Recent original studies and metanalytic reviews on the epidemiological and clinical features of lean MASLD.
| Author, year, [ref] | Methods | Findings | Conclusion | Limitations |
|---|---|---|---|---|
| Miller et al., 2026 [36] | This multicenter analysis included 3,386 adults with biopsy-confirmed MASLD enrolled in prospective NASH Clinical Research Network studies between 2004 and 2024. Centrally reviewed liver histology was compared across BMI categories, including ethnicity-adjusted thresholds for Asian participants. Clinical risk factors, metabolic comorbidities, PNPLA3 genotype and histological outcomes, including steatohepatitis, NAFLD Activity Score, advanced fibrosis and cirrhosis, were assessed using regression and standard group-comparison methods. | Normal-weight and overweight MASLD showed comparable histological severity, with identical rates of definite steatohepatitis, advanced fibrosis and cirrhosis. Histological severity increased mainly at BMI ≥ 30 kg/m2, suggesting obesity as a threshold for greater liver injury rather than a linear gradient across lower BMI categories. In normal-weight MASLD, advanced fibrosis was associated with older age, type 2 diabetes, hypertension, metabolic syndrome and enrichment for homozygous PNPLA3 G-allele carriage. Liver-related events were infrequent during follow-up. | Normal-weight and overweight MASLD appear histologically similar, whereas more severe disease emerges predominantly with obesity. These findings suggest that lean and overweight MASLD may represent a shared non-obese phenotype and challenge the use of “lean MASLD” as a distinct clinical category. Metabolic and genetic factors, rather than BMI alone, seem to drive fibrosis risk in normal-weight individuals. | The biopsy-based design may introduce referral and selection bias, as participants were recruited from tertiary centers and clinical studies. The normal-weight subgroup was small, limiting precision for subgroup analyses and rare liver-related outcomes. Event analyses were constrained by few liver-related events and limited adjustment for confounders, and the cohort composition may restrict generalizability to broader community-based or ethnically diverse MASLD populations. |
| Wang et al., 2026 [37] | Lean NHANES adults were assessed for MASLD using CAP and metabolic criteria; MLM-selected predictors were validated in an independent NHANES cohort with survey-weighted and subgroup analyses. | WWI emerged as the leading predictor of lean MASLD, showing an independent association and a non-linear risk increase above ~11 cm/√kg. | In lean adults, WWI may refine MASLD risk stratification by capturing central-adiposity-related risk beyond BMI. | The cross-sectional NHANES design, CAP-based steatosis assessment, residual confounding, limited generalizability, and the need for prospective validation limit the methodology of the study. |
| Chowdhary et al., 2026 [52] | This multicenter retrospective TriNetX cohort study identified 1,130,297 adults with non-cirrhotic MASLD from 2010 to 2024, including n = 229,084 (20.27%) with lean MASLD (BMI < 25 kg/m2). Incident fibrosis, cirrhosis, portal hypertensive complications, HCC, liver transplantation and all-cause mortality were assessed over 3-, 5-, and 10-year follow-up. | After propensity matching, lean MASLD showed higher risks of fibrosis, cirrhosis, portal hypertensive complications, HCC, liver transplantation and mortality than non-lean MASLD across follow-up intervals. In lean MASLD, older age, diabetes, hypoalbuminemia, thrombocytopenia and elevated AST independently predicted adverse hepatic outcomes. | Lean MASLD is associated with substantial long-term hepatic morbidity and mortality. These findings suggest that BMI-centered screening and risk-stratification strategies may overlook clinically relevant disease in lean individuals, supporting more inclusive approaches to surveillance. | Key limitations include the retrospective EHR-based design, potential selection and coding bias, residual confounding despite matching, and possible misclassification of MASLD status and liver outcomes. The BMI-based definition of lean MASLD may not reflect body composition or visceral adiposity. Differential surveillance and limited generalizability across populations and health care settings should also be considered. |
| Al Ta’ani et al., 2026 [38] | TriNetX data were used to compare CV, cerebrovascular and mortality outcomes in lean MASLD, defined as BMI < 25 kg/m2, after propensity-score matching and Cox modelling over up to 7 years. | After matching 67,519 patients per group, lean MASLD was associated with higher 7-year risks of HF, CVE, CBVE and all-cause mortality, with HRs ranging from 1.21 to 1.48 and 95% CIs excluding unity; excess risk was also evident at 1, 3 and 5 years. | Lean MASLD is associated with greater CV and mortality risk than non-lean MASLD. | The study is limited by the retrospective TriNetX design, reliance on coded EHR data, potential misclassification and residual confounding, limited granularity on liver disease severity and event adjudication, and uncertain generalizability across health-care settings. |
| Kasar et al., 2025 [53] | This NHANES 2017–2020 cross-sectional analysis included 1,707 lean adults with metabolic dysfunction, using ethnicity-specific BMI thresholds. Diet quality was assessed with the HEI-2015. | Higher diet quality was associated with lower odds of lean MASLD; participants in the highest HEI quartile had 66% lower odds than those in the lowest quartile. Protein and refined-grain intake were positively associated with MASLD, whereas oil intake appeared protective. | Better overall diet quality was associated with reduced lean MASLD risk, highlighting diet composition as a potentially modifiable factor in lean individuals and supporting further mechanistic and longitudinal studies. | The cross-sectional NHANES design precludes causal inference or assessment of temporality between diet quality and lean MASLD. Dietary intake was self-reported and may be affected by recall or reporting bias, while residual confounding from genetic, psychosocial or socioeconomic factors cannot be excluded. In addition, the available dietary and diagnostic measures may not fully capture nutrient subtype effects or the histological spectrum of MASLD. Finally, generalizability beyond US populations and the study-defined lean BMI category may be limited. |
| Njei et al., 2025 [39] | NHANES 2017–March 2020 data were used to estimate nationally representative lean MASLD prevalence among adults with valid elastography, excluding viral hepatitis and significant alcohol use; MASLD was defined by CAP ≥ 274 dB/m and lean status by BMI < 25 kg/m2 (< 23 kg/m2 in Asian individuals). | Lean MASLD affected an estimated 6.4% of U.S. adults, corresponding to 3.8 million people, of whom 34.6% were undiagnosed; prevalence rose markedly with glycemic dysfunction, reaching 33% in lean individuals with diabetes. | Lean MASLD is common and frequently underdiagnosed among adults in the USA, highlighting the need for refined screening and risk-stratification strategies in metabolically at-risk lean populations. | Limitations include the cross-sectional NHANES design, CAP-based MASLD ascertainment, self-reported diagnostic status, residual confounding, and limited precision for subgroup analyses in lean MASLD. |
| Zhang et al., 2025 [54] | In this population-based NHANES cohort of 14,740 adults, advanced fibrosis was assessed using NFS or FIB-4. | Over a median follow-up of 7.3 years, lean MASLD was associated with higher all-cause and diabetes-specific mortality than non-lean MASLD, whereas CV- and cancer-specific mortality did not differ significantly. Advanced fibrosis modified this risk, with particularly high mortality rates observed among lean individuals with fibrosis. Elevated FIB-4 or NFS identified lean MASLD patients at increased mortality risk. | Lean MASLD patients exhibited greater risks of all-cause and diabetes-specific mortality than non-lean MASLD patients. Noninvasive fibrosis assessment is recommended in lean MASLD patients. | The use of laboratory-based markers may have introduced misclassification or selection bias, while the limited number of participants and deaths within subgroups may have reduced statistical precision. Residual confounding, (e.g., underreported alcohol intake), cannot be excluded, and reliance on ICD-coded causes of death may have led to outcome misclassification. LRM was unavailable. Finally, the findings should be generalized cautiously to other populations. |
| Tesfai et al., 2026 [55] | Cross-sectional analysis of prospectively enrolled 312 adults with MASLD. Lean MASLD subjects (n = 44 = 14.1%) were age- and sex-matched with non-lean MASLD. Fibrosis assessment included vibration-controlled transient elastography, magnetic resonance elastography and liver biopsy. A GRS, summating the effect alleles of PNPLA3 and TM6SF2 minus the protective HSD17B13 genotype, was estimated to consider inherited genetic risk across BMI categories. Results were validated in an external Latin American cohort. | A high GRS compared to a low GRS was associated with a greater prevalence of significant fibrosis (36.5% vs. 25.2%; p = 0.043). Among individuals with a high GRS, fibrosis prevalence did not differ significantly between lean and non-lean patients (31.3% vs. 37.1%; p = 0.645), a pattern that was recapitulated in the Latin American cohort. | The prevalence of significant fibrosis and the effect of GRS were similar in lean and non-lean MASLD, highlighting that lean MASLD patients may have a comparable risk to overweight and obese MASLD. | Limitations include the small number of lean MASLD participants, which limits statistical power and precludes robust subgroup analyses. Although deep phenotyping supports accurate case classification, prior BMI trajectories, differences in fibrosis assessment methods between cohorts, limited representation of Asian populations, and the absence of longitudinal liver stiffness data may affect interpretation and generalizability. |
| Huo et al., 2026 [40] | This prospective multicohort study included 153,192 patients with MASLD from UKB, 29,700 from the Kailuan cohort, and 3,329 from CKB. Lean MASLD was defined as BMI< 23 kg/m2 in Kailuan and CKB and < 25 kg/m2 in UKB. Primary endpoints were LREs, all-cause mortality, LRM, CVD mortality, CVD, HCC, and EHC. | Across 181,191 non-lean and 5,030 lean patients with MASLD followed for a median of 14.2 years, lean MASLD was associated with higher risks of LREs (HR 2.14), all-cause mortality (HR 1.26), LRM (HR 2.31) and CVD mortality (HR 1.22), but lower incident CVD risk (HR 0.89) and no clear increase in HCC or EHC. | Lean MASLD appears to confer an excess burden of adverse liver outcomes and all-cause mortality, while risks of HCC and EHC remain comparable and incident CVD risk is lower. | This long-term prospective multicohort study is observational and cannot establish causality; heterogeneity in cohort design, MASLD ascertainment, covariates and outcome capture, together with limited data on body composition, lifestyle, genetic risk, disease severity and treatment exposure, may contribute to residual confounding. |
| Hsu et al., 2025 [56] | This Taiwan Biobank study integrated genomic data with ultrasound-defined steatosis to identify lean MASLD among individuals with BMI < 23 kg/m2. GWAS was performed in 1,720 lean MASLD cases and 5,331 lean controls, adjusting for demographic and ancestry-related covariates, with conditional analyses used to define independent loci and replication undertaken in the All of Us cohort. | Lean MASLD was characterized by adverse metabolic features and a strong genetic signal at the PNPLA3–SAMM50 locus. The PNPLA3 variant rs9625962 emerged as the lead independent association and was replicated in the All of Us cohort, with risk conferred across BMI strata. In lean MASLD, higher metabolic risk scores also correlated with FIB-4 and NFS, linking metabolic burden to fibrosis risk. | In the Taiwan Han Chinese population, rs9625962 showed pleiotropic effects on MASLD across BMI categories, triglyceride levels and liver-test profiles. Replication in the All of Us cohort supports the robustness and broader relevance of these variants, underscoring a genetic contribution to lean MASLD pathogenesis. | Key limitations include incomplete ascertainment of MASLD-related metabolic criteria, missing viral hepatitis and lifestyle data, limited generalizability beyond the cohort in Taiwan, and reliance on BMI alone to define lean status. |
| Sato-Espinoza et al., 2025 [57] | Case-control study including 177 lean non-alcoholic SLD cases (47 cryptogenic, 130 MASLD), 677 matched lean controls, and 3,090 overweight/obese SLD cases in the Mayo Clinic Biobank and Tapestry databases. | Lean non-alcoholic SLD was associated with an intermediate metabolic and genetic risk profile between lean controls and overweight or obese SLD, with diabetes emerging as an independent predictor. Within lean SLD, MASLD was linked to GCKR risk-allele homozygosity, whereas FIB-4 retained good performance for identifying advanced fibrosis. | Diabetes is a key predictor of non-alcoholic SLD in lean individuals, supporting SLD screening irrespective of BMI. GCKR variation may contribute to lean MASLD, whereas the determinants of cryptogenic SLD remain unresolved. | Its retrospective design supports association rather than causation, and case ascertainment, diagnostic variability, and self-reported alcohol intake may have introduced misclassification. Limited diversity, the absence of an external validation cohort, and the use of a single BMI threshold further constrain generalizability. |
| Njei et al., 2025 [58] | Cross-sectional study including 860 lean adults (median age 53 years; 48% female; 38% White; mean BMI 22.5 kg/m2). | The age-adjusted prevalence of MASLD was 8.9%. Among individuals with MASLD, 84% had a high visceral adiposity index (> 1.92). The percent change in CAP associated with HOMA-IR, QUICKI, and HOMA-β was 11.62, 95% CI: 6.31–16.93), 0.76 (95% CI: 0.26–1.26), and –162.72 (95% CI: –260.94 to –64.50), respectively. HOMA-IR (aOR: 2.60, 95% CI: 1.36–4.98), QUICKI (aOR: 1.06, 95% CI: 1.01–1.11), and HOMA-β (aOR: 0.04, 95% CI: 0.01–0.56) were each associated with MASLD. HOMA-IR was associated with advanced liver fibrosis (aOR: 1.51, 95% CI: 1.19–2.15). | Lean MASLD affected 8.9% of US adults and was strongly linked to visceral adiposity and insulin resistance; HOMA-IR was also independently associated with advanced fibrosis. | The cross-sectional design precludes causal inference, and generalizability beyond the US population remains uncertain. Insulin resistance was assessed using surrogate fasting indices rather than dynamic methods, residual confounding cannot be excluded, and MASLD and advanced fibrosis were diagnosed non-invasively. |
| Rotaru et al., 2025 [41] | This observational study assessed 1,438 adults with metabolic and hepatic profiling, including VCTE-CAP. In normal-weight participants (BMI < 25 kg/m2), CUN-BAE and other non-invasive indices were evaluated, and multivariable regression identified predictors of steatosis and fibrosis. | Among normal-weight individuals, MASLD prevalence was 19.28% and was accompanied by substantial metabolic dysfunction; CUN-BAE independently predicted steatosis, whereas conventional non-invasive scores showed only moderate discrimination in lean phenotypes. | In Eastern Europe lean MASLD is clinically relevant, supporting the integration of visceral adiposity measures and advanced non-invasive liver assessment to improve detection and risk stratification. | Interpretation should consider the small single-center lean cohort, VCTE-based MASLD assessment, cross-sectional design and possible residual confounding related to lifestyle, treatment exposure, disease activity, body composition, genetic risk and metabolic factors. |
| Ghani et al., 2025 [59] | Seventy-five thousand nine hundred twenty-one patients (with 5% lean individuals) were included in this retrospective multicenter study on patients with MASLD at the Banner and the University of Michigan Health systems from 2012 to 2023. | After adjusting for confounders, lean individuals had a higher mortality, higher incidence of CVD, and higher incidence of MALO. | Lean individuals have a higher mortality, higher cardiovascular burden, and higher incidence of MALO compared to non-lean individuals. Further studies are warranted to explore lean patients with MASLD, and interventions are needed to decrease mortality in this patient population. | In this retrospective cohort study all diagnoses (MASLD, CVD, metabolic diseases, cirrhosis, MACE, and MALO, were mainly based on ICD codes. Unmeasured confounders (e.g., diet, physical activity, body composition, genetic variants) which were not available can exist. Finally, causes of death data are unavailable. |
| Njei et al., 2025 [42] | This retrospective VHA cohort study included 15,155 patients with newly diagnosed compensated MASLD-cirrhosis between January 2008 and May 2021, comprising 1,597 lean and 13,558 non-lean individuals. Associations with hepatic decompensation, MACE and all-cause mortality were assessed using multivariable Cox models, with Fine and Gray competing-risk regression where appropriate. | In this predominantly male VHA cohort (95%; median age 67 years), lean MASLD-cirrhosis was associated with lower diabetes prevalence than non-lean disease (46.7% vs. 73.9%) but higher all-cause mortality (aHR 1.64) and cardiovascular mortality (aHR 1.40), despite a lower risk of hepatic decompensation (aSHR 0.67). | Lean MASLD-cirrhosis was associated with higher all-cause and cardiovascular mortality despite lower hepatic decompensation risk, more favorable CMP, and similar MACE rates. | Restricted generalizability of the findings results from an older, predominantly male Veterans Affairs cohort, potential misclassification from routine clinical data, and residual confounding related to sarcopenia, frailty, alcohol exposure, lifestyle, medication use and cirrhosis severity. |
| Nso et al., 2024 [43] | This meta-analysis included 21 studies comparing lean (n = 7,153; mean age 52.9 years; 56% male) and non-lean MASLD (n = 23,514; mean age 53.2 years; 63% male), with pooled ORs and 95% CIs estimated for CV, mortality and metabolic outcomes. | Lean MASLD was associated with higher cardiovascular mortality than non-lean MASLD (OR 1.5, 95% CI 1.2–1.8), while CVD odds were lower by 40% and MACE did not differ materially; all-cause mortality was directionally higher, and hypertension and dyslipidemia were each 30% less frequent. | Despite a more favorable CMP and comparable MACE rates, lean MASLD is associated with increased CV mortality. | As a meta-analysis of observational studies, these findings are limited by heterogeneity in MASLD definitions, BMI cut-offs, outcome ascertainment, follow-up duration, and covariate adjustment across included cohorts. Residual confounding, publication bias, and the use of aggregate rather than individual participant data may affect the precision and causal interpretation of the pooled CV risk estimates. |
aHR: adjusted hazard ratio; aOR: adjusted odds ratio; aSHR: adjusted subdistribution hazard ratio; BMI: body mass index; CAP: controlled attenuation parameter; CBVE: cerebrovascular events; CKB: China Kadoorie Biobank; CIs: confidence intervals; CMP: cardiometabolic profile; CUN-BAE: Clínica Universidad de Navarra-Body Adiposity Estimator; CV: cardiovascular; CVD: cardiovascular disease; CVE: cardiovascular events; EHC: extrahepatic cancer; GCKR: glucokinase regulator; GRS: genetic risk score; HEI: Healthy Eating Index; HF: heart failure; HRs: hazard ratios; HCC: hepatocellular carcinoma; HOMA-IR: homeostasis model assessment of insulin resistance; LREs: liver-related events; LRM: liver-related mortality; MACE: major adverse cardiovascular events; MALO: major adverse liver outcome; MASLD: metabolic dysfunction-associated steatotic liver disease; MLM: machine learning model; NASH: nonalcoholic steatohepatitis; NFS: NAFLD fibrosis score; NHANES: National Health and Nutrition Examination Survey; SLD: steatotic liver disease; OR: odds ratio; UKB: UK Biobank; USA: United States of America; VCTE: vibration-controlled transient elastography.