Abstract
OBJECTIVE
Metabolic dysfunction–associated steatotic liver disease (MASLD) is the most common chronic liver disease in children and is linked to type 2 diabetes. This study evaluates whether longitudinal changes in liver chemistries—γ-glutamyl transferase (GGT), aspartate aminotransferase (AST), and alanine aminotransferase (ALT)—can serve as biomarkers of increased type 2 diabetes risk in children with MASLD.
RESEARCH DESIGN AND METHODS
This multicenter longitudinal cohort study followed 1,035 children with biopsy-confirmed MASLD, without type 2 diabetes at baseline, for a mean of 3.9 years. Liver chemistries were measured annually, and type 2 diabetes was diagnosed based on fasting glucose, HbA1c, and clinical diagnosis. Extended Cox models with inverse probability weighting were used to evaluate associations between liver enzyme trajectories and type 2 diabetes risk.
RESULTS
The cumulative incidence of type 2 diabetes was 12.3%. Increases in GGT (hazard ratio [HR] 1.55; 95% CI 1.34–1.80), AST (HR 1.31; 95% CI 1.20–1.43), and ALT (HR 1.13; 95% CI 1.07–1.20) were associated with a higher risk of developing type 2 diabetes in the independent models. In the mutual model with all three liver chemistries, only GGT and AST remained significant.
CONCLUSIONS
A 30-unit increase in GGT over time was associated with a substantially higher risk of developing type 2 diabetes in children with MASLD. Together with AST, GGT may provide clinicians with concrete, routinely available parameters to monitor for early risk stratification. Further validation in independent cohorts is needed to confirm these findings and inform clinical application.
Graphical Abstract
Introduction
Metabolic dysfunction–associated steatotic liver disease (MASLD) is the most prevalent chronic liver disease in children, affecting ∼9.6% of individuals aged 2–19 years and up to 26% of children with obesity (1,2). MASLD encompasses a spectrum of conditions, ranging from hepatic steatosis to metabolic dysfunction–associated steatohepatitis, progressive fibrosis, and cirrhosis. Beyond liver involvement, MASLD is a multisystem disease strongly linked to metabolic conditions such as obesity, insulin resistance, dyslipidemia, and cardiovascular disease (3,4).
Children with MASLD face a heightened risk of developing type 2 diabetes, which exacerbates both metabolic and liver-related complications (5,6). The incidence of type 2 diabetes in children with MASLD far exceeds that of the general pediatric population (7,8). Prediabetes similarly has a prevalence rate of >20% among children with MASLD, reflecting the broader category of intermediate hyperglycemia that represents an early stage on the trajectory from normal glucose metabolism to overt diabetes (9). Dysglycemia, in the form of prediabetes or type 2 diabetes, is associated with a higher likelihood of disease progression to metabolic dysfunction–associated steatohepatitis and fibrosis (10,11). Despite the well-established link between MASLD and type 2 diabetes, limited research has explored dynamic predictors that could identify children at higher risk of developing type 2 diabetes.
In adults, elevated liver enzymes such as alanine aminotransferase (ALT) and γ-glutamyl transferase (GGT) have been identified as independent predictors of type 2 diabetes (12–21). Similar findings have begun to emerge in pediatric populations. For example, a study by Koutny et al. (22) found that elevated ALT levels in children with overweight and obesity were independently associated with a significantly increased risk of dysglycemia, regardless of BMI, age, and sex, highlighting ALT’s potential as an early predictor of glycemic deterioration. However, no studies have examined whether longitudinal changes in liver chemistries, including ALT, aspartate aminotransferase (AST), and GGT, can predict type 2 diabetes onset in children with MASLD.
This study aimed to address these gaps in understanding by evaluating the relationship between longitudinal changes in the liver chemistries ALT, AST, and GGT and the incidence of type 2 diabetes in children with MASLD. We applied inverse probability treatment and censoring weighting (IPTCW) to more accurately model liver enzyme trajectories and their association with type 2 diabetes risk. These methods provide a framework for capturing dynamic changes over time in an observational setting. Additionally, we examined the combined influence of these markers using traditional epidemiological methods to derive clinically relevant insights into overall type 2 diabetes risk. We hypothesized that increasing liver enzyme levels over time would be strongly associated with a higher risk of type 2 diabetes, potentially serving as early indicators for targeted interventions in this high-risk pediatric population.
Research Design and Methods
Study Population
This study included children with MASLD who were enrolled in the Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) database studies between 2004 and 2020 (Database, Pediatric Database 2, and Database 3). Participants were recruited from 15 pediatric care centers across the U.S. Institutional review board approval was obtained from each participating center, and written informed consent was provided by a parent or guardian. Assent was also obtained from children aged 8 years and older. Eligible participants were under 18 years of age, had biopsy-proven steatotic liver disease, met the diagnostic criteria for MASLD, and had a minimum of 9 months of follow-up data. Children with a diagnosis of type 2 diabetes at baseline were excluded from the study.
MASLD Diagnosis
MASLD was diagnosed based on the presence of hepatic steatosis, confirmed by liver histology, along with at least one of the following five cardiometabolic risk factors: 1) BMI ≥85th percentile for age and sex (BMI z-score ≥1); 2) fasting serum glucose ≥100 mg/dL, HbA1c ≥39 mmol/mol, or a diagnosis/treatment for type 2 diabetes; 3) blood pressure ≥95th percentile for children under 13 years or ≥130/85 mmHg for those aged 13 years and older, or treatment with antihypertensive medication; 4) plasma triglycerides ≥100 mg/dL for children under 10 years or ≥150 mg/dL for those aged 10 years and older, or treatment with lipid-lowering therapy; and 5) HDL cholesterol ≤40 mg/dL or treatment with lipid-lowering therapy.
Liver Chemistries
The primary exposure variables in this study were the longitudinal changes in liver chemistries, ALT, AST, and GGT. These liver enzyme levels were measured from fasting blood samples collected at baseline and at each subsequent annual visit throughout the follow-up period.
Outcome Variable: Type 2 Diabetes Definition
At annual research visits, type 2 diabetes screening was conducted using fasting glucose and HbA1c measurements, along with a thorough review of participants’ medical histories. Type 2 diabetes was defined based on the criteria established by the American Diabetes Association, consistent with those used in large epidemiological studies (19). Incident type 2 diabetes was diagnosed if participants met at least one of the following criteria during a research visit: 1) fasting serum glucose ≥126 mg/dL, 2) HbA1c ≥48 mmol/mol, or 3) a self-reported type 2 diabetes diagnosis that was subsequently confirmed through medical record review. Oral glucose tolerance testing (2-h plasma glucose) was not performed, reflecting standard practice in pediatric clinical care.
Measurement of Covariates
Demographic data were collected through structured interviews at enrollment, including age, sex, race, and ethnicity. Sex was categorized as female or male. Race categories included American Indian/Alaska Native, Asian, Black, mixed, Native Hawaiian/Pacific Islander, White, and unknown. Ethnicity was categorized as either Hispanic or non-Hispanic. Race and ethnicity were self-identified by the parent using standard National Institutes of Health classifications and were included because of well-documented differences in MASLD and BMI associated with race and ethnicity. Anthropometric measurements, including height, weight, and waist and hip circumferences, were taken at enrollment and during annual research visits. Weight was measured to the nearest 0.1 kg, and height was measured to the nearest 0.1 cm, with participants wearing light clothing and no shoes. BMI was calculated as weight (kg) divided by height squared (m2), and BMI z-scores were computed based on age, sex, height, and weight based on extended growth reference data for children and adolescents with obesity from 1988 to 2016 (23).
Laboratory assessments included fasting serum glucose (mg/dL), insulin (µU/mL), HbA1c (mmol/mol), hemoglobin, platelets, total cholesterol (mg/dL), HDL (mg/dL), LDL (mg/dL), and triglycerides (mg/dL). Baseline liver histology measures included steatosis grade (grade 0: <5%; grade 1: 5–33%; grade 2: 34–66%; grade 3: >66%), lobular inflammation (less than two, two to four, or more than four foci per field at 20× magnification), portal inflammation (none, mild, or more than mild), hepatocellular ballooning (none, few, or many), and fibrosis stage (stage 0: no fibrosis; stage 1a: mild zone 3 perisinusoidal fibrosis; stage 1b: moderate zone 3 perisinusoidal fibrosis; stage 1c: portal/periportal fibrosis only; stage 2: zone 3 perisinusoidal and periportal fibrosis; stage 3: bridging fibrosis; stage 4: cirrhosis).
Statistical Analysis
Participant characteristics were described based on incident type 2 diabetes status. Means and SDs were calculated for continuous variables, while categorical variables were summarized using frequencies and proportions. Baseline differences in covariates by type 2 diabetes status were assessed using ANOVA and t tests for continuous variables, and using χ2 tests for categorical variables. The incident type 2 diabetes rate was determined by dividing the number of new cases by the number of participants without type 2 diabetes at baseline, while cumulative type 2 diabetes incidence was calculated as the number of cases per total person-years at risk. All analyses were conducted using SAS Studio version 9.4 (SAS Institute Inc., Cary, NC), with statistical significance set at P < 0.05.
To estimate the hazard ratios (HRs) and 95% CIs for the associations between each liver chemistry measure (per 30-unit increase of ALT, AST, and GGT) and type 2 diabetes incidence, we employed three independent extended Cox models: unadjusted, multivariable, and IPTCW models. Multivariable models were adjusted for baseline variables (sex, race, ethnicity, steatosis amount, fibrosis stage, hepatocellular ballooning, and portal inflammation) and time-varying covariates (age at visit and BMI z-score). The use of separate models for each liver chemistry marker allowed for an independent evaluation of their effects on type 2 diabetes risk, providing a more detailed understanding of their individual contributions. This approach also minimized potential confounding that could arise from analyzing these correlated liver markers together.
Inverse Probability Weighting Methodology
Treatment weights were constructed using the quintile binning method for continuous exposures, following the methodology described by Naimi et al. (24). This method was chosen because it does not assume a specific distribution of the exposure and provides relatively unbiased effect estimates with a smaller mean squared error compared with other methods for heteroscedastic exposures. At each visit, liver chemistries were ranked into 10 quantiles, and treatment weights were generated by fitting a conditional cumulative logistic model to the ranked values. The denominator of the model included age at visit, sex, race, ethnicity, steatosis amount, fibrosis stage, hepatocellular ballooning, portal inflammation, and BMI z-score. To stabilize treatment weights, the numerator includes 1 divided by 10 because 10 quantiles were constructed.
Censoring weights were generated using a logistic regression model to account for potential loss to follow-up. The numerator of the model included sex, race, ethnicity, and clinic center, while the denominator incorporated age at visit, sex, race, ethnicity, steatosis amount, fibrosis stage, hepatocellular ballooning, portal inflammation, clinic center, and BMI z-score. Combined weights were calculated by multiplying the treatment and censoring weights for each visit. The resulting combined weights were then applied to adjust odds ratios, thereby reducing, but not eliminating, bias from measured confounding and informative censoring relative to traditional multivariable models. Consistent with recommendations from Carlin and Moreno-Betancur (25), our analyses are intended to strengthen risk prediction rather than to make causal inferences.
Mutually Adjusted Model
Additionally, a mutually adjusted extended Cox model was employed, incorporating all three time-varying liver chemistries (per 30-unit increase of ALT, AST, and GGT) simultaneously to estimate odds ratios and 95% CIs. This approach offers a straightforward epidemiological interpretation of the relative contributions of each liver chemistry marker, while adjusting for potential confounding between them. By including all three markers in a single model, we aimed to provide clinically relevant estimates of their independent effects on type 2 diabetes risk—critical for informing clinical decision-making and understanding MASLD progression in pediatric patients.
Sensitivity Analyses
Sensitivity analyses were conducted to evaluate the robustness of the associations between liver chemistries and incident type 2 diabetes. These analyses included 1) additional adjustments for baseline PNPLA3 genotype and time-varying metformin use to account for potential confounding by genetic and medication-related factors—metformin, widely used in pediatrics for conditions related to insulin resistance beyond diabetes, was considered to address its potential influence on glycemic outcomes—and 2) multiple imputation using chained equations to address missing data, assuming a missing-at-random pattern.
We imputed 1,000 data sets using 1,234 seeds and 50 iterations. The discriminant method was applied for categorical variables, specifically, ethnicity (0.2% missing) and fibrosis stage (0.3% missing) at baseline, while predictive mean matching via regression was used for continuous variables, including BMI z-scores, which had 7.5% missing data at both baseline and follow-up. We quantified missingness for variables included in our analytic models; for other variables collected in the cohort, missingness was not tabulated, because they were not used in the models.
Data and Resource Availability
Deidentified data for NASH CRN studies are made available by the National Institute of Diabetes and Digestive and Kidney Diseases Central Repository at https://repository.niddk.nih.gov/home/.
Results
Study Population
A total of 1,479 children were enrolled in NASH CRN between 2004 and 2020. After applying inclusion and exclusion criteria, which removed participants who did not meet MASLD criteria, had less than 9 months of follow-up, or had type 2 diabetes at baseline, the final analytic cohort comprised 1,035 children (Supplementary Fig. 1). The mean follow-up period was 3.9 years ± 2.9 years. The mean age at enrollment was 12.8 ± 2.7 years, and 73.5% of participants were male. The cohort was predominantly Hispanic (72.8%).
Incident Type 2 Diabetes
During follow-up, 127 participants (12.3%) developed type 2 diabetes, yielding an incidence rate of 31.1 cases per 1,000 person-years (95% CI 26.2–37.1). Compared with those who did not develop type 2 diabetes, affected participants were more likely to be female (37.8% vs. 24.9%, P = 0.002) and non-Hispanic (35.4% vs. 26.0%, P = 0.03). The incidence was also higher among participants of Asian (4.7% vs. 1.3%, P = 0.01) and Native Hawaiian/Pacific Islander (2.4% vs. 0.3%, P = 0.01) descent (Table 1). Incidence rates per 1,000 person-years were highest among Native Hawaiian/Pacific Islander (197.1; 95% CI 50.1–536.4) and Asian participants (102.8; 95% CI 41.7–213.8), followed by Black (44.4; 95% CI 11.3–121.0), American Indian/Alaska Native (31.8; 95% CI 14.8–60.4), and White participants (31.3; 95% CI 25.0–38.8).
Table 1.
Baseline characteristics among 1,046 children with MASLD
| Incident type 2 diabetes | ||||
|---|---|---|---|---|
| No | Yes | Total | P value | |
| n | 908 | 127 | 1,035 | |
| Age (years), mean (SD) | 12.8 (2.7) | 13.1 (2.8) | 12.8 (2.7) | 0.29* |
| Follow-up/time to type 2 diabetes (years), mean (SD) | 4.0 (2.8) | 3.7 (3.0) | 3.9 (2.9) | 0.40* |
| Sex, n (%) | ||||
| Female | 226 (24.9) | 48 (37.8) | 274 (26.5) | 0.00† |
| Male | 682 (75.1) | 79 (62.2) | 761 (73.5) | |
| Ethnicity, n (%) | ||||
| Hispanic | 670 (74.0) | 82 (64.6) | 752 (72.8) | 0.0† |
| Non-Hispanic | 236 (26.0) | 45 (35.4) | 281 (27.2) | |
| Race, n (%) | ||||
| American Indian/Alaska Native | 52 (5.7) | 8 (6.3) | 60 (5.8) | 0.01† |
| Asian | 12 (1.3) | 6 (4.7) | 18 (1.7) | |
| Black | 17 (1.9) | 3 (2.4) | 20 (1.9) | |
| Mixed | 25 (2.8) | 3 (2.4) | 28 (2.7) | |
| Native Hawaiian/Pacific Islander | 3 (0.3) | 3 (2.4) | 6 (0.6) | |
| White | 589 (64.9) | 80 (63.0) | 669 (64.6) | |
| Not reported | 210 (23.1) | 24 (18.9) | 234 (22.6) | |
| Weight (kg), mean (SD) | 81.1 (25.1) | 87.8 (27.8) | 82.0 (25.5) | 0.006* |
| Height, (cm), mean (SD) | 158.0 (14.2) | 159.5 (13.5) | 158.2 (14.1) | 0.27* |
| BMI z-score, mean (SD) | 2.4 (0.8) | 2.6 (0.9) | 2.4 (0.8) | 0.004‡ |
| Waist (cm), mean (SD) | 102.9 (15.1) | 107.8 (15.7) | 103.5 (15.3) | <0.001* |
| Systolic blood pressure (mmHg), mean (SD) | 119.9 (13.1) | 122.0 (15.1) | 120.2 (13.4) | 0.14‡ |
| Diastolic blood pressure (mmHg), mean (SD) | 67.6 (9.5) | 68.4 (9.4) | 67.7 (9.5) | 0.36* |
| GCT (units/L), median (IQR) | 34.5 (27.0) | 35.0 (34.0) | 35.0 (27.0) | 0.13§ |
| AST (units/L), median (IQR) | 48.0 (33.0) | 52.0 (44.0) | 49.0 (35.0) | 0.09§ |
| ALT (units/L), median (IQR) | 81.0 (74.0) | 89.0 (92.0) | 82.0 (77.5) | 0.04§ |
| Total cholesterol (mg/dL), mean (SD) | 164.7 (36.5) | 163.9 (38.5) | 164.6 (36.8) | 0.83* |
| HDL cholesterol (mg/dL), mean (SD) | 39.6 (9.4) | 37.8 (8.8) | 39.4 (9.3) | 0.04* |
| LDL cholesterol (mg/dL), mean (SD) | 97.0 (29.8) | 95.3 (29.9) | 96.8 (29.8) | 0.57* |
| Triglycerides, (mg/dL), median (IQR) | 131.0 (89.0) | 134.0 (100.0) | 132.0 (90.0) | 0.25§ |
| HbA1c (mmol/mol), mean (SD) | 34.0 (1.2) | 37.0 (1.2) | 36.0 (1.2) | <0.001* |
| Fasting serum glucose (mg/dL), mean (SD) | 87.6 (9.9) | 89.0 (11.5) | 87.7 (10.1) | 0.18‡ |
| Insulin (mU/mL), median (IQR) | 25.0 (25.0) | 29.7 (27.3) | 25.9 (26.0) | 0.07§ |
| HOMA-IR, median (IQR) | 97.3 (100.8) | 113.4 (102.9) | 98.9 (100.4) | 0.07§ |
| Hemoglobin (g/dL), mean (SD) | 13.9 (1.2) | 13.8 (1.0) | 13.9 (1.2) | 0.32‡ |
| Platelets (×103 cells/µL), mean (SD) | 294.2 (67.1) | 285.1 (58.8) | 293.2 (66.2) | 0.19* |
| NAFLD activity score, mean (SD) | 4.1 (1.4) | 4.5 (1.5) | 4.1 (1.4) | 0.002* |
| Steatosis amount, n (%) | ||||
| 5–33% | 239 (26.3) | 30 (23.6) | 269 (26.0) | 0.66† |
| >33–66% | 278 (30.6) | 37 (29.1) | 315 (30.4) | |
| >66% | 391 (43.1) | 60 (47.2) | 451 (43.6) | |
| Lobular inflammation: number of foci under ×20 magnification, n (%) | ||||
| None | 3 (0.3) | 0 (0.0) | 3 (0.3) | 0.10† |
| Less than two | 528 (58.1) | 62 (48.8) | 590 (57.0) | |
| Two to four | 329 (36.2) | 53 (41.7) | 382 (36.9) | |
| More than four | 48 (5.3) | 12 (9.4) | 60 (5.8) | |
| Hepatocellular ballooning, n (%) | ||||
| None | 594 (65.4) | 71 (55.9) | 665 (64.3) | 0.002† |
| Few | 229 (25.2) | 31 (24.4) | 260 (25.1) | |
| Many | 85 (9.4) | 25 (19.7) | 110 (10.6) | |
| Portal inflammation, n (%) | ||||
| None | 95 (10.5) | 11 (8.7) | 106 (10.3) | 0.71† |
| Mild | 681 (75.1) | 95 (74.8) | 776 (75.0) | |
| More than mild | 131 (14.4) | 21 (16.5) | 152 (14.7) | |
| Fibrosis stage, n (%) | ||||
| 0: none | 307 (33.9) | 38 (29.9) | 345 (33.4) | 0.01† |
| 1a: mild, zone 3 perisinusoidal | 59 (6.5) | 4 (3.1) | 63 (6.1) | |
| 1b: moderate, zone 3 perisinusoidal | 29 (3.2) | 8 (6.3) | 37 (3.6) | |
| 1c: portal/periportal only | 286 (31.6) | 29 (22.8) | 315 (30.5) | |
| 2: zone 3 and periportal, any combination | 110 (12.2) | 26 (20.5) | 136 (13.2) | |
| 3: bridging | 104 (11.5) | 21 (16.5) | 125 (12.1) | |
| 4: cirrhosis | 10 (1.1) | 1 (0.8) | 11 (1.1) | |
| Steatohepatitis diagnosis, n (%) | ||||
| No | 281 (30.9) | 33 (26.0) | 314 (30.3) | 0.002† |
| Borderline zone 1 pattern | 316 (34.8) | 31 (24.4) | 347 (33.5) | |
| Borderline zone 3 pattern | 138 (15.2) | 21 (16.5) | 159 (15.4) | |
| Definite | 173 (19.1) | 42 (33.1) | 215 (20.8) | |
| PNPLA3 genotype, n (%) | ||||
| CC | 84 (11.5) | 23 (20.4) | 107 (12.7) | 0.01† |
| CG | 239 (32.8) | 40 (35.4) | 279 (33.1) | |
| GG | 406 (55.7) | 50 (44.2) | 456 (54.2) | |
HOMA-IR, HOMA of insulin resistance; MCV, mean corpuscular volume; RBC, red blood cell; WBC, white blood cell; WHR, waist-to-hip ratio.
*Equal-variance two-sample t test.
†χ2 P value.
‡Unequal-variance two-sample t test.
§Wilcoxon rank sum two-sided test.
At baseline, children who developed type 2 diabetes had higher mean BMI z-scores (2.6 vs. 2.4, P = 0.004), median ALT levels (89.0 units/L vs. 81.0 units/L, P = 0.04), mean HbA1c (5.5% vs. 5.3%, P < 0.0001), and lower mean HDL cholesterol (37.8 mg/dL vs. 39.6 mg/dL, P = 0.04). Children who developed type 2 diabetes were less likely to carry the GG variant of the PNPLA3 genotype compared with those who did not (44.2% vs. 55.7%, P = 0.01), with a higher proportion carrying the CC variant (20.4% vs. 11.5%).
Baseline histologic findings showed that children who developed type 2 diabetes had more severe liver disease compared with those who did not. A significantly higher proportion of children with incident type 2 diabetes exhibited hepatocellular ballooning (19.7% vs. 9.4%, P = 0.002), moderate to severe fibrosis (37.8% vs. 24.7% with stage 2 or higher, P = 0.01), and definite steatohepatitis (33.1% vs. 19.1%, P = 0.0008).
Longitudinal Liver Chemistry Trajectories and Risk of Type 2 Diabetes
Longitudinal increases in liver chemistries were significantly associated with an increased risk of type 2 diabetes. Table 2 presents the results from the extended Cox models. In unadjusted models, the hazard of developing type 2 diabetes increased by 34% (95% CI 22–47) for each 30-unit increase in GGT, 15% (95% CI 11–19) for ALT, and 29% (95% CI 22–38) for AST. These associations were strengthened after adjustment for IPTCW, with HRs of 1.55 (95% CI 1.34–1.80) for GGT, 1.13 (95% CI 1.07–1.20) for ALT, and 1.31 (95% CI 1.20–1.43) for AST. In multivariable models adjusting for age at visit, sex, race, ethnicity, fibrosis stage, steatosis severity, portal inflammation, hepatocellular ballooning, and BMI z-score, the associations remained significant: GGT (HR 1.32; 95% CI 1.19–1.47), ALT (HR 1.17; 95% CI 1.12–1.23), and AST (HR 1.31; 95% CI 1.22–1.41).
Table 2.
Extended Cox analysis for risk of incident type 2 diabetes by individual time-varying liver chemistry among 1,035 children with MASLD
| Independent models* | Mutual models* | ||||
|---|---|---|---|---|---|
| Models | Unadjusted | IPTCW | Multivariate† | Model 1‡ | Model 2§ |
| GGT | 1.34 (1.22–1.47) | 1.55 (1.34–1.80) | 1.32 (1.19–1.47) | 1.27 (1.15–1.41) | 1.23 (1.09–1.40) |
| ALT | 1.15 (1.11–1.19) | 1.13 (1.07–1.20) | 1.17 (1.12–1.23) | 0.96 (0.85–1.08) | 0.99 (0.86–1.14) |
| AST | 1.29 (1.22–1.38) | 1.31 (1.20–1.43) | 1.31 (1.22–1.41) | 1.30 (1.08–1.57) | 1.28 (1.05–1.57) |
*Unit change is 30 for GGT, ALT, and AST.
†Multivariate model is adjusted for baseline variables (sex, race, ethnicity, steatosis amount, fibrosis stage, hepatocellular ballooning, portal inflammation) and time-varying variable (age at visit, BMI z-score).
‡Model 1 includes all three time-varying liver chemistries (GGT, ALT, AST).
§Model 2 includes model 1 with baseline variables (sex, race, ethnicity, steatosis amount, fibrosis stage, hepatocellular ballooning, portal inflammation) and time-varying variable (age at visit, BMI z-score). Bold font indicates significant HRs.
To translate these associations into more clinically interpretable terms, we estimated the relative hazard of type 2 diabetes based on percent changes in liver chemistries between visits at varying baseline levels (Table 3 and Fig. 1). For GGT, a 10% increase between visits was associated with a relative hazard of 1.07 when starting from a baseline of 45 units/L, and increased progressively to 1.42 with a 40% rise from a baseline of 60 units/L. Similar patterns were observed for AST, with a 10% increase between visits from a baseline of 40 units/L yielding a hazard of 1.04, rising to 1.33 with a 40% increase from a baseline of 80 units/L. In contrast, ALT demonstrated more modest associations: a 10% increase from a baseline of 60 units/L corresponded to a hazard of 1.02, increasing to 1.22 with a 40% rise from a baseline of 120 units/L.
Table 3.
Incident risk of type 2 diabetes based on change in liver chemistries
| Baseline | Follow-up | Change, % | Absolute change | Type 2 diabetes risk |
|---|---|---|---|---|
| GGT* | ||||
| 30 | 33 | 10 | 3 | 1.04 |
| 36 | 20 | 6 | 1.09 | |
| 39 | 30 | 9 | 1.14 | |
| 42 | 40 | 12 | 1.19 | |
| 45 | 50 | 10 | 4.5 | 1.07 |
| 54 | 20 | 9 | 1.14 | |
| 59 | 30 | 13.5 | 1.22 | |
| 63 | 40 | 18 | 1.30 | |
| 60 | 66 | 10 | 6 | 1.09 |
| 72 | 20 | 12 | 1.19 | |
| 78 | 30 | 18 | 1.30 | |
| 84 | 40 | 24 | 1.42 | |
| ALT† | ||||
| 60 | 66 | 10 | 6 | 1.02 |
| 72 | 20 | 12 | 1.05 | |
| 78 | 30 | 18 | 1.08 | |
| 84 | 40 | 24 | 1.10 | |
| 80 | 88 | 10 | 8 | 1.03 |
| 96 | 20 | 16 | 1.07 | |
| 104 | 30 | 24 | 1.10 | |
| 112 | 40 | 32 | 1.14 | |
| 120 | 132 | 10 | 12 | 1.05 |
| 144 | 20 | 24 | 1.10 | |
| 156 | 30 | 36 | 1.16 | |
| 168 | 40 | 48 | 1.22 | |
| AST‡ | ||||
| 40 | 44 | 10 | 4 | 1.04 |
| 48 | 20 | 8 | 1.07 | |
| 52 | 30 | 12 | 1.11 | |
| 56 | 40 | 16 | 1.15 | |
| 60 | 66 | 10 | 6 | 1.06 |
| 72 | 20 | 12 | 1.11 | |
| 78 | 30 | 18 | 1.18 | |
| 84 | 40 | 24 | 1.24 | |
| 80 | 88 | 10 | 8 | 1.07 |
| 96 | 20 | 16 | 1.15 | |
| 104 | 30 | 24 | 1.24 | |
| 112 | 40 | 32 | 1.33 |
*Type 2 diabetes risk conversion formula based on IPTCW HRs in Table 2: HR = 1.55(PercentChange*Baseline)/30.
†Type 2 diabetes risk conversion formula based on IPTCW HRs in Table 2: HR = 1.13(PercentChange*Baseline)/30.
‡Type 2 diabetes risk conversion formula based on IPTCW HRs in Table 2: HR = 1.31(PercentChange*Baseline)/30.
Figure 1.
Line graph depicting the relationship between percent change in liver enzymes between visits and hazard for incident type 2 diabetes. This illustrates that the relationship between change in a laboratory parameter and the corresponding hazard depends on the baseline value. Shown are examples for a hypothetical patient with baseline values of GGT 45 units/L, ALT 80 units/L, and AST 60 units/L.
Mutually Adjusted Liver Chemistry Analysis
Table 2 presents the results of the survival analysis for time-varying liver chemistries, including GGT, ALT, and AST modeled simultaneously. In model 1, which included only the three liver chemistries, both GGT (HR 1.27; 95% CI 1.15–1.41) and AST (HR 1.30; 95% CI 1.08–1.57) were significantly associated with an increased risk of type 2 diabetes, while ALT was not (HR 0.96; 95% CI 0.85–1.08). In model 2, which adjusted for baseline covariates (sex, race, ethnicity, fibrosis stage, steatosis severity, portal inflammation, and hepatocellular ballooning) and time-varying variables (age at visit and BMI z-score), the associations for GGT (HR 1.23; 95% CI 1.09–1.40) and AST (HR 1.28; 95% CI 1.05–1.57) remained statistically significant, while ALT continued to show no significant association with type 2 diabetes risk.
Other Clinical Predictors
In additional analyses, we evaluated lipid-based predictors of type 2 diabetes. The risk of type 2 diabetes increased by 7% (95% CI 2–13) for each 30-unit increase in triglycerides and by 8% (95% CI 3–14) for each 1-unit increase in the TG-to–HDL cholesterol ratio in IPTCW models (data not shown).
Sensitivity Analyses
In the independent models, the associations between each liver chemistry and type 2 diabetes remained significant after adjusting for PNPLA3 and metformin use (model 2): GGT (HR 1.59; 95% CI 1.36–1.87), AST (HR 1.29; 95% CI 1.14–1.46), and ALT (HR 1.13; 95% CI 1.05–1.22) (Supplementary Table 1). We also evaluated effect modification by PNPLA3 genotype in these independent models and found no significant interactions with GGT (P = 0.45), ALT (P = 0.82), or AST (P = 0.29) (data not shown). In the mutually adjusted model, only GGT (HR 1.30; 95% CI 1.13–1.49) retained a significant association with type 2 diabetes risk after adjusting for PNPLA3 and metformin use, in addition to the covariates already accounted for in Table 2. No sex differences were observed in the associations between liver chemistries and type 2 diabetes in either the independent or mutually adjusted models (data not shown).
The results of the multiple imputation by chained equations analysis were consistent with those from the complete case analysis (Supplementary Table 2). In the independent models using IPTCW, the hazard of type 2 diabetes increased by 46% (95% CI 30–63) for each 30-unit increase in GGT, 14% (95% CI 7–21) for ALT, and 31% (95% CI 19–44) for AST. Similarly, independent models with multivariate adjustment yielded comparable findings, with HRs of 1.33 (95% CI 1.18–1.50) for GGT, 1.17 (95% CI 1.12–1.22) for ALT, and 1.32 (95% CI 1.24–1.41) for AST. In the mutually adjusted model, GGT (HR 1.25; 95% CI 1.11–1.40) and AST (HR 1.32; 95% CI 1.08–1.60) remained significantly associated with type 2 diabetes risk, while ALT did not (HR 0.97; 95% CI 0.85–1.11).
Conclusions
This study evaluated the association between longitudinal changes in the liver chemistries ALT, AST, and GGT and the risk of developing type 2 diabetes in children with MASLD. Among 1,035 children with biopsy-proven MASLD followed for a median of 3.9 years, the cumulative incidence of type 2 diabetes was 12.2%, corresponding to an incidence rate of 3,111 per 100,000 person-years. Increases in GGT and AST were strongly associated with incident type 2 diabetes, even after accounting for confounding factors and the interdependence of liver markers. Although ALT demonstrated modest associations in independent models, its predictive value diminished when GGT and AST were included simultaneously. These findings suggest that monitoring changes in GGT and AST could help identify children at highest risk for type 2 diabetes, offering an opportunity for earlier and potentially more effective intervention.
Our study provides compelling evidence that rising GGT and AST levels over time are strongly associated with an increased likelihood of developing type 2 diabetes in children with MASLD. These findings align with prior studies in adults, such as those by Schneider et al. (13) and Chen et al. (12), which identified baseline elevations of liver enzymes as predictors of diabetes risk. By using time-varying Cox models, we extend previous research by illuminating the dynamic nature of liver enzyme trajectories, a critical aspect overlooked by static baseline measurements. The observed increases in GGT may reflect its involvement in systemic oxidative stress, inflammation, and insulin resistance, pathways central to type 2 diabetes pathogenesis as suggested by Zheng et al. (14) and Kaneko et al. (15). AST, traditionally viewed as a marker of liver injury, further emphasizes the interplay between hepatic dysfunction and metabolic health, potentially indicating progressive liver damage and heightened metabolic vulnerability (26–28). While ALT has been highlighted in studies such as Morinaga et al. (16) and Lan et al. (18), its association with type 2 diabetes in our study was weaker than that of GGT and AST in mutually adjusted models, suggesting that it may lack the specificity and independence of GGT and AST in children. Collectively, these findings highlight the importance of longitudinally monitoring liver chemistries—particularly GGT and AST—in pediatric MASLD to better anticipate metabolic disease progression and inform timely interventions.
These findings have significant clinical implications. Monitoring GGT and AST trajectories provides a dynamic approach to type 2 diabetes risk stratification in children with MASLD, surpassing the utility of static baseline assessments. Because current pediatric guidelines emphasize ALT for liver injury (29,30), expanding them to include GGT and AST could better identify children at heightened metabolic risk. These markers not only inform liver health but also reflect systemic metabolic dysfunction, making them ideal for dual-risk assessment. Their associations persist despite adjustment for BMI and other confounders, highlighting their independent predictive value for type 2 diabetes. Moreover, the progressive elevations in GGT and AST noted in this study suggest that they may serve as early warning signals, allowing clinicians to intervene—through lifestyle modifications, targeted pharmacotherapy, or enhanced metabolic monitoring—before type 2 diabetes develops. Looking ahead, MASLD management guidelines could be updated to integrate GGT and AST measurements routinely, focusing on children who stand to benefit most from early, proactive intervention strategies.
This study has several strengths that bolster the reliability and clinical applicability of its findings. By using advanced epidemiological methods—including time-varying Cox models with IPTCW—we improved risk prediction and gained a nuanced understanding of the dynamic relationship between liver chemistries and type 2 diabetes risk, an uncommon approach in pediatric research. The large, multicenter NASH CRN cohort, comprising children with biopsy-confirmed MASLD and extensive follow-up, offered a solid foundation for analyzing how liver enzyme trajectories evolve over time. Examining ALT, AST, and GGT independently and in combination yielded actionable insights for clinical practice. Additionally, the study’s focus on children with MASLD, an underrepresented population in metabolic and liver disease research, addresses critical gaps in understanding pediatric-specific metabolic risks. Finally, because liver chemistries are inexpensive, widely available, and routinely measured, these findings can readily be integrated into clinical workflows, enhancing early risk stratification and intervention.
This study also had several limitations. First, despite employing inverse probability weighting to reduce bias, residual confounding may still influence the observed association because we could not account for unmeasured confounders (e.g., lifestyle and socioeconomic factors). Second, because type 2 diabetes was defined using fasting plasma glucose, HbA1c, and medical history without oral glucose tolerance testing, incidence was likely underestimated. This reflects real-world pediatric practice in the U.S., where oral glucose tolerance testing is not routinely performed, although it is commonly used in some European countries, and is unlikely to have materially biased the observed associations (31). Another limitation is the generalizability of the findings, as the cohort was predominantly Hispanic, a population with a high risk for both MASLD and type 2 diabetes. Notably, however, our results revealed that non-Hispanic children were more likely to develop incident type 2 diabetes, an observation that warrants further investigation. Validation of these findings in diverse populations of children with MASLD will help to ensure their broader applicability.
In conclusion, longitudinal increases in GGT and AST are strongly associated with an elevated risk of incident type 2 diabetes in children with MASLD, highlighting their value as dynamic biomarkers of metabolic dysfunction. Systematic tracking of these markers could help clinicians identify high-risk children earlier, facilitating timely intervention to mitigate type 2 diabetes progression. Future research should evaluate the integration of these biomarkers into predictive models and assess the impact of reducing liver chemistry levels on type 2 diabetes risk, paving the way for improved clinical management and a deeper understanding of liver-metabolic interactions in children.
This article contains supplementary material online at https://doi.org/10.2337/figshare.30569516.
Article Information
Acknowledgments. The authors thank members of the Nonalcoholic Steatohepatitis Clinical Research Network Pediatric Clinical Centers, which includes the following: Paula M. Hertel, Donna Garner, Telma Gomez, Krupa R. Mysore, Paushpala Sen, Mary Elizabeth Tessier, Nicole Triggs, and Cynthia M. Tsai at Baylor College of Medicine, Houston, TX; Stavra Xanthakos, Ana Catalina Arce-Clachar, Kristin Bramlage, Kim Cecil, Nicole Chaaban, Marialena Mouzaki, Ann Popelar, and Andrew Trout at Cincinatti Children’s Hospital Medical Center, Cincinnati, OH; Miriam Vos, Adina Alazraki, Christina Carapia Chaparro, and Jorge Jara-Garra of Emory University, Atlanta, GA; Jean P. Molleston, Oscar W. Cummings, Kathryn Harlow Adams, Ashley Hartman, Kelley S. Jackson, Chaowapong Jarasvaraparn, Sandie Kennedy, Ann Klipsch, Wendy Morlan, Emily Ragozzino, and Kyla Tolliver at Indiana University School of Medicine/Riley Hospital for Children, Indianapolis, IN; Mark H. Fishbein, Angela Anthony, Catherine Chapin (also at Saint Louis University, St Louis, MO), Ajay K. Jain, Danielle Carpenter, and Paige Puricelli at Northwestern University Feinberg School of Medicine/Ann & Robert H. Lurie Children’s Hospital of Chicago; Jeffrey B. Schwimmer, Amy Alba, Cynthia Behling, Nidhi Goyal, Michael S. Middleton, Rebecca Morfin, Kimberly Newton, Claude Sirlin, Jaret Skonieczny, Patricia Ugalde-Nicalo, and Karenina Valdez at University of California San Diego, San Diego, CA; Ryan Gill at University of California San Francisco, San Francisco, CA; Niviann Blondet, Randolph Otto, Matthew Yeh, and Melissa Young at University of Washington Medical Center and Seattle Children’s Hospital, Seattle, WA; David E. Kleiner at Resource Centers National Cancer Institute, Bethesda, MD; and Jeanne M. Clark, David M. Shade, Peggy Adamo, Patricia Belt, Jennifer M. DeSanto, Jill Meinert, Laura Miriel, Emily P. Mitchell, Carrie Shade, Jacqueline Smith, Alice Sternberg, Annette Wagoner, Laura A. Wilson, Tinsay Woreta, and Katherine P. Yates at Data Coordinating Center, Johns Hopkins University, Bloomberg School of Public Health, Baltimore, MD.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Duality of Interest. R.M.S. is a consultant for Travere Therapeutics, Inc. C.B. is a consultant for 89 Bio, Akero Therapeutics, Boehringer Ingelheim, Novo Nordisk, and Pathology Institute. J.B.S. receives grant support to University of California San Diego from Seraphina and Intercept and is a consultant for Merck. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. N.Q.N.T., K.P.N., L.W., R.M.S., G.B., J.A., N.C., and J.B.S. were involved with the conceptualization of this work. N.Q.N.T. and L.W. contributed to data curation. K.P.N., L.W., S.A.X., C.B., and J.B.S. supervised the investigation. N.Q.N.T., R.M.S., G.B., J.A., N.C., and J.B.S. designed the methods and contributed to the data analysis for this study. N.Q.N.T., L.F.C., and J.B.S. drafted the manuscript. All authors contributed to the editing of the manuscript and have read and approved the final manuscript. J.B.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Handling Editors. The journal editors responsible for overseeing the review of the manuscript were John B. Buse and Amalia Gastaldelli.
Funding Statement
NASH CRN is supported by National Institute of Diabetes and Digestive and Kidney Diseases (grants U01DK061713, U01DK061718, U01DK061728, U01DK061732, U01DK061734, U01DK061737, U01DK061738, U01DK061730, and U24DK061730). Additional support is received from the National Center for Advancing Translational Sciences (grants UL1TR000077, UL1TR000150, UL1TR000006, UL1TR000448, UL1TR000100, UL1TR000004, UL1TR000423, and UL1TR000454).
Footnotes
See accompanying article, p. 578.
Supporting information
References
- 1. Schwimmer JB, Deutsch R, Kahen T, Lavine JE, Stanley C, Behling C. Prevalence of fatty liver in children and adolescents. Pediatrics 2006;118:1388–1393 [DOI] [PubMed] [Google Scholar]
- 2. Yu EL, Golshan S, Harlow KE, et al. Prevalence of nonalcoholic fatty liver disease in children with obesity. J Pediatr 2019;207:64–70 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Targher G, Byrne CD, Tilg H. MASLD: a systemic metabolic disorder with cardiovascular and malignant complications. Gut 2024;73:691–702 [DOI] [PubMed] [Google Scholar]
- 4. Sandireddy R, Sakthivel S, Gupta P, Behari J, Tripathi M, Singh BK. Systemic impacts of metabolic dysfunction-associated steatotic liver disease (MASLD) and metabolic dysfunction-associated steatohepatitis (MASH) on heart, muscle, and kidney related diseases. Front Cell Dev Biol 2024;12:1433857. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Putri RR, Casswall T, Danielsson P, Marcus C, Hagman E. Steatotic liver disease in pediatric obesity and increased risk for youth-onset type 2 diabetes. Diabetes Care 2024;47:2196–2204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Nobili V, Mantovani A, Cianfarani S, et al. Prevalence of prediabetes and diabetes in children and adolescents with biopsy-proven non-alcoholic fatty liver disease. J Hepatol 2019;71:802–810 [DOI] [PubMed] [Google Scholar]
- 7. Newton KP, Wilson LA, Crimmins NA, et al.; Nonalcoholic Steatohepatitis Clinical Research Network . Incidence of type 2 diabetes in children with nonalcoholic fatty liver disease. Clin Gastroenterol Hepatol 2023;21:1261–1270 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Divers J, Mayer-Davis EJ, Lawrence JM, et al. Trends in incidence of type 1 and type 2 diabetes among youths – selected counties and Indian reservations, United States, 2002–2015. MMWR Morb Mortal Wkly Rep 2020;69:161–165 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Newton KP, Hou J, Crimmins NA, et al.; Nonalcoholic Steatohepatitis Clinical Research Network . Prevalence of prediabetes and type 2 diabetes in children with nonalcoholic fatty liver disease. JAMA Pediatr 2016;170:e161971. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Pedicelli S, Fintini D, Ravà L, et al. Prevalence of prediabetes in children and adolescents by class of obesity. Pediatr Obes 2022;17:e12900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Newfield RS, Graves CL, Newbury RO, et al. Non-alcoholic fatty liver disease in pediatric type 2 diabetes: metabolic and histologic characteristics in 38 subjects. Pediatr Diabetes 2019;20:41–47 [Google Scholar]
- 12. Chen SC-C, Tsai SP, Jhao J-Y, Jiang W-K, Tsao CK, Chang L-Y. Liver fat, hepatic enzymes, alkaline phosphatase and the risk of incident type 2 diabetes: a prospective study of 132,377 adults. Sci Rep 2017;7:4649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Schneider ALC, Lazo M, Ndumele CE, et al. Liver enzymes, race, gender and diabetes risk: the Atherosclerosis Risk in Communities (ARIC) study. Diabet Med 2013;30:926–933 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Zheng X, Cao C, He Y, Wang X, Wu J, Hu H. Association between nonalcoholic fatty liver disease and incident diabetes mellitus among Japanese: a retrospective cohort study using propensity score matching. Lipids Health Dis 2021;20:59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kaneko K, Yatsuya H, Li Y, et al. Association of gamma-glutamyl transferase and alanine aminotransferase with type 2 diabetes mellitus incidence in middle-aged Japanese men: 12-year follow up. J Diabetes Investig 2019;10:837–845 [Google Scholar]
- 16. Morinaga A, Iwanaga K, Maki K, et al. Association between serum ALT levels and incidence of new-onset diabetes in general population of Japanese: a longitudinal observational study (ISSA-CKD). BMJ Open 2023;13:e074007 [Google Scholar]
- 17. Folli F, Pontiroli AE, Zakaria AS, Centofanti L, Tagliabue E, La Sala L. Alanine transferase levels (ALT) and triglyceride-glucose index are risk factors for type 2 diabetes mellitus in obese patients. Acta Diabetol 2024;61:435–440 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Lan Q, Zhang Y, Lin F, et al. Association between serum aminotransferases and risk of new-onset cardiometabolic disease in a healthy chinese population: a cohort study. Front Public Health 2022;10:902393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Nguyen QM, Srinivasan SR, Xu J-H, et al. Elevated liver function enzymes are related to the development of prediabetes and type 2 diabetes in younger adults: the Bogalusa Heart Study. Diabetes Care 2011;34:2603–2607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Ford ES, Schulze MB, Bergmann MM, Thamer C, Joost H-G, Boeing H. Liver enzymes and incident diabetes: findings from the European Prospective Investigation into Cancer and Nutrition (EPIC)-Potsdam Study. Diabetes Care 2008;31:1138–1143 [DOI] [PubMed] [Google Scholar]
- 21. Choi S-H, Kim B-T, Shin J, Kim K-N. Combined effect of serum alanine aminotransferase and gamma-glutamyltransferase on incidence of diabetes mellitus: a longitudinal study. Medicine (Baltimore) 2020;99:e18963. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Koutny F, Stein R, Kiess W, Weghuber D, Körner A. Elevated transaminases potentiate the risk for emerging dysglycemia in children with overweight and obesity. Pediatr Obes 2021;16:e12822. [DOI] [PubMed] [Google Scholar]
- 23. Wei R, Ogden CL, Parsons VL, Freedman DS, Hales CM. A method for calculating BMI z-scores and percentiles above the 95th percentile of the CDC growth charts. Ann Hum Biol 2020;47:514–521 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Naimi AI, Moodie EEM, Auger N, Kaufman JS. Constructing inverse probability weights for continuous exposures: a comparison of methods. Epidemiology 2014;25:292–299 [DOI] [PubMed] [Google Scholar]
- 25. Carlin JB, Moreno‐Betancur M. On the uses and abuses of regression models: a call for reform of statistical practice and teaching. Stat Med 2025;44:e10244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Zhang X, Yang S, Chen J, Su Z. Unraveling the regulation of hepatic gluconeogenesis. Front Endocrinol (Lausanne) 2018;9:802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Onyango AN. Excessive gluconeogenesis causes the hepatic insulin resistance paradox and its sequelae. Heliyon 2022;8:e12294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Holeček M. Roles of malate and aspartate in gluconeogenesis in various physiological and pathological states. Metabolism 2023;145:155614. [DOI] [PubMed] [Google Scholar]
- 29. Vos MB, Abrams SH, Barlow SE, et al. NASPGHAN clinical practice guideline for the diagnosis and treatment of nonalcoholic fatty liver disease in children: recommendations from the Expert Committee on NAFLD (ECON) and the North American Society of Pediatric Gastroenterology, Hepatology and Nutrition (NASPGHAN). J Pediatr Gastroenterol Nutr 2017;64:319–334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Hampl SE, Hassink SG, Skinner AC, et al. Clinical practice guideline for the evaluation and treatment of children and adolescents with obesity. Pediatrics 2023;151:e2022060640. [DOI] [PubMed] [Google Scholar]
- 31. Tucker LA. . Limited agreement between classifications of diabetes and prediabetes resulting from the OGTT, hemoglobin A1c, and fasting glucose tests in 7412 U.S. Adults. J Clin Med 2020;9:2207. [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.