Branched-chain amino acids (BCAAs) and aromatic AAs (AAAs) are associated with increased risk for type 2 diabetes in adults. Studies in youth show conflicting results. We hypothesized that an AA metabolomic signature can be defined to identify youth at risk for β-cell failure and the development of type 2 diabetes. We performed targeted AA metabolomics analysis on 127 adolescents (65 girls; 15.5 [SD ±1.9] years old, Tanner stage II–V) with normal weight or obesity across the spectrum of glycemia, with assessment of AA concentrations by mass spectrometry, at fasting, and steady state of a hyperinsulinemic-euglycemic clamp, with determination of insulin sensitivity (IS) per fat-free mass (FFM). We measured insulin secretion during a 2-h hyperglycemic clamp and calculated the disposition index per FFM (DIFFM), a measure of β-cell function. Our results showed that concentration of glycine (Gly) and the glutamine (Gln)-to-glutamate (Glu) ratio were lower, whereas BCAA, tyrosine, and lysine (Lys) concentrations were higher in the groups with obesity and dysglycemia compared with those with normal weight. Gly and Gln-to-Glu ratio were positively related to IS and DIFFM, with opposite relationships observed for BCAAs, AAAs, and Lys. We conclude that a metabolic signature of low Gly concentration and low Gln-to-Glu ratio, and elevated BCAAs, AAAs, and Lys concentrations may constitute a biomarker to identify youth at risk for β-cell failure.

Article Highlights
  • Branched-chain amino acids (BCAAs) and aromatic amino acids (AAAs) are associated with increased risk for type 2 diabetes in adults. This relationship is unclear in youth, with studies showing conflicting results.

  • We aimed to determine if an amino acid (AA) profile could be identified to characterize youth at risk for type 2 diabetes.

  • We show that a metabolic signature of low glycine concentration, low glutamine-to-glutamate ratio, and elevated BCAA, AAA and lysine concentrations, as well as increased concentration of short-chain, AA-derived acylcarnitines, may constitute a biomarker profile to identify youth at risk for β-cell failure.

  • Our findings provide support for use of the identified AA profile in metabolomic studies aiming at preventive and therapeutic interventions in youth-onset prediabetes and type 2 diabetes.

Type 2 diabetes in youth is now recognized as having a more aggressive course than in adults, with more profound insulin resistance (1) and rapid deterioration of β-cell function compared with adults (2,3). There is a critical need to identify the metabolic pathways involved, to improve risk stratification, and provide better treatment targets.

Branched-chain amino acids (BCAAs) and related metabolites are associated with obesity and insulin resistance (4) and are prognostic for the development of type 2 diabetes in adults. Elevated plasma levels of the BCAAs leucine (Leu), isoleucine (Ile), and valine (Val) and the aromatic amino acids (AAAs) phenylalanine (Phe) and tyrosine (Tyr) were associated with up to fivefold risk of type 2 diabetes in longitudinal cohorts (5,6). The data from pediatric studies are limited and conflicting. Fasting levels of BCAAs were positively related to insulin sensitivity (IS) in youth with obesity with and without dysglycemia (7), with no adverse effect of BCAAs on β-cell function (8). In contrast, other studies showed increased BCAAs in children with obesity and a positive association with insulin resistance (9,10).

The divergence of findings in pediatrics may be related to different study populations as well as conditions of study and methodology. Fasting metabolite concentrations may not adequately reflect early alterations of insulin-mediated substrate utilization that characterizes insulin resistance in youth (11). Given that hyperinsulinemia is a determinant of amino acid (AA) uptake in muscle (12), we hypothesized that dynamic, insulin-stimulated conditions may better reflect the defect in AA metabolism in youth, and that an adverse AA metabolomic signature characterizes youth with prediabetes and type 2 diabetes. We thus evaluated fasting and insulin-stimulated concentrations of BCAA, AAA, glycine (Gly), and short-chain AA-derived acylcarnitines (acylCNs) in youth with obesity across the spectrum of glucose regulation compared with normal-weight counterparts and characterized the relationship of these metabolites with in vivo IS and secretion.

Study Design and Participants

The study comprise 127 adolescents (n = 65 girls), 15.5 SD ±1.9) years of age, Tanner stage II–V, 57% Hispanic, 31% non-Hispanic Black (NHB), and 12% non-Hispanic White (NHW). Of the participants, 30 had normal weight (NW) and normal glucose tolerance (NGT; the NW-NGT group) and 97 had overweight/obesity (OW; BMI ≥85th percentile for age and sex). On the basis of oral glucose tolerance test (OGTT) results, participants with OW were categorized as having NGT (OW-NGT; n = 33), prediabetes (n = 34), or type 2 diabetes (n = 30) (13). Youth with type 2 diabetes had adequate glycemic control, with HbA1c <8.5% (mean ± SD: HbA1c 6.24% ± 0.61%; 44.8 ± 6.7 mmol/mol) and mean diabetes duration 15.34 (SD ± 16.9) months, maintained with lifestyle modifications (n = 6), metformin (n = 16), insulin and metformin (n = 6), or insulin alone (n = 2). Oral hypoglycemic agents and long-acting insulin were discontinued 24 h before the OGTT (48 h before the clamp) (11). Participants were excluded in the presence of other diseases or chronic medication that could interfere with endocrine function or pregnancy.

All procedures received approval from the institutional review boards of the Baylor College of Medicine. Parental informed consent and child assent were obtained before any research procedure. Studies were performed in the Metabolic Research Unit (MRU) at the Children’s Nutrition Research Center. Participants underwent an OGTT; a body composition assessment including fat mass and fat-free mass (FFM) by a dual-energy X-ray absorptiometry scan and subcutaneous abdominal fat tissue (SAT) and visceral abdominal fat tissue (VAT) by MRI scan; and a hyperinsulinemic-euglycemic clamp (HEC), and a hyperglycemic clamp (in random order, 2 weeks apart) after a 12-h overnight fast. Participants were asked to refrain from heavy exercise for 24 h before presentation to the MRU. They were admitted overnight to the MRU and consumed a standardized weight maintenance diet (55% carbohydrate, 30% protein, and 15% fat) the day before the clamp studies.

OGTT

Participants ingested dextrose 1.75 g/kg body weight (maximum, 75.0 g). Blood samples were obtained at −15 min, 0 min before and at 15, 30, 60, 90, and 120 min after the ingestion to determine plasma glucose and insulin concentrations. AA concentrations were measured at fasting (0 min) and at 120 min to assess the dynamic change in AAs in response to physiologic insulin response to the oral glucose challenge.

HECs and Hyperglycemic Clamp

A 3-h HEC (100 mg/dL) was performed after a 10- to 12-h overnight fast to evaluate in vivo IS and clearance, as previously described (14). Basal hepatic glucose production was evaluated by the use of [6,6-2H2] glucose started 120 min before starting the HEC. AA and acylCN concentrations were measured during fasting and at steady state (SS) of the HEC. A 2-h hyperglycemic clamp (∼225 mg/dL) was performed 2–3 weeks apart from the HEC to measure insulin secretion (2).

Calculations

Substrate turnover at baseline was calculated during the last 30 min of the fasting 2-h isotopic infusion according to SS tracer dilution equations to determine hepatic glucose production and hepatic IS (14). Insulin-stimulated glucose disposal and IS per FFM (ISFFM) were calculated during the last 30 min of the HEC. The metabolic clearance rate of insulin was calculated by dividing the insulin infusion rate by the increase in circulating insulin concentrations during the SS of the HEC, as described previously (15). First-phase insulin was calculated as the mean insulin concentrations from 2.5 to 12.5 min of the hyperglycemic clamp, and the disposition index (DI) per FFM (DIFFM) was calculated as ISFFM × first-phase insulin (16).

Biochemical Measurements

Plasma glucose was measured with a glucose analyzer (Yellow Springs Instrument) and insulin by electrochemiluminescence immunoassays (Elecsys 2010; Roche Diagnostics). The isotopic enrichment of [6,6-2H2] glucose was measured by gas chromatography–mass spectrometry (11). Targeted metabolic profiling of AAs and acylCNs was performed using liquid chromatography–mass spectrometry (TSQ Altis; Thermo Scientific) (17). We used a stable isotope tracer as an internal standard for each individual AA measured, which minimizes inter- and intra-assay variability. The inter- and intra-assay coefficient of variation values were <4% for all AAs, except ornithine (5.6% and 5.8%, respectively) and Tyr (7.2% and 8.4%, respectively).

Statistical Analyses

Statistical analyses were performed using ANOVA followed by post hoc Bonferroni correction (or Kruskal-Wallis test for nonparametric variables) for multiple group comparison. Categorical variables were compared using a χ2 test. ANCOVA was used to adjust for group differences in race/ethnicity and covariates of interest (sex and Tanner stage). Bivariate relationships were examined using Pearson or Spearman correlation. The relationship of AAs to IS and β-cell function was further assessed in linear regression models, controlling for the covariates of sex, race/ethnicity, and Tanner stage (and percent body fat for ISFFM). Analyses were performed using SPSS, version 28 (IBM Corp., Armonk, NY). Data are presented as mean ± SD or as percentages. P values with post hoc Bonferroni correction for multiple comparisons are presented. A two-tailed P value ≤0.025 was considered statistically significant.

Principle component analysis (PCA) was also performed to reduce the number of correlated metabolites into clusters of fewer components (18) and to evaluate the contribution of AAs to the major principle components (PCs). We also performed linear regression analysis to examine the relationship of PCs to IS and β-cell function.

Data and Resource Availability

Data are available upon appropriate request to the corresponding author.

Physical and Metabolic Characteristics

The participants did not differ with respect to age or Tanner stage (Table 1). There was a greater proportion of boys in the NW-NGT group, and a greater number of Hispanic youth in the type 2 diabetes group. Therefore, we adjusted for sex, race/ethnicity, and Tanner stage in subsequent analyses. Peripheral and hepatic IS values were lower in youth with prediabetes and type 2 diabetes compared with the OW-NGT and NW-NGT groups. The insulin metabolic clearance rate was lower in the groups with obesity compared with the NW-NGT group. Insulin clearance was strongly related to ISFFM (r = 0.8; P < 0.001). First-phase insulin secretion and DI values were significantly lower in the groups with dysglycemia compared with the OW-NGT group (Table 1).

Table 1

Study population physical and metabolic characteristics

NW-NGTOW-NGTPrediabetesType 2 diabetesP value*
Physical characteristic      
 Age (years) 15.9 ± 1.7 15.2 ± 2.0 15.4 ± 2.1 15.7 ± 1.8 0.5 
 Sex (male/female) 22/8 11/22 18/16 11/19 0.006 
 Race/ethnicity (NHB/NHW/H) 16/6/8 3/6/24 12/3/19 9/0/21 0.001 
 Tanner stage (II–III/IV–V) 2/28 4/29 3/31 2/28 0.7 
 BMI (kg/m222.3 ± 3.1 30.7 ± 4.8 34.6 ± 6.0 35.5 ± 5.0 <0.001a,b,c,d,e 
 BMI z score 0.48 ± 0.71 1.89 ± 0.43 2.21 ± 0.40 2.23 ± 0.38 <0.001a,b,c,d 
 Waist circumference (cm) 74.2 ± 7.3 96.2 ± 13.7 104.5 ± 13.8 105.8 ± 13.0 <0.001a,b,c,d,e 
 Fat mass (kg) 12.1 ± 5.5 31.0 ± 8.7 36.5 ± 11.8 37.9 ± 9.8 <0.001a,b,c,d 
 FFM (kg) 51.6 ± 10.4 49.5 ± 10.2 57.8 ± 12.2 56.0 ± 9.1 0.007e 
 Body fat (%) 18.4 ± 7.0 37.2 ± 5.2 37.5 ± 7.0 39.1 ± 6.1 <0.001a,b,c 
 VAT (cm2) 31 ± 17.6 74.1 ± 33 87.7 ± 28.8 93.3 ± 36.6 <0.001a,b,c 
 SAT (cm2) 123.2 ± 76.1 391.4 ± 119.3 451.5 ± 156.6 493.9 ± 152.9 <0.001a,b,c 
Metabolic characteristic      
 HbA1c (%) [mmol/mol] 5.54 ± 0.25 [37.1 ± 2.8] 5.46 ± 0.22 [36.2 ± 2.4] 5.67 ± 0.36 [38.5 ± 3.9] 6.24 ± 0.61 [44.8 ± 6.7] <0.001a,d,f 
 Adiponectin (ng/mL) 26.4 ± 15.5 18.2 ± 14.4 15.7 ± 10.6 10.8 ± 6.9 <0.001a,b 
 Hepatic glucose production (mg/kg/min) 2.6 ± 0.4 2.1 ± 0.4 2.2 ± 0.6 2.1 ± 0.3 <0.001a,b,c 
 Hepatic IS (mg/kg/min per μU/mL) 42.0 ± 16.7 27.0 ± 16.7 19.1 ± 9.1 17.2 ± 8.8 <0.001a,b,c,d 
 Peripheral IS (mg/kgFFM/min per μU/mL) 11.0 ± 4.0 4.4 ± 2.2 3.0 ± 1.3 2.8 ± 1.7 <0.001a,b,c 
 Insulin metabolic clearance rate (mL/kgFFM/min) 15.6 ± 4.5 13.5 ± 3.4 11.9 ± 2.5 13.5 ± 3.7 <0.001b 
 First-phase insulin (μU/mL) 72.8 ± 34.8 238.8 ± 172.6 171.5 ± 85.9 93.9 ± 93.0 <0.001b,c,d 
 DI (mg/kgFFM/min) 733.4 ± 388.8 773.4 ± 315.0 504.6 ± 293.3 232.9 ± 252.9 <0.001a,d,e 
NW-NGTOW-NGTPrediabetesType 2 diabetesP value*
Physical characteristic      
 Age (years) 15.9 ± 1.7 15.2 ± 2.0 15.4 ± 2.1 15.7 ± 1.8 0.5 
 Sex (male/female) 22/8 11/22 18/16 11/19 0.006 
 Race/ethnicity (NHB/NHW/H) 16/6/8 3/6/24 12/3/19 9/0/21 0.001 
 Tanner stage (II–III/IV–V) 2/28 4/29 3/31 2/28 0.7 
 BMI (kg/m222.3 ± 3.1 30.7 ± 4.8 34.6 ± 6.0 35.5 ± 5.0 <0.001a,b,c,d,e 
 BMI z score 0.48 ± 0.71 1.89 ± 0.43 2.21 ± 0.40 2.23 ± 0.38 <0.001a,b,c,d 
 Waist circumference (cm) 74.2 ± 7.3 96.2 ± 13.7 104.5 ± 13.8 105.8 ± 13.0 <0.001a,b,c,d,e 
 Fat mass (kg) 12.1 ± 5.5 31.0 ± 8.7 36.5 ± 11.8 37.9 ± 9.8 <0.001a,b,c,d 
 FFM (kg) 51.6 ± 10.4 49.5 ± 10.2 57.8 ± 12.2 56.0 ± 9.1 0.007e 
 Body fat (%) 18.4 ± 7.0 37.2 ± 5.2 37.5 ± 7.0 39.1 ± 6.1 <0.001a,b,c 
 VAT (cm2) 31 ± 17.6 74.1 ± 33 87.7 ± 28.8 93.3 ± 36.6 <0.001a,b,c 
 SAT (cm2) 123.2 ± 76.1 391.4 ± 119.3 451.5 ± 156.6 493.9 ± 152.9 <0.001a,b,c 
Metabolic characteristic      
 HbA1c (%) [mmol/mol] 5.54 ± 0.25 [37.1 ± 2.8] 5.46 ± 0.22 [36.2 ± 2.4] 5.67 ± 0.36 [38.5 ± 3.9] 6.24 ± 0.61 [44.8 ± 6.7] <0.001a,d,f 
 Adiponectin (ng/mL) 26.4 ± 15.5 18.2 ± 14.4 15.7 ± 10.6 10.8 ± 6.9 <0.001a,b 
 Hepatic glucose production (mg/kg/min) 2.6 ± 0.4 2.1 ± 0.4 2.2 ± 0.6 2.1 ± 0.3 <0.001a,b,c 
 Hepatic IS (mg/kg/min per μU/mL) 42.0 ± 16.7 27.0 ± 16.7 19.1 ± 9.1 17.2 ± 8.8 <0.001a,b,c,d 
 Peripheral IS (mg/kgFFM/min per μU/mL) 11.0 ± 4.0 4.4 ± 2.2 3.0 ± 1.3 2.8 ± 1.7 <0.001a,b,c 
 Insulin metabolic clearance rate (mL/kgFFM/min) 15.6 ± 4.5 13.5 ± 3.4 11.9 ± 2.5 13.5 ± 3.7 <0.001b 
 First-phase insulin (μU/mL) 72.8 ± 34.8 238.8 ± 172.6 171.5 ± 85.9 93.9 ± 93.0 <0.001b,c,d 
 DI (mg/kgFFM/min) 733.4 ± 388.8 773.4 ± 315.0 504.6 ± 293.3 232.9 ± 252.9 <0.001a,d,e 

Values are reported as mean ± SD. P values are reported for ANOVA, Kruskal-Wallis, or χ2 tests. H, Hispanic.

*

Post hoc P values <0.05

a

NW-NGT vs. type 2 diabetes

b

NW-NGT vs. prediabetes

c

NW-NGT vs. OW-NGT

d

OW-NGT vs. type 2 diabetes

e

OW-NGT vs. prediabetes

f

prediabetes vs. type 2 diabetes.

VAT and SAT data were available for 94 study participants (17 in NW-NGT group, 29 in OW-NGT group, 27 with prediabetes, 21 with type 2 diabetes). Hyperglycemic clamp data were not available for 15 participants (three in the NW-NGT group, four in the OW-NGT group, six with prediabetes, and one with type 2 diabetes), due to difficulty with venous access or with scheduling a second study visit.

AA Concentrations in Youth With Type 2 Diabetes and Prediabetes

There was a distinct AA profile in youth with type 2 diabetes and prediabetes compared with similarly obese youth with NGT and with NW peers, both fasting and under the hyperinsulinemic conditions of the OGTT and the HEC (Supplementary Fig. 1). We further examined the concentrations of individual AAs in the four participant groups and their relationship to the outcomes of interest of IS and β-cell function (DIFFM). In the fasting state and at SS of the HEC (Fig. 1A and B), Gly concentrations were lower, and concentrations of the BCAAs Val, Leu, and Ile were higher in the groups with type 2 diabetes and prediabetes compared with the NW-NGT group (P < 0.001). Concentrations of the AAAs Tyr and tryptophan were higher in the groups with obesity compared with the NW-NGT group. Lys concentrations were higher in the groups with obesity and dysglycemia, and fasting alanine (Ala) concentration was higher in the groups with obesity, compared with the NW-NGT group. The glutamine (Gln)-to-glutamate (Glu) ratio was lower in participants with type 2 diabetes and prediabetes compared with the OW-NGT and NW-NGT groups. This was primarily related to higher Glu levels (P < 0.001) in the groups with obesity and dysglycemia (Fig. 1A and B). Gly concentration was negatively associated with percent body fat, whereas the BCAAs, AAAs, and Lys were positively associated with percent body fat, waist circumference, SAT, and VAT. Glu concentration was positively associated, whereas the Gln-to-Glu ratio was negatively correlated, with percent body fat and VAT. Gly concentration and the Gln-to-Glu ratio were positively associated, whereas the BCAAs, Tyr, and Lys concentrations were negatively related with adiponectin (Table 2).

Figure 1

AA profiles in the postabsorptive fasting state (A) and under insulin-stimulated conditions (B) at the SS of the HEC. Post hoc P values <0.025: aNW-NGT vs. type 2 diabetes, bNW-NGT vs. prediabetes, cNW-NGT vs. OW-NGT, dOW-NGT vs. type 2 diabetes, eOW-NGT vs. prediabetes, fprediabetes vs. type 2 diabetes. Asn, asparagine; Asp, aspartic acid; Cys, cysteine; His, histidine; Met, methionine; Orn, ornithine; Pro, proline; Ser, serine; Thr, threonine; Trp, tryptophan.

Figure 1

AA profiles in the postabsorptive fasting state (A) and under insulin-stimulated conditions (B) at the SS of the HEC. Post hoc P values <0.025: aNW-NGT vs. type 2 diabetes, bNW-NGT vs. prediabetes, cNW-NGT vs. OW-NGT, dOW-NGT vs. type 2 diabetes, eOW-NGT vs. prediabetes, fprediabetes vs. type 2 diabetes. Asn, asparagine; Asp, aspartic acid; Cys, cysteine; His, histidine; Met, methionine; Orn, ornithine; Pro, proline; Ser, serine; Thr, threonine; Trp, tryptophan.

Close modal
Table 2

Relationship of AAs to adiposity measures, adiponectin, peripheral and hepatic IS, and insulin clearance

ConditionBMI (kg/m2)Waist circumference (cm)Body fat (%)VAT (cm2)SAT (cm2)Adiponectin (ng/mL)Hepatic IS (mg/kg/min per μU/mL)Peripheral IS (mg/kgFFM/min per μU/mL)Insulin clearance rate (mL/kgFFM/min)
Gly Fasting −0.423 −0.375 −0.410 −0.287 −0.366 0.375 0.378 0.507 0.386 
SS-Eu −0.371 −0.322 −0.376 — −0.283 0.384 0.284 0.438 0.356 
Gln Fasting — — −0.259 — — — — 0.228 — 
SS-Eu — — — — — — — — — 
Glu Fasting 0.451 0.445 0.395 0.331 0.372 −0.277 −0.471 −0.513 −0.397 
SS-Eu 0.421 0.421 0.294 0.341 0.317 −0.402 −0.448 −0.483 −0.338 
Gln/Glu Fasting −0.483 −0.475 −0.482 −0.383 −0.431 0.246 0.467 0.569 0.438 
SS-Eu −0.376 −0.362 −0.299 −0.280 −0.285 0.369 0.422 0.489 0.432 
Val Fasting — 0.222 — — — — −0.219 −0.212 — 
SS-Eu 0.404 0.438 0.237 0.403 0.385 −0.332 −0.445 −0.430 −0.365 
Leu Fasting — — — — — — — — — 
SS-Eu 0.286 0.306 — 0.279 — −0.318 −0.421 −0.389 −0.257 
Ile Fasting 0.236 0.232 — — — — — −0.228 — 
SS-Eu 0.420 0.429 0.235 0.396 0.385 −0.410 −0.522 −0.519 −0.338 
Tyr Fasting 0.530 0.502 0.437 0.442 0.439 −0.213 −0.409 −0.463 −0.329 
SS-Eu 0.393 0.379 0.230 0.314 0.387 −0.291 −0.435 −0.424 −0.497 
Phe Fasting — — — — — — — — — 
SS-Eu 0.341 0.370 0.234 0.437 0.385 — −0.342 −0.370 −0.225 
Lys Fasting 0.378 0.379 0.285 0.426 0.361 — −0.281 −0.296 −0.234 
SS-Eu 0.494 0.468 0.374 0.457 0.425 −0.283 −0.411 −0.422 −0.297 
Ala Fasting 0.242 0.232 0.211 — — — −0.379 −0.342 — 
SS-Eu 0.224 0.223 — — — — — — — 
ConditionBMI (kg/m2)Waist circumference (cm)Body fat (%)VAT (cm2)SAT (cm2)Adiponectin (ng/mL)Hepatic IS (mg/kg/min per μU/mL)Peripheral IS (mg/kgFFM/min per μU/mL)Insulin clearance rate (mL/kgFFM/min)
Gly Fasting −0.423 −0.375 −0.410 −0.287 −0.366 0.375 0.378 0.507 0.386 
SS-Eu −0.371 −0.322 −0.376 — −0.283 0.384 0.284 0.438 0.356 
Gln Fasting — — −0.259 — — — — 0.228 — 
SS-Eu — — — — — — — — — 
Glu Fasting 0.451 0.445 0.395 0.331 0.372 −0.277 −0.471 −0.513 −0.397 
SS-Eu 0.421 0.421 0.294 0.341 0.317 −0.402 −0.448 −0.483 −0.338 
Gln/Glu Fasting −0.483 −0.475 −0.482 −0.383 −0.431 0.246 0.467 0.569 0.438 
SS-Eu −0.376 −0.362 −0.299 −0.280 −0.285 0.369 0.422 0.489 0.432 
Val Fasting — 0.222 — — — — −0.219 −0.212 — 
SS-Eu 0.404 0.438 0.237 0.403 0.385 −0.332 −0.445 −0.430 −0.365 
Leu Fasting — — — — — — — — — 
SS-Eu 0.286 0.306 — 0.279 — −0.318 −0.421 −0.389 −0.257 
Ile Fasting 0.236 0.232 — — — — — −0.228 — 
SS-Eu 0.420 0.429 0.235 0.396 0.385 −0.410 −0.522 −0.519 −0.338 
Tyr Fasting 0.530 0.502 0.437 0.442 0.439 −0.213 −0.409 −0.463 −0.329 
SS-Eu 0.393 0.379 0.230 0.314 0.387 −0.291 −0.435 −0.424 −0.497 
Phe Fasting — — — — — — — — — 
SS-Eu 0.341 0.370 0.234 0.437 0.385 — −0.342 −0.370 −0.225 
Lys Fasting 0.378 0.379 0.285 0.426 0.361 — −0.281 −0.296 −0.234 
SS-Eu 0.494 0.468 0.374 0.457 0.425 −0.283 −0.411 −0.422 −0.297 
Ala Fasting 0.242 0.232 0.211 — — — −0.379 −0.342 — 
SS-Eu 0.224 0.223 — — — — — — — 

Values reported are unadjusted Spearman correlation coefficients. Values in nonbolded type are significant at P < 0.025. Bolded values indicate P < 0.01. —, nonsignificant relationship at P ≥ 0.025. SS-Eu, steady state of the hyperinsulinemic-euglycemic clamp.

AAs were not significantly different in girls compared with boys in the total group except for fasting Phe (65.8 ± 20.6 vs. 75.7 ± 17.7 μmol/L; P = 0.007) and SS-Leu (113.0 ± 26.7 vs. 124.9 ± 29.7 μmol/L; P = 0.025) concentrations in girls compared with boys, respectively. The Gln-to-Glu ratio was also lower in girls (9.8 ± 3.1 vs. 12.1 ± 6.8; P = 0.007). When the NW-NGT group and overweight participants were examined separately, fasting levels of BCAAs, Gln, Phe, aspargine, histidine, and Gly were significantly lower in girls compared with boys only in the NW-NGT group. In those with overweight, SS-Leu was lower in girls compared with boys. In the overweight subgroups, girls with type 2 diabetes had lower fasting Val (292.8 ± 11.05 vs. 328.8 ± 15.6 μmol/L; P = 0.07) and Leu (231.45 ± 9.5 vs. 272.27 ± 16.3 μmol/L; P = 0.045) concentrations, with no significant sex differences in fasting metabolites noted in the OW-NGT and prediabetes groups.

Relationship of AA and Short-Chain AcylCNs to IS and β-Cell Function

Fasting and SS-Gly and Gln-to-Glu ratio positively related, whereas fasting and SS-BCAAs, SS-AAAs, and fasting and SS-Lys negatively related to peripheral ISFFM, hepatic IS, and insulin clearance in the total study population (Table 2). These relationships remained significant after adjustment for sex, race/ethnicity, and Tanner stage (Fig. 2A). Fasting Ala levels correlated inversely with peripheral and hepatic IS (Table 2). The relationship of SS-AAs to ISFFM remained significant after further adjustment for percent body fat (except for Lys and Ala, and attenuated for SS-Gly). Race/ethnicity was not a significant determinant of the relationship between AAs and ISFFM after adjusting for percent body fat. Sex differences were noted in the ISFFM regression models for Val (β = −0.25; P = 0.004), Leu (β = −0.3; P = 0.001), Ile (β = −0.25; P = 0.003), and Gln-to-Glu ratio (β = 0.21; P = 0.008). After adjustment for sex, race/ethnicity, Tanner stage, SS-Gly concentration, and Gln-to-Glu ratio were positively associated, whereas SS-Val, Leu, and Ile concentrations were inversely associated, with DIFFM (Fig. 2B). SS-Phe (β = −0.28; P = 0.005) and SS-Lys (β = −0.24; P = 0.027) were also inversely related to DIFFM. There was no significant relationship between Ala or arginine and DIFFM.

Figure 2

Relationship of AA concentrations at steady-state of the HEC with peripheral IS (A) and DI (B). Linear regression models were adjusted for the covariates sex, race/ethnicity, and Tanner stage. For each AA, β and P values are reported as the independent variable in the linear regression model. An effect for sex and an effect for race/ethnicity were noted in the ISFFM regression models, and a Tanner stage effect was noted in the DIFFM regression model, as reported in Results.

Figure 2

Relationship of AA concentrations at steady-state of the HEC with peripheral IS (A) and DI (B). Linear regression models were adjusted for the covariates sex, race/ethnicity, and Tanner stage. For each AA, β and P values are reported as the independent variable in the linear regression model. An effect for sex and an effect for race/ethnicity were noted in the ISFFM regression models, and a Tanner stage effect was noted in the DIFFM regression model, as reported in Results.

Close modal

When the groups with normal weight and overweight were examined separately, the relationships of the AAs with ISFFM and DIFFM remained significant in the groups with overweight, but were attenuated in the NW-NGT group except for the SS-Leu relationship with ISFFM (r = −0.45; P = 0.019).

The BCAA-derived isobutyryl (C4)-CN and isovaleryl (C5)-CN concentrations were higher in the groups with prediabetes and type 2 diabetes (Supplementary Fig. 2). SS–propionylcarnitine (C3-CN) (r = −0.22; P = 0.021), SS–C5-CN (r = −0.24; P = 0.009) and fasting and SS–C4-CN (r = −0.3, P = 0.001; and r = −0.39, P < 0.001, respectively) values were negatively related to ISFFM. These relationships remained significant after adjustment for sex, race/ethnicity, and Tanner stage; there was no relationship between acylCN and DIFFM.

PCA

PCA analysis revealed six principal components that each contributed >5% to the variance in metabolites at SS-HEC. Together, they explained 74% of the variance of all AAs and acylCNs. PC1 and PC2 explained most of the variance (47%) (Supplementary Table 1). In a linear regression analysis, PC1 (β = −0.16; P = 0.02), in addition to percent body fat (β = −0.8; P < 0.001), and sex (β = 0.24; P = 0.03) contributed to the variance in ISFFM, independent of other PCs, Tanner stage, or race/ethnicity (R2 = 0.58; P < 0.001). PC1 and PC2 did not contribute independently to the variance in DIFFM as the primary outcome in the regression analysis.

In this study, we demonstrate that youth with type 2 diabetes and prediabetes have a distinct AA profile with elevated levels of BCAAs and AAAs along with reduced Gly concentration and Gln-to-Glu ratio, compared with normoglycemic youth. These differences in AA profiles are related to IS. Importantly, our findings show that Gly and Gln-to-Glu ratio are positively related to β-cell function with opposite relationships observed for BCAAs and AAAs. We conclude that 1) reduced IS in youth is associated with impairment in AA metabolism, and 2) a metabolic signature of low Gly, low Gln-to-Glu ratio, and elevated BCAA and AAA concentrations is associated with β-cell dysfunction.

Our findings of a distinct AA profile and elevated concentration of the BCAA-derived acylCNs (4) in youth with dysglycemia, compared with normoglycemic youth of similar adiposity and with NW peers, point to impaired protein metabolism in youth with obesity that is more pronounced in the more severe insulin resistance states of prediabetes and type 2 diabetes and who also manifest reduced insulin clearance. Our findings are consistent with findings from studies of adults (4,5,19) and clarify inconsistencies in the pediatric literature (7) that may be related to differences in the study population and in the experimental design. Other pediatric studies support elevation of BCAAs and AAAs in relation to obesity and insulin resistance (9,10). Untargeted metabolomics studies in children identified BCAAs and related products in youth with obesity compared with peers of normal weight (9,20), including children as young as 6 to 10 years of age (20). In a large cohort of Hispanic children, concentrations of BCAAs and their catabolites C3-CN and butyrylcarnitine (C4-CN) were significantly elevated in relation to obesity (10). These metabolites were related to indices of insulin resistance and an adverse cardiometabolic profile (9,20,21). Elevated concentrations of BCAA metabolites were found in youth with nonalcoholic fatty liver disease (22). Authors of a study that evaluated BCAA response to the OGTT similarly reported elevation in BCAAs after glucose challenge in insulin-resistant adolescents (23). Overall, these findings are consistent with an impaired antiproteolytic effect of insulin in the setting of obesity and insulin resistance (24). In turn, chronic elevation of circulating AAs may directly promote insulin resistance, possibly via disruption of insulin signaling in skeletal muscle (25). Our data expand these findings to children with obesity across the glycemic spectrum and using gold-standard methodologies of in vivo measurement of IS and secretion. Importantly, we show that elevated BCAA and AAA concentrations and reduced Gly concentration and Gln-to-Glu ratio are also related to β-cell dysfunction as reflected by the DI, consistent with the metabolite pattern associated with incident diabetes in adults (5,6). The reciprocal relationship between high levels of BCAAs and Glu and low levels of Gly has been related to mitochondrial substrate overload, with improvement in IS and restoration of Gly level upon BCAA restriction (26). High extracellular Glu levels may contribute to loss of β-cell function (27), whereas Gln level improved incretin effect (28). Lys was related to peripheral and hepatic IS in our study. The elevation of Lys, a precursor of aminoadipic acid, may be a compensatory mechanism to the insulin resistance, because aminoadipic acid improves insulin secretion in animal studies (29).

Sex differences were noted in the relationship between AAs with IS. These sex differences are overall consistent with the study by Newbern et al. (30), who reported important sex effects in the relationship between BCAAs and their byproducts with markers of insulin resistance (HOMA of insulin resistance and adiponectin) in adolescents with obesity. However, unlike their findings of higher levels of fasting AAs in adolescent boys compared with girls with obesity, we only noted these sex differences in fasting AAs in the group with normal weight. Of interest, in subgroup analysis, we also found fasting concentrations of BCAAs (i.e., Val and Leu) were higher in male than in female youth with type 2 diabetes. Differences in the study population, with ours including youth with prediabetes and type 2 diabetes, outcome measures, and methodology may account for these divergent findings. Nevertheless, these highlight the importance of future studies powered to examine sex differences in metabolic pathways involved in the pathogenesis of youth-onset type 2 diabetes.

The strengths of our study include a targeted metabolomic examination of AA profile and acylCN by liquid chromatography–mass spectrometry using stable isotopic internal standards for all reported metabolites; evaluation of both fasting and dynamic response of these metabolites; and in vivo measurement of IS and secretion in a well-phenotyped cohort of youth across the glycemic spectrum. This allowed us to demonstrate a metabolic signature reflective of β-cell dysfunction in youth. Flux measurements and future intervention studies can further clarify mechanisms and confirm our findings. We had some imbalance in race/ethnicity distribution among the four glycemic categories examined. This was adjusted for in the statistical analyses. Our results did not indicate major race or ethnicity effects on the relationship between AA metabolites with the outcome measures of IS and DI. We did not systematically evaluate diet and activity patterns in our participants. However, all participants consumed a standardized diet the day before the clamp studies, to minimize acute dietary effects on the AA profile. Although the cross-sectional design did not allow us to imply causation, our findings provide strong support for an AA metabolic signature in youth with prediabetes and type 2 diabetes, related not only to insulin resistance but also to β-cell function.

In summary, reduced suppression of BCAAs in response to insulin may be the earliest manifestation of a defect in AA metabolism in youth. The resultant prolonged exposure to elevated AA levels may contribute to β-cell stress. A metabolic signature of low Gly, low Gln-to-Glu ratio, and elevated BCAA, AAA, and Lys concentrations may constitute a biomarker to identify youth at risk for β-cell failure. Longitudinal and intervention studies in high-risk youth are needed to verify these findings.

This article contains supplementary material online at https://doi.org/10.2337/figshare.24975603.

Acknowledgments. The investigators thank the youth volunteers and their parents for participation in this study; the nurses and staff of the Metabolic Research Unit at the Children’s Nutrition Research Center; Elizabeth Johnson and Janette Rodela for research coordination efforts; Roman Shypailo and Maryse Laurent for assistance with body composition assessment; and Adam Gillum for help with illustrations. The authors thank Dr. Christopher Newgard for his review of the manuscript and helpful input.

Funding. This study was supported by the U.S. Department of Agriculture, Agricultural Research Service Current Research Information System Award 3092-51000-057 to F.B.

The funding source had no involvement in the study design, collection, analysis and interpretation of data, writing of the report, or in the decision to submit the article for publication.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.B. designed and conducted the study, obtained funding, analyzed the data, and wrote the manuscript. H.E.-A. contributed to data analysis and writing of the manuscript. M.M. and S.S. contributed to laboratory analyses and reviewed the manuscript. M.P., R.K., and C.C. contributed to data analyses and reviewed the manuscript. All authors approved the manuscript in its final version. F.B. 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.

Prior Presentation. This work was presented in abstract format (348-OR) at the American Diabetes Association 82nd Scientific Sessions, New Orlean, LA, 3–6 June, 2022.

1.
RISE Consortium
.
Metabolic contrasts between youth and adults with impaired glucose tolerance or recently diagnosed type 2 diabetes: I. Observations using the hyperglycemic clamp
.
Diabetes Care
2018
;
41
:
1696
1706
2.
Bacha
F
,
Gungor
N
,
Lee
S
,
Arslanian
SA.
Progressive deterioration of β-cell function in obese youth with type 2 diabetes
.
Pediatr Diabetes
2013
;
14
:
106
111
3.
TODAY Study Group
.
Postintervention effects of varying treatment arms on glycemic failure and beta-cell function in the TODAY trial
.
Diabetes Care
2021
;
44
:
75
80
4.
Newgard
CB
,
An
J
,
Bain
JR
, et al
.
A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance
.
Cell Metab
2009
;
9
:
311
326
5.
Wang
TJ
,
Larson
MG
,
Vasan
RS
, et al
.
Metabolite profiles and the risk of developing diabetes
.
Nat Med
2011
;
17
:
448
453
6.
Palmer
ND
,
Stevens
RD
,
Antinozzi
PA
, et al
.
Metabolomic profile associated with insulin resistance and conversion to diabetes in the Insulin Resistance Atherosclerosis Study
.
J Clin Endocrinol Metab
2015
;
100
:
E463
E468
7.
Mihalik
SJ
,
Michaliszyn
SF
,
de las Heras
J
, et al
.
Metabolomic profiling of fatty acid and amino acid metabolism in youth with obesity and type 2 diabetes: evidence for enhanced mitochondrial oxidation
.
Diabetes Care
2012
;
35
:
605
611
8.
Michaliszyn
SF
,
Sjaarda
LA
,
Mihalik
SJ
, et al
.
Metabolomic profiling of amino acids and β-cell function relative to insulin sensitivity in youth
.
J Clin Endocrinol Metab
2012
;
97
:
E2119
E2124
9.
McCormack
SE
,
Shaham
O
,
McCarthy
MA
, et al
.
Circulating branched-chain amino acid concentrations are associated with obesity and future insulin resistance in children and adolescents
.
Pediatr Obes
2013
;
8
:
52
61
10.
Butte
NF
,
Liu
Y
,
Zakeri
IF
, et al
.
Global metabolomic profiling targeting childhood obesity in the Hispanic population
.
Am J Clin Nutr
2015
;
102
:
256
267
11.
Bacha
F
,
Bartz
SK
,
Puyau
M
,
Adolph
A
,
Sharma
S.
Metabolic flexibility across the spectrum of glycemic regulation in youth
.
JCI Insight
2021
;
6
:
e146000
12.
Bonadonna
RC
,
Saccomani
MP
,
Cobelli
C
,
DeFronzo
RA.
Effect of insulin on system A amino acid transport in human skeletal muscle
.
J Clin Invest
1993
;
91
:
514
521
13.
American Diabetes Association
.
2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes—2019
.
Diabetes Care
2019
;
42
(
Suppl. 1
):
S13
S28
14.
Bacha
F
,
Tomsa
A
,
Bartz
SK
, et al
.
Nonalcoholic fatty liver disease in Hispanic youth with dysglycemia: risk for subclinical atherosclerosis?
J Endocr Soc
2017
;
1
:
1029
1040
15.
Arslanian
SA
,
Bacha
F
,
Saad
R
,
Gungor
N.
Family history of type 2 diabetes is associated with decreased insulin sensitivity and an impaired balance between insulin sensitivity and insulin secretion in white youth
.
Diabetes Care
2005
;
28
:
115
119
16.
Bacha
F
,
Gungor
N
,
Arslanian
SA.
Measures of beta-cell function during the oral glucose tolerance test, liquid mixed-meal test, and hyperglycemic clamp test
.
J Pediatr
2008
;
152
:
618
621
17.
Marini
JC.
Quantitative analysis of 15N-labeled positional isomers of glutamine and citrulline via electrospray ionization tandem mass spectrometry of their dansyl derivatives
.
Rapid Commun Mass Spectrom
2011
;
25
:
1291
1296
18.
O’Rourke
NHL.
A Step-by-Step Approach to Using SAS for Factor Analysis and Structural Equation Modeling.
Cary, NC
,
SAS Publishing
,
2014
, p.
1
42
19.
Shah
SH
,
Bain
JR
,
Muehlbauer
MJ
, et al
.
Association of a peripheral blood metabolic profile with coronary artery disease and risk of subsequent cardiovascular events
.
Circ Cardiovasc Genet
2010
;
3
:
207
214
20.
Perng
W
,
Gillman
MW
,
Fleisch
AF
, et al
.
Metabolomic profiles and childhood obesity
.
Obesity (Silver Spring)
2014
;
22
:
2570
2578
21.
Cosentino
RG
,
Churilla
JR
,
Josephson
S
, et al
.
Branched-chain amino acids and relationship with inflammation in youth with obesity: a randomized controlled intervention study
.
J Clin Endocrinol Metab
2021
;
106
:
3129
3139
22.
Goffredo
M
,
Santoro
N
,
Tricò
D
, et al
.
A branched-chain amino acid-related metabolic signature characterizes obese adolescents with non-alcoholic fatty liver disease
.
Nutrients
2017
;
9
:
642
23.
Tricò
D
,
Prinsen
H
,
Giannini
C
, et al
.
Elevated α-hydroxybutyrate and branched-chain amino acid levels predict deterioration of glycemic control in adolescents
.
J Clin Endocrinol Metab
2017
;
102
:
2473
2481
24.
Jensen
MD
,
Haymond
MW.
Protein metabolism in obesity: effects of body fat distribution and hyperinsulinemia on leucine turnover
.
Am J Clin Nutr
1991
;
53
:
172
176
25.
Krebs
M
,
Krssak
M
,
Bernroider
E
, et al
.
Mechanism of amino acid-induced skeletal muscle insulin resistance in humans
.
Diabetes
2002
;
51
:
599
605
26.
White
PJ
,
Lapworth
AL
,
An
J
,
Wang
L
, et al
.
Branched-chain amino acid restriction in Zucker-fatty rats improves muscle insulin sensitivity by enhancing efficiency of fatty acid oxidation and acyl-glycine export
.
Mol Metab
2016
;
5
:
538
551
27.
Newsholme
P
,
Procopio
J
,
Lima
MMR
,
Pithon-Curi
TC
,
Curi
R.
Glutamine and glutamate--their central role in cell metabolism and function
.
Cell Biochem Funct
2003
;
21
:
1
9
28.
Reimann
F
,
Williams
L
,
da Silva Xavier
G
,
Rutter
GA
,
Gribble
FM.
Glutamine potently stimulates glucagon-like peptide-1 secretion from GLUTag cells
.
Diabetologia
2004
;
47
:
1592
1601
29.
Wang
TJ
,
Ngo
D
,
Psychogios
N
, et al
.
2-Aminoadipic acid is a biomarker for diabetes risk
.
J Clin Invest
2013
;
123
:
4309
4317
30.
Newbern
D
,
Gumus Balikcioglu
P
,
Balikcioglu
M
, et al
.
Sex differences in biomarkers associated with insulin resistance in obese adolescents: metabolomic profiling and principal components analysis
.
J Clin Endocrinol Metab
2014
;
99
:
4730
4739
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.