This is the first systematic review and meta-analysis examining the global prevalence, and metabolic and clinical outcomes of patients with lean-MAFLD in comparison with overweight/obese-MAFLD patients. As hypothesized, we found that lean-MAFLD patients have an equivalent metabolic burden when compared with overweight/obese-MAFLD and this likely accounts for their equal risk for the extrahepatic outcomes examined (cardiovascular, cancer-related, and all-cause mortality). On the other hand, lean-MAFLD is phenotypically distinct with liver disease onset at a lower set-point BMI-driven by disparate and unique mechanisms compared to overweight/obese-MAFLD. This contributes toward their increased liver-related mortality compared to overweight/obese-MAFLD. Our study answers previous controversey among lean NAFLD cohorts that report higher liver-related mortality despite having a better metabolic profile [9]. In addition, it highlights the strengths of the MAFLD criteria that is able to positively and correctly identify lean/normal individuals who are metabolically unhealthy with a high risk of adverse extrahepatic and hepatic outcomes.

A key contributor to hepatic inflammation in lean-MAFLD is a dysregulation of metabolic adaptation. Early in disease, bile acids levels are consistently increased in lean when compared with overweight/obese-MAFLD patients and dampens liver inflammation. However, over time, endotoxemia contributed by differences in gut microbial dysbiosis among lean-MAFLD but not healthy individuals or overweight/obese-MAFLD patients results in alterations in the epigenomic and transcriptome including upregulation of Toll-Like Receptor 4 (TLR4). This contributes to a failure of bile acid mediated anti-inflammatory signaling resulting in increased cytokine activation driving liver inflammation and long-term adverse hepatic sequelae [10]. In addition, telomere attrition mediated at least in part by increased reactive oxygen species generation is more pronounced among lean-MAFLD cohorts when compared with overweight/obese patients and might contribute toward increased mortality [11].

Another contributory factor to the increased liver-related mortality is the greater degree of sarcopenia found among MAFLD patients who are lean [12]. This may be exacerbated to a greater extent with the onset of cirrhosis, which in turn is associated with an increased risk of adverse liver-related outcomes [13]. This is a key issue as weight loss is the typical initial recommendation for the management of MAFLD and careful consideration needs to be given for patients who are lean with MAFLD. Our findings are particularly relevant in the current era of potent pharmacological weight loss drugs where there are growing concerns for worsening sarcopenia associated with rapid weight loss. This was recently shown using data from the dual x-ray absorptiometry (DEXA) subgroup of the STEP-1 trial evaluating Semaglutide for weight loss. That sudy reported that although there was a 8.36 kg fat mass loss in the semaglutide group, there was an accompanying 5.26 kg loss of lean muscle mass (comprising ~ 40% of total weight loss) [14]. In the SURMOUNT-1 trial evaluating Tirzeptide for weight loss, the investigators reported a -33.9% fat mass loss but a worrying 10.9% lean mass loss [15]. While mouse models purport that GLP1-receptor agonists (GLP1-RA) ameliorate sarcopenia and muscle atrophy [16], this is not reflected in human data. Similar effects of weight loss-associated lean muscle loss have been reported in a recent meta-analysis of bariatric surgery patients. That study reported a greater than 8 kg of lean body mass loss during the first year post bariatric surgery [17]. The findings suggest that lean muscle loss is associated with the rapidity of weight loss rather than a drug-class specific effect. Thus, perhaps, the positive pleiotropic effects of GLP1-receptor agonists need to be balanced against the risk of sacropenia and the need to achieve more gradual weight loss targets in lean-MAFLD patients, particularly those with more advanced fibrosis.

An unresolved issue remains that there are no current, nor in-development treatments that specifically benefit patients with lean-MAFLD. The ENLIVEN (Pegozafermin, fibroblast growth factor 21 (FGF21) analog) [18] and SEMA-NASH (Semeglutide) [19] trials outrightly exclude patients with BMI < 25 kg/m2, while other trials, such as MASTERO NASH (Resmetirom, thyroid hormone receptor-β agonist) [20], HARMONY (Efruxifermin, FGF21 analog) [21], FALCON (Pegbelfermin, FGF21 analog) [22], and FLINT (Obeticholic Acid, Farsenoid X nuclear receptor ligand) [23], have an average BMI > 35 kg/m2 across treatment and control groups, suggesting predominantly overweight/obese participants. The underrepresentation of lean-MAFLD patients in these trials limits the generalizability of the therapeutic agents and sets a pressing need for trials inclusive of patients with lean-MAFLD. Practically, patients with lean-MAFLD will require therapies that target visceral and hepatic fat loss, inflammation, and fibrosis, while preserving lean muscle mass and improving liver-related and metabolic outcomes. This might require a more nuanced approach to treatment with dose reduction of weight loss agents such as GLP1-RA (where needed), possibly in combination with agents that also improve lipid and glycemic profiles but having minimal effects on weight such as resmetirom [20] and FGF21 analogues [22].

We found that the global prevalence of lean-MAFLD (1.94%) was similar among Western (2.83%) and Asian (2.00%) cohorts. This is half the estimated global prevalence of lean NAFLD (5.1%) from another meta-analysis [24]. The difference is likely attributed to the more stringent lean-MAFLD criteria identifying the group of lean patients having a greater metabolic burden. In a study using the NHANES III cohort (1999–2015), patients who were MAFLD ( – ) NAFLD ( +) (hence were lean NAFLD patients who did not have diabetes and who had less than 2 metabolic risk factors) were shown to have a lower metabolic burden, lower BMI (21.8 ± 0.2), lower waist circumference, lower proportions of significant and advanced hepatic fibrosis, with lower rates of all-cause, cardiovascular, cancer, and other cause mortatily when compared with patients who were MAFLD ( +) NAFLD ( +) and patients who were MAFLD ( +) NAFLD ( – ). This indicates that lean NAFLD patients who did not satisfy the MAFLD criteria have significantly more benign outcomes [25]. In another NHANES cohort (1999–2018) which compared lean patients with or without NAFLD and/or MAFLD, lean patients who were MAFLD ( +) NAFLD ( – ) and MAFLD ( +) NAFLD ( +) were shown to have higher all-cause and cardiovascular mortality when compared with lean-MAFLD ( – ) NAFLD ( +) patients [26]. Another recent meta-analysis also demonstrated that the metabolic burden of lean NAFLD patients is lower than that of overweight/obese NAFLD patients [27]. These findings imply a dilutional effect of metabolic risk factors in studies reporting on lean NAFLD, where patients with the more benign MAFLD ( – ) NAFLD ( +) are mixed with lean-MAFLD ( +) patients. In sum, by not applying the MAFLD criteria, lean fatty liver patients who have a higher metabolic burden might potentially be masked and underreported in lean NAFLD cohorts; conversely, the lean-MAFLD criteria identifies the proportion of lean fatty liver patients who are at risk of worse clinical outcomes and require more pro-active management.

In real-world practice, MAFLD is not a static disease and patients may accrue metabolic risk and potentially have changes in phenotype either in a positive or negative direction. This was examined in a longitudinal observational study by Younes et al [28], where liver biopsy proven lean caucasian NAFLD patients from a multicentre cohort were followed for a median of 94 months. In that study, the majority of lean patients remained lean (77.5%) and yet developed incident diabetes and had complications including cardiovascular events, extrahepatic malignancy, and liver-related events and HCC similar to their overweight and obese patients. This was not explained by concomitant weight gain nor PNLP3A genotype. Importantly, the development of diabetes was an independent predictor of cardiovascular events and extrahepatic malignancy, indicating the importance of metabolic risk factors in the prognosis of patients with MAFLD.

Another issue is the impact of the metabolic dysfunction-associated steatotic liver disease (MASLD) criteria is the MAFLD criteria among lean patients. The diagnosis of MASLD is established when a patient meets any one of the five cardiometabolic criteria [29]. This approach aims to emphasize the role of metabolic dysfunction while ensuring that the majority of individuals with NAFLD are reclassified under MASLD. However, the key limitation of the MASLD criteria, especially toward lean patients, is the equal weighting of a single metabolic criterion for diagnosis which does not account for differences in clinical outcomes or treatment. For instance, lean individuals with diabetes—a recognized driver of disease progression—are considered equivalent to those lean individuals with isolated hypertension. This lack of differentiation has several implications. First, from an epidemiological point of view, patients with a higher metabolic burden and higher risk of adverse outcomes may become underrepresented when grouped under a single umbrella of MASLD. Further, it would be impossible to effectively report individual phenotypes (e.g., MASLD: lean with hypertension), especially when patients may have more than one metabolic risk factor. Finally, this suboptimal classification will have an impact on the design of future clinical trials for lean patients as the inclusion of individuals with varying metabolic burdens may obscure differences in treatment efficacy, potentially leading to inadequate reporting of treatment outcomes. The lean-MAFLD criteria, on the hand, identify a proportion of lean patients with a higher metabolic burden requiring urgent clinical attention. In our manuscript, we demonstrated that lean-MAFLD patients had similar clinical outcomes for all-cause, cancer-related, and cardiac-related mortality when compared with overweight/obese-MAFLD patients, with higher liver-related mortality among lean-MAFLD patients. We believe that the lean-MAFLD better risk stratifies lean patients based on their higher metabolic burden.

Our study has several limitations. First, the Doi plot and LKF index (5.51) suggest publication bias when evaluating the global prevalence of lean-MAFLD (appendix). During our systematic review, we found that the majority of the MAFLD cohorts screened and excluded from the study did not sub-classify lean-MAFLD as a separate subgroup. In addition, we found that there were six studies that did not use ethnic appropriate cutoffs (of waist circumference and BMI) in defining lean-MAFLD and were thus also excluded. To compensate for this, we used the trim-and-fill method to approximate the true prevalence of lean-MAFLD. Second, as expected with global data, there was high heterogeneity detected in the majority of our analyses. We recognize that heterogeneity is often viewed with concern, we would like to emphasize that it does not diminish the utility of the analysis itself. In principle, we believe that the MAFLD criteria of having at least two metabolic risk factors in patients with lean-MAFLD inherently sub-select a group of lean patients of a higher metabolic burden. While heterogeneity may indicate a greater degree of dispersion around the true effect size, we believe that the effect sizes from our analysis accurately portray the higher metabolic burden associated with lean-MAFLD and are comparable to patients with overweight/obese-MAFLD. Further, heterogeneity is anticipated in global epidemiological studies and has been well reported [30]. This is likely to be contributed by multiethnic societies and reflects differences in genetic, dietary, and environmental factors unique to each country, even when geographically close. Statistically, we have sought to compensate for the heterogeneity encountered by applying random-effects models throughout our analysis. Third, there were insufficient data to compare metabolic and outcome data of lean NAFLD, and healthy lean patients. Finally, we lacked outcome data on incident development of metabolic risk factors such as new onset diabetes or dyslipidemia which may add insight into the prognostic value of the MAFLD criteria.

In summary, our findings highlight the ultility of the MAFLD criteria, which is able to correctly classify patients that share an equivalent metabolic burden and a similar risk of extrahepatic adverse outcomes, regardless of anthropomorphic differences between lean and overweight/obese patients. We report increased liver-related mortality among lean-MAFLD patients which is likely contributed by dysregulation of metabolic adaptation, telomere shortening, and sarcopenia. Nonetheless, there remains challenges in raising awareness in correctly identifying metabolically unhealthy lean-MAFLD patients. Finally, there is an urgent need for directed and specific clinical trials for lean-MAFLD patients that target fat loss and inflammation explicity in the liver, while preserving lean muscle mass and improving both metabolic and liver-related outcomes.