Intrahepatic fat, irrespective of ethnicity, is associated with reduced endogenous insulin clearance and hepatic insulin resistance in obese youths: a cross-sectional and longitudinal study from the Yale Pediatric NAFLD cohort.

Abstract

Aims.

Non-Alcoholic Fatty Liver Disease (NAFLD) has been associated with reduced endogenous insulin clearance (EIC) and hepatic insulin resistance (HIRI). These relationships, however, might be differentially affected by the ethnic background, as populations of African ancestry are typically featured with reduced intrahepatic fat content (HFF%) but impaired EIC. Therefore, we evaluated the influence of the ethnicity on the relationships between HFF%, EIC and HIRI.

Methods.

The HFF% was quantified by magnetic resonance imaging in a multi-ethnic cohort of 632 obese youths from 7 to 18 years of age at baseline and after a 2-year follow-up. Insulin secretion rate (ISR), EIC and HIRI were estimated by modeling glucose, insulin, and C-peptide data during 9-point oral glucose tolerance tests.

Results.

African American youths exhibited the lowest HFF% and a prevalence of NAFLD less than half of Caucasians and one-third of Hispanics. Furthermore, African Americans had lower EIC and glucose-stimulated ISR, despite similar HIRI and plasma insulin levels, compared with Caucasians and Hispanics. EIC and HIRI were markedly reduced in individuals with NAFLD and declined across group-specific HFF% tertiles in all ethnic groups. Consistently, the HFF% correlated with EIC and HIRI, irrespective of the ethnic background, after adjustment for age, sex, ethnicity, adiposity, waist-hip ratio, pubertal status, and plasma glucose levels. An increased HFF% at follow-up was associated with decreased EIC and increased HIRI across all groups.

Conclusions.

Intrahepatic lipid accumulation is associated with reduced insulin clearance and hepatic insulin sensitivity in obese youths, irrespective of their ethnic background.

Introduction

A reduced insulin clearance has been proposed as a key pathogenic mechanism underlying prediabetes and type 2 diabetes (1; 2). The liver is the primary site of endogenous insulin extraction, accounting for the majority (up to 80%) of secreted insulin removal during its first-pass transit through the portal system (3). Although a wide interindividual variability exists, hepatic insulin clearance is typically reduced in insulin-resistant obese adults (4; 5) and youths (6; 7). This has been regarded as a compensatory mechanism for insulin resistance to allow for a larger fraction of the secreted insulin to reach the systemic circulation. In the long term, this process however may lead to chronic hyperinsulinemia (5), a major determinant in the etiology of diabetes (1; 2; 8).

Insulin clearance may be affected by many factors, such as Non-Alcoholic Fatty Liver Disease (NAFLD) (9–14) and ethnicity (15–19). In youths, the relationship between these factors has been often overlooked. In fact, although pediatric NAFLD is associated with a greater prevalence of prediabetes and type 2 diabetes (20; 21), whether this effect is mediated by a lower insulin clearance and/or hepatic insulin resistance, as in adults (9–14), is still debated (22). Ethnic differences in insulin clearance and NAFLD are well documented. Compared with Caucasians and Hispanics, children and adolescents of African ancestry typically show reduced insulin clearance (15–19), which is associated, however, with a lower intrahepatic fat content and prevalence of NAFLD (21; 23; 24). This paradox has led to the hypothesis of a dissociation between liver metabolic abnormalities and hepatic fat content in African Americans (14; 23).

In this study, we aim to evaluate whether intrahepatic fat accumulation contributes to impaired hepatic insulin clearance across different ethnic groups during the fasting state and after a glucose challenge. To these ends, we analyzed cross-sectional and longitudinal data from a large and carefully phenotyped cohort of overweight and obese youths representing the three most prevalent racial groups in the United States.

Methods

We analyzed data from the Yale Pediatric NAFLD cohort from 2012 to 2019, a cohort recruited as part of a long-term project aimed at studying the impact of intrahepatic fat accumulation on glucose homeostasis (21; 25). Inclusion criteria were age between 7 and 18 years and BMI greater than the 85^th percentile for age and sex. Exclusion criteria were known hepatic diseases (other than NAFLD), alcohol consumption, and use of medications affecting glucose or lipid metabolism. A detailed medical and family history was obtained from all participants, and a physical examination was performed including measurement of the waist-hip ratio (WHR). The age- and sex-adjusted BMI (BMI z-score) was calculated (26). We ascertained ethnic origin by a self-reporting questionnaire, with Caucasians, African Americans, and white Hispanics as the major categories. Subjects with persistent elevation of ALT levels for more than 6 months underwent appropriate blood tests to exclude autoimmune hepatitis, Wilson’s disease, alpha-1-antitrypsin deficiency, hepatitis B and C, and iron overload. Metabolic studies were performed at the Hospital Research Unit (HRU) of the Yale New Haven Hospital. Intrahepatic fat content was quantified by a validated magnetic resonance imaging (MRI) method, while endogenous insulin clearance and hepatic insulin resistance were assessed during 3-hour, 9-point oral glucose tolerance test (OGTT) at baseline and after a follow-up of 2 years. Subjects received standard nutritional guidance and recommendations for physical activity during the follow-up. Overall, the study group was composed by 632 adolescents, including 229 (36.2%) Caucasians, 172 (27.2%) African Americans, and 231 (36.6%) Hispanics. The study was approved by the Yale University Human Investigation Committee. Written parental informed consent and child assent were obtained from all participants before enrollment.

OGTT.

All participants underwent a 3-hour OGTT (1.75 g glucose/kg body weight, up to 75 g) at baseline and after the 2-year follow-up evaluation. Subjects arrived in the morning after a 12-hour overnight fast. Blood samples for glucose, insulin, and C-peptide measurement were drawn at 0, 10, 20, 30, 60, 90, 120, 150 and 180 min from oral glucose ingestion.

MRI.

Liver MRI studies at baseline and at follow-up were performed on a Siemens Sonata 1.5 Tesla system using an advanced magnitude-based liver fat quantification MRI technique, the 2-point Dixon (2PD), as modified by Fishbein et al. (27). This method is based on phase-shift imaging where Hepatic Fat Fraction (HFF%) is calculated from the signal difference between the vectors resulting from in-phase and out-of-phase signals. Five regions of interest were drawn on each image, and the mean pixel signal intensity level was recorded (27). The pulse sequence was a T1-weighted fast low angle shot gradient echo. Slices were acquired using a 400-cm field of view (echo time 4.76, repetition time 100, 4 excitations, 90-degree flip angle, matrix 256×128, bandwidth 140). Non-alcoholic fatty liver was defined as HFF >5.5%, according to previous studies (21; 28). We previously validated the modified 2PD method in our cohort against proton nuclear magnetic resonance (r=0.954, p<0.0001) (29) and liver biopsy (r=0.836, p=0.0001) (21), which is the gold standard for diagnosing NAFLD, conducted according to standard procedures (21; 29). Fast-MRI was able to track longitudinal changes in liver fat content in obese adolescent with NAFLD (30).

Biochemical analyses.

Plasma glucose was determined by the glucose oxidase method (Beckman Instruments, Brea, CA). Plasma insulin was measured by the Linco radioimmunoassay (St. Charles, MO). Plasma C-peptide levels were determined with an assay by Diagnostic Products (Los Angeles, CA). Lipid levels were determined with an Auto-Analyzer (model 747-200, Roche Diagnostics, Indianapolis, IN). Liver enzymes were measured using standard automated kinetic enzymatic assays.

Calculations.

Insulin secretion rate (ISR) was estimated by C-peptide deconvolution (31). Beta cell function parameters were calculated by mathematical modeling of ISR and glucose concentrations, as previously reported (18; 32; 33). Briefly, this model describes the relationship between ISR and glucose as the sum of two components (32). The first component represents the dependence of ISR on glucose concentration through a dose–response function relating the two variables. From the dose–response, beta cell glucose sensitivity (the slope) is calculated. The dose–response is modulated by a potentiation factor, accounting for various physiological mechanisms (e.g. prolonged hyperglycaemia, non-glucose substrates, gastrointestinal hormones, neural modulation). The potentiation factor averages 1 during the test and expresses relative potentiation or inhibition of ISR; its excursion is quantified by the ratio between the mean values in the intervals 160-180 min and 0-20 min (potentiation factor ratio). The second ISR component represents the dependence of ISR on the rate of change of glucose concentration and is determined by a single parameter (beta cell rate sensitivity), which is related to early (first-phase) insulin release (32).

Endogenous insulin clearance was calculated as the ratio between fasting ISR and plasma insulin levels (ISR_fast/I_fast) and as the ratio of their areas under the curve (AUC) over the duration of the OGTT (ISR_AUC/ I_AUC), as previously reported (18; 34; 35). This method provides an estimate of endogenous insulin clearance under less controlled but more physiological conditions compared with other techniques using intravenous glucose and insulin infusions. We validated this method against hepatic insulin clearance estimated by mathematical modeling of clamp data (2) in a subset of 47 participants who underwent a hyperinsulinemic-euglycemic clamp at baseline (r=0.56, p<0.0001).

Whole-body insulin sensitivity was assessed by the OGTT-derived whole-body insulin sensitivity index (WBISI), which has been validated against the euglycemic-hyperinsulinemic clamp in obese adolescents (36; 37). The WBISI was calculated as 10,000 / square root of [(fasting glucose x fasting insulin) x (mean glucose x mean insulin during OGTT)] (36). The hepatic insulin resistance index (HIRI) was calculated as the product of the AUCs of plasma glucose and plasma insulin during the first 30 minutes of the OGTT (Glucose AUC_0–30 × Insulin AUC_0–30), as proposed by DeFronzo (38). A strong correlation has been reported between the HIRI and a direct measure of hepatic insulin resistance (r=0.64, p<0.0001) provided by the product of basal endogenous glucose production (measured with labeled glucose) and fasting plasma insulin (38).

Statistical analyses.

Continuous variables are presented as means ± standard deviations (SD), unless otherwise stated, and nominal variables are reported as counts and/or percentages. Variables with a skewed distribution are presented as median [interquartile range] and were log-transformed in multivariable analyses to approximate univariate normality, except for HFF% for which a square root transformation was used. Differences between ethnic groups and tertiles of percent changes in HFF% or HIRI at follow-up were tested by ANOVA or Kruskal–Wallis tests. Post-hoc pairwise comparisons were performed using Tukey or Steel-Dwass post-hoc tests, respectively. Nominal variables were compared using χ² tests. Two-way ANOVA was used to compare plasma glucose, insulin, and ISR profiles during the OGTT between groups. Correlations between variables were tested using Pearson correlation or Spearman rank correlation, as appropriate. Multivariable linear regression models were used to account for known potential confounders, including age, sex, ethnicity, BMI z-score, WHR, pubertal status, and plasma glucose levels. To assess whether the relationship of HFF% with insulin clearance and HIRI is modulated by ethnicity, we added an interaction factor (HFF% × ethnicity) to regression models. Statistical tests were performed using JMP Pro 13.2.1 (SAS Institute Inc., Cary, NC) using a two-sided α level of 0.05.

Results

Clinical and anthropometric features of the study population are shown in Table 1. Hispanic youth were younger than African Americans and Caucasians while African Americans showed higher BMI compared with the other groups (Table 1). There were no statistically significant differences in sex, BMI z-score, or Tanner stage among the three ethnic groups. African Americans and Hispanics exhibited the lowest and highest HFF% and prevalence of NAFLD, respectively (p<0.001). Ethnic differences in HFF% and prevalence of NAFLD remained significant after adjustments for age, sex, BMI z-score, developmental status, and plasma glucose levels (p<0.001 for both).

Table 1.

Clinical and metabolic characteristics of study subjects in the three ethnic groups.

	Caucasians	African Americans	Hispanics	p
CLINICAL FEATURES
Number (%)	229 (36.2)	172 (27.2)	231 (36.6)	-
Age (years)	13.8 ± 3.3 a	13.7 ± 3.2 a	12.8 ± 2.7	0.0004
Sex (M/F) [%]	39.3/60.7	38.4/61.6	47.6/52.4	0.10
Tanner Stage (I-III/IV-V) [%]	33/67	29/71	39/61	0.10
Body Mass Index (kg/m2)	32.0 ± 6.7	33.8 ± 7.1 a,b	31.8 ± 6.5	0.009
Body Mass Index z-score	2.1 ± 0.7	2.2 ± 0.6	2.2 ± 0.5	0.06
Waist-Hip Ratio	0.93 ± 0.09	0.92 ± 0.11 a	0.95 ± 0.09	0.01
GLUCOSE METABOLISM
Whole-Body Insulin Sensitivity Index	1.8 [1.1-2.6]	1.6 [1.1-2.5]	1.6 [1.1-2.4]	0.26
Fasting Plasma Glucose (mmol/L)	5.0 ± 0.5	5.1 ± 0.5	5.1 ±0.4	0.67
2-hour Plasma Glucose (mmol/L)	6.9 ± 1.8	6.8 ±1.8	6.8 ± 1.5	0.93
Fasting Plasma Insulin (pmol/L)	189 [131-292]	193 [140-287]	189 [140-287]	0.77
2-hour Plasma Insulin (pmol/L)	931 [571-1726]	1022 [553-1659]	952 [560-1666]	0.99
LIPID PROFILE
Total Cholesterol (mg/dL)	156 ± 31	156 ± 26	156 ± 38	0.98
HDL Cholesterol (mg/dL)	45 ± 11	47 ± 11 a,b	42 ± 10	<0.0001
LDL Cholesterol (mg/dL)	88± 27	94 ± 23	91 ± 33	0.13
Triglycerides (mg/dL)	90 [65-136]	64 [49-121] a,b	102 [75-145]	<0.0001
LIVER VARIABLES
Hepatic Fat Fraction (%)	3.3 [0.0-11.3] a	0.6 [0.0-2.9] a,b	6.7 [0.9-18.0]	<0.0001
Non-Alcoholic Fatty Liver Disease (n, (%))	83 (36.2)	27 (15.7)	127 (55.0)	<0.0001
Alanine Transaminase (U/L)	20 [13-32]	15 [11-22] a,b	23 [16-33]	<0.0001
Aspartate Transaminase (U/L)	22 [18-27]	20 [18-25] a	24 [20-30]	0.001

Data are mean±SD or median [interquartile range] for normally or non-normally distributed variables, respectively. Differences were tested using ANOVA or Kruskal–Wallis tests followed by post-hoc pairwise comparisons.

^ap<0.05 vs Hispanics,

^bp<0.05 vs Caucasians.

Ethnic differences in glucose tolerance and beta cell function.

Fasting and 2-hour plasma glucose, insulin, and ISR did not differ across the three groups (Table 1, Figure 1). During the OGTT, however, African Americans exhibited lower glucose excursions (p<0.0001) (Figure 1A), which were mirrored by lower ISR profiles (p=0.005) (Figure 1B), compared with the other groups. Despite different ISR, plasma insulin responses were similar across ethnic groups (Figure 1C). Beta cell rate sensitivity was higher in African Americans and Hispanics than in Caucasians, while beta cell glucose sensitivity and potentiation were similar in all participants (Figure 1D–F). Whole-body insulin sensitivity did not differ across groups in univariable analysis (Table 1) or after adjustment for age and adiposity (p=0.12).

Figure 1.

Plasma glucose (A), insulin secretion rate (B), plasma insulin (C), and beta cell function variables (D-F) measured during an oral glucose tolerance test (OGTT) in Caucasian (C; n=229), African American (AA; n=172) and Hispanic (H; n=231) obese adolescents from the Yale Pediatric NAFLD cohort. Data are mean ± SEM. Differences in A-C were tested by two-way ANOVA including group, time, and an interaction term between group and time as factors. Differences in D-F were tested by Kruskall-Wallis test followed by post-hoc pairwise comparisons.

Ethnic differences in insulin clearance.

Insulin clearance was evaluated during the fasting state and after the ingestion of 75g glucose during the OGTT. During the fasting state, endogenous insulin clearance was lower in African Americans compared to the other ethnic groups, while there was no difference between Caucasians and Hispanics (Figure 2A). Differences in fasting insulin clearance remained significant after adjustments for age, sex, BMI z-score, WHR, developmental status, HFF%, and plasma glucose levels (p=0.0005).

Figure 2.

Fasting endogenous insulin clearance by group (A) and by presence of Non-Alcoholic Fatty Liver (NAFL; B) and by group-specific tertile in hepatic fat content (HFF%; C); endogenous insulin clearance during an oral glucose tolerance test (OGTT) by group (D) and by NAFL (E) and group-specific HFF% tertile (F); percent change in post-glucose vs fasting endogenous insulin clearance by group (G) and by NAFL (H) and group-specific HFF% tertile (I) in Caucasian (C; n=229), African American (AA; n=172) and Hispanic (H; n=231) obese adolescents from the Yale Pediatric NAFLD cohort. Data are mean ± SEM. Differences were tested by Kruskall-Wallis test followed by post-hoc pairwise comparisons.

After glucose ingestion, insulin clearance was lower in African Americans than in Caucasians (p=0.02) and intermediate in Hispanics (Figure 2D). Group differences remained statistically significant after adjustments for age, sex, BMI z-score, WHR, developmental status, HFF%, and plasma glucose levels (p=0.03).

Post-glucose insulin clearance was associated with fasting insulin clearance in all ethnic groups (Caucasians: r=0.75; African Americans: r=0.69; Hispanics: r=0.67; p<0.0001 for all). The percent reduction in insulin clearance during the OGTT was different among groups (p=0.02), being smaller in African Americans than Hispanics (p=0.02) and intermediate in Caucasians (Figure 2G). Ethnic differences lost statistical significance after adjustments for age, sex, BMI z-score, WHR, developmental status, HFF%, and plasma glucose levels (p=0.12).

Association between fatty liver and insulin clearance.

Fasting insulin clearance declined significantly in presence of NAFLD (Figure 2B) and from the lower to the upper group-specific HFF% tertile (Figure 2C) only in Caucasians. Conversely, fasting insulin clearance was similar between subjects with and without NAFLD in African Americans and Hispanics. After glucose ingestion, insulin clearance was markedly lower in the groups of individuals with NAFLD (Figure 2E) and declined across HFF% tertiles (Figure 2F) in all the ethnic groups. Regression models using continuous variables demonstrated the effect of HFF% on both fasting (std. β= −0.11, p=0.01) and post-glucose insulin clearance (std. β= −0.23, p<0.0001), without significant interactions between HFF% and the ethnic group (p=0.13 and p=0.15). The association of HFF% with post-glucose insulin clearance remained significant after adjustments for age, sex, ethnicity, BMI z-score, WHR, pubertal status, and plasma glucose levels (std. β= −0.13, p=0.003).

The percent reduction in insulin clearance from the fasting to the absorptive phase was greater in individuals with NAFLD (Figure 2H) and across HFF% tertiles (Figure 2I) within all the three ethnic groups. A regression model confirmed the effect of HFF% on the post-glucose decline in insulin clearance (std. β= −0.19, p<0.0001), without significant interactions between HFF% and the ethnic group (p=0.99). The association of HFF% with post-glucose changes in insulin clearance remained significant after adjustments for age, sex, ethnicity, BMI z-score, WHR, pubertal status, and plasma glucose levels (std. β= −0.11, p=0.015).

Association between fatty liver and hepatic insulin resistance.

The hepatic insulin resistance, assessed by the HIRI, did not differ across groups in univariable analysis (Figure 3A) or after adjustment for age and BMI (p=0.57). The HIRI increased markedly in presence of NAFLD (Figure 3B) and across group-specific HFF% tertiles (Figure 3C). Regression models using HFF% as a continuous variable demonstrated the effect of HFF% on HIRI (std. β= 0.11, p<0.0001), without a significant interaction between HFF% and the ethnic group (p=0.18). The association between HFF% and HIRI remained significant after adjustments for age, sex, ethnicity, BMI z-score, WHR, pubertal status, and plasma glucose levels (std. β=0.17, p=0.0001).

Figure 3.

Hepatic insulin resistance index by group (A) and by NAFL (B) and group-specific HFF% tertile (C) in Caucasian (C; n=229), African American (AA; n=172) and Hispanic (H; n=231) obese adolescents from the Yale Pediatric NAFLD cohort. Correlations of hepatic insulin resistance and endogenous insulin clearance at fasting (D), during an oral glucose tolerance test (OGTT) (E), and its percent variation from the fasting to the absorptive phase (F) in the same cohort. Grouped data are mean ± SEM. Differences were tested by Kruskall-Wallis test followed by post-hoc pairwise comparisons. Correlations were tested using Spearman rank correlation.

Association between insulin clearance and hepatic insulin resistance.

The HIRI showed a negative correlation with insulin clearance at fasting (r= −0.37, p<0.0001) (Figure 3D) and after glucose ingestion (r= −0.67, p<0.0001) (Figure 3E), and with its percent reduction from the fasting to the absorptive phase (r= −0.50, p<0.0001) (Figure 3E). These relationships were not influenced by the ethnicity (interaction factors p=0.70, p=0.69, and p=0.66, respectively) and remained significant in models adjusted for age, sex, ethnicity, BMI z-score, WHR, pubertal status, and plasma glucose levels (std. β= −0.38, p<0.0001; std. β= −0.64, p<0.0001; and std. β= −0.42, p<0.0001, respectively).

Longitudinal changes in hepatic fat, insulin clearance and hepatic insulin resistance.

A total of 89 adolescents had repeated assessments of HFF% and model-derived insulin clearance after a median follow-up of 2.0 [1.6-3.4] years, including 36 (40.5%) Caucasians, 23 (25.8%) African Americans and 30 (33.7%) Hispanics. Their baseline demographic and metabolic characteristics were not significantly different from those of the original cross-sectional cohort (age 13.8 ± 2.5 years, 39 boys/50 girls, BMI z-score 2.3 ± 0.4, fasting plasma glucose 5.2 ± 0.5 mmol/L). At follow-up, the three ethnic groups showed small and similar median changes in HFF% (0.0 [−2.7–3.6]%, p=0.55), fasting insulin clearance (0.1 [−0.1–0.3] L/min/m², p=0.55), insulin clearance after glucose ingestion (0.0 [−0.2–0.2] L/min/m², p=0.97), percent changes in insulin clearance from fasting to post-glucose conditions (0.02 [−0.19–0.14, p=0.95), and HIRI (30 [−604–666], p=0.84).

After stratification by tertile of percent changes in HFF% at follow-up, we observed that OGTT insulin clearance increased in youths with a reduction in HFF% greater than 38% (first tertile), while it decreased in those with an increase in HFF% greater than 27% (third tertile; p=0.007) (Figure 4A). HIRI was substantially unchanged in the first tertile but increased by almost 60% in the third tertile (p=0.04) (Figure 4B). Percent changes in fasting insulin clearance and in its variations under post-glucose conditions were not different across tertiles (p=0.35 and p=0.40, respectively).

Figure 4.

Percent changes at follow-up in endogenous insulin clearance during an oral glucose tolerance test (OGTT) (A) and hepatic insulin resistance index (HIRI) (B) across tertiles of percent changes in hepatic fat content (HFF%) in obese adolescents from the Yale Pediatric NAFLD cohort (n=89). Percent changes at follow-up in endogenous insulin clearance at fasting (C) and during an OGTT (D) across tertiles of percent changes in HIRI in the same cohort. Data are mean ± SEM. Differences were tested by Kruskall-Wallis test followed by post-hoc pairwise comparisons.

After stratification by tertile of percent changes in HIRI at follow-up, both fasting and OGTT insulin clearance markedly increased in the first tertile (HIRI −25% or lower), only slightly changed in the second tertile (HIRI −25% to +28%), and decreased in the third tertile (HIRI +28% or higher) (Figure 4C–D). Percent variations in insulin clearance under post-glucose conditions were not different across tertiles (p=0.35).

Discussion

The current study provides evidence in support of the interaction between fatty liver and ethnicity in the development of abnormalities of insulin metabolism. We report that intrahepatic fat accumulation is associated with reduced glucose-stimulated insulin clearance and hepatic insulin resistance across all ethnic groups and that insulin clearance is affected by ethnicity in obese youth. In fact, African American adolescents have markedly lower endogenous insulin clearance in the fasting state, lower reduction in insulin clearance after carbohydrate consumption, and enhanced beta cell rate sensitivity, compared with Caucasians and Hispanics, despite higher body mass and similar insulin sensitivity.

Insulin clearance plays a major role in glucose metabolism and insulin resistance, contributing to hyperinsulinemia in subjects with insulin resistance. The liver is responsible for the removal of a large part (50%-80%) of the secreted insulin during its first-pass transit through the portal system (3). The reason for this apparently futile insulin cycle from the beta cell to the liver is difficult to explain from an evolutionary perspective (1). Based on the current understanding, however, it may relate to the need of preventing the negative metabolic effects of chronic hyperinsulinemia (1; 8). Among the factors linked to impaired insulin clearance, the intrahepatic fat accumulation has been associated with reduced hepatic insulin extraction in most studies (9–14) but data in youths were limited (22). Herein, we identified the hepatic fat content as an independent factor affecting endogenous insulin clearance in obese youths both in cross-sectional and in longitudinal analyses. These findings are consistent with data in adults showing that insulin clearance is impaired in metabolically unhealthy obese adults (who are more likely to show fatty liver) than in metabolically healthy obese (39) and that there is an inverse correlation between NAFLD and insulin clearance (11). It has been long known that liver failure is associated with hyperinsulinemia, probably due to impairment of insulin clearance (40). Our data show that impairment of insulin clearance can occur at a young age in adolescents with fatty liver and that changes in intrahepatic fat content parallel changes in insulin clearance over time. The latter evidence suggests that one of the mechanisms leading to a reduction of hyperinsulinemia, when intrahepatic fat content is reduced, might be an amelioration of hepatic insulin extraction. It has been demonstrated that ingestion of glucose or a high carbohydrate diet induce a reduction of insulin clearance (41; 42), probably owing higher insulin needs. In presence of high intrahepatic fat content, as in patients with NAFLD, hepatic insulin extraction is lower and there is a more significant reduction of insulin clearance after a glucose challenge. Given the high degree of insulin resistance in these adolescents, the lower insulin clearance probably results in an enhancement of hepatic lipogenesis causing further intrahepatic fat accumulation and perpetuating, in a vicious circle, insulin resistance.

Intrahepatic fat accumulation only partly explains the inter-individual variability in insulin clearance. Evidence from us (21; 24) and others (17; 18; 23) demonstrates that individuals of African ancestry have 10-15% lower insulin clearance compared with Caucasians and Hispanics. Intrahepatic fat content affects insulin clearance during fasting and after carbohydrate ingestion in Caucasians, while we did not observe any statistically significant difference in insulin clearance during the fasting state between African American and Hispanic youth with and without fatty liver. These data suggest that intrahepatic fat content plays a major role in modulating hepatic insulin clearance in the fasting state only among Caucasian obese youth, while its role in regulating fasting insulin clearance in African Americans and Hispanics may be marginal. Nevertheless, the drop in insulin clearance in subjects with high intrahepatic fat content (third HFF% tertile) after carbohydrate ingestion was comparable among the three ethnic groups (p>0.10), showing a similar effect of intrahepatic fat on post-glucose insulin extraction.

Interestingly, we confirmed the relationship between intrahepatic fat and hepatic insulin resistance in obese youths (22; 43) and we demonstrated that it is not influenced by the ethnic background. In addition, our cross-sectional and longitudinal data expose the association between hepatic insulin resistance and alterations in endogenous insulin clearance, again without evidence of differences related to the ethnicity, thereby supporting the suggested pathogenetic link between the two conditions (11; 44; 45).

In this study, we reported other important metabolic differences between African American adolescents and their counterparts. We described higher beta cell responsiveness to glucose changes in African Americans, which is consistent with previous studies using the OGTT (18) and the hyperglycemic clamp (17; 46). Given that whole-body insulin sensitivity is comparable between groups, the combination of earlier insulin secretory response and reduced insulin clearance may contribute to explain lower plasma glucose peaks in African Americans (47). Lower prevailing glucose levels, in turn, may be responsible for the reduced total glucose-stimulated insulin secretion (18).

To our best knowledge, this is the first study that examined the influence of the ethnic background on the relationship between intrahepatic fat content and insulin clearance in a large cohort of youths with a wide range of hepatic fat content and insulin resistance. Strengths of this study include the presence of longitudinal data and the detailed clinical and metabolic characterization of participants, using an advanced magnitude-based MRI technique for liver fat quantification and data from 9-point OGTTs for the estimation of insulin secretion and clearance and the modeling of specific beta cell function components. The method used to assess endogenous insulin clearance is relatively easy and non-invasive, meaning that it can be adopted outside the research setting also in a pediatric population, and has been validated against a more complex model-derived estimate of hepatic insulin clearance provided by the hyperinsulinemic-euglycemic clamp (2). This OGTT-derived method, which was previously adopted in pediatric studies by our group (34; 35) and others (18), provides an accurate estimate of endogenous insulin clearance under more physiological fasting and fed conditions compared with other techniques using intravenous glucose and insulin infusions.

We acknowledge that this study has some limitations. Longitudinal analysis should be interpreted with caution given the small sample size and that wide changes in insulin clearance and resistance seem to occur in African Americans without substantial changes in HFF% (Supplementary Appendix, Figure S1). The extreme phenotype of our cohort of obese youths allowed us to evaluate the effect of intrahepatic fat accumulation in a population with a very low prevalence of fatty liver (i.e., African Americans); further studies are needed to confirm whether these findings extend to different populations.

In conclusion, we demonstrated that intrahepatic lipid accumulation is associated with reduced endogenous insulin clearance and hepatic insulin sensitivity in obese youths, irrespective of their ethnic background, in cross-sectional and longitudinal analyses adjusted for multiple confounding factors. These findings support a pathogenetic link between liver steatosis and hepatic metabolic abnormalities contributing to glucose intolerance.