ABSTRACT
-
Background/Aims
Previous studies have identified a substantial degree of agreement between the non-alcoholic fatty liver disease (NAFLD) and metabolic dysfunction-associated steatotic liver disease (MASLD) populations, but the same notion may not apply to normal-weight patients with a lower cardiometabolic risk burden. This study aims to investigate the cardiometabolic risk factor (CMRF) distributions between normal-weight and overweight/obese MASLD, the agreement between historical NAFLD and MASLD, and to compare the risk of liver-related events (LREs) and all-cause mortality in normal-weight versus overweight or obese MASLD.
-
Methods
This study included participants with steatotic liver disease (SLD) from five cohorts in China (Hong Kong), South Korea, and the United States. Participants were recruited from settings including both hospitals and communities. Individuals were classified into normal-weight and overweight/obese groups.
-
Results
This study included 33,793 participants with SLD from five cohorts, of whom 20,893 and 20,701 patients met the diagnosis of NAFLD and MASLD, respectively. Normal-weight patients with NAFLD demonstrated a lower CMRF distribution compared to those with overweight/obese NAFLD. In the community-based cohorts, the proportions with 0 CMRF ranged from 9.0 to 26.7% among normal-weight NAFLD patients, representing the discrepancy between MASLD and NAFLD definitions. Compared with the overweight/obese MASLD, the normal-weight MASLD had increased all-cause mortality (normal-weight vs. overweight/obese, 23.44 and 13.80 per 1,000 person-years; P<0.001) but not LREs (2.81 and 2.59 per 1,000 person-years; P=0.54) in the Hong Kong Clinical Data Analysis and Reporting System cohort.
-
Conclusions
Normal-weight individuals with NAFLD demonstrated a lower distribution of CMRFs, resulting in the incomplete agreement between historical NAFLD and MASLD.
-
Keywords: Nonalcoholic fatty liver disease; Hepatic steatosis; Cardiometabolic risk factors
Study Highlights
• In community settings, the proportions of NAFLD patients presenting with 0 CMRF ranged from 9.0 to 26.7% among normal-weight patients, not fulfilling the criteria of MASLD.
• Caution is warranted when applying the findings from previous studies based on NAFLD to the current definition of MASLD, particularly when considering the population with normal body mass index.
• The normal-weight MASLD population was associated with increased all-cause mortality but not increased adverse liver outcomes compared to the overweight/obese MASLD population, attributable to the underlying chronic extrahepatic disease.
Graphical Abstract
INTRODUCTION
In 2023, the American and European liver societies recommended changing the nomenclature of the previous non-alcoholic fatty liver disease (NAFLD) to the current metabolic dysfunction-associated steatotic liver disease (MASLD) [
1]. Unlike NAFLD, which is a negative diagnosis by excluding the presence of other chronic liver diseases and confirming the presence of hepatic steatosis, MASLD is a diagnosis based on the presence of both hepatic steatosis and cardiometabolic risks, highlighting the importance of metabolic dysfunction.
The major difference between the definitions of NAFLD and MASLD is that the latter requires the presence of at least one cardiometabolic risk factor (CMRF). Concerns have been raised regarding whether these two definitions identify the same group of patients. While both our previous work [
2] and other studies [
3,
4] have identified the significant overlap between NAFLD and MASLD across diverse cohorts [
5], this issue necessitates reconsideration in populations with normal body mass index (BMI). Normal-weight patients with steatotic liver disease (SLD), defined as having a BMI<25 kg/m
2 (or <23 kg/m
2 in Asians), must fulfil at least one of the four other CMRFs or have increased waist circumference to be classified as having MASLD. Consequently, the pronounced overlap between NAFLD and MASLD may not apply to this specific population.
The primary aim of the study was to investigate the CMRF distributions between normal-weight and overweight MASLD, and the discrepancy between NAFLD and MASLD. The study also aimed to compare the characteristics, and long-term risk of liver-related complications and all-cause mortality between normal-weight and overweight MASLD across various populations.
MATERIALS AND METHODS
Study population
This study included participants from five cohorts in China (Hong Kong [HK]), South Korea and the United States (US). Individuals with SLD, i.e., hepatic steatosis, were identified by liver biopsy, magnetic resonance spectroscopy (MRS), ultrasonography, International Classification of Diseases, 9th Revision (ICD-9) coding, or the controlled attenuation parameter (CAP) measured by vibration-controlled transient elastography (VCTE) in datasets with corresponding measurements. The exclusion criteria included chronic viral hepatitis by history and/or hepatitis serology, excessive alcohol consumption, incomplete data for assessing CMRFs and missing or substandard-quality measurements to identify patients with MASLD. The details of harmonization in definitions, inclusion, and data processing for cohorts are shown in
Supplementary Figure 1 and
Supplementary Table 1.
The HK biopsy cohort included patients who underwent biopsy between 2006 and 2024. The HK MRS cohort randomly selected individuals from the community for MRS screening between 2008 and 2010 [
6]. The HK Clinical Data Analysis and Reporting System (CDARS) cohort consisted of individuals diagnosed with NAFLD according to the CDARS from 2000 to 2021. CDARS is a territory-wide electronic health database embedded in public hospitals in HK, covering approximately 80% of the local population. It captured comprehensive anonymized clinical data, including demographics, hospitalizations, and death status and causes. In the HK CDARS cohort, CMRF assessments were based on patients’ electronic medical records before or within 2 years of NAFLD diagnosis.
For the three HK datasets, excessive alcohol intake was recorded qualitatively, and participants with excessive alcohol intake were excluded. Significant/advanced fibrosis was defined as ≥F2/≥F3 fibrosis based on histology in the biopsy cohort, liver stiffness measurement (LSM) by VCTE≥8.0/12.0 kPa in the MRS cohort, and Fibrosis-4 index (FIB-4)≥1.3 (2.0 for age≥65 years)/2.67 for CDARS cohort, respectively. For each HK cohort, the study protocol conformed to the ethical guidelines of the Declaration of Helsinki, as reflected in a priori approval by the Joint Chinese University of HK-New Territories East Cluster Clinical Research Ethics Committee (Ref. No.: 2006.439-T, 2008.212, and 2020.359). Written consent was obtained from participants in the HK biopsy cohort and HK MRS cohort. Informed consent was waived for HK CDARS cohort because it was a registry study using de-identified data.
National Health and Nutrition Examination Survey cohort
The National Health and Nutrition Examination Survey (NHANES) is a continuous, nationally representative survey of the non-institutionalized US civilian population, conducted over 2-year cycles. This study used the database derived from this survey between 2017 and 2020, the period when the VCTE measurements were available.
Individuals without records of alcohol intake were excluded from the NHANES cohort. Participants with CAP by VCTE≥248 dB/m were defined as having SLD, as recommended in the updated guidelines [
7]. Significant/advanced fibrosis was defined by LSM by VCTE≥8.0/12.0 kPa. A sensitivity analysis was conducted for the NHANES cohort using a CAP threshold of 274 dB/m to define steatosis [
8], considering the higher cut-off values from meta-analyses or similar cohorts with reported sensitivities of 80–90% [
8-
10].
The Korean magnetic resonance elastography (MRE) cohort is a health check-up cohort in South Korea with available MRE measurements [
11]. The data were collected from health promotion centers in South Korea, which provide health check-up examinations for the general Korean population and are equipped with MRE. Quantitative alcohol consumption data were recorded, and those with missing alcohol intake data were excluded. Hepatic steatosis was detected using ultrasonography. Significant/advanced fibrosis was defined using the MRE threshold values of 3.0 and 3.6 kPa [
11]. The study protocol followed the Declaration of Helsinki and the Istanbul Principles guidelines and was approved by the Institutional Review Board of Hanyang University Hospital (HYUH 2023-10-020) with a waiver for informed consent.
Patients were divided into different groups based on the following BMI criteria: BMI<23 (25 Western) kg/m
2 for normal weight, BMI≥23 and <25 (25–30 Western) kg/m
2 for overweight, and BMI≥25 (30 Western) kg/m
2 for obesity. CMRF1 is defined as a BMI≥25 kg/m
2 (≥23 kg/m
2 for Asians) or a waist circumference≥94 cm (M) or 80 cm (F) (≥90 cm [M] or 80 cm [F] for Asians). Abdominal obesity is defined using the same waist circumference threshold as that of the CMRF1. CMRF2 is identified by a fasting serum glucose≥5.6 mmol/L, hemoglobin A
1c≥5.7%, a history of diagnosis for type 2 diabetes or the use of any antidiabetic medication. CMRF3 is defined based on a blood pressure of ≥130/85 mmHg, a history of diagnosis for hypertension, or antihypertensive treatment. CMRF4 and CMRF5 are defined as plasma triglycerides≥1.7 mmol/L, plasma high-density lipoprotein-cholesterol≤1.0 mmol/L (M) or ≤1.3 mmol/L (F), or the use of lipid-lowering treatments. The data availability to ascertain the CMRFs was shown for two cohorts from public/healthcare databases, namely HK CDARS cohort and NHANES cohort (
Supplementary Table 2).
Participants with hepatic steatosis and a daily alcohol intake of <30 g (M)/20 g (F) without other chronic liver diseases were considered to have NAFLD. Participants with NAFLD who simultaneously met at least one CMRF criterion were defined as having MASLD. The presence of NAFLD and the absence of CMRF would be classified as cryptogenic SLD. By definition, all overweight or obese individuals with NAFLD fulfilled the definition of MASLD as their BMI met the CMRF1 criteria. Participants having MASLD who presented with moderate alcohol consumption were classified as MetALD. Participants with steatosis who consumed alcohol>60 g (M)/50 g (F) were classified as having alcohol-related liver disease (ALD).
Outcomes
The primary outcome in the cross-sectional analysis was the CMRF distributions between normal-weight and overweight MASLD, and the discrepancy between NAFLD and MASLD (proportions of 0 CMRF). In the longitudinal analysis, the primary outcome was liver-related events (LREs), defined as the earliest events among hepatic decompensation, hepatocellular carcinoma (HCC), liver transplantation and liver-related deaths. Hepatic decompensation was defined as the diagnosis or complications of decompensated cirrhosis, based on ICD-9 coding. The related diagnoses and procedures included ascites, paracentesis, esophageal or gastric variceal bleeding with liver diseases, hepatic encephalopathy, hepatorenal syndrome, hepatopulmonary syndrome, and spontaneous bacterial peritonitis. The secondary outcome was all-cause mortality. The follow-up duration was defined as the period from the date of diagnosis for MASLD to the earliest date of recorded LREs (for the primary outcome), death (for the secondary outcome), or the last follow-up date (July 1st, 2024). Only the HK CDARS cohort captured the follow-up data and was examined in the longitudinal analysis. Patients who developed LREs or died within 180 days of their MASLD diagnosis date were excluded from analyses on LRE and mortality. Patients with a previous diagnosis of malignancy and whose recorded death date coincided with their MASLD diagnosis date were excluded from the analyses.
Statistical analysis
Data were presented as number (%) for categorical variables and as mean±standard deviation (SD) or median (1st quartile–3rd quartile [Q1–Q3]) for continuous variables. Chi-square tests or Fisher’s exact tests were performed to compare the distribution of categorical variables, as appropriate. The Student’s t-tests were performed to compare the distribution of continuous variables as mean±SD, and Mann–Whitney U-tests were performed to compare the distribution of continuous variables reported as medians with interquartile ranges. Pearson’s chi-squared test was conducted to compare the distribution of FIB-4 categories between the normal-weight and overweight/obese groups across all cohorts. The distributions of CMRF numbers and individual CMRFs were examined for all cohorts with normal-weight vs. overweight/obese NAFLD (which included MASLD populations), respectively. The proportions of patients with 0 CMRF represented the discrepancy between MASLD and NAFLD. A subgroup analysis on MetALD/ALD was performed to examine the impact of alcohol consumption on the distribution of BMI categories in the NHANES cohort.
The cumulative incidence of all-cause mortality and LREs was estimated using the Kaplan-Meier and Aalen-Johansen methods, respectively. The incidence rate of all-cause mortality was calculated as the number of deaths per 1,000 person-years. Subdistribution hazard ratios and 95% confidence intervals (CIs) for the covariates associated with the risk of developing LRE were estimated using the Fine-Gray subdistribution hazard model, considering non-liver-related mortality as a competing risk. Stepwise adjusted Cox regression model and Fine-Gray models were conducted to estimate the hazard ratios (HRs) and 95% CIs of covariates associated with all-cause mortality or LRE. Extrahepatic comorbidities were considered within a 180-day lag period from the diagnosis date of MASLD based on disease-specific ICD-9 diagnosis codes. A lag analysis with a 2-year period was further conducted for all-cause mortality to address potential reverse causation in normal-weight groups. Proportional hazards assumptions were fulfilled for all hazards models (P-values for Schoenfeld residual test>0.05).
RESULTS
Overview of the five cohorts
After excluding incomplete data, a total of 33,793 participants with SLD from five different cohorts were included in the analysis. Participants meeting the definition of MASLD (N=20,701) were examined for clinical characteristics and outcomes. In the HK biopsy and CDARS cohorts, 461 and 15,265 individuals were from hospital settings, while 268, 1,432, and 3,275 individuals in the HK MRS cohort, NHANES cohort and Korean MRE cohort were from community (or health check-up) settings, respectively (
Table 1,
Supplementary Table 3). The average age of participants at the time of diagnosis of MASLD ranged from 47.8 to 58.2 years across five cohorts, with mean BMI values ranging from 25.5 to 32.8 kg/m
2. The median FIB-4 index values ranged from 0.82 to 1.17. The prevalence of missing baseline characteristics in three population-based cohorts is shown in
Supplementary Table 4.
Normal-weight individuals consistently exhibited a lower prevalence of hepatic steatosis than overweight/obese individuals across all cohorts. The prevalence of SLD in the normal-weight and overweight/obese groups ranged from 11.6 to 34.0% and from 50.5 to 77.9%, respectively. The age-standardized prevalence of MASLD was 4.9–15.2% and 21.4–54.0% in normal and overweight groups (
Supplementary Table 5). The prevalence of MASLD in HK CDARS cohort was not calculable as the participant inclusion criteria were based solely on the ICD-9 diagnosis code for NAFLD within the dataset. In the HK biopsy cohort, the proportion of patients meeting the definition of MASLD was 14.1% (95% CI 10.2–18.1%) in the normal-weight group and 54.4% (95% CI 50.9–57.9%) in the overweight/obese group. In the community-based cohorts, the prevalence of MASLD ranged from 7.7 to 10.1% in the normal-weight population, compared to a significantly higher prevalence ranging from 28.0 to 47.3% in their overweight counterparts.
Among normal-weight individuals with NAFLD from hospital settings, all patients in the HK biopsy cohort had ≥1 CMRF(s) and fulfilled the MASLD criteria, whereas 2.1% (37/1,770) of NAFLD patients in the CDARS cohort did not have any CMRF and were classified as having cryptogenic SLD (
Fig. 1A). Among the three community-based cohorts with normal-weight NAFLD, the proportions of individuals having 0 CMRF were 9.0% (6/67), 24.5% (40/163), and 26.7% (109/408) in the HK MRS cohort, NHANES cohort, and Korean MRE cohort, respectively; these individuals did not meet the criteria for MASLD. All overweight or obese individuals with NAFLD met at least 1 CMRF(s) and fulfilled the definition of MASLD naturally (
Fig. 1B). Cohen’s κ (95% CI) and BMI-stratified confusion matrix for the NHANES cohort and Korean MRE cohort have also been presented to suggest the agreement between the historical NAFLD and MASLD. (
Table 2) Across all cohorts, the discrepancy between MASLD and NAFLD populations was observed solely in normal-weight patients, with more significant dif-ferences noted among those from community/health check-ups. The sensitivity analysis in the NHANES cohort using CAP≥274 dB/m for defining steatosis yielded similar results, with 16.7% of normal-weight individuals meeting the criteria for NAFLD, but not MASLD (
Supplementary Fig. 2).
In both the hospital-based and community-based cohorts, participants with normal-weight MASLD were older, had lower alanine aminotransferase levels, and presented with a lower waist circumference than those with overweight MASLD (
Table 1,
Supplementary Table 6). The prevalence of advanced fibrosis was comparable between the normal-weight and overweight/obese groups in the HK biopsy cohort, the HK MRS cohort, and the Korean MRE cohort, as indicated by LSM using VCTE and MRE. However, in the NHANES cohort, participants in the normal-weight MASLD group were less likely to have advanced fibrosis (
Table 1).
The distribution of CMRFs was examined across all five cohorts from the hospital or community/health checkup settings (
Fig. 2). In cohorts with normal-weight MASLD, the prevalence of CMRF1 ranged from 4% to 50%, based on waist circumference. In the HK biopsy and HK CDARS cohorts from hospital settings, CMRF2 presented the highest proportion among CMRF2–5 in both the normal-weight and overweight/obese groups. The prevalence of CMRF3 ranked first among CMRF2–5 in all community-based/health checkup cohorts except for the Korean MRE cohort. Patients with overweight/obese MASLD consistently exhibited a higher total number of CMRFs than those with normal-weight MASLD in all cohorts. Hospital-based cohorts exhibited a higher number of CMRFs than community or health check-up cohorts (
Fig. 2).
In the NHANES cohort where alcohol consumption records were available, the distribution of BMI categories within subgroups of MetALD and ALD was examined (
Supplementary Table 7). A similar pattern in the distribution of BMI categories was observed in these two cohorts. Obese individuals ranged from 61.9% to 64.2% across each subgroup, representing the major composition among individuals with MetALD or ALD.
LRE
No significant difference in the cumulative incidence of LRE was observed across the three BMI categories in the hospital-based HK CDARS cohort (Gray’s test,
P=0.54,
Fig. 3). The incidence rates of LRE in participants with normal-weight and overweight/obese MASLD were 2.81 and 2.59 per 1,000 person-years, respectively (
Supplementary Table 8).
In hierarchically adjusted analyses, neither overweight nor obesity was significantly associated with LRE in the non-adjusted, age-and sex-adjusted and fully adjusted models by age, sex, CMRF2–5, FIB-4, and extrahepatic diseases in the CDARS cohort (fully adjusted HRs: overweight 0.92 [95% CI 0.55–1.55],
P=0.75; obese 1.29 [95% CI 0.85–1.96],
P=0.23;
Table 3).
All-cause mortality
Normal-weight, overweight, and obese participants exhibited various 15-year cumulative all-cause mortality rates, with the highest observed in the normal-weight MASLD group in the HK CDARS cohorts (log-rank test,
P<0.001;
Supplementary Fig. 3). At a median follow-up duration of 4.6 (Q1–Q3 3.9–7.4) years, the incidence rate of all-cause mortality among participants with normal-weight MASLD in the CDARS cohort was 23.44 (95% CI 20.56–26.72) per 1,000 person-years, which was significantly higher than that of their overweight counterparts (13.80 [95% CI 13.03–14.61] per 1,000 person-years,
P<0.001;
Supplementary Table 8).
In multivariable analyses with hierarchical adjustments, both overweight and obesity were found to be consistently associated with a lower risk of all-cause mortality (fully adjusted HRs: overweight 0.69 [95% CI 0.58–0.83],
P<0.001; obese 0.74 [95% CI 0.64–0.86],
P<0.001;
Table 4,
Supplementary Fig. 2). Similar HRs were shown in the lag analysis with 2-year time window to overcome potential reverse causation (fully adjusted: overweight: 0.77 [95% CI 0.63– 0.94], P = 0.009; obese: 0.81 [95% CI 0.68–0.95], P = 0.01;
Supplementary Table 9).
DISCUSSION
In this study, we examined the CMRF number distributions between normal-weight and overweight/obese NAFLD across five cohorts from HK, South Korea, and the US, in hospitals, communities, and health checkup set-tings. Special attention was given to the proportions of 0 CMRF, which represented the discrepancy between the NAFLD and MASLD populations. While overweight/obese NAFLD presented at least 1 CMRF(s) and fulfilled the criteria of NAFLD naturally, normal-weight patients consistently showed an incomplete overlap between NAFLD and MASLD. In hospital settings, normal-weight NAFLD and MASLD populations reached 98–100% overlap; however, the discrepancy between NAFLD and MASLD was more evident in community settings. The proportion of patients with absent CMRF among normal-weight NAFLD populations ranged from 9.0% to 26.7%, and these individuals would be classified as having cryptogenic SLD.
The prevalence of NAFLD among the normal-weight population ranged from 10.3% to 15.1% in our study, consistent with 9.0–10.6% as reported by previous meta-analyses conducted in different populations [
12,
13]. Both the Asian and Western cohorts (the NHANES cohort) showed a significant discrepancy of up to 2.8% between the prevalence of normal-weight NAFLD and normal-weight MASLD, which could be explained by absence of CMRFs. Notably, these findings underscore the need for caution when applying findings from previous studies on normal-weight NAFLD to the MASLD population with a normal BMI, particularly in settings where a lower metabolic disease burden is expected.
Normal-weight and overweight/obese individuals with MASLD from various cohorts shared a similar distribution of CMRF2–5, with the primary difference in CMRF1 due to the definition. Notably, only 8% in the Korean MRE cohort met the CMRF1 criteria, a significantly lower percentage compared to that in other cohorts. However, previous research on the same cohort has suggested a similarly low prevalence of abdominal obesity of less than 2% in the lean Korean population [
14]. The low prevalence of abdominal obesity within normal-weight MASLD thus seems reasonable and contributed to the distinct CMRF1 distribution. Across different cohorts, those from hospital settings presented a greater number of CMRFs compared to those from community settings, and the discrepancy was more pronounced when examining the proportions of total number of CMRFs≥3–5 (
Supplementary Fig. 4).
Patients with normal-weight MASLD had higher FIB-4 levels than those with overweight/obese MASLD in both hospital and community settings, which is consistent with previous findings on normal-weight NAFLD [
15]. Paradoxically, in cohorts with VCTE-LSM, MRE-LSM, or histology results, normal-weight individuals had a similar or lower prevalence of advanced fibrosis than their overweight counterparts. The older age of the normal-weight MASLD group contributed to the discrepancy in FIB-4, along with the low accuracy of FIB-4 in community settings [
16,
17]. Conversely, overweight or obese individuals with excess percutaneous fat could confound LSM measurements by VCTE and the fibrosis stage [
18,
19]. Based on more accurate MRE-LSM measurements in the Korean MRE cohort and liver histology in the HK biopsy cohort, normal-weight MASLD and overweight/obese MASLD shared comparable risks of significant/advanced fibrosis.
In this study, we examined the longitudinal clinical outcomes in the HK CDARS cohort. The overweight or obese group was associated with similar risk of LRE development compared with the normal-weight group. Reasons included the fact that the population recruited in the HK CDARS cohort was based on ICD-9 coding for NAFLD diagnosis, which selected patients with more advanced liver disease regardless of BMI subgroup. Consequently, the severity of fibrosis as reflected by FIB-4, instead of BMI, was the key predictor of LRE.
Notably, patients with normal-weight MASLD had a higher risk of all-cause mortality compared with the overweight/obese patients, despite the normal-weight group exhibiting a lower metabolic disease burden than the overweight/obese MASLD in all datasets (
Supplementary Fig. 2). This could be attributed to heterogeneity in the normal-weight MASLD population which included both really “normal” and underweight individuals. The latter group may represent a distinct category with advanced underlying diseases that affect the prognosis. We compared the distribution of nonhepatic comorbidities, and the normal-weight group exhibited a higher prevalence of chronic extrahepatic diseases, despite having a lower metabolic burden and a lower prevalence of hypertension and diabetes (
Supplementary Table 10). This is in line with previous findings that metabolic dysfunction-associated fatty liver disease in individuals with normal weight was correlated with an increased risk of several extrahepatic manifestations [
20]. Interestingly, even in the fully adjusted model including extrahepatic diseases as covariates, the normal-weight group was consistently associated with higher risk of all-cause mortality; this finding aligned with results from other adjusted models and extended to 2-year lag analyses. In addition, the classification of “normal-weight” by BMI may mask the presence of the central (waist-based) obesity, of which the prevalence reached ~45% even within the normal-weight group. Central obesity is strongly associated with insulin resistance and metabolic dysfunction, contributing to the heightened mortality risk for the group as well. Notably, given our findings were driven primarily by one specific cohort, HKCDARS cohort, the results should be interpreted with caution and need further validation in diverse populations.
Notably, contradictory findings emerged that obesity was associated with a reduced risk of all-cause mortality but a similar or elevated risk of LREs compared to normal-weight individuals. The discrepancy stems from the distinct sets of risk factors driving LREs versus all-cause mortality. The risk of all-cause mortality is driven primarily by extrahepatic risk including the aforementioned subtle risks, which are more prevalent in the normal-weight cohort. In contrast, the risk of LRE is fueled by liver disease progression exacerbated by the adiposity-related metabolic burden. Excess adiposity amplifies insulin resistance and lipotoxicity, leading to higher rates of hepatic progression in the obese individuals with MASLD [
21]. Nevertheless, in this hospital-based cohort where all patients received the clinical diagnosis of NAFLD at recruitment, the elevated risk of LRE dominated by metabolic disturbances in the overweight group could be mitigated and modified through sequential covariate adjustments.
This study has the strength of including patients with normal-weight or overweight/obese MASLD across diverse populations and settings to examine the clinical characteristics, disease burden, and adverse outcomes in the normal-weight versus the overweight/obese populations with MASLD. Notably, this is the first study to investigate the differences between normal-weight MASLD and normal-weight NAFLD among different ethnic populations in the East and West. Furthermore, HK CDARS cohort had longitudinal data and enabled us to observe the long-term adverse liver and non-liver outcomes for both normal-weight and overweight/obese MASLD.
The limitations of this study include the heterogeneity across cohorts. In the CDARS cohort, only patients with NAFLD were initially included, resulting in incomplete recruitment of the SLD population. Nevertheless, this between-cohort heterogeneity did not affect the investigation of proportions of overlap between two nomenclatures. In the NHANES cohort, although the validated accuracy of CAP in predicting steatosis has been established [
7], the cutoff of CAP used for diagnosing hepatic steatosis may affect the results. We identified SLD using a cutoff of the median CAP value of 248 dB/m, and the sensitivity analysis with a higher threshold yielded similar results. While heterogeneity in recruitment and diagnostics introduces complexity, the consistency of key findings across cohorts, lack of effect modification by diagnostic modality, and supportive sensitivity analyses suggest our conclusions are robust. Second, records of quantitative alcohol consumption were not available in the HK datasets. Since both NAFLD and MASLD used the same alcohol consumption thresholds (<30 g/d for men and <20 g/d for women), any potential misclassification due to incomplete alcohol records would affect both definitions similarly. This methodological symmetry mitigates concerns that incomplete alcohol data would substantially bias the observed agreement between the two definitions. However, the impact of imperfect alcohol assessment remains an important consideration for the field. Another limitation is the under-ascertainment of obesity and CMRFs across cohorts derived from electronic health systems. There was a risk of under-captured CMRFs resulting from incomplete clinical records like waist circumference, fasting blood glucose, triglycerides, and subsequent misclassification of high-risk, overweight individuals into the normal-weight group. The latter would artificially inflate the diagnostic discrepancy between NAFLD and MASLD in normal-weight individuals and lead to an overestimation of adverse outcomes attributed to the normal-weight phenotype. Although we employed a two-year time window to define CMRFs to mitigate this bias, the findings should still be interpreted with caution. Last, longitudinal data were not available for the NHANES cohort in this study. The prognosis of normal-weight MASLD in the Western populations remains to be investigated.
In conclusion, the reclassification from NAFLD to MASLD introduces a notable discrepancy within the normal-weight population, leading to a reduced overlap between NAFLD and MASLD in normal-weight individuals compared with their overweight or obese counterparts. Caution is warranted when applying the findings based on NAFLD to the current MASLD, especially for populations with normal BMI and a lower cardiometabolic burden.
FOOTNOTES
-
Data Availability
The datasets were generated and/or analyzed by corresponding authors, who obtained the permission to access them during this study.
-
Authors’ contributions
Sherlot Juan Song: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing–original draft, Project administration. Eileen Laureal Yoon: Data curation, Investigation, Resources, Writing–review & editing. Vincent Wai-Sun Wong: Conceptualization, Methodology, Data curation, Writing - Review & Editing, Supervision. Ae Jeong Jo: Data curation, Software, Investigation. Grace LaiHung Wong: Data curation, Resources, Writing - Review & Editing, Project administration. Jimmy Che-To Lai: Data curation, Investigation, Project administration. Dae Won Jun: Data curation, Investigation, Resources, Writ ing–review & editing. Terry Cheuk-Fung Yip: Conceptualization, Methodology, Data curation, Writing - Review & Editing, Supervision, Project administration.
-
Acknowledgements
This work was partially supported by the Health and Medical Research Fund of Health Bureau of HKSAR Government, Hong Kong, China (reference number: 20210891).
-
Conflicts of Interest
Dr Song and Dr Yoon had no conflicts of interest to declare. Dr V. Wong reported receiving personal consultant fees from Boehringer Ingelheim, Echosens, Gilead Sciences, grants from Gilead Sciences, personal consultant fees from Intercept, Inventiva, Novo Nordisk, personal consultant fees from Pfizer, Sagimet Biosciences, TARGET PharmaSolutions Visirna, and being a cofounder of Illuminatio outside the submitted work. Dr G. Wong reported receiving personal consultant fees from AstraZeneca, Echosens, GlaxoSmithKline, Janssen and Virion Biotherapeutics and research grants from Gilead Sciences Research outside the submitted work. Dr Jo, Dr Lai and Dr Jun had no conflicts of interest to declare. Dr Yip reported serving as an advisory committee member and a speaker for Gilead Sciences outside the submitted work.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (
http://www.e-cmh.org).
Supplementary Table 3.
Baseline characteristics of participants with MASLD in the HK Biopsy Cohort, HK CDARS Cohort, HK MRS Cohort, NHANES Cohort, and Korean MRE Cohort
Supplementary Table 4.
Prevalence of missing data for baseline characteristics in patients with metabolic dysfunction-associated steatotic liver disease from the HK CDARS Cohort, NHANES Cohort, and Korean MRE Cohort
Supplementary Table 9.
The 2-year lag analysis of hierarchical adjusted model for all-cause mortality in the HK CDARS Cohort Cox regression models on all-cause mortality
Supplementary Figure 1.
Flowcharts for all cohorts detailing harmonization in definitions and data processing. (A) HK biopsy Cohort. (B) HK MRS Cohort. (C) HK CDARS Cohort. (D) NHANES cohort. (E) Korean MRE Cohort. HK, Hong Kong; MRS, magnetic resonance spectroscopy; CDARS, Clinical Data Analysis and Reporting System; NHANES, National Health and Nutrition Examination Survey; MRE, magnetic resonance elastography; CMRF, cardiometabolic risk factors as defined in the methods section; NAFLD, non-alcoholic fatty liver disease; MASLD, metabolic dysfunction-associated steatotic liver disease; VCTE, vibration-controlled transient elastography.
Supplementary Figure 2.
Sensitivity analysis for the normal-weight NHANES Cohort. Steatosis was defined based on the controlled attenuation parameter (CAP) measurement at a cutoff of 274 dB/m for the normal-weight NHANES Cohort. Patients without available alcohol drinking records were excluded. CMRF, cardiometabolic risk factors as defined in the methods section. NHANES, National Health and Nutrition Examination Survey; MASLD, metabolic dysfunction-associated steatotic liver disease. NAFLD, non-alcoholic fatty liver disease.
Supplementary Figure 3.
All-cause mortality among patients with MASLD in the HK CDARS Cohort. MASLD, metabolic dysfunction-associated steatotic liver disease; HK, Hong Kong; CDARS, Clinical Data Analysis and Reporting System.
Supplementary Figure 4.
Bar charts of total CMRF numbers in five cohorts with normal-weight vs. overweight/obese MASLD. (A) Cohorts with normal-weight MASLD. (B) Cohorts with overweight or obese MASLD. CMRF, cardiometabolic risk factors as defined in the methods section; MASLD, metabolic dysfunction-associated steatotic liver disease; HK, Hong Kong; CDARS, Clinical Data Analysis and Reporting System; MRS, magnetic resonance spectroscopy; NHANES, National Health and Nutrition Examination Survey; MRE, magnetic resonance elastography.
Figure 1.CMRF distributions of patients with normal-weight NAFLD (A) and overweight or obese (B) NAFLD across five cohorts. (i) HK Biopsy Cohort. (ii) HK MRS Cohort. (iii) NHANES Cohort. (iv) Korean MRE Cohort. (v) HK CDARS cohort. For HK cohorts and Korean MRE Cohort, the cut-off of 23 kg/m2 was used to define normal-weight versus overweight or obese NAFLD; for NHANES cohort, 25 kg/m2 was used as the threshold. CMRF, cardiometabolic risk factors; NAFLD, non-alcoholic fatty liver disease; HK, Hong Kong; MRS, magnetic resonance spectroscopy; NHANES, National Health and Nutrition Examination Survey; MRE, magnetic resonance elastography; CDARS, Clinical Data Analysis and Reporting System.
Figure 2.Prevalence of metabolic risk factors among patients with normal-weight MASLD (A) and overweight or obese MASLD (B) across five cohorts. For the HK and Korean cohorts, the body mass index cut-off of 23 kg/m2 was used to define normal-weight versus overweight or obese MASLD; for the NHANES cohort, the cut-off of 25 kg/m2 was used as the threshold. For normal-weight MASLD, waist circumference≥90 (Asian) 94 (Western) cm in men or >80 cm in women was defined as meeting CMRF1. MASLD, metabolic dysfunction-associated steatotic liver disease; CMRF, cardiometabolic risk factors; HDL, high-density lipoprotein; HK, Hong Kong; CDARS, Clinical Data Analysis and Reporting System; MRS, magnetic resonance spectroscopy; NHANES, National Health and Nutrition Examination Survey; MRE, magnetic resonance elastography.
Figure 3.Liver-related events among patients with MASLD in HK CDARS Cohort. MASLD, metabolic dysfunction-associated steatotic liver disease. Gray’s test was used for statistical comparison. Levels of significance: P=0.05. MASLD, metabolic dysfunction-associated steatotic liver disease; HK, Hong Kong; CDARS, Clinical Data Analysis and Reporting System.
Table 1.Baseline characteristics of participants with MASLD in the HK CDARS cohort, NHANES cohort, and Korean MRE cohort
Table 1.
|
HK CDARS cohort (N=15,265)
|
NHANES cohort (N=1,399)
|
Korean MRE cohort (N=3,275)
|
|
Normal-weight MASLD (n=1,733) |
Overweight/obese MASLD (n=13,532) |
P-value |
Normal-weight MASLD (n=123) |
Overweight/obese MASLD (n=1,276) |
P-value |
Normal-weight MASLD (n=299) |
Overweight/obese MASLD (n=2,976) |
P-value |
|
Age (yr) |
62.9±11.9 |
57.9±12.0 |
<0.001 |
58.7±15.3 |
52.5±16.6 |
<0.001 |
51.2±9.4 |
47.5±9.7 |
<0.001 |
|
Sex, male (%) |
605 (34.9) |
5,968 (44.1) |
<0.001 |
75 (60.9) |
739 (57.9) |
0.57 |
241 (80.6) |
2,708 (91.0) |
<0.001 |
|
BMI (kg/m2) |
21.4±1.5 |
28.9±4.3 |
<0.001 |
23.4±1.2 |
33.7±7.0 |
<0.001 |
21.8±1.1 |
26.9±2.7 |
<0.001 |
|
Waist circumference (cm) |
82.2±6.5 |
97.1±10.3 |
<0.001 |
88.9±6.1 |
110.5±14.5 |
<0.001 |
80.3±5.1 |
91.3±7.2 |
<0.001 |
|
ALT (U/L) |
33 (21–57) |
44 (28–67) |
<0.001 |
18 (13–24) |
21 (15–30) |
<0.001 |
24 (17–37) |
33 (23–49) |
<0.001 |
|
AST (U/L) |
31 (23–51) |
35 (25–52) |
<0.001 |
20 (17–24) |
20 (16–24) |
0.81 |
26 (21–32) |
28 (23–37) |
0.36 |
|
Albumin (g/dL) |
4.1±0.6 |
4.1±0.5 |
<0.001 |
4.2±0.3 |
4.1±0.3 |
<0.001 |
4.6±0.3 |
4.6±0.3 |
0.86 |
|
Creatinine (µmol/L) |
69 (59–84) |
72 (61–86) |
<0.001 |
75 (63–88) |
77 (65–91) |
0.19 |
/ |
/ |
/ |
|
CMRF (%) |
|
|
|
|
|
|
|
|
|
|
1-Overweight |
44.7 |
100 |
<0.001 |
51.2 |
100 |
<0.001 |
8.4 |
100 |
<0.001 |
|
BMI-based criteria |
0 |
100 |
/ |
0 |
100 |
/ |
0 |
100 |
/ |
|
Waist-based criteria |
44.7 |
66.7 |
<0.001 |
51.2 |
93.7 |
<0.001 |
8.4 |
59.1 |
<0.001 |
|
2-Diabetes |
93.0 |
95.4 |
<0.001 |
50.4 |
52.6 |
0.71 |
67.6 |
63.6 |
0.17 |
|
3-Hypertension |
79.4 |
86.0 |
<0.001 |
61.8 |
62.0 |
>0.99 |
26.4 |
36.6 |
<0.001 |
|
4-High triglycerides |
87.9 |
89.9 |
0.01 |
53.7 |
42.5 |
0.02 |
55.2 |
61.7 |
0.03 |
|
5-Low HDL |
85.9 |
89.6 |
<0.001 |
52.8 |
58.6 |
0.30 |
38.1 |
37.9 |
0.93 |
|
MAFLD (%) |
97.0 |
100 |
<0.001 |
70.7 |
100 |
<0.001 |
/ |
/ |
/ |
|
FIB-4 |
1.38 (0.92–2.19) |
1.15 (0.77–1.80) |
<0.001 |
1.19 (0.84–1.66) |
0.94 (0.63–1.37) |
<0.001 |
1.12 (0.79–1.47) |
0.94 (0.73–1.28) |
<0.001 |
|
FIB-4 categories (%) |
|
|
<0.001 |
|
|
0.15 |
|
|
<0.001 |
|
<1.3 (2) |
46.6 |
58.0 |
|
74.1 |
84.1 |
|
65.2 |
76.3 |
|
|
1.3–2.67 |
35.8 |
30.0 |
|
21.2 |
14.7 |
|
29.1 |
21.9 |
|
|
≥2.67 |
17.6 |
12.0 |
|
1.7 |
1.3 |
|
5.7 |
1.8 |
|
|
LSM (kPa) |
/ |
/ |
/ |
4.4 (3.8–5.2) |
5.5 (4.5–6.9) |
<0.001 |
/ |
/ |
/ |
|
≥8.0 (%) |
/ |
/ |
/ |
3.3 |
15.9 |
0.003 |
/ |
/ |
/ |
|
≥12.0 (%) |
/ |
/ |
/ |
0 |
6.5 |
0.007 |
/ |
/ |
/ |
|
MRE (kPa) |
/ |
/ |
/ |
/ |
/ |
/ |
2.3 (2.0–2.6) |
2.4 (2.1–2.6) |
0.03 |
|
≥3.0 (%) |
/ |
/ |
/ |
/ |
/ |
/ |
7.4 |
8.7 |
0.42 |
|
≥3.6 (%) |
/ |
/ |
/ |
/ |
/ |
/ |
3.7 |
2.1 |
0.08 |
Table 2.
Concordance between historical NAFLD and MASLD in the NHANES cohort and Korean MRE cohort
*
(A) Cohen’s κ (95% CI)
Table 2.
|
Cohort |
Subgroup |
Cohen's κ (95% CI) |
|
NHANES |
BMI<25 kg/m2
|
0.756 (0.649–0.863) |
|
BMI≥25 kg/m2
|
1.000 (0.964–1.000) |
|
Korean MRE |
BMI<23 kg/m2
|
0.754 (0.690–0.818) |
|
BMI≥23 kg/m2
|
1.000 (0.967–1.000) |
(B) BMI-stratified Confusion matrices
Table
|
Cohort/subgroup |
BMI<23 (25) kg/m2
|
BMI≥23 (25) kg/m2
|
|
NHANES |
|
NAFLD (–) |
NAFLD (+) |
|
NAFLD (–) |
NAFLD (+) |
|
MASLD (–) |
172 |
40 |
MASLD (–) |
1,751 |
0 |
|
MASLD (+) |
0 |
123 |
MASLD (+) |
0 |
1,276 |
|
Korean MRE |
|
NAFLD (–) |
NAFLD (+) |
|
NAFLD (–) |
NAFLD (+) |
|
MASLD (–) |
539 |
109 |
MASLD (–) |
455 |
0 |
|
MASLD (+) |
0 |
299 |
MASLD (+) |
0 |
2,976 |
Table 3.Hierarchical adjusted hazards of BMI categories associated with liver-related events in participants with MASLD in the HK CDARS cohort
Table 3.
|
BMI categories§
|
Crude
|
Adjusted 1*
|
Adjusted 2†
|
Fully-adjusted‡
|
|
SHRs (95% CI) |
P-value |
SHRs (95% CI) |
P-value |
SHRs (95% CI) |
P-value |
SHRs (95% CI) |
P-value |
|
Normal weight |
Reference |
- |
Reference |
- |
Reference |
- |
Reference |
- |
|
Overweight |
0.81 (0.48–1.36) |
0.41 |
0.87 (0.52–1.47) |
0.61 |
0.92 (0.55–1.55) |
0.76 |
0.92 (0.55–1.55) |
0.75 |
|
Obese |
0.98 (0.65–1.48) |
0.93 |
1.30 (0.86–1.97) |
0.22 |
1.30 (0.86–1.97) |
0.21 |
1.29 (0.85–1.96) |
0.23 |
Table 4.Hierarchical adjusted hazards of BMI categories associated with all-cause mortality in participants with MASLD in the HK CDARS cohort
Table 4.
|
BMI categories§
|
Crude
|
Adjusted 1*
|
Adjusted 2†
|
Fully-adjusted‡
|
|
HRs (95% CI) |
P-value |
HRs (95% CI) |
P-value |
HRs (95% CI) |
P-value |
HRs (95% CI) |
P-value |
|
Normal weight |
Reference |
- |
Reference |
- |
Reference |
- |
Reference |
- |
|
Overweight |
0.65 (0.54–0.78) |
<0.001 |
0.66 (0.54–0.79) |
<0.001 |
0.67 (0.56–0.81) |
<0.001 |
0.69 (0.58–0.83) |
<0.001 |
|
Obese |
0.53 (0.46–0.61) |
<0.001 |
0.73 (0.63–0.85) |
<0.001 |
0.72 (0.62–0.83) |
<0.001 |
0.74 (0.64–0.86) |
<0.001 |
Abbreviations
alcohol-related liver disease
controlled attenuation parameter
Clinical Data Analysis and Reporting System
cardiometabolic risk factor
International Classification of Diseases
liver stiffness measurement
metabolic dysfunction-associated steatotic liver disease
magnetic resonance elastography
magnetic resonance spectroscopy
non-alcoholic fatty liver disease
National Health and Nutrition Examination Survey
vibration-controlled transient elastography
REFERENCES
- 1. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023;78:1966-1986.
- 2. Song SJ, Lai JC, Wong GL, Wong VW, Yip TC. Can we use old NAFLD data under the new MASLD definition? J Hepatol 2024;80:e54-e56.
- 3. Chen L, Tao X, Zeng M, Mi Y, Xu L. Clinical and histological features under different nomenclatures of fatty liver disease: NAFLD, MAFLD, MASLD and MetALD. J Hepatol 2024;80:e64-e66.
- 4. Hagström H, Vessby J, Ekstedt M, Shang Y. 99% of patients with NAFLD meet MASLD criteria and natural history is therefore identical. J Hepatol 2024;80:e76-e77.
- 5. Danpanichkul P, Suparan K, Prasitsumrit V, Ahmed A, Wijarnpreecha K, Kim D. Long-term outcomes and risk modifiers of metabolic dysfunction-associated steatotic liver disease between lean and non-lean populations. Clin Mol Hepatol 2025;31:74-89.
- 6. Wong VW, Chu WC, Wong GL, Chan RS, Chim AM, Ong A, et al. Prevalence of non-alcoholic fatty liver disease and advanced fibrosis in Hong Kong Chinese: a population study using proton-magnetic resonance spectroscopy and transient elastography. Gut 2012;61:409-415.
- 7. European Association for the Study of the Liver (EASL); European Association for the Study of Diabetes (EASD); European Association for the Study of Obesity (EASO). EASL-EASD-EASO Clinical Practice Guidelines on the management of metabolic dysfunction-associated steatotic liver disease (MASLD). J Hepatol 2024;81:492-542.
- 8. Eddowes PJ, Sasso M, Allison M, Tsochatzis E, Anstee QM, Sheridan D, et al. Accuracy of FibroScan controlled attenuation parameter and liver stiffness measurement in assessing steatosis and fibrosis in patients with nonalcoholic fatty liver disease. Gastroenterology 2019;156:1717-1730.
- 9. Siddiqui MS, Vuppalanchi R, Van Natta ML, Hallinan E, Kowdley KV, Abdelmalek M, et al.; NASH Clinical Research Network. Vibration-controlled transient elastography to assess fibrosis and steatosis in patients with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol 2019;17:156-163.e2.
- 10. Petroff D, Blank V, Newsome PN, Voican CS, Thiele M, et al. Assessment of hepatic steatosis by controlled attenuation parameter using the M and XL probes: an individual patient data meta-analysis. Lancet Gastroenterol Hepatol 2021;6:185-198.
- 11. Oh JH, Ahn SB, Cho S, Nah EH, Yoon EL, Jun DW. Diagnostic performance of non-invasive tests in patients with MetALD in a health check-up cohort. J Hepatol 2024;81:772-780.
- 12. Ye Q, Zou B, Yeo YH, Li J, Huang DQ, Wu Y, et al. Global prevalence, incidence, and outcomes of non-obese or lean non-alcoholic fatty liver disease: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol 2020;5:739-752.
- 13. Zhang J, Huang X, Dong L, Yang Y, Kong D. Epidemiology of lean/non-obese nonalcoholic fatty liver disease in China: a systematic review and meta-analysis. Saudi Med J 2023;44:848-863.
- 14. Oh H, Kwak SY, Jo G, Lee J, Park D, Lee DH, et al. Adiposity and mortality in Korean adults: a population-based prospective cohort study. Am J Clin Nutr 2021;113:142-153.
- 15. Nabi O, Lapidus N, Boursier J, de Ledinghen V, Petit JM, Kab S, et al. Lean individuals with NAFLD have more severe liver disease and poorer clinical outcomes (NASH-CO Study). Hepatology 2023;78:272-283.
- 16. Graupera I, Thiele M, Serra-Burriel M, Caballeria L, Roulot D, Wong GL, et al. Low accuracy of FIB-4 and NAFLD fibrosis scores for screening for liver fibrosis in the population. Clin Gastroenterol Hepatol 2022;20:2567-2576.e6.
- 17. Kjaergaard M, Lindvig KP, Thorhauge KH, Andersen P, Hansen JK, Kastrup N, et al. Using the ELF test, FIB-4 and NAFLD fibrosis score to screen the population for liver disease. J Hepatol 2023;79:277-286.
- 18. Wong VW, Irles M, Wong GL, Shili S, Chan AW, Merrouche W, et al. Unified interpretation of liver stiffness measurement by M and XL probes in non-alcoholic fatty liver disease. Gut 2019;68:2057-2064.
- 19. Wong GL, Chan HL, Choi PC, Chan AW, Lo AO, Chim AM, et al. Association between anthropometric parameters and measurements of liver stiffness by transient elastography. Clin Gastroenterol Hepatol 2013;11:295-302.e1-3.
- 20. Eslam M, El-Serag HB, Francque S, Sarin SK, Wei L, Bugianesi E, et al. Metabolic (dysfunction)-associated fatty liver disease in individuals of normal weight. Nat Rev Gastroenterol Hepatol 2022;19:638-651.
- 21. Vesković M, Šutulović N, Hrnčić D, Stanojlović O, Macut D, Mladenović D. The interconnection between hepatic insulin resistance and metabolic dysfunction-associated steatotic liver disease-the transition from an adipocentric to liver-centric approach. Curr Issues Mol Biol 2023;45:9084-9102.
, Terry Cheuk-Fung Yip1,2
