 4.4226.29 ± 4.60 < 0.001 AST (U/L)24.43 ± 4.9124.99 ± 5.1125.04 ± 4.8623.97 ± 4.7723.90 ± 4.85 < 0.001 GGT (U/L)24.80 ± 8.4427.81 ± 8.7425.86 ± 8.7022.38 ± 7.6023.93 ± 7.88 < 0.001 CAP (dB/m)175.34 ± 36.17195.33 ± 42.74184.61 ± 42.52170.15 ± 39.03177.17 ± 40.41 < 0.001 CRP (mg/dl)1.87 ± 1.672.45 ± 2.041.95 ± 1.731.50 ± 1.271.73 ± 1.50 < 0.001 Type 2 diabetes Mellitus, n (%)460 (9.1%)112 (10.3%)111 (8.9%)116 (8.2%)121 (9.2%) < 0.001 Hypertension, n (%)354 (7.0%)102 (9.4%)110 (8.8%)68 (4.8%)74 (5.6%)0.111Planetary health diet index score50.78 ± 5.6743.66 ± 2.4748.57 ± 1.0951.35 ± 0.4158.16 ± 4.01 < 0.001 a Data are presented as mean ± standard deviation for continuous variables and number (percent) for non-continuous variablesb P-value was obtained from linear regression analysis for continuous variable, and χ2 test for categorical variables, respectively, according to the category of planetary health diet indexWSI, Wealth Score Index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HOMA-IR, Homeostasis Model Assessment for Insulin Resistance; LDL, Low-Density Lipoprotein cholesterol; HDL, High-Density Lipoprotein cholesterol; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; GGT, Gamma-glutamyl transferase; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter Table 1Baseline characteristics of participants across quartiles of planetary health diet index.aCharacteristicsTotal population (n = 5058)Q1 (n = 1085)Q2 (n = 1245)Q3 (n = 1418)Q4 (n = 1310)P-trend bAge41.95 ± 9.9042.89 ± 9.9542.30 ± 10.3741.66 ± 9.5141.14 ± 9.75 < 0.001 Men, n (%)2665 (52.7%)527 (48.6%)728 (58.5%)781 (55.1%)629 (48.0%) < 0.001 Smoker, n (%)599 (11.9%)123 (11.3%)158 (12.7%)160 (11.3%)158 (12.1%)0.657Education, n (%)0.537Low928 (18.3%)209 (19.3%)241 (19.4%)263 (18.5%)215 (16.4%)Middle2984 (59.0%)627 (57.8%)726 (58.3%)839 (59.2%)792 (60.5%)High1146 (22.7%)249 (22.9%)278 (22.3%)316 (22.3%)303 (23.1%)WSI71.41 ± 14.2869.99 ± 14.0671.24 ± 14.7172.68 ± 14.1471.37 ± 14.09 0.004 Physical activity (MET/H/week)24.01 ± 2.7523.70 ± 2.6923.87 ± 2.7124.26 ± 2.8024.11 ± 2.76 < 0.001 Weight (kg)84.90 ± 13.4786.48 ± 14.0886.47 ± 13.5683.00 ± 12.9484.14 ± 13.09 < 0.001 Height (cm)166.74 ± 6.14166.70 ± 6.14167.07 ± 6.24166.81 ± 6.02166.38 ± 6.160.093Waist circumference (cm)84.82 ± 14.5387.84 ± 15.2684.98 ± 14.7682.80 ± 13.7484.35 ± 14.12 < 0.001 Body mass index (kg/m2)27.31 ± 4.5127.60 ± 4.5927.74 ± 4.5326.79 ± 4.3927.21 ± 4.47 < 0.001 Fat mass index (kg/m2)7.28 ± 3.228.07 ± 3.417.39 ± 3.266.69 ± 2.927.18 ± 3.17 < 0.001 SBP (mmHg)118.33 ± 7.85118.64 ± 7.99118.80 ± 7.99117.87 ± 7.45118.11 ± 7.99 0.010 DBP (mmHg)72.64 ± 9.2272.98 ± 9.4872.58 ± 9.1772.39 ± 9.1472.67 ± 9.150.375Fasting blood sugar (mg/dl)97.92 ± 14.6999.05 ± 14.8897.89 ± 14.6696.76 ± 14.6498.27 ± 14.520.087Fasting Insulin (µU/mL)8.42 ± 2.739.23 ± 2.938.58 ± 2.837.90 ± 2.488.17 ± 2.54 < 0.001 HOMA-IR1.29 ± 0.541.47 ± 0.581.34 ± 0.541.15 ± 0.471.23 ± 0.50 < 0.001 Triglyceride (mg/dl)138.31 ± 16.69141.00 ± 17.28138.85 ± 17.19136.03 ± 15.37138.05 ± 16.73 < 0.001 Total cholesterol (mg/dl)170.52 ± 26.73172.74 ± 28.71172.34 ± 26.43168.15 ± 24.99169.52 ± 26.89 < 0.001 LDL-C (mg/dl)95.00 ± 15.2496.63 ± 15.2595.77 ± 15.3193.71 ± 14.7894.30 ± 15.52 < 0.001 HDL-C (mg/dl)44.45 ± 7.1043.52 ± 7.1144.05 ± 7.2745.12 ± 6.9844.88 ± 6.96 < 0.001 ALT (U/L)26.86 ± 4.6627.44 ± 4.8427.59 ± 4.6626.31 ± 4.4226.29 ± 4.60 < 0.001 AST (U/L)24.43 ± 4.9124.99 ± 5.1125.04 ± 4.8623.97 ± 4.7723.90 ± 4.85 < 0.001 GGT (U/L)24.80 ± 8.4427.81 ± 8.7425.86 ± 8.7022.38 ± 7.6023.93 ± 7.88 < 0.001 CAP (dB/m)175.34 ± 36.17195.33 ± 42.74184.61 ± 42.52170.15 ± 39.03177.17 ± 40.41 < 0.001 CRP (mg/dl)1.87 ± 1.672.45 ± 2.041.95 ± 1.731.50 ± 1.271.73 ± 1.50 < 0.001 Type 2 diabetes Mellitus, n (%)460 (9.1%)112 (10.3%)111 (8.9%)116 (8.2%)121 (9.2%) < 0.001 Hypertension, n (%)354 (7.0%)102 (9.4%)110 (8.8%)68 (4.8%)74 (5.6%)0.111Planetary health diet index score50.78 ± 5.6743.66 ± 2.4748.57 ± 1.0951.35 ± 0.4158.16 ± 4.01 < 0.001 a Data are presented as mean ± standard deviation for continuous variables and number (percent) for non-continuous variablesb P-value was obtained from linear regression analysis for continuous variable, and χ2 test for categorical variables, respectively, according to the category of planetary health diet indexWSI, Wealth Score Index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HOMA-IR, Homeostasis Model Assessment for Insulin Resistance; LDL, Low-Density Lipoprotein cholesterol; HDL, High-Density Lipoprotein cholesterol; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; GGT, Gamma-glutamyl transferase; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter Baseline characteristics of participants across quartiles of planetary health diet index.a a Data are presented as mean ± standard deviation for continuous variables and number (percent) for non-continuous variables b P-value was obtained from linear regression analysis for continuous variable, and χ2 test for categorical variables, respectively, according to the category of planetary health diet index WSI, Wealth Score Index; SBP, systolic blood pressure; DBP, diastolic blood pressure; HOMA-IR, Homeostasis Model Assessment for Insulin Resistance; LDL, Low-Density Lipoprotein cholesterol; HDL, High-Density Lipoprotein cholesterol; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; GGT, Gamma-glutamyl transferase; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter Higher adherence to the PHDI was significantly associated with lower weight, waist circumference, and body mass index (all P-trend < 0.001). Correspondingly, fasting insulin levels and HOMA-IR decreased progressively across quartiles (both P-trend < 0.001). Favorable lipid profile trends were observed, with reductions in triglycerides, total cholesterol, LDL-C and increases in HDL-C across increasing PHDI quartiles (all P-trend < 0.001). Liver enzymes—including ALT, AST, and GGT—as well as C-reactive protein concentrations, showed significant decreasing trends with higher PHDI adherence (all P-trend < 0.001). Mean CAP values, indicative of liver steatosis, also decreased across quartiles, from 195.33 ± 42.74 dB/m in Q1 to 177.17 ± 40.41 dB/m in Q4 (P-trend < 0.001). Dietary intakes across PHDI quartiles To verify the construct validity of the PHDI in this population, we examined dietary composition across adherence quartiles. This analysis confirmed that the PHDI effectively discriminates between participants with diets that align with the theoretical EAT-Lancet food group recommendations. As shown in Table 2, total energy intake decreased significantly across increasing quartiles of PHDI (P-trend < 0.001). Participants in higher quartiles had a greater percentage of energy from carbohydrates and protein and a lower percentage from fat (all P-trend < 0.001). Table 2Dietary intake of participants across quartiles of planetary health diet index.aQ1 (n = 1085)Q2 (n = 1245)Q3 (n = 1418)Q4 (n = 1310)P-trend bTotal energy intake (Kcal)2257.99 ± 360.382275.12 ± 341.892197.09 ± 319.802134.70 ± 341.67< 0.001Carbohydrate intake (% of energy)60.42 ± 3.6160.57 ± 3.5060.92 ± 3.4261.30 ± 3.78 < 0.001 Protein intake (% of energy)12.93 ± 1.3913.20 ± 1.3413.68 ± 1.2213.71 ± 1.39 < 0.001 Fat intake (% of energy)28.70 ± 4.0228.31 ± 3.7227.56 ± 3.6127.15 ± 4.02 < 0.001 Whole grain (gram/day)11.09 ± 7.6317.67 ± 6.4927.36 ± 9.0623.61 ± 9.66 < 0.001 Refined grain (gram/day)391.56 ± 77.60387.35 ± 74.60367.11 ± 66.91370.89 ± 71.83 < 0.001 Legumes (gram/day)59.22 ± 19.0563.84 ± 20.6669.05 ± 19.0465.59 ± 19.52 < 0.001 Fruits (gram/day)342.43 ± 66.37354.90 ± 62.58363.60 ± 62.02357.28 ± 61.65 < 0.001 Vegetables (gram/day)489.67 ± 104.26500.88 ± 99.79523.57 ± 99.77512.86 ± 100.52 < 0.001 Nuts (gram/day)3.91 ± 4.135.48 ± 4.245.85 ± 3.985.50 ± 3.91 < 0.001 Fish and seafood (gram/day)9.44 ± 5.029.44 ± 5.539.83 ± 5.546.96 ± 5.91 < 0.001 Eggs (gram/day)30.03 ± 13.1831.55 ± 13.5631.07 ± 13.6129.57 ± 14.100.201Dairy (gram/day)97.72 ± 28.18102.71 ± 29.34104.82 ± 29.29102.44 ± 28.07 < 0.001 Cheese (gram/day)23.32 ± 9.5924.05 ± 9.1726.07 ± 8.3825.65 ± 9.12 < 0.001 Vegetable oils (gram/day)7.06 ± 4.926.97 ± 5.286.35 ± 4.954.42 ± 4.90 < 0.001 Animal fats (gram/day)21.36 ± 13.2619.35 ± 11.7916.14 ± 10.8815.82 ± 12.19 < 0.001 Chicken (gram/day)32.92 ± 12.9534.73 ± 13.5835.66 ± 13.5534.28 ± 13.58 0.007 Red meat (gram/day)17.55 ± 8.6617.94 ± 8.5716.05 ± 8.6415.82 ± 8.72 < 0.001 Added sugar (gram/day)48.34 ± 22.3447.75 ± 23.2941.74 ± 21.1136.34 ± 24.37 < 0.001 Trans-fatty acid (% of energy)10.57 ± 1.8010.30 ± 1.6510.06 ± 1.6010.12 ± 1.85 < 0.001 Saturated fatty acid (% of energy)0.25 ± 0.130.24 ± 0.130.21 ± 0.120.22 ± 0.13 < 0.001 Total Monounsaturated Fatty Acids (MUFA) (gram/day)22.56 ± 5.8722.20 ± 5.3720.65 ± 5.0919.37 ± 5.25 < 0.001 Total Polyunsaturated Fatty Acids (PUFA) (gram/day)12.02 ± 3.4012.44 ± 3.3712.01 ± 3.2211.01 ± 3.11 < 0.001 α-Linolenic Acid (ALA) (gram/day)0.63 ± 0.280.63 ± 0.290.58 ± 0.270.52 ± 0.26 < 0.001 Eicosapentaenoic Acid (EPA) (gram/day)0.015 ± 0.0040.016 ± 0.0040.016 ± 0.0040.014 ± 0.004 < 0.001 Docosahexaenoic Acid (DHA) (gram/day)0.042 ± 0.0120.043 ± 0.0130.044 ± 0.0130.039 ± 0.013 < 0.001 Vitamin E (mg/day)8.32 ± 1.728.59 ± 1.778.39 ± 1.737.86 ± 1.67 < 0.001 Vitamin A (IU/day)1048.35 ± 359.701054.60 ± 350.331046.80 ± 347.521042.41 ± 347.11 < 0.001 Vitamin B12 (mcg/day)8.00 ± 3.458.11 ± 3.497.89 ± 3.427.84 ± 3.320.086Choline (mg/day)299.79 ± 53.49309.94 ± 53.57310.36 ± 54.60299.97 ± 56.100.850Niacin (mg/day)21.03 ± 3.5321.64 ± 3.4421.63 ± 3.4121.08 ± 3.550.970Vitamin D (IU/day)58.56 ± 17.8159.74 ± 18.5060.08 ± 18.3651.71 ± 18.76 < 0.001 Vitamin C (mg/day)133.74 ± 27.51137.70 ± 27.60141.28 ± 26.61137.81 ± 26.69 < 0.001 Folate (mcg/day)527.30 ± 78.06546.31 ± 78.96554.12 ± 77.70547.84 ± 79.05 < 0.001 Iron (mg/day)14.63 ± 2.0115.10 ± 1.9815.27 ± 1.9815.01 ± 2.07 < 0.001 Zinc (mg/day)9.53 ± 1.439.78 ± 1.409.69 ± 1.409.54 ± 1.460.567Magnesium (mg/day)308.95 ± 37.54318.37 ± 37.58319.54 ± 36.69315.21 ± 38.35 < 0.001 Calcium (mg/day)846.39 ± 108.07866.00 ± 106.52869.51 ± 107.80858.95 ± 109.20 0.007 Selenium (mcg/day)91.36 ± 17.4494.57 ± 17.2494.54 ± 17.0091.43 ± 17.190.800a Data are presented as mean ± standard deviationb P-value was obtained from linear regression analysis according to the category of planetary health diet index Table 2Dietary intake of participants across quartiles of planetary health diet index.aQ1 (n = 1085)Q2 (n = 1245)Q3 (n = 1418)Q4 (n = 1310)P-trend bTotal energy intake (Kcal)2257.99 ± 360.382275.12 ± 341.892197.09 ± 319.802134.70 ± 341.67< 0.001Carbohydrate intake (% of energy)60.42 ± 3.6160.57 ± 3.5060.92 ± 3.4261.30 ± 3.78 < 0.001 Protein intake (% of energy)12.93 ± 1.3913.20 ± 1.3413.68 ± 1.2213.71 ± 1.39 < 0.001 Fat intake (% of energy)28.70 ± 4.0228.31 ± 3.7227.56 ± 3.6127.15 ± 4.02 < 0.001 Whole grain (gram/day)11.09 ± 7.6317.67 ± 6.4927.36 ± 9.0623.61 ± 9.66 < 0.001 Refined grain (gram/day)391.56 ± 77.60387.35 ± 74.60367.11 ± 66.91370.89 ± 71.83 < 0.001 Legumes (gram/day)59.22 ± 19.0563.84 ± 20.6669.05 ± 19.0465.59 ± 19.52 < 0.001 Fruits (gram/day)342.43 ± 66.37354.90 ± 62.58363.60 ± 62.02357.28 ± 61.65 < 0.001 Vegetables (gram/day)489.67 ± 104.26500.88 ± 99.79523.57 ± 99.77512.86 ± 100.52 < 0.001 Nuts (gram/day)3.91 ± 4.135.48 ± 4.245.85 ± 3.985.50 ± 3.91 < 0.001 Fish and seafood (gram/day)9.44 ± 5.029.44 ± 5.539.83 ± 5.546.96 ± 5.91 < 0.001 Eggs (gram/day)30.03 ± 13.1831.55 ± 13.5631.07 ± 13.6129.57 ± 14.100.201Dairy (gram/day)97.72 ± 28.18102.71 ± 29.34104.82 ± 29.29102.44 ± 28.07 < 0.001 Cheese (gram/day)23.32 ± 9.5924.05 ± 9.1726.07 ± 8.3825.65 ± 9.12 < 0.001 Vegetable oils (gram/day)7.06 ± 4.926.97 ± 5.286.35 ± 4.954.42 ± 4.90 < 0.001 Animal fats (gram/day)21.36 ± 13.2619.35 ± 11.7916.14 ± 10.8815.82 ± 12.19 < 0.001 Chicken (gram/day)32.92 ± 12.9534.73 ± 13.5835.66 ± 13.5534.28 ± 13.58 0.007 Red meat (gram/day)17.55 ± 8.6617.94 ± 8.5716.05 ± 8.6415.82 ± 8.72 < 0.001 Added sugar (gram/day)48.34 ± 22.3447.75 ± 23.2941.74 ± 21.1136.34 ± 24.37 < 0.001 Trans-fatty acid (% of energy)10.57 ± 1.8010.30 ± 1.6510.06 ± 1.6010.12 ± 1.85 < 0.001 Saturated fatty acid (% of energy)0.25 ± 0.130.24 ± 0.130.21 ± 0.120.22 ± 0.13 < 0.001 Total Monounsaturated Fatty Acids (MUFA) (gram/day)22.56 ± 5.8722.20 ± 5.3720.65 ± 5.0919.37 ± 5.25 < 0.001 Total Polyunsaturated Fatty Acids (PUFA) (gram/day)12.02 ± 3.4012.44 ± 3.3712.01 ± 3.2211.01 ± 3.11 < 0.001 α-Linolenic Acid (ALA) (gram/day)0.63 ± 0.280.63 ± 0.290.58 ± 0.270.52 ± 0.26 < 0.001 Eicosapentaenoic Acid (EPA) (gram/day)0.015 ± 0.0040.016 ± 0.0040.016 ± 0.0040.014 ± 0.004 < 0.001 Docosahexaenoic Acid (DHA) (gram/day)0.042 ± 0.0120.043 ± 0.0130.044 ± 0.0130.039 ± 0.013 < 0.001 Vitamin E (mg/day)8.32 ± 1.728.59 ± 1.778.39 ± 1.737.86 ± 1.67 < 0.001 Vitamin A (IU/day)1048.35 ± 359.701054.60 ± 350.331046.80 ± 347.521042.41 ± 347.11 < 0.001 Vitamin B12 (mcg/day)8.00 ± 3.458.11 ± 3.497.89 ± 3.427.84 ± 3.320.086Choline (mg/day)299.79 ± 53.49309.94 ± 53.57310.36 ± 54.60299.97 ± 56.100.850Niacin (mg/day)21.03 ± 3.5321.64 ± 3.4421.63 ± 3.4121.08 ± 3.550.970Vitamin D (IU/day)58.56 ± 17.8159.74 ± 18.5060.08 ± 18.3651.71 ± 18.76 < 0.001 Vitamin C (mg/day)133.74 ± 27.51137.70 ± 27.60141.28 ± 26.61137.81 ± 26.69 < 0.001 Folate (mcg/day)527.30 ± 78.06546.31 ± 78.96554.12 ± 77.70547.84 ± 79.05 < 0.001 Iron (mg/day)14.63 ± 2.0115.10 ± 1.9815.27 ± 1.9815.01 ± 2.07 < 0.001 Zinc (mg/day)9.53 ± 1.439.78 ± 1.409.69 ± 1.409.54 ± 1.460.567Magnesium (mg/day)308.95 ± 37.54318.37 ± 37.58319.54 ± 36.69315.21 ± 38.35 < 0.001 Calcium (mg/day)846.39 ± 108.07866.00 ± 106.52869.51 ± 107.80858.95 ± 109.20 0.007 Selenium (mcg/day)91.36 ± 17.4494.57 ± 17.2494.54 ± 17.0091.43 ± 17.190.800a Data are presented as mean ± standard deviationb P-value was obtained from linear regression analysis according to the category of planetary health diet index Dietary intake of participants across quartiles of planetary health diet index.a a Data are presented as mean ± standard deviation b P-value was obtained from linear regression analysis according to the category of planetary health diet index Higher PHDI scores were characterized by greater intakes of whole grains, legumes, fruits, vegetables, nuts, dairy, and cheese (all P-trend < 0.001), and lower intakes of refined grains, animal fats, red meat, added sugars, trans-fatty acids, saturated fatty acids, monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA), alpha-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA) (all P-trend < 0.001). Fish and seafood intake decreased in Q4 compared with Q3 (P-trend < 0.001), while egg intake did not differ significantly across quartiles (P-trend = 0.201). Micronutrient intakes including vitamin C, folate, iron, magnesium, and calcium increased significantly across quartiles (all P-trend < 0.01). Vitamin D intake decreased in the highest quartile (P-trend < 0.001). No significant trends were observed for vitamin B12, choline, niacin, zinc, or selenium (all P-trend > 0.05). Association between PHDI and MASLD incidence During follow-up, 562 incident cases of MASLD were identified, yielding a 5-year cumulative incidence of 11.1%. The number of cases across increasing quartiles of PHDI was 197/1085 (Q1), 164/1245 (Q2), 92/1418 (Q3), and 109/1310 (Q4) (Table 3). Table 3Odds ratios (95% confidence intervals) for MASLD incidence according to the proportion of planetary health diet indexQ1Q2Q3Q4P-trendContinuous (Per SD increment)P-valueMedian Score44485157Cases/populations197/ 1085164/ 124592/ 1418109/ 1310Crude model1.00 (Ref)0.68 (0.55, 0.86)0.31 (0.24, 0.41)0.41 (0.32, 0.53) < 0.001 0.66 (0.60, 0.73) < 0.001 Model 1a1.00 (Ref)0.71 (0.56, 0.89)0.33 (0.25, 0.43)0.43 (0.33, 0.55) < 0.001 0.67 (0.61, 0.74) < 0.001 Model 2b1.00 (Ref)0.85 (0.66, 1.09)0.49 (0.37, 0.66)0.51 (0.39, 0.67) < 0.001 0.75 (0.68, 0.83) < 0.001 Model 3c1.00 (Ref)0.84 (0.65, 1.08)0.51 (0.38, 0.68)0.51 (0.39, 0.68) < 0.001 0.76 (0.68, 0.84) < 0.001 Model 4d1.00 (Ref)0.89 (0.69, 1.15)0.61 (0.45, 0.81)0.66 (0.50, 0.88) < 0.001 0.84 (0.76, 0.94) 0.002 MASLD: Metabolic dysfunction-associated steatotic liver diseaseSignificant p-values are highlighted in bolda Model 1 adjusted for age and sexb Model 2 additionally adjusted for Model 1 and physical activity, wealth score index, education level, smoking, and fat mass indexc Model 3 additionally adjusted for Model 2 and hypertension, and type 2 diabetes mellitusd Model 4 additionally adjusted for Model 3 and total energy intake Table 3Odds ratios (95% confidence intervals) for MASLD incidence according to the proportion of planetary health diet indexQ1Q2Q3Q4P-trendContinuous (Per SD increment)P-valueMedian Score44485157Cases/populations197/ 1085164/ 124592/ 1418109/ 1310Crude model1.00 (Ref)0.68 (0.55, 0.86)0.31 (0.24, 0.41)0.41 (0.32, 0.53) < 0.001 0.66 (0.60, 0.73) < 0.001 Model 1a1.00 (Ref)0.71 (0.56, 0.89)0.33 (0.25, 0.43)0.43 (0.33, 0.55) < 0.001 0.67 (0.61, 0.74) < 0.001 Model 2b1.00 (Ref)0.85 (0.66, 1.09)0.49 (0.37, 0.66)0.51 (0.39, 0.67) < 0.001 0.75 (0.68, 0.83) < 0.001 Model 3c1.00 (Ref)0.84 (0.65, 1.08)0.51 (0.38, 0.68)0.51 (0.39, 0.68) < 0.001 0.76 (0.68, 0.84) < 0.001 Model 4d1.00 (Ref)0.89 (0.69, 1.15)0.61 (0.45, 0.81)0.66 (0.50, 0.88) < 0.001 0.84 (0.76, 0.94) 0.002 MASLD: Metabolic dysfunction-associated steatotic liver diseaseSignificant p-values are highlighted in bolda Model 1 adjusted for age and sexb Model 2 additionally adjusted for Model 1 and physical activity, wealth score index, education level, smoking, and fat mass indexc Model 3 additionally adjusted for Model 2 and hypertension, and type 2 diabetes mellitusd Model 4 additionally adjusted for Model 3 and total energy intake Odds ratios (95% confidence intervals) for MASLD incidence according to the proportion of planetary health diet index MASLD: Metabolic dysfunction-associated steatotic liver disease Significant p-values are highlighted in bold a Model 1 adjusted for age and sex b Model 2 additionally adjusted for Model 1 and physical activity, wealth score index, education level, smoking, and fat mass index c Model 3 additionally adjusted for Model 2 and hypertension, and type 2 diabetes mellitus d Model 4 additionally adjusted for Model 3 and total energy intake In the crude model, higher PHDI was inversely associated with odds of MASLD. Compared with Q1, the odds ratios (ORs) (95% CIs) were 0.68 (0.55–0.86) for Q2, 0.31 (0.24–0.41) for Q3, and 0.41 (0.32–0.53) for Q4 (P-trend < 0.001). After adjustment for age and sex (Model 1), the associations remained largely unchanged. Further adjustment for lifestyle and socioeconomic factors (Model 2), as well as hypertension and type 2 diabetes mellitus (Model 3), slightly attenuated the associations; however, significant inverse associations persisted for Q3 and Q4. In the fully adjusted model (Model 4), additionally controlling for total energy intake, participants in Q3 and Q4 had 39% (OR: 0.61; 95% CI: 0.45–0.81) and 34% (OR: 0.66; 95% CI: 0.50–0.88) lower odds of MASLD, respectively, compared with Q1 (P-trend < 0.001). Each one standard deviation increase in PHDI score was associated with a 16% lower odds of MASLD (OR: 0.84; 95% CI: 0.76–0.94; P = 0.002). Figure 2 shows a strong inverse association between the PHDI and MASLD risk (p < 0.01). Higher PHDI scores correspond to lower odds of MASLD, with a largely linear inverse relationship (P for non-linearity = 0.08) and a suggestive plateau at higher PHDI scores. Fig. 2Multivariable adjusted restricted cubic spline curve for the association between the planetary health diet index (PHDI) and the odds of metabolic dysfunction–associated steatotic liver disease (MASLD). The solid line represents the odds ratio, and the shaded area represents the 95% confidence interval. The model was adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index, hypertension, type 2 diabetes mellitus, and total energy intake. Knots were placed at the 5th, 35th, 65th, and 95th percentiles of the PHDI distribution. The reference value is set at the 10th percentile of PHDI. The y-axis is displayed on a logarithmic scale Multivariable adjusted restricted cubic spline curve for the association between the planetary health diet index (PHDI) and the odds of metabolic dysfunction–associated steatotic liver disease (MASLD). The solid line represents the odds ratio, and the shaded area represents the 95% confidence interval. The model was adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index, hypertension, type 2 diabetes mellitus, and total energy intake. Knots were placed at the 5th, 35th, 65th, and 95th percentiles of the PHDI distribution. The reference value is set at the 10th percentile of PHDI. The y-axis is displayed on a logarithmic scale Supplementary Table S3 presents the continuous associations between each PHDI component and the risk of MASLD. In fully adjusted models, significant inverse associations were found for legumes (OR = 0.42), whole grains (OR = 0.69), fish and seafood (OR = 0.77), total vegetables (OR = 0.49), fruits (OR = 0.73), dairy (OR = 0.88), vegetable oils (OR = 0.73), poultry (OR = 0.65), dark green (OR = 0.66), and red and orange vegetables (OR = 0.66) (all p < 0.001 except dairy p = 0.032). Positive associations were observed for tubers (OR = 1.30), red meat (OR = 1.33), and added sugars (OR = 1.40) (all p < 0.001). Nuts, eggs, and animal fats showed no significant associations. Subgroup analyses Subgroup analyses (Table 4) revealed significant effect modification by age, sex, BMI, wealth score index, and physical activity (all P for interaction < 0.05). The inverse association between PHDI and MASLD was more evident among participants aged < 45 years, men, and those with BMI < 25 kg/m². Table 4Subgroup analysis of the association between planetary health diet index and the odds of MASLD.aQ1Q2Q3Q4P-trendP for interaction Age < 45 years old1.00 (Ref)0.87 (0.60, 1.25)0.52 (0.34, 0.80)0.61 (0.41, 0.91) < 0.001 0.021 > 45years old1.00 (Ref)0.90 (0.62, 1.29)0.72 (0.47, 1.08)0.73 (0.49, 1.09) 0.003 Sex Men1.00 (Ref)0.94 (0.65, 1.35)0.61 (0.40, 0.94)0.62 (0.40, 0.96) 0.007 < 0.001 Women1.00 (Ref)0.83 (0.57, 1.21)0.62 (0.41, 0.94)0.71 (0.48, 1.05) 0.038 Body mass index < 25 kg/m21.00 (Ref)0.74 (0.39, 1.39)0.52 (0.27, 1.00)0.39 (0.19, 0.79) 0.004 < 0.001 > 25 kg/m21.00 (Ref)0.92 (0.69, 1.22)0.65 (0.47, 0.90)0.75 (0.55, 1.03) 0.019 Smoking Yes1.00 (Ref)0.57 (0.27, 1.20)0.37 (0.16, 0.88)0.29 (0.12, 0.68) 0.002 0.773No1.00 (Ref)0.94 (0.71, 1.24)0.64 (0.47, 0.88)0.74 (0.54, 1.00) 0.009 Hypertension Yes1.00 (Ref)0.80 (0.47, 1.38)0.80 (0.47, 1.38)0.45 (0.24, 0.85) 0.007 0.800No1.00 (Ref)0.91 (0.68, 1.22)0.61 (0.43, 0.85)0.73 (0.53, 1.01) 0.011 Type 2 diabetes mellitus Yes1.00 (Ref)1.68 (0.87, 3.24)1.03 (0.46, 2.29)0.69 (0.31, 1.53)0.265 < 0.001 No1.00 (Ref)0.77 (0.58, 1.02)0.54 (0.39, 0.75)0.65 (0.48, 0.88) 0.001 Wealth score index < Median (68.86)1.00 (Ref)0.82 (0.53, 1.26)0.62 (0.39, 0.99)0.70 (0.44, 1.11) 0.003 0.006 > Median (68.86)1.00 (Ref)0.94 (0.68, 1.30)0.59 (0.40, 0.86)0.65 (0.45, 0.93)0.063 Physical activity < Median (23.74)1.00 (Ref)0.96 (0.72, 1.30)0.63 (0.44, 0.88)0.68 (0.49, 0.94) 0.003 < 0.001 > Median (23.74)1.00 (Ref)0.81 (0.47, 1.37)0.58 (0.32, 1.04)0.67 (0.37, 1.19)0.089MASLD: Metabolic dysfunction-associated steatotic liver diseasea Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake. For subgroup analyses, the stratifying variable was included as a main effect with a product (interaction) term, and the corresponding continuous covariate was omitted from the adjustment set for that subgroup to avoid over-adjustment (e.g., age was not included when stratifying by age) Table 4Subgroup analysis of the association between planetary health diet index and the odds of MASLD.aQ1Q2Q3Q4P-trendP for interaction Age < 45 years old1.00 (Ref)0.87 (0.60, 1.25)0.52 (0.34, 0.80)0.61 (0.41, 0.91) < 0.001 0.021 > 45years old1.00 (Ref)0.90 (0.62, 1.29)0.72 (0.47, 1.08)0.73 (0.49, 1.09) 0.003 Sex Men1.00 (Ref)0.94 (0.65, 1.35)0.61 (0.40, 0.94)0.62 (0.40, 0.96) 0.007 < 0.001 Women1.00 (Ref)0.83 (0.57, 1.21)0.62 (0.41, 0.94)0.71 (0.48, 1.05) 0.038 Body mass index < 25 kg/m21.00 (Ref)0.74 (0.39, 1.39)0.52 (0.27, 1.00)0.39 (0.19, 0.79) 0.004 < 0.001 > 25 kg/m21.00 (Ref)0.92 (0.69, 1.22)0.65 (0.47, 0.90)0.75 (0.55, 1.03) 0.019 Smoking Yes1.00 (Ref)0.57 (0.27, 1.20)0.37 (0.16, 0.88)0.29 (0.12, 0.68) 0.002 0.773No1.00 (Ref)0.94 (0.71, 1.24)0.64 (0.47, 0.88)0.74 (0.54, 1.00) 0.009 Hypertension Yes1.00 (Ref)0.80 (0.47, 1.38)0.80 (0.47, 1.38)0.45 (0.24, 0.85) 0.007 0.800No1.00 (Ref)0.91 (0.68, 1.22)0.61 (0.43, 0.85)0.73 (0.53, 1.01) 0.011 Type 2 diabetes mellitus Yes1.00 (Ref)1.68 (0.87, 3.24)1.03 (0.46, 2.29)0.69 (0.31, 1.53)0.265 < 0.001 No1.00 (Ref)0.77 (0.58, 1.02)0.54 (0.39, 0.75)0.65 (0.48, 0.88) 0.001 Wealth score index < Median (68.86)1.00 (Ref)0.82 (0.53, 1.26)0.62 (0.39, 0.99)0.70 (0.44, 1.11) 0.003 0.006 > Median (68.86)1.00 (Ref)0.94 (0.68, 1.30)0.59 (0.40, 0.86)0.65 (0.45, 0.93)0.063 Physical activity < Median (23.74)1.00 (Ref)0.96 (0.72, 1.30)0.63 (0.44, 0.88)0.68 (0.49, 0.94) 0.003 < 0.001 > Median (23.74)1.00 (Ref)0.81 (0.47, 1.37)0.58 (0.32, 1.04)0.67 (0.37, 1.19)0.089MASLD: Metabolic dysfunction-associated steatotic liver diseasea Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake. For subgroup analyses, the stratifying variable was included as a main effect with a product (interaction) term, and the corresponding continuous covariate was omitted from the adjustment set for that subgroup to avoid over-adjustment (e.g., age was not included when stratifying by age) Subgroup analysis of the association between planetary health diet index and the odds of MASLD.a MASLD: Metabolic dysfunction-associated steatotic liver disease a Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake. For subgroup analyses, the stratifying variable was included as a main effect with a product (interaction) term, and the corresponding continuous covariate was omitted from the adjustment set for that subgroup to avoid over-adjustment (e.g., age was not included when stratifying by age) Regarding wealth status, the association differed by wealth score index (P for interaction = 0.006). Although the magnitude of the inverse association in the highest quartile appeared slightly stronger among participants with higher wealth score index (Q4 vs. Q1: OR = 0.65), the linear trend across quartiles was statistically significant only among those with lower wealth score index (P-trend = 0.003). Regarding physical activity, the protective association was observed primarily among less active individuals (below median MET-h/week; P-trend = 0.003), while no significant association was detected in the more active subgroup (P-trend = 0.089; P for interaction < 0.001). No significant interaction was observed for smoking or hypertension. While the inverse association was significant among participants without type 2 diabetes mellitus, no significant association was observed among those with diabetes (P for interaction < 0.001). Sensitivity analyses Sensitivity analyses (Table 5) generally supported the robustness of the findings. Excluding participants with > 10% weight gain, higher PHDI remained significantly associated with lower MASLD odds (P-trend = 0.013; per SD OR: 0.84, 95% CI: 0.71–0.99). Adjustments for baseline triglyceride-to-HDL ratio, dietary saturated fatty acids, and dietary sugar intake also maintained significant inverse associations (all P-trend ≤ 0.002). However, additional adjustment for baseline HOMA-IR or C-reactive protein attenuated the associations and made them non-significant (P-trend > 0.05), indicating that insulin resistance and systemic inflammation may partially mediate the relationship between PHDI and MASLD. Table 5Sensitivity analysis for developing MASLD according to the proportion of planetary health diet index. aQ1Q2Q3Q4P-trendContinuous (Per SD increment)P-valueExcluding participants with weight gain of > 10% from baseline1.00 (Ref)1.00 (0.71, 1.40)0.63 (0.43, 0.94)0.70 (0.48, 1.02)0.0130.84 (0.71, 0.99)0.041Additional adjusted for baseline CAP score1.00 (Ref)1.06 (0.80, 1.40)0.86 (0.62, 1.18)0.80 (0.60, 1.08)0.0850.92 (0.82, 1.03)0.143Additional adjusted for baseline HOMA-IR1.00 (Ref)1.10 (0.83, 1.46)0.93 (0.67, 1.29)0.88 (0.65, 1.21)0.3340.95 (0.85, 1.07)0.401Additional adjusted for baseline TG to HDL ratio1.00 (Ref)0.89 (0.69, 1.16)0.62 (0.46, 0.83)0.66 (0.49, 0.87) < 0.001 0.84 (0.76, 0.94) 0.002 Additional adjusted for baseline C-reactive protein1.00 (Ref)1.25 (0.93, 1.67)0.99 (0.71, 1.37)0.87 (0.63, 1.19)0.2540.95 (0.85, 1.07)0.417Additional adjusted for dietary saturated fatty acid1.00 (Ref)0.88 (0.68, 1.14)0.60 (0.45, 0.81)0.66 (0.50, 0.88) < 0.001 0.84 (0.76, 0.94) 0.002 Additional adjusted for dietary fiber1.00 (Ref)1.05 (0.80–1.37)0.85 (0.62–1.16)0.88 (0.66–1.19)0.270.95 (0.86–1.06)0.39Additional adjusted for dietary sugar1.00 (Ref)0.90 (0.69, 1.16)0.63 (0.47, 0.85)0.70 (0.52, 0.93) 0.002 0.86 (0.78, 0.96) 0.008 MASLD: Metabolic dysfunction-associated steatotic liver diseaseSignificant p-values are highlighted in bolda Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake Table 5Sensitivity analysis for developing MASLD according to the proportion of planetary health diet index. aQ1Q2Q3Q4P-trendContinuous (Per SD increment)P-valueExcluding participants with weight gain of > 10% from baseline1.00 (Ref)1.00 (0.71, 1.40)0.63 (0.43, 0.94)0.70 (0.48, 1.02)0.0130.84 (0.71, 0.99)0.041Additional adjusted for baseline CAP score1.00 (Ref)1.06 (0.80, 1.40)0.86 (0.62, 1.18)0.80 (0.60, 1.08)0.0850.92 (0.82, 1.03)0.143Additional adjusted for baseline HOMA-IR1.00 (Ref)1.10 (0.83, 1.46)0.93 (0.67, 1.29)0.88 (0.65, 1.21)0.3340.95 (0.85, 1.07)0.401Additional adjusted for baseline TG to HDL ratio1.00 (Ref)0.89 (0.69, 1.16)0.62 (0.46, 0.83)0.66 (0.49, 0.87) < 0.001 0.84 (0.76, 0.94) 0.002 Additional adjusted for baseline C-reactive protein1.00 (Ref)1.25 (0.93, 1.67)0.99 (0.71, 1.37)0.87 (0.63, 1.19)0.2540.95 (0.85, 1.07)0.417Additional adjusted for dietary saturated fatty acid1.00 (Ref)0.88 (0.68, 1.14)0.60 (0.45, 0.81)0.66 (0.50, 0.88) < 0.001 0.84 (0.76, 0.94) 0.002 Additional adjusted for dietary fiber1.00 (Ref)1.05 (0.80–1.37)0.85 (0.62–1.16)0.88 (0.66–1.19)0.270.95 (0.86–1.06)0.39Additional adjusted for dietary sugar1.00 (Ref)0.90 (0.69, 1.16)0.63 (0.47, 0.85)0.70 (0.52, 0.93) 0.002 0.86 (0.78, 0.96) 0.008 MASLD: Metabolic dysfunction-associated steatotic liver diseaseSignificant p-values are highlighted in bolda Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake Sensitivity analysis for developing MASLD according to the proportion of planetary health diet index. a MASLD: Metabolic dysfunction-associated steatotic liver disease Significant p-values are highlighted in bold a Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs). Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, fat mass index hypertension, type 2 diabetes mellitus, and total energy intake Mediation analysis Mediation analysis was conducted to evaluate the extent to which biochemical and dietary factors mediate the association between the PHDI and the odds of MASLD, adjusting for potential confounders (Table 6). The mediated effect through the controlled attenuation parameter (CAP) was significant, with an indirect effect estimate of -0.00152 (95% CI: -0.00247, -0.00071; P < 0.001), accounting for 53.8% of the total effect of PHDI on MASLD odds. The direct effect through pathways other than CAP was not statistically significant (P = 0.14). Table 6The mediation effect of biochemical and dietary variables on the association between planetary health diet index and the odds of MASLD.aMediatorDirect effectIndirect effect (Mediated)Total effectProportion mediated (%)CAPEstimate-0.00130-0.00152-0.0028253.895% CI(-0.00380, 0.00032)(-0.00247, -0.00071)(-0.00586, -0.00071)P-value0.14 < 0.001 < 0.01 < 0.01 HOMA-IREstimate-0.00070-0.00160-0.0023169.395% CI(-0.00317, 0.00076)(-0.00259, -0.00075)(-0.00537, -0.00030)P-value0.42 < 0.001 0.01 0.01 CRPEstimate-0.00122-0.00096-0.0021844.095% CI(-0.00427, 0.00048)(-0.00171, -0.00035)(-0.00565, -0.00012)P-value0.22 < 0.001 0.03 0.03 Dietary fiberEstimate-0.00097-0.00244-0.0034271.495% CI(-0.00712, 0.00090)(-0.00341, -0.00157)(-0.00721, -0.00082)P-value0.39 < 0.001 < 0.01 < 0.01 a Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, body mass index, hypertension, type 2 diabetes mellitus, and total energy intakeHOMA-IR, Homeostasis Model Assessment for Insulin Resistance; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter Table 6The mediation effect of biochemical and dietary variables on the association between planetary health diet index and the odds of MASLD.aMediatorDirect effectIndirect effect (Mediated)Total effectProportion mediated (%)CAPEstimate-0.00130-0.00152-0.0028253.895% CI(-0.00380, 0.00032)(-0.00247, -0.00071)(-0.00586, -0.00071)P-value0.14 < 0.001 < 0.01 < 0.01 HOMA-IREstimate-0.00070-0.00160-0.0023169.395% CI(-0.00317, 0.00076)(-0.00259, -0.00075)(-0.00537, -0.00030)P-value0.42 < 0.001 0.01 0.01 CRPEstimate-0.00122-0.00096-0.0021844.095% CI(-0.00427, 0.00048)(-0.00171, -0.00035)(-0.00565, -0.00012)P-value0.22 < 0.001 0.03 0.03 Dietary fiberEstimate-0.00097-0.00244-0.0034271.495% CI(-0.00712, 0.00090)(-0.00341, -0.00157)(-0.00721, -0.00082)P-value0.39 < 0.001 < 0.01 < 0.01 a Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, body mass index, hypertension, type 2 diabetes mellitus, and total energy intakeHOMA-IR, Homeostasis Model Assessment for Insulin Resistance; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter The mediation effect of biochemical and dietary variables on the association between planetary health diet index and the odds of MASLD.a a Models were adjusted for age, sex, physical activity, wealth score index, education level, smoking, body mass index, hypertension, type 2 diabetes mellitus, and total energy intake HOMA-IR, Homeostasis Model Assessment for Insulin Resistance; CRP, C-reactive protein; CAP, Controlled Attenuation Parameter HOMA-IR showed a substantial mediation effect, with the indirect effect estimate of -0.00160 (95% CI: -0.00259, -0.00075; P < 0.001) and a proportion mediated of 69.3%. Similar to CAP, the direct effect was not significant (P = 0.42), underscoring the importance of insulin resistance as a mediator. CRP also significantly mediated the association, with an indirect effect of -0.00096 (95% CI: -0.00171, -0.00035; P < 0.001), representing 44.0% mediation. The direct effect was non-significant (P = 0.22). Dietary fiber demonstrated the largest mediation effect among the variables studied, with an indirect effect of -0.00244 (95% CI: -0.00341, -0.00157; P < 0.001), mediating 71.4% of the total effect. The direct effect was not significant (P = 0.39), indicating that dietary fiber substantially contributes to the protective association between PHDI and MASLD. Discussion In this large population-based study of Iranian adults, it is found that higher adherence to the PHDI was significantly associated with a reduced risk of incident MASLD. Over a 5-year follow-up period, participants in the highest quartiles of PHDI adherence exhibited a 34% lower odds of developing MASLD compared to those in the lowest quartile, after adjusting for confounders. Each standard deviation increase in PHDI score was associated with a 16% reduction in MASLD risk. These inverse associations were consistent across several sensitivity analyses but were partially attenuated after adjustment for biomarkers of insulin resistance and inflammation, suggesting these pathways may mediate the protective effect. Notably, the inverse association was more pronounced among participants aged < 45 years, men, those with BMI < 25 kg/m², lower physical activity, and higher wealth status, and was significant only among those without type 2 diabetes. Similar studies from Western populations support our findings. Zhang et al. conducted a large cohort study that examined adherence to the ELD using three indices: the PHDI by Willett et al., the EAT-Lancet index (ELI) by Stubbendorff et al., and the ELD-I by Kesse-Guyot et al. They found that higher scores across these indices significantly reduced the risk of MASLD after a median follow-up of 10.5 years [9]. In another study by Wu et al. using the UK Biobank, individuals in the highest tertile of ELD adherence had a 33% lower risk of MASLD after 13.3 years. This study utilized the EAT-Lancet Diet Score (ELDS) by Knuppel et al. and demonstrated that the protective effect of the ELD was independent of genetic predisposition, reinforcing that the diet itself is beneficial, rather than the population’s genetics [10]. However, both studies used the same dataset, highlighting a need for further research in different populations. Restricted cubic spline analysis revealed a significant inverse dose–response relationship between PHDI and MASLD odds, with the curve showing a steep linear decline across lower-to-moderate PHDI scores and a suggestive plateau at higher adherence levels (PHDI ≥ ~ 45), though this departure from linearity did not reach statistical significance (P = 0.08). A similar linear dose–response pattern was reported by both Zhang et al. [9] and Wu et al. [10], who found no significant non-linearity between plant-based or EAT-Lancet diet scores and MASLD risk. Collectively, these findings suggest a graded hepatoprotective effect of planetary health–aligned dietary patterns, with the possibility of attenuating marginal benefit at very high levels of adherence. Various indices have been developed to evaluate adherence to the ELD, which can be categorized into three main groups. Binary scoring is a straightforward “pass or fail” system where participants receive 1 point for meeting a specific intake recommendation and 0 points for not meeting it. This method offers a simplified perspective that is useful for broad surveys. In contrast, proportional scoring assigns points based on how closely a person’s intake aligns with the recommended range (for example, consuming half the target yields half the points). This approach captures greater dietary variability and is advantageous for research focused on precision. Graded scoring, a type of proportional system, utilizes a fixed, bounded scale (e.g., 0 to 10 points per food component) to reward partial adherence with more nuance than binary scoring. This strikes a balance between detail and a structured framework [21]. For instance, the EAT-Lancet Index (ELI) by Stubbendorff et al. employs graded scoring, the ELD-I by Kesse-Guyot et al. uses proportional scoring, and the EAT-Lancet Diet Score (ELDS) by Knuppel et al. is based on binary scoring. Consequently, it can be concluded that the ELD is beneficial for MASLD, irrespective of the index utilized; however, some indices demonstrate a stronger association, which may depend on the population and the scoring system used. The observed trends in dietary intake across increasing PHDI quartiles are consistent with the index’s scoring structure. Higher scores reward greater consumption of “Adequacy” components (whole grains, legumes, fruits, vegetables, nuts, dairy), explaining their positive trends, while penalizing “Moderation” components (red meat, animal fats, added sugars), accounting for their decline. These food shifts naturally drive the increased energy from carbohydrates and protein and the reduced fat intake. Specific findings, such as the plateau in fish intake and stable vitamin B12 and zinc levels, reflect the PHDI’s emphasis on moderate, balanced consumption—where high fish intake is penalized, and dairy compensates for nutrients lost from reduced meat intake. Notably, the decline in vitamin D in higher quartiles corresponds with reduced fish intake, a key source of this vitamin. Concerns about potential shortfalls of vitamin A, B12, calcium, iron, and zinc due to limited animal food access in ELD exist, but these do not undermine its role in a sustainable diet [11]. Efforts to address these shortfalls include enhancing the bioavailability of nutrients from plant-based foods, promoting indigenous practices, fortification, cultural inclusiveness, and integrating a One Health approach [22]. Various dietary animal models are used to study MASLD, emphasizing how nutritional deficiencies replicate human pathology. A high-fat diet disrupts metabolic pathways, activating sterol regulatory element-binding proteins (SREBPs), impairing AMP-activated protein kinase (AMPK) signaling, and triggering inflammation via nuclear factor-kappa B (NF-κB), leading to liver damage [23]. Choline is crucial for fat export from the liver; its absence causes fat accumulation in hepatocytes, contributing to fibrosis. Other models, like the high cholesterol diet and methionine-choline deficiency, help study inflammation and unique cell death pathways [23]. Dietary therapy for MASLD includes various approaches, ranging from structured eating patterns—like the Mediterranean diet, ketogenic diet, and intermittent fasting—to targeted supplementation with substances like coffee and vitamins, as well as microbiome modulation through prebiotics and probiotics. These strategies collectively aim to improve metabolic profiles, reduce inflammation, and reverse liver damage by regulating lipid metabolism, restoring the gut-liver axis, and inducing epigenetic modifications [23, 24]. This study provides novel, detailed insight into the biological pathways linking a sustainable diet to MASLD through comprehensive mediation analyses. A substantial proportion of the association was transmitted through metabolic, inflammatory, and dietary mechanisms, with insulin resistance and dietary fiber emerging as dominant mediators. HOMA‑IR accounted for nearly 70% of the total effect, consistent with the central role of impaired insulin signaling in hepatic de novo lipogenesis and MASLD development [25, 26]. The finding that dietary fiber mediated over 70% of the association aligns with its established roles in improving glycemic control, enhancing insulin sensitivity, and promoting beneficial short‑chain fatty acid production via the gut microbiota [27, 28]. Systemic inflammation, reflected by circulating CRP, mediated approximately 44% of the association, further supporting the contribution of chronic low‑grade inflammation to metabolic liver disease [25]. In addition, hepatic steatosis severity assessed by CAP accounted for more than half of the total mediated effect, suggesting that adherence to this dietary pattern may influence MASLD risk partly through mechanisms that directly limit hepatic fat deposition. Importantly, once these mediating pathways were taken into account, the direct association between PHDI and MASLD was no longer statistically significant, indicating that the protective effect of this dietary pattern may operate primarily through improvements in metabolic homeostasis, inflammatory regulation, and dietary quality. It should also be acknowledged that these mediators are biologically interconnected and may represent overlapping pathways rather than independent mechanisms. For instance, increased dietary fiber intake can improve insulin sensitivity and attenuate systemic inflammation [27, 28], while insulin resistance itself contributes to hepatic lipid accumulation and inflammatory activation [25]. Therefore, the estimated mediation proportions should not be interpreted as additive but rather as reflecting intertwined metabolic processes through which plant‑forward, sustainable dietary patterns such as the EAT–Lancet diet may exert beneficial effects on liver health. The favorable influence of the ELD on reducing inflammation has prompted further studies to explore the mediating role of inflammation in the association between ELD and chronic diseases across multiple organ systems. Although specific studies on inflammatory biomarkers linking ELD to MASLD are currently lacking, evidence from other chronic conditions—including chronic kidney disease, lung cancer, inflammatory bowel disease, and heart failure—consistently demonstrates that inflammatory pathways partially mediate the health benefits of this dietary pattern [29–33]. These findings collectively support the biological plausibility of our mediation results. Our subgroup analyses revealed several noteworthy patterns. The diet was particularly protective for younger individuals (under 45 years) and those with a normal weight (BMI < 25 kg/m²), groups that typically have a lower baseline risk for MASLD. This may reflect a window of opportunity wherein dietary interventions are most effective before the establishment of advanced metabolic dysregulation. The more pronounced benefit in men aligns with the higher prevalence and risk of MASLD in males [34], suggesting a greater potential for risk reduction through dietary modification. The protective effect was observed exclusively among participants without type 2 diabetes, which is consistent with the notion that advanced, pharmacologically managed insulin resistance represents a state less amenable to dietary reversal alone [35]. The stronger inverse association among less physically active individuals suggests that a healthy diet may compensate for lower activity levels, whereas among those who are already highly active, diet may offer limited additional benefit due to an already low baseline risk. Interestingly, the inverse association was statistically significant only among participants with lower wealth status. This finding, in line with evidence that sustainable food baskets can be more affordable than current Iranian dietary patterns [36], suggests that the ELD may serve as an accessible and equitable prevention strategy for disadvantaged populations. Strengths and Limitations To our knowledge, this is the first prospective study in the Middle East, benefiting from a large cohort design that enables the examination of the long-term impact of a sustainable diet on metabolic health. Additionally, it uniquely analyzes the mediating effect of common inflammatory markers in the relationship between ELD and MASLD. By adopting a broader concept of MASLD, rather than relying solely on traditional definitions such as fatty liver or NAFLD, this research presents a significant advancement. However, it is important to acknowledge the limitations of the study. Dietary intakes were assessed using an FFQ, which may be subject to recall bias. Although the PHDI is classified as a proportional index that benefits from scoring all dietary intakes based on their degree of adherence to recommendations, future research could explore different diet quality indices to determine which scoring system better predicts the incidence of chronic diseases (21). Additionally, alcohol consumption and drug use were not recorded in the data, which prevented us from adjusting for these important confounders. Furthermore, this cohort study collected data exclusively from urban areas, and the use of convenience sampling may limit generalizability to rural populations and introduce selection bias. Considering the substantial differences between urban and rural regions in Iran (37), further studies are recommended to examine the interaction between urban and rural residency in relation to these associations. The absence of interim follow-up assessments precluded the exclusion of early incident cases, which limits our ability to completely rule out reverse causality. However, the exclusion of participants with substantial weight gain (> 10%) partially mitigates this concern. Finally, the concurrent measurement of exposure and mediators at baseline in the mediation analysis assumes that these mediators represent a stable intermediate physiological state, which is an inherent limitation of single-time-point mediation models. Conclusion Greater adherence to the EAT–Lancet dietary pattern, reflected by higher PHDI scores, was associated with a significantly lower risk of incident MASLD over five years among Iranian adults, with evidence of a largely linear inverse dose–response relationship, although a suggestive plateau was observed at very high adherence levels. Mediation analyses suggest that this association may operate through interconnected metabolic pathways, including improvements in insulin resistance, reductions in hepatic fat accumulation, lower systemic inflammation, and higher dietary fiber intake. Collectively, these findings provide prospective evidence that dietary patterns aligned with the EAT–Lancet framework may contribute to MASLD prevention in Middle Eastern populations and support the potential role of sustainable, plant‑forward diets as a feasible public health strategy for addressing the growing burden of metabolic liver disease. Supplementary Information Supplementary Material 1 Supplementary Material 1