Abstract
Aims
The impact of steatotic liver disease (SLD) subtypes on incident type 2 diabetes remains unclear. We investigated the association of metabolic dysfunction‐associated steatotic liver disease (MASLD), MetALD (MASLD with moderate alcohol consumption), and alcohol‐related liver disease (ALD) with incident type 2 diabetes in Korean adults aged 20–39 years.
Materials and Methods
Using the Korean National Health Insurance Service database, we included 6 250 145 adults aged 20–39 who underwent health examinations between 2009 and 2012. SLD subtypes were classified by fatty liver index and alcohol intake. Incident type 2 diabetes was assessed using Cox proportional hazards models.
Results
During a median follow‐up of 10.6 years, 72 028 participants developed type 2 diabetes. Incidence rates per 1000 person‐years were higher in MASLD (10.56), MetALD (9.96), and ALD (12.49) than in non‐SLD (1.29). Multivariable analyses showed significantly increased risks of type 2 diabetes (adjusted hazard ratios [95% CI]: 2.81 [2.78–2.85] for MASLD, 2.92 [2.87–2.98] for MetALD, and 3.99 [3.88–4.09] for ALD).
Conclusion
All SLD subtypes significantly increased the risk of type 2 diabetes in young adults, with the greatest risk in ALD, followed by MetALD and MASLD. Early detection and management of SLD are essential to reduce the future burden of diabetes.
Keywords: diabetes, metabolic, steatosis, young
1. INTRODUCTION
The worldwide increase in type 2 diabetes among young adults represents a significant public health challenge, conferring a disproportionate burden of lifetime morbidity. 1 Steatotic liver disease (SLD) has emerged as a central component of this risk, with multiple cohort studies and meta‐analyses showing that non‐alcoholic fatty liver disease (NAFLD) is associated with a substantially higher incidence of type 2 diabetes compared with individuals without hepatic steatosis. 2 , 3 , 4 , 5 In Korean young adults, this association has been reported as NAFLD was linked to an approximately five‐fold higher incidence of type 2 diabetes over long‐term follow‐up. 6
Recently, an international consensus introduced metabolic dysfunction‐associated steatotic liver disease (MASLD), which underscores the importance of metabolic dysfunction in the pathophysiology of disease. 7 This new framework also defines distinct categories, such as MetALD, for patients with coexisting metabolic risk and moderate alcohol intake, thereby separating them from purely alcohol‐related liver disease (ALD) and enabling more precise risk stratification. 8 However, alcohol consumption has complex dose–response relationships with glycemia: moderate intake has been linked to lower diabetes risk in several meta‐analyses, whereas excessive use abolishes any potential benefit and increases risk, particularly in Asian men. 9 , 10 Consequently, the relative contribution of each SLD subtype to type 2 diabetes development in young adults is not well defined.
Therefore, we aimed to investigate incident type 2 diabetes across distinct SLD subtypes (MASLD, MetALD, and ALD) in a nationwide cohort of Korean adults aged 20–39 years.
2. METHODS
2.1. Study design and data source
We performed a retrospective cohort study using the Korean National Health Insurance Service (NHIS) database, which links biennial health examinations with administrative claims for virtually all Korean residents. The database integrates a wide range of health records for each participant. This information encompasses demographic characteristics, anthropometric measurements (height, weight, waist circumference, systolic and diastolic blood pressures), and key biochemical parameters. The laboratory data include fasting glucose, a full lipid panel, liver function tests, and creatinine. Furthermore, data on lifestyle behaviours (smoking status, alcohol intake, physical activity levels), medical diagnoses based on the International Classification of Diseases, 10th Revision (ICD‐10), and details of prescribed treatments are systematically collected. 11 The specific diagnostic codes used for our analysis are detailed in Table S1.
Individuals aged 20–39 who underwent a national health screening between 2009 and 2012 were included. From an initial pool of 6 891 401 individuals, we excluded participants with missing data for key variables, a history of liver cancer or liver transplantation, or those with preexisting concomitant liver disease, a current diagnosis of cryptogenic SLD, or specific aetiology of SLD (ICD code, Table S1). To accurately identify incident diabetes, we also excluded individuals with pre‐existing diabetes, defined as a baseline fasting plasma glucose ≥126 mg/dL or a prior diagnosis of diabetes during a pre‐baseline wash‐out period. Furthermore, to minimise reverse causality, participants who developed diabetes within the first year of follow‐up (lag period) were excluded. After these exclusions, the final study population consisted of 6 250 145 individuals (Figure 1).
FIGURE 1.
Flow chart.
This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H‐2401‐130‐1505). The requirement to obtain informed consent from patients was waived by the IRB due to the use of de‐identified data.
2.2. Assessment of hepatic steatosis and SLD subtype classification
Hepatic steatosis was ascertained using the fatty liver index (FLI), a validated surrogate marker calculated from body mass index (BMI), waist circumference, triglyceride levels, and gamma‐glutamyl transferase (GGT) activity. 12 Consistent with prior NHIS studies in young adults and to maximise sensitivity for steatosis, we defined SLD as FLI ≥30. 13 , 14
MASLD was defined as the presence of SLD without any other identifiable causes of steatosis in combination with one or more of the following cardiometabolic risks 15 : (i) a BMI ≥23 kg/m2 or a WC ≥90 cm for men and ≥85 cm for women 16 ; (ii) a blood pressure ≥130/85 mmHg or treatment with antihypertensive medication; (iii) a fasting plasma triglyceride level ≥150 mg/dL or the use of lipid‐lowering therapy; (iv) a plasma high‐density lipoprotein cholesterol level <40 mg/dL for men and <50 mg/dL for women or the use of cholesterol‐lowering treatment; and (v) a fasting glucose level ≥100 mg/dL. Because individuals with preexisting diabetes were excluded, the diabetes‐related criterion for MASLD was not applied (diagnosis of type 2 diabetes or use of antidiabetic medication).
MetALD was defined as meeting the criteria for MASLD with moderate alcohol consumption (30–60 g/day for men, 20–50 g/day for women). 7 ALD indicated steatosis in the setting of heavy alcohol use exceeding these thresholds (and 60 g/day for men or 50 g/day for women) or a diagnosis of alcohol abuse/alcohol‐related liver disease (ICD‐code, Table S1).
2.3. Outcome definition
The primary outcome was the incidence of new‐onset type 2 diabetes. We used a validated definition previously employed in the NHIS study, identifying incident cases as a new diagnosis of type 2 diabetes (ICD‐10 codes E11–E14) with a concurrent prescription for an antidiabetic medication. 6 Participants were followed from their baseline examination date until the first event of type 2 diabetes diagnosis, death, or the end of the study period on December 31, 2021.
2.4. Covariates
At enrollment, data were obtained using a standardised, self‐administered questionnaire. This questionnaire captured smoking status (never, former, or current), physical activity, and alcohol consumption. Physical activity was classified as engaging in moderate‐intensity exercise for ≥30 min on ≥5 days per week or vigorous‐intensity exercise for ≥20 min on ≥3 days per week. Alcohol consumption was categorised as none, mild (<30 g/day for men, <20 g/day for women), moderate (≥30 g/day for men, ≥20 g/day for women), or excessive (≥60 g/day for men, ≥50 g/day for women). Low‐income status was defined as the lowest quantile based on income‐related insurance contributions.
Comorbid conditions were ascertained from ICD‐10 diagnosis codes, prescription records from the preceding year, and results of routine health screenings. Hypertension was defined as codes I10–I13 or I15 together with a prescription for antihypertensive agents, or a measured systolic/diastolic blood pressure ≥140/90 mmHg. Type 2 diabetes was defined by codes E11–E14 with prescriptions for oral hypoglycemic agents or insulin, or a fasting plasma glucose ≥126 mg/dL. Dyslipidemia was defined by code E78 with lipid‐lowering therapy or a total cholesterol ≥240 mg/dL. The estimated glomerular filtration rate (eGFR) was calculated from the serum creatinine levels via a formula derived from the Modification of Diet in Renal Disease Study. 17
2.5. Statistical analysis
Baseline characteristics were summarised across SLD categories and compared using appropriate statistical tests. Continuous variables are described as mean ± standard deviation, and categorical variables are presented as number (%). Comparisons of continuous variables were made using Student's t‐test; the chi‐squared test was used for categorical variables. Because triglyceride, aspartate aminotransferase (AST), alanine aminotransferase (ALT), and GGT showed skewed distributions, these variables were log‐transformed before analysis.
We calculated incidence rates for diabetes per 1000 person‐years. To assess the association between SLD subtypes and incident diabetes, we used Cox proportional hazards regression to compute hazard ratios (HRs) and 95% confidence intervals (CIs), with the non‐SLD group as the reference.
Multivariable models were adjusted for a set of clinically relevant covariates selected a priori: age, sex, BMI, low income, smoking status, regular exercise, fasting glucose, total cholesterol, systolic blood pressure, eGFR, and daily alcohol consumption. Sex, income, smoking, and regular exercise were modelled as categorical variables, whereas all other covariates were treated as continuous variables. The proportional hazards assumption was verified using Schoenfeld residuals.
Restricted cubic spline (RCS) analyses were performed for BMI and fasting glucose using three knots placed at the 10th, 50th, and 90th percentiles to evaluate potential nonlinear associations with diabetes risk. The RCS models were adjusted for the same covariates as the main Cox models.
Several sensitivity analyses were conducted to assess the robustness of our findings: (1) exclusion of BMI from the model to evaluate potential overadjustment bias and expansion model that further adjusted for lipid‐lowering and antihypertensive medications to evaluate whether cardiometabolic treatment influenced the associations; (2) application of an alternative FLI threshold of ≥60; (3) use of the Hepatic Steatosis Index (HSI) as an alternative surrogate for hepatic steatosis; and (4) Fine‐Gray competing risk regression accounting for mortality as a competing event. The HSI was calculated as 8 × (ALT/AST ratio) ± BMI (+2 for type 2 diabetes, +2 for female sex), with SLD defined as HSI >36 based on previous validation studies. 18
To address potential alcohol misclassification, two additional analyses were performed: a two‐year landmark analysis that reassessed alcohol categories and SLD subtypes at 2 years after baseline, and a stable‐consumption subgroup analysis restricted to participants who maintained consistent alcohol intake patterns across consecutive examinations.
In addition, we conducted pre‐specified subgroup analyses stratified by age (<30 vs. ≥30 years), sex, BMI (<25 vs. ≥25 kg/m2), smoking status, impaired fasting glucose (IFG) status, and ALT quartiles. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA), and a two‐sided p‐value <0.05 was considered statistically significant.
3. RESULTS
3.1. Study population
Among a total of 6 250 145 participants, 4 561 978 were classified as non‐SLD, 1381732 as MASLD, 228586 as MetALD, and 77 849 as ALD. Table 1 summarises the baseline characteristics of the study population according to the presence and subtype of SLD. Individuals with MASLD, MetALD, and ALD were older than those without SLD (mean ages, 32.6, 32.4, and 32.3 years vs. 30.1 years; p < 0.001). The proportion of men was significantly higher in the MASLD (90.7%), MetALD (95.4%), and ALD (96.0%) groups compared with the non‐SLD group (46.9%, p < 0.001). Socioeconomic and lifestyle factors also differed across groups. A lower household income was more common among non‐SLD participants compared with SLD groups (p < 0.001). Current smoking was markedly more prevalent in the ALD (70.2%) and MetALD (70.5%) groups, followed by MASLD (52.2%), compared with 26.8% in non‐SLD (p < 0.001). Regarding physical activity and comorbidities, regular exercise was more frequent in the SLD groups than in non‐SLD (p < 0.001). Hypertension and dyslipidemia were significantly more prevalent in SLD groups, particularly ALD (22.8% and 17.6%, respectively), compared with non‐SLD (3.4% and 3.1%, p < 0.001). Anthropometric and metabolic indices were consistently higher in participants with SLD (p < 0.001).
TABLE 1.
Baseline characteristics of the study population.
| Non‐SLD | MASLD | MetALD | ALD | p‐value | |
|---|---|---|---|---|---|
| (n = 4 561 978) | (n = 1 381 732) | (n = 228 586) | (n = 77 849) | ||
| Age, years | 30.14 ± 5.04 | 32.59 ± 4.37 | 32.38 ± 4.43 | 32.33 ± 4.51 | <0.001 |
| Male, n (%) | 2 140 805 (46.93) | 1 253 339 (90.71) | 218 027 (95.38) | 74 723 (95.98) | <0.001 |
| Income_low a , n (%) | 1 067 301 (23.40) | 219 941 (15.92) | 33 174 (14.51) | 12 630 (16.22) | <0.001 |
| Smoking, n (%) | <0.001 | ||||
| Never | 2 949 526 (64.65) | 459 880 (33.28) | 30 449 (13.32) | 10 895 (14.00) | |
| Former | 389 532 (8.54) | 201 187 (14.56) | 37 085 (16.22) | 12 332 (15.84) | |
| Current | 1 222 920 (26.81) | 720 665 (52.16) | 161 052 (70.46) | 54 622 (70.16) | |
| Alcohol consumption, n (%) | <0.001 | ||||
| None | 1 946 539 (42.67) | 405 795 (29.37) | 0 (0) | 3730 (4.79) | |
| Mild | 2 294 863 (50.30) | 975 937 (70.63) | 0 (0) | 12 138 (15.59) | |
| Moderate | 272 986 (5.98) | 0 (0) | 228 586 (100) | 4354 (5.59) | |
| Excessive | 47 590 (1.04) | 0 (0) | 0 (0) | 57 627 (74.02) | |
| Regular exercise, n (%) | 566 801 (12.42) | 187 043 (13.54) | 33 567 (14.68) | 12 686 (16.30) | <0.001 |
| Hypertension, n (%) | 156 276 (3.43) | 219 874 (15.91) | 46 092 (20.16) | 17 750 (22.80) | <0.001 |
| Dyslipidemia, n (%) | 139 954 (3.07) | 210 048 (15.20) | 33 479 (14.65) | 13 705 (17.60) | <0.001 |
| BMI, kg/m2 | 21.49 ± 2.47 | 26.91 ± 3.09 | 26.52 ± 3.04 | 26.73 ± 3.21 | <0.001 |
| WC, cm | 73.24 ± 7.32 | 88.44 ± 7.18 | 87.96 ± 7.24 | 88.49 ± 7.68 | <0.001 |
| SBP, mmHg | 114.66 ± 11.98 | 124.99 ± 12.68 | 127.11 ± 12.97 | 127.53 ± 13.04 | <0.001 |
| DBP, mmHg | 71.73 ± 8.62 | 78.58 ± 9.35 | 80.21 ± 9.63 | 80.48 ± 9.7 | <0.001 |
| Glucose, mg/dL | 88.20 ± 10.03 | 92.68 ± 11.24 | 93.8 ± 11.45 | 94.05 ± 11.66 | <0.001 |
| Total cholesterol, mg/dL | 177.2 ± 29.99 | 203.09 ± 35.16 | 202.28 ± 35.14 | 202.46 ± 36.04 | <0.001 |
| HDL cholesterol, mg/dL | 60.0 ± 20.68 | 49.84 ± 24.27 | 53.19 ± 20.37 | 53.41 ± 21.11 | <0.001 |
| LDL cholesterol, mg/dL | 100.77 ± 31.15 | 116.04 ± 39.72 | 108.67 ± 39.2 | 107.39 ± 40.69 | <0.001 |
| eGFR, mL/min/1.73 m2 | 102.72 ± 21.73 | 99.08 ± 21.12 | 99.98 ± 20.83 | 100.32 ± 20.56 | <0.001 |
| TG b , mg/dL | 76.53 (76.50–76.56) | 173.2 (173.06–173.35) | 183.08 (182.70–183.47) | 187.72 (187.03–188.42) | <0.001 |
| AST b , IU/L | 19.59 (19.58–19.59) | 26.55 (26.53–26.57) | 28.43 (28.39–28.48) | 29.98 (29.89–30.07) | <0.001 |
| ALT b , IU/L | 15.60 (15.59–15.61) | 33.82 (33.78–33.85) | 33.88 (33.80–33.95) | 36.13 (35.98–36.27) | <0.001 |
Note: Data are presented as means ± standard deviations for continuous variables and n (%) for categorical variables.
Abbreviations: ALD, alcohol‐associated liver disease; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; MASLD, metabolic dysfunction‐associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol‐related liver disease; SBP, systolic blood pressure; TG, triglyceride; WC, waist circumference.
Lower 25% of income.
Geometric means (95% confidence interval).
3.2. Risk of incident diabetes according to SLD subtype
During a median follow‐up period of 10.6 years, 72 028 new‐onset diabetes cases occurred, yielding an incidence rate of 1.29 per 1000 person‐years. In contrast, substantially higher incidence rates were observed in all SLD subgroups: 10.56 in MASLD, 9.96 in MetALD, and 12.49 in ALD per 1000 person‐years (Table 2). Kaplan–Meier cumulative incidence curves for diabetes according to SLD subtypes were constructed and are presented in Figure 2. In age‐ and sex‐adjusted models, the HRs for incident diabetes were 8.18 (95% CI, 8.10–8.27) for MASLD, 7.83 (95% CI, 7.71–7.94) for MetALD, and 9.91 (95% CI, 9.71–10.11) for ALD, compared with the non‐SLD group. After adjustment for potential confounders including age, sex, income, smoking status, regular physical activity, systolic blood pressure, fasting blood glucose level, total cholesterol level, eGFR, and BMI, the risk remained significantly elevated, with HRs of 2.81 (95% CI, 2.78–2.85) for MASLD, 2.92 (95% CI, 2.87–2.98) for MetALD, and 3.99 (95% CI, 3.88–4.09) for ALD. Taken together, individuals with SLD exhibited a markedly increased risk of diabetes compared with those without SLD, and the risk was highest among those with ALD, followed by MetALD and MASLD.
TABLE 2.
Risk of diabetes development by steatotic liver disease subtype.
| Number | Events | Duration (PYs) | Incidence rate (per 1000 PY) | Hazard ratio (95% confidence interval) | ||
|---|---|---|---|---|---|---|
| Age, sex‐adjusted | Multivariable a | |||||
| Non‐SLD | 4 561 978 | 72 028 | 55 825 938 | 1.29 | 1 (Ref.) | 1 (Ref.) |
| MASLD | 1381 732 | 172 503 | 16 335 197 | 10.56 | 8.18 (8.10, 8.27) | 2.81 (2.78, 2.85) |
| MetALD | 228 586 | 26 980 | 2 708 609 | 9.96 | 7.83 (7.71, 7.94) | 2.92 (2.87, 2.98) |
| ALD | 77 849 | 11 330 | 907 446 | 12.49 | 9.91 (9.71, 10.11) | 3.99 (3.88, 4.09) |
Abbreviations: ALD, alcohol‐associated liver disease; MASLD, metabolic dysfunction‐associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol‐related liver disease; PY, person years; SLD, steatotic liver disease.
Adjusted for age, sex, low income, smoking, exercise, total cholesterol, fasting glucose, systolic blood pressure, estimated glomerular filtration fraction, body mass index, and alcohol consumption.
FIGURE 2.
Cumulative incidence of new‐onset type 2 diabetes among participants stratified by steatotic liver disease subtype. ALD, alcohol‐associated liver disease; MASLD, metabolic dysfunction‐associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol‐related liver disease; SLD, steatotic liver disease.
Sensitivity analyses confirmed the robustness of our findings. Exclusion of BMI from the covariate set yielded consistent results (Table S2). Using a stricter FLI threshold of ≥60, the associations remained consistent with adjusted HRs (95% CIs) of 2.24 (2.22–2.26) for MASLD, 2.16 (2.12–2.20) for MetALD, and 2.79 (2.71–2.86) for ALD versus non‐SLD (Table S3). Similarly, defining SLD using the HSI produced comparable results, indicating that the associations were not dependent on a specific steatosis index (Table S4). Fine‐Gray competing risk analyses yielded subdistribution HRs (95% CIs) of 2.82 (2.67–2.98) for MASLD, 2.92 (2.75–3.11) for MetALD, and 3.98 (3.71–4.26) for ALD, indicating minimal impact of competing mortality (Table S5). Collectively, these analyses demonstrated that the elevated diabetes risk associated with SLD was robust across multiple alternative model specifications.
In addition, we further assessed the potential influence of changes in alcohol consumption over time. The 2‐year landmark analysis showed adjusted HRs of 2.97 (2.93–3.02) for MASLD, 3.20 (3.11–3.28) for MetALD, and 4.49 (4.34–4.65) for ALD. Among participants with stable alcohol consumption, associations persisted with comparable effect estimates (Tables S6 and S7).
RCS analyses revealed nonlinear, progressively increasing associations of both BMI and fasting glucose with diabetes risk. The risk rose steadily across the full range of each variable, without evidence of threshold or U‐shaped effects (Figure S1).
3.3. Subgroup analysis
In stratified analyses, the association between SLD and incident diabetes was consistently observed across age, sex, BMI, and smoking status. The relative risk was particularly elevated among younger adults (<30 years; P for interaction = 0.001) and among leaner individuals, as further demonstrated in analyses using Asian‐specific BMI categories (<23, 23–25, 25–30, and ≥30 kg/m2; P for interaction <0.001). The risk was highest in participants with BMI ≤23 kg/m2 and gradually decreased with increasing BMI (P for interaction <0.001), indicating that adults with SLD face a substantial diabetes risk even in the lean BMI range (Table 3, Figure S2).
TABLE 3.
Subgroup analysis of incident diabetes risk among young adults with SLD.
| Variables | Subgroups | Number of events | Follow‐up duration (person‐years) | Incidence rate (per 1000 p‐y) | Adjusted HR a (95% CI) | P for interaction |
|---|---|---|---|---|---|---|
| Age < 30 years | Non‐SLD | 25 530 | 26 934 434 | 0.95 | 1 (Ref.) | 0.001 |
| MASLD | 40 274 | 4 276 973 | 9.42 | 3.12 (3.07, 3.18) | ||
| MetALD | 6584 | 789 918 | 8.34 | 3.12 (3.03, 3.21) | ||
| ALD | 2788 | 271 382 | 10.27 | 4.24 (4.05, 4.43) | ||
| Age ≥ 30 years | Non‐SLD | 46 498 | 28 891 503 | 1.61 | 1 (Ref.) | |
| MASLD | 132 229 | 12 058 224 | 10.97 | 2.70 (2.66, 2.73) | ||
| MetALD | 20 396 | 1 918 691 | 10.63 | 2.82 (2.77, 2.88) | ||
| ALD | 8542 | 636 064 | 13.43 | 3.85 (3.74, 3.97) | ||
| Male | Non‐SLD | 37 582 | 26 316 592 | 1.43 | 1 (Ref.) | 0.782 |
| MASLD | 150 551 | 14 876 850 | 10.12 | 2.81 (2.77, 2.84) | ||
| MetALD | 25 600 | 2 587 872 | 9.89 | 2.95 (2.89, 3.01) | ||
| ALD | 10 845 | 872 267 | 12.43 | 4.03 (3.92, 4.14) | ||
| Female | Non‐SLD | 34 446 | 29 509 346 | 1.17 | 1 (Ref.) | |
| MASLD | 21 952 | 1 458 346 | 15.05 | 2.86 (2.81, 2.92) | ||
| MetALD | 1380 | 120 736 | 11.43 | 2.54 (2.41, 2.68) | ||
| ALD | 485 | 35 179 | 13.79 | 3.39 (3.09, 3.71) | ||
| BMI <25 kg/m2 | Non‐SLD | 54 393 | 51 057 874 | 1.07 | 1 (Ref.) | 0.033 |
| MASLD | 25 253 | 4 525 455 | 5.58 | 2.77 (2.73, 2.81) | ||
| MetALD | 4799 | 880 582 | 5.45 | 2.88 (2.79, 2.97) | ||
| ALD | 1890 | 282 394 | 6.69 | 3.99 (3.79, 4.19) | ||
| BMI ≥25 kg/m2 | Non‐SLD | 17 635 | 4 768 064 | 3.70 | 1 (Ref.) | |
| MASLD | 147 250 | 11 809 741 | 12.47 | 1.98 (1.94, 2.01) | ||
| MetALD | 22 181 | 1 828 026 | 12.13 | 2.04 (1.99, 2.09) | ||
| ALD | 9440 | 625 052 | 15.10 | 2.76 (2.68, 2.85) | ||
| Never‐smoker | Non‐SLD | 41 145 | 36 026 174 | 1.14 | 1 (Ref.) | 0.267 |
| MASLD | 55 105 | 5 416 662 | 10.17 | 2.99 (2.94, 3.03) | ||
| MetALD | 3262 | 360 116 | 9.06 | 3.08 (2.97, 3.20) | ||
| ALD | 1622 | 126 843 | 12.79 | 4.51 (4.28, 4.75) | ||
| Former‐smoker | Non‐SLD | 6411 | 4 816 242 | 1.33 | 1 (Ref.) | |
| MASLD | 21 131 | 2 417 690 | 8.74 | 2.73 (2.66, 2.81) | ||
| MetALD | 3676 | 446 371 | 8.24 | 2.80 (2.68, 2.92) | ||
| ALD | 1616 | 145 927 | 11.07 | 3.93 (3.72, 4.16) | ||
| Current smoker | Non‐SLD | 24 472 | 14 983 522 | 1.63 | 1 (Ref.) | |
| MASLD | 96 267 | 8 500 845 | 11.32 | 2.67 (2.63, 2.71) | ||
| MetALD | 20 042 | 1 902 122 | 10.54 | 2.80 (2.74, 2.86) | ||
| ALD | 8092 | 634 676 | 12.75 | 3.76 (3.64, 3.87) |
Abbreviations: ALD, alcohol‐associated liver disease; BMI, body mass index; CI, confidence interval; HR, hazard ratio; MASLD, metabolic dysfunction‐associated steatotic liver disease; MetALD, metabolic dysfunction and alcohol‐related liver disease; SLD, steatotic liver disease.
Adjusted for age, sex, low income, smoking, exercise, total cholesterol, fasting glucose, systolic blood pressure, estimated glomerular filtration fraction, body mass index, and alcohol consumption.
In analyses stratified by IFG status, the associations remained significant in both normoglycemic and IFG groups, with higher absolute incidence among participants with IFG (P for interaction <0.001; Table S8).
In analyses stratified by ALT quartiles, diabetes incidence was highest in the top quartile (≥28 IU/L) across SLD subtypes, whereas relative hazards were greatest in the lowest quartile and attenuated with increasing ALT; associations remained significant across all strata (Table S9).
4. DISCUSSION
In this nationwide cohort of adults aged 20–39 years, individuals with SLD are at significantly elevated risk of incident type 2 diabetes within a decade. The risk was greatest for ALD, followed by MetALD, and MASLD. These findings extend prior evidence that NAFLD is a strong predictor of incident diabetes 2 , 6 and support the clinical value of the updated steatosis nomenclature, which focuses on metabolic dysfunction. 7 , 8 , 15 The effect sizes observed here among young adults exceed those typically reported in middle‐aged cohorts in meta‐analyses, likely reflecting a low baseline incidence in the reference group and the enrichment of insulin resistance among those with steatosis. 2
From a mechanistic perspective, hepatic triglyceride accumulation promotes hepatic insulin resistance, impairs suppression of gluconeogenesis, and induces accumulation of lipotoxic intermediates such as diacylglycerol that interfere with insulin receptor signalling; concomitant adipose–liver inflammatory crosstalk further aggravates systemic insulin resistance and β‐cell stress. 19 , 20 These biological pathways align with the persistent excess diabetes risk observed after adjustment for conventional cardiometabolic risk factors.
The differential patterns across SLD subtypes highlight the central role of metabolic dysfunction in the development of type 2 diabetes. MASLD, defined by steatosis with at least one metabolic abnormality, conferred nearly threefold adjusted risk, consistent with previous reports that hepatic steatosis is both a marker and mediator of whole‐body insulin resistance. 21 MetALD exhibited a similar risk profile, indicating that the presence of metabolic dysfunction cancels any modest glycemic benefit of light or moderate alcohol consumption in general populations. 9 By contrast, ALD significantly increased diabetes risk, in line with evidence that higher alcohol doses worsen insulin sensitivity and increase type 2 diabetes risk in a dose‐responsive manner. 10 Collectively, these data suggest that the intersection of hepatic steatosis and metabolic dysfunction is the dominant driver of diabetes risk, while heavy alcohol consumption adds further risk in susceptible individuals.
Our findings are consistent with prior studies showing a significant association between NAFLD and incident type 2 diabetes across different ethnic groups, although hazard ratios in East Asian cohorts tend to be larger, 22 possibly reflecting greater insulin‐secretory defects at lower levels of adiposity. 23 , 24 The present study extends this literature by separating SLD into MASLD, MetALD, and ALD, an approach endorsed by current guidance and relevant to risk communication. In this study, young adults with MASLD or MetALD constitute a high‐risk group for type 2 diabetes. These findings support prioritising intensive lifestyle modification and periodic glycemic screening. In individuals with ALD, counselling to reduce alcohol intake with metabolic risk assessment is warranted.
The pronounced male predominance in SLD, particularly in MASLD and MetALD, is consistent with previous epidemiologic studies. Recent Korean cohort data reported 74% male prevalence in MASLD and 91% in MetALD, consistent with global epidemiological patterns. 25 , 26 These sex differences likely reflect hormonal influences, visceral adiposity distribution, alcohol consumption patterns, smoking behaviours, and gut microbiome composition. 27
The inverse association of lower income with SLD and the higher prevalence of SLD among those reporting regular exercise likely reflects screening participation bias and reverse causality rather than causal relationships. Similar socioeconomic gradients in MASLD prevalence have been reported internationally, warranting further investigation of underlying mechanisms. 28 , 29
The age‐stratified heterogeneity in relative risk likely reflects differential baseline diabetes incidence and metabolic exposure duration. Among participants aged <30 years, the higher relative risk may indicate early‐stage, potentially reversible metabolic dysfunction. In contrast, those aged 30–39 years have accumulated longer metabolic exposure, potentially attenuating relative risk differences. Age‐related variations in hormonal status, adipose tissue distribution, and lifestyle factors may further contribute to these patterns. 30
Strengths of our study include the large sample size, standardised classification of SLD subtypes and outcomes, and the use of prespecified models aligned with prior research. However, several limitations warrant careful interpretation. First, the FLI is an indirect measure of steatosis, and misclassification may be possible; if this error is nondifferential, it would bias the associations towards null. 31 Nevertheless, the FLI effectively captures metabolic dysfunction underlying both hepatic fat accumulation and diabetes risk. The consistency of our findings using FLI ≥60 supports their robustness beyond threshold selection. To further ensure that our findings were not dependent on a single steatosis index, we repeated the analyses using the HSI, and the associations between SLD subtypes and diabetes risk remained robust and consistent. Second, misclassification of alcohol consumption is also possible due to reliance on self‐report, and alcohol intake may have changed over time 32 ; nevertheless, two sensitivity analyses including landmark and stable‐consumption approaches yielded consistent findings, indicating that temporal changes in alcohol intake do not significantly bias our results. Third, despite multivariable adjustment for major confounders, residual confounding from unmeasured factors such as dietary patterns, genetic variants, and longitudinal weight changes cannot be excluded. Causal inference should therefore remain conservative. Fourth, fibrosis‐based stratification could not be performed in this study due to unavailability of platelet counts. Future research incorporating laboratory or imaging‐derived fibrosis measures is needed to determine whether fibrosis severity modifies the association between SLD subtypes and diabetes risk. Fifth, incident diabetes was defined using a validated claims‐based definition that requires both a new ICD‐10 diagnosis and a concurrent prescription for antidiabetic medication. 33 , 34 While this approach provides high specificity for clinically recognised diabetes, it may still miss cases detectable only through post‐challenge hyperglycemia assessed by oral glucose tolerance testing. 35 Finally, our findings derive from a single ethnic cohort, necessitating external validation in diverse populations to establish generalisability.
In conclusion, young adults with SLD have a significantly increased risk of type 2 diabetes compared with those without SLD, and the risk was highest among those with ALD, followed by MetALD and MASLD. The elevated diabetes risk observed even in individuals younger than 30 years highlights the long‐term metabolic consequences of early‐onset SLD. These findings underscore the need for early metabolic risk assessment and lifestyle modification in young adults, with proactive monitoring potentially preventing progression to overt diabetes.
AUTHOR CONTRIBUTIONS
The corresponding authors (Eun Ju Cho and Kyungdo Han) had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Eun Ju Cho and Kyungdo Han. Provision of study materials or patients: Goh Eun Chung, Su Jong Yu. Collection and assembly of data: Su Jong Yu, Yoon Jun Kim, Jung‐Hwan Yoon. Data analysis and interpretation: Kyungdo Han, Jeayeon Park. Manuscript writing: Goh Eun Chung, Su Jong Yu, Kyungdo Han, and Eun Ju Cho. Final approval of manuscript: All authors.
FUNDING INFORMATION
This work was supported by the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health and Welfare, the Ministry of Food and Drug Safety) (NTIS, KMDF‐RnD 1711196788), and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS‐2025‐24523730).
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
Supporting information
Data S1: Supporting Information.
ACKNOWLEDGEMENTS
This study was performed using a database from the Korean National Health Insurance System (NHIS).
Chung GE, Yu SJ, Park J, et al. Associations between steatotic liver disease subtypes and incident diabetes in young Korean adults: A nationwide cohort study. Diabetes Obes Metab. 2026;28(3):1947‐1955. doi: 10.1111/dom.70381
Contributor Information
Kyungdo Han, Email: hkd917@naver.com.
Eun Ju Cho, Email: creatioex@gmail.com.
DATA AVAILABILITY STATEMENT
The dataset supporting the conclusions of this article is available on the homepage of National Health Insurance Sharing Service [http://nhiss.nhis.or.kr/bd/ab/bdaba021eng.do]. To gain access to the data, a completed application form, a research proposal and the applicant's approval document from the institutional review board should be submitted to and reviewed by the inquiry committee of research support in NHIS. Currently, use of NHIS data are allowed only for Korean researchers.
REFERENCES
- 1. Luk A, Wild SH, Jones S, et al. Early‐onset type 2 diabetes: the next major diabetes transition. Lancet. 2025;405(10497):2313‐2326. [DOI] [PubMed] [Google Scholar]
- 2. Mantovani A, Petracca G, Beatrice G, Tilg H, Byrne CD, Targher G. Non‐alcoholic fatty liver disease and risk of incident diabetes mellitus: an updated meta‐analysis of 501 022 adult individuals. Gut. 2021;70(5):962‐969. [DOI] [PubMed] [Google Scholar]
- 3. Chang Y, Jung HS, Yun KE, Cho J, Cho YK, Ryu S. Cohort study of non‐alcoholic fatty liver disease, NAFLD fibrosis score, and the risk of incident diabetes in a Korean population. Am J Gastroenterol. 2013;108(12):1861‐1868. [DOI] [PubMed] [Google Scholar]
- 4. Shen X, Cai J, Gao J, et al. Nonalcoholic fatty liver disease and risk of diabetes: a prospective study in China. Endocr Pract. 2018;24(9):823‐832. [DOI] [PubMed] [Google Scholar]
- 5. Fukuda T, Hamaguchi M, Kojima T, et al. The impact of non‐alcoholic fatty liver disease on incident type 2 diabetes mellitus in non‐overweight individuals. Liver Int. 2016;36(2):275‐283. [DOI] [PubMed] [Google Scholar]
- 6. Chung GE, Cho EJ, Yoon JW, et al. Nonalcoholic fatty liver disease increases the risk of diabetes in young adults: a nationwide population‐based study in Korea. Metabolism. 2021;123:154866. [DOI] [PubMed] [Google Scholar]
- 7. Rinella ME, Lazarus JV, Ratziu V, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. Hepatology. 2023;78(6):1966‐1986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Hsu CL, Loomba R. From NAFLD to MASLD: implications of the new nomenclature for preclinical and clinical research. Nat Metab. 2024;6(4):600‐602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Koppes LL, Dekker JM, Hendriks HF, Bouter LM, Heine RJ. Moderate alcohol consumption lowers the risk of type 2 diabetes: a meta‐analysis of prospective observational studies. Diabetes Care. 2005;28(3):719‐725. [DOI] [PubMed] [Google Scholar]
- 10. Han M. The dose‐response relationship between alcohol consumption and the risk of type 2 diabetes among Asian men: a systematic review and meta‐analysis of prospective cohort studies. J Diabetes Res. 2020;2020:1032049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Cheol Seong S, Kim YY, Khang YH, et al. Data resource profile: the National Health Information Database of the National Health Insurance Service in South Korea. Int J Epidemiol. 2017;46(3):799‐800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Bedogni G, Bellentani S, Miglioli L, et al. The fatty liver index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol. 2006;6:33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Park J, Chung GE, Yu SJ, et al. All‐cause and disease‐specific mortality in young adults with MASLD: a nationwide cohort study. JHEP Rep. 2025;7(9):101477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Cho EJ, Jung GC, Kwak MS, et al. Fatty liver index for predicting nonalcoholic fatty liver disease in an asymptomatic Korean population. Diagnostics (Basel). 2021;11(12):2233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kanwal F, Neuschwander‐Tetri BA, Loomba R, Rinella ME. Metabolic dysfunction‐associated steatotic liver disease: update and impact of new nomenclature on the American Association for the Study of Liver Diseases practice guidance on nonalcoholic fatty liver disease. Hepatology. 2024;79(5):1212‐1219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Yang YS, Han BD, Han K, Jung JH, Son JW. Obesity fact sheet in Korea, 2021: trends in obesity prevalence and obesity‐related comorbidity incidence stratified by age from 2009 to 2019. J Obes Metab Syndr. 2022;31(2):169‐177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. National Kidney Foundation . K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002;39(2 Suppl 1):S1‐S266. [PubMed] [Google Scholar]
- 18. Lee JH, Kim D, Kim HJ, et al. Hepatic steatosis index: a simple screening tool reflecting nonalcoholic fatty liver disease. Dig Liver Dis. 2010;42(7):503‐508. [DOI] [PubMed] [Google Scholar]
- 19. Petersen MC, Shulman GI. Roles of diacylglycerols and ceramides in hepatic insulin resistance. Trends Pharmacol Sci. 2017;38(7):649‐665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Heeren J, Scheja L. Metabolic‐associated fatty liver disease and lipoprotein metabolism. Mol Metab. 2021;50:101238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 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(11):9084‐9102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Mantovani A, Byrne CD, Bonora E, Targher G. Nonalcoholic fatty liver disease and risk of incident type 2 diabetes: a meta‐analysis. Diabetes Care. 2018;41(2):372‐382. [DOI] [PubMed] [Google Scholar]
- 23. Yabe D, Seino Y, Fukushima M, Seino S. β cell dysfunction versus insulin resistance in the pathogenesis of type 2 diabetes in east Asians. Curr Diab Rep. 2015;15(6):602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Cho YM. Characteristics of the pathophysiology of type 2 diabetes in Asians. Annals of Laparoscopic and Endoscopic Surgery. 2017;2:2. [Google Scholar]
- 25. Lee DY, Ko JH, Jang HN, et al. Characteristics of metabolic dysfunction‐associated Steatotic liver disease and its risk for hepatic fibrosis in 476,124 Korean adults: a cross‐sectional study. Endocrinol Metab (Seoul). 2025;40(4):648‐652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Kan C, Zhang K, Wang Y, et al. Global burden and future trends of metabolic dysfunction‐associated Steatotic liver disease: 1990‐2021 to 2045. Ann Hepatol. 2025;30(2):101898. [DOI] [PubMed] [Google Scholar]
- 27. Joo SK, Kim W. Sex differences in metabolic dysfunction‐associated steatotic liver disease: a narrative review. Ewha Med J. 2024;47(2):e17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Cho Y, Kim H, Jeong J, et al. Trends in metabolic dysfunction‐associated steatotic liver disease by household income, 2007‐2022: a national representative study in South Korea. Medicine (Baltimore). 2025;104(43):e45296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Moon JH, Jeong S, Jang H, et al. Young adults with metabolic dysfunction‐associated Steatotic liver disease have an increased risk of early‐onset cancer: a nationwide cohort study differentiating the risk of 23 site‐specific cancers. Clin Gastroenterol Hepatol. 2025;23:2540‐2549.e8. [DOI] [PubMed] [Google Scholar]
- 30. Ou MY, Zhang H, Tan PC, Zhou SB, Li QF. Adipose tissue aging: mechanisms and therapeutic implications. Cell Death Dis. 2022;13(4):300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Castellana M, Donghia R, Guerra V, et al. Performance of fatty liver index in identifying non‐alcoholic fatty liver disease in population studies. A meta‐analysis. J Clin Med. 2021;10(9):1877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Knott CS, Coombs N, Stamatakis E, Biddulph JP. All cause mortality and the case for age specific alcohol consumption guidelines: pooled analyses of up to 10 population based cohorts. BMJ. 2015;350:h384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Baek JH, Park YM, Han KD, Moon MK, Choi JH, Ko SH. Comparison of operational definition of type 2 diabetes mellitus based on data from Korean National Health Insurance Service and Korea National Health and nutrition examination survey. Diabetes Metab J. 2023;47(2):201‐210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Park J, Jung JH, Park H, et al. Association between exercise habits and incident type 2 diabetes mellitus in patients with thyroid cancer: nationwide population‐based study. BMC Med. 2024;22(1):251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Ma RC, Chan JC. Type 2 diabetes in east Asians: similarities and differences with populations in Europe and the United States. Ann N Y Acad Sci. 2013;1281(1):64‐91. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: Supporting Information.
Data Availability Statement
The dataset supporting the conclusions of this article is available on the homepage of National Health Insurance Sharing Service [http://nhiss.nhis.or.kr/bd/ab/bdaba021eng.do]. To gain access to the data, a completed application form, a research proposal and the applicant's approval document from the institutional review board should be submitted to and reviewed by the inquiry committee of research support in NHIS. Currently, use of NHIS data are allowed only for Korean researchers.