New-onset T2DM patients were diagnosed at the Department of Endocrinology, the First People's Hospital of Yunnan Province, Kunming, China, from December 2016 to June 2018. Ninety-seven T2DM patients diagnosed with T2DM within 1 year were included in the analysis (53 male/44 female, 43.3 ± 11.2 years of age). The diagnosis and classification of T2DM were based on 1999 WHO standards[35]: (1) Fasting plasma glucose concentration ≥ 7.0 mmol (or ≥ 126 mg/dL); (2) ≥ 11.1 mmol (or ≥ 200 mg/dL) 2 h after a 75 g oral glucose load; and (3) HbA1c ≥ 6.5%. The exclusion criteria were as follows: (1) A history of diabetes that was diagnosed more than 12 mo prior; and (2) acute complications of diabetes and severe liver and/or kidney dysfunction or other serious health conditions. The study procedures were conducted in accordance with the Helsinki Declaration of 1975 and were approved by the Medical Ethics Review Committee of the First People's Hospital of Yunnan Province [No. 2016(001)].

Anthropometric measures of height, body weight and waist circumference (WC) were used to determine BMI and body weight status. The 2004 WHO classifications for the Asian/Chinese population were used for this study: (1) NW (BMI < 24); (2) OW (BMI ≥ 24 and < 28); and (3) OB (BMI ≥ 28), which differ from the standards of United States and European populations (NW: BMI < 25 kg/m², OW: BMI ≥ 25, OB: ≥ 30). Visceral adipose tissue (VAT) at the level of the umbilicus was measured via an abdominal dual BIA machine following the manufacturer’s protocol (DUALSCAN HDS-2000, Omron Health care Co., Kyoto, Japan).

Fasting blood samples of the T2DM patients were collected and analyzed following the standard operation protocol. Briefly, overnight fasting (> 8 h) venous blood samples were collected and centrifuged immediately to separate the serum.

The frozen serum samples were thawed on ice and at 4°C followed by deproteinization by the addition of acetonitrile (1:3 ratio of serum to acetonitrile). The sample was then vortexed for 2 min followed by centrifugation at 12000 rpm for 10 min at 4°C to remove any precipitate from the supernatant before ultra-performance liquid chromatography (UPLC)/triple stage quadrupole mass spectrometer (TSQ/MS) analysis. A quality control sample consisting of 6 reference standards (isoleucine, leucine, valine, tyrosine, tryptophan, phenylalanine) was prepared and run after each of the 15 serum samples.

A 5 μL aliquot of extracted serum sample was injected into the UPLC column (DionexUltiMate3000-UPLC, United States) TSQ/MS (Thermo TSQ Endura, United States). Separation was achieved on a Luna Omega Polar C18 column (2.1 mm × 100 mm, 1.6 μm, Phenomenex) held at 40°C. The samples were eluted with A (water with 0.1% formic acid) and B (acetonitrile), and the gradient program was 1% B over 0-0.5 min, 1%-20% B over 0.5-9 min, 20%-75% B over 9-11 min, and 75%-99% B over 11-16 min. The composition was held at 99% B for 0.5 min and finally returned to 1% B at 20 min. The flow rate was 0.3 mL/min. Mass spectrometry was performed in positive ion electrospray (ESI +) mode. The temperature for the ion transfer tube and vaporizer was set at 350°C and 300°C, respectively. The pressures for the sheath gas, aux gas and sweep gas were set at 40, 15 and 1 Arb, respectively. The positive ion voltage was set to 3.5 kV. All the compounds were detected in selective reaction monitoring mode.

The data are presented as the mean ± standard deviation or median [25% quartile range (QR), 75% QR]. The effects of BMI and sex were analyzed using ANOVA for quantitative variables with a normal distribution or using the Wilcoxon rank sum test for nonnormally distributed parameters (SPSS 24, IBM). Post hoc multiple comparisons were assessed using Bonferroni correction. Pearson’s correlation coefficients were calculated among the variables. A 2-sided P≤0.05 was considered indicative of statistical significance.

Ninety-seven (97) T2DM patients, including 57 males (59%) and 40 females (41%), all within 1 year of T2DM diagnosis, were identified and included in this study. Of them, 33 (34%) were NW-T2DM (BMI = 22.23 ± 1.60), 42 (43.2%) were OW-T2DM (BMI = 25.90 ± 1.08), and 22 (22.8%) were OB-T2DM (BMI = 30.96 ± 2.59), based on the WHO body weight classification (Table 1).

No group differences were found in T2DM diagnosis, sex distribution, height, medication, or systolic and diastolic blood pressures (Table 1). The OB-T2DM group was significantly younger (36.77 ± 10.32 years) than the NW-T2DM and OW-T2DM groups (43.64 ± 6.95 and 42.21 ± 9.76 years, respectively, P < 0.05, each). The OB-T2DM group also showed greater values than the NW-T2DM and OW-T2DM groups in body weight, WC, WC-to-height ratio (WHR), and BMI (P < 0.01, each, Table 1).

The male and female patients showed similarities in T2DM onset age (30.4 ± 77.0 vs 43.3 ± 91.7 d), BMI (25.9 ± 4.2 vs 25.8 ± 2.9), WHR (0.535 ± 0.05 vs 0.555 ± 0.05), systolic blood pressure (125.46 ± 16.4 vs 125.88 ± 16.1 mmHg), diastolic blood pressure (84.0 ± 14.3 vs 81.2 ± 7.3 mmHg), and antidiabetic medication. The males were, however, significantly younger (39.7 ± 8.8 vs 44.0 ± 9.6 year), heavier (76.7 ± 13.7 vs 64.9 ± 8.2 kg), and taller (171.6 ± 6.3 vs 159.2 ± 4.9 cm) with greater WC (91.7 ± 8.7 vs 88.3 ± 7.7 cm, P < 0.05, all) than the females.

Two-way ANOVA of the effects of BMI and sex showed that among the 23 serum amino acids measured by using UPLC/TSQ-MS, 8 amino acids were significantly greater in obese than in NW and/or OW subjects (alanine, asparagine, carnosine, glutamate, hydroxyproline, proline, tyrosine, and tryptophan, P < 0.05, each) (Figure 1, Table 2). Valine, isoleucine, and phenylalanine were marginally greater in obese than in NW and/or OW subjects (P ≤ 0.1, each). Ten amino acids were significantly greater in male subjects than in female subjects (serine, asparagine, glutamate, lysine, hydroxyproline, citrulline, isoleucine, leucine, tyrosine, and tryptophan, P < 0.05, each). Histidine, methionine, phenylalanine, and threonine were marginally greater in male subjects than in female subjects (P < 0.1, each). Significant BMI and sex interactions were found for proline only.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 One-way and two-way ANOVA of the effects of body mass index and sex on serum amino acid concentrations in male and female new-onset type-2 diabetes patients in differing body mass index categories. A: Valine; B: Isoleucine; C: Leucine; D: Tyrosine; E: Tryptophan; F: Phenylalanine; G: Histidine; H: Arginine; I: Taurine; J: Threonine; K: Proline; L: Methionine; M: Alanine; N: Glutamate; O: Hydroxyproline; P: Asparagine; Q: Glutamine; R: Lysine; S: Citrulline; T: Homocysteine; U: Serine. N: Normal weight; P: Overweight; O: Obese group. ^aP < 0.05 vs normal weight; ^bP < 0.01 vs normal weight; ^cP < 0.05 vs overweight; NS: Not significantly different (P > 0.05); B: Body mass index; S: Sex/sex; Int: BMI × sex interaction.

Further 1-way ANOVA of BMI effects largely confirmed the 2-way ANOVA results. Within-sex 1-way ANOVA showed that most of the BMI differences were present among the male subjects only (alanine, asparagine, glutamate, valine, isoleucine, leucine, tyrosine, and tryptophan) (P ≤ 0.05, each) (Figure 1, Table 2), except that carnosine was significantly greater in the female obese group and proline was significantly lower in the female OW group than in the other two groups. Covariance analysis showed no difference in serum amino acids after controlling for day of T2DM diagnosis or controlling for metformin equivalent dose. of T2DM medicine. However, covariance analysis showed significant effects of BMI on taurine, citrulline and proline after controlling for age (P ≤ 0.05, each, Table 2).

Two-way ANOVA showed no significant effects of BMI and sex on triglycerides, total cholesterol, and low-density lipoprotein. However, high-density lipoprotein (HDL) was significantly affected by BMI, with significantly lower HDL levels found in obese T2DM patients than in their NW and OW counterparts (P < 0.01, P < 0.05, respectively), especially among male patients (Table 2).

Although multiple amino acids were significantly correlated with VAT, body weight and BMI and negatively correlated with HDL, in pooled samples (Table 3), the correlations with VAT were the strongest (tyrosine, r = 0.524, P < 0.0001; phenylalanine, r = 0.508, P < 0.0001; tryptophan r = 0.373, P < 0.01; isoleucine, r = 0.443, P = 0.002; leucine, r = 0.396, P = 0.006, valine, r = 0.375, P < 0.01; methionine, r = 0.379, P < 0.01; alanine, r = 0.429, P = 0.003; glutamate, r = 0.398, P < 0.01; lysine, r = 0.39, P < 0.01; hydroxyproline, r = 0293; P < 0.05, threonine, r = 0.388, P < 0.01; proline, r = 0.314, P < 0.05; and histidine, r = 0.278, P < 0.06), followed by body weight, BMI and HDL (Table 3). Taurine was negatively correlated with VAT (r = -0.367, P = 0.01), and HDL was negatively correlated with isoleucine (r = -0.309, P < 0.01), leucine (r = -0.276, P < 0.01) and phenylalanine (r = -0.307 P < 0.01). Within-BMI category correlation analysis showed no significant correlations in the NW group (Table 3). For the OW group, phenylalanine was highly correlated with VAT (r = 0.607, P = 0.002), and BCAAs and AAAs were significantly or marginally correlated with body weight (tyrosine, r = 0.343, P < 0.05; tryptophan r = 0.339, P < 0.05; isoleucine, r = 0.289, P = 0.063; leucine, r = 0.32, P = 0.039, valine, r = 0.286, P = 0.067, respectively), and HDL was negatively correlated with isoleucine (r = -0.417, P < 0.01), leucine (r = -0.345, P < 0.05) and phenylalanine (r = -0.323,P < 0.05) (Table 3). For the obese group, alanine was significantly correlated with VAT (r = 0.64, P < 0.05); homocysteine was significantly correlated with body weight (r = 0.475, P < 0.05) and BMI (r = 0.43, P < 0.05)); HDL was negatively correlated with asparagine (r = -0.478, P < 0.01) and hydroxyproline (r = -0.436, P < 0.05); and BCAAs and AAAs were negatively correlated with waist-to-height ratio (isoleucine, r = -0.54, P < 0.01; leucine, r = -0.51, P < 0.05; and tryptophan, r = -0.512, P < 0.05). These observations support VAT as the major source of elevated BCAAs/AAAs in obese T2DM patients[36].

While elevated BCAAs/BCKAs may induce and interact with FFA accumulation in obese T2DM candidates, T2DM development in lean/NW populations may arise from different mechanisms. Unlike the gold standard for predicting different subtypes of T2DM diagnosis (i.e., fasting glucose and HbA1c), so far, there is no established standard for T2DM prediction. Although BCAAs/AAAs is a promising predictor, its utility may be affected by a range of factors including race/ethnicity, age, sex, body weight/BMI, and subtypes of T2DM. The heterogeneous elevation of BCAAs/AAAs among new-onset T2DM patients indicates limitations and restricted utility of BCAAs/AAAs as the predictor for different subtypes of T2DM. While the greater BCAAs/AAAs elevation in obese T2DM patients would support its’ prediction in individuals with OB(-propensity) which account for a large portion of T2DM population, the lower level, or a lack of BCAAs/AAAs elevation in normal-weight and reproductive-aged females would diminish its’ predicting power in these individuals.

A longitudinal 12-year follow up study of old adults (56 ± 8 years) showed that when BMI-matched obese non-T2DM individuals (n = 189/group, BMI = 30) were compared, the highest quartile of individual with elevated baseline BCAAs/AAAs had a 2- to 3.5-fold higher odds of risk of developing diabetes per SD increment over a 12-year follow-up period based on individual BCAAs/AAAs, or a 5- to 7-fold higher odds of developing diabetes if all BCAAs/AAAs were combined, in comparison with those individuals whose plasma amino acid levels were in the lowest quartile[16]. However, such increments in odds of risk were reduced to 1.3 and 2.0, respectively if the obese T2D candidates were compared with NW controls (n = 400, BMI = 25) randomly selected form a larger pool. Thus, how the controls are selected can lead to very different outcomes.

Similarly, a meta-analysis show that MUO, MUOW and MUNW individuals would show similar 4-fold risk increase of developing T2DM if each of them is compared with healthy counterparts of their corresponding BMI category (vs MHO, MHOW, and MHNW, respectively), but MUO and MUOW would have 2-3-fold higher risks than the MU-NW group if each of them is compared with MH-NW[45]. Thus, BCAAs/AAAs elevation would be greater in obese T2DM candidates than in NW T2DM candidates if all were compared with NW controls, but similar BCAAs/AAAs elevations across different BMI categories if obese, OW and NW T2DM candidates were compared with same BMI categories controls, respectively.

T2DM is a highly heterogeneous disease that include latent autoimmune diabetes in adults (LADA, defined by the presence of glutamic acid decarboxylase antibodies (GADA), maturity onset diabetes of the young (MODY, defined by gene mutations that disrupt insulin production) and neonatal diabetes, in addition to insulin resistant and BMI-related subgroups.

A recent data-driven cluster analysis of 14755 European T2DM patients using six variables (GADA, age at diagnosis, BMI, HbA1c, β-cell function and insulin resistance) resulted in 5 well-separated novel subgroups of adult-onset diabetes with distinct outcomes: A cluster of more severe insulin resistant individuals associated with higher risk of diabetic kidney disease; insulin deficiency cluster associated with highest risk of retinopathy; relatively young insulin deficient individuals with poor glycemic control (high HbA1c) and; a larger group of elderly patients with benign disease course[74]. That finding has been confirmed and extended by another cluster analysis of 2316 Chinese T2D patients and 685 United States T2D patients using five variables (age at diagnosis, BMI, HbA1c/glucose, β-cell function and insulin resistance) that resulted in 4 clusters: Half of the patients were elders with milder metabolic derangements; 25% of the patients had the highest BMI values but average blood glucose, β-cell function and insulin resistance; 14% of the patients had severe insulin deficiency and highest blood glucose; 8% of the patients were elders with severe insulin resistance and β-cell dysfunction[75]. Similar results of cluster analysis were reported in 55777 individuals with prediabetes[76]. However, none of these studies included BCAAs/AAAs as the study variable. Thus, the contributions of BCAAs/AAAs to different clusters or subtypes of T2DM remain unknown.

Nevertheless, BCAAs/AAAs elevation, if standardized based on different age and BMI sub-groups, could be useful in screening future prediabetes and OB-related diabetes in infants and adolescents as blood BCAAs/AAAs were found significantly correlated with BMI standard deviation score, fasting glucose, HbA1c, triglycerides, cystatin C and creatinine in 2191 healthy participants aged 3 months to 18 years[77]. A 7.5-year longitudinal study of 396 nondiabetic Finnish girls showed that serum BCAA profile in childhood (11.2 ± 0.8 years at baseline) were associated with insulin resistance during pubertal development (significant both before and after menarche) independent of adiposity, and it predicted dysregulated glycemic and triglyceride levels in adulthood[78-80]. Blood BCAA/AAA were also found significantly elevated in OW and obese prepubertal children than in healthy controls[39-41,81].

The global prevalence of prediabetes in children and adults have reached alarming levels[81], with an annualized diabetes conversion rate of 5%–10%[82]. And 500 million adults in China (50% of China’s adult population) and 98 million adult Americans (38.0% of the United States adult population) have prediabetes (https://www.cdc.gov/diabetes/data/statistics-report/index.html#anchor_23827), the world T2DM population could double by 2050. In the United States, the total cost of diagnosed diabetes in 2022 was $413 billion, including $106 billion in indirect costs (https://www.cdc.gov/diabetes/health-equity/diabetes-by-the-numbers.html). Because preventive lifestyle modification can reduce the risk of diabetes by up to 70%[83], identification of “at-risk” individuals 10-20 years prior to T2DM onset based on BCAAs/AAAs elevation would offer plenty time for lifestyle modifications.

While BCAAs/AAAs elevation alone may not predict all subtypes of T2DM, its combined use with other anthropogenic, metabolic, and genetic biomarkers such as visceral adiposity index, muscle mass index[47-50], fasting glucose, GADA (associated with LADA), genetic polymorphisms (associated with MODY), and metabolic parameters associated with insulin deficiency, diabetic kidney disease and retinopathy should be evaluated in future studies.

This study has limitations. The lack of BMI-matched healthy control individuals makes it impossible to evaluate a potential low grade hyperaminoacidemia/BCAA/AAA elevation that may exist in NW-T2DM patients. Only fasting samples, not postprandial/oral glucose tolerance test samples or BCKAs, were studied, which may have missed the dynamic changes in BCAA catabolism. The moderate sample size of this cross-sectional study could not discern the causality of the findings. The lack of lifestyle data limits exploration of the influences of diet and social, psychological, and physical activities on BCAA/AAA. Anti-diabetic medication taken by some T2DM patients may have compromised the results[84,85]. In addition, no genetic, race/ ethnicity influences can be derived from this study, although sex differences and significant correlations between BCAAs/AAAs and anthropometric parameters were demonstrated.

Future studies should overcome the above limitations of this study. To date, most of the published BCAAs/AAAs findings were based on comparisons between obese T2DM patients and nonobese individuals, whereas the BCAAs/AAAs data of the lean/NW-T2DM groups were largely unreported. This publication bias should be addressed by stratified analyses for age, BMI, adiposity, sex, and genotype as discussed above and by comparisons between metabolically healthy vs metabolically unhealthy individuals of different BMI categories.

Type-2 diabetes mellitus (T2DM) is a major cause of comorbidity and mortality in society and was responsible for more than 4.2 million annual deaths in 2019 alone. The current world population of T2DM (approximately 450 million) is expected to double to 1 billion soon after 2050. This T2DM pandemic, however, can be curbed or prevented if the population at risk of T2DM can be identified and prophylactic actions be taken long before the onset of T2DM.

Research over the past decades has indicated that elevated branched-chain amino acids (BCAAs: Isoleucine, leucine, valine) and aromatic amino acids (AAAs: Tyrosine, tryptophan, phenylalanine) show high sensitivity and specificity (both > 97%) in predicting diabetes in animals and can successfully predict T2DM nearly 20 years before T2DM onset in select human populations. However, these findings have not been widely translated into clinical utilization due to unidentified factors.

We hypothesized that body weight and sex are potential confounding factors that could affect BCAAs/AAAs as general T2DM predictors. As the first step, the aim of our study was to determine the effects of body weight and sex on BCAAs/AAAs in new-onset T2DM individuals.

Fasting blood samples were collected from 97 new-onset T2DM patients (53 male/44 female, 43.3 ± 11.2 years of age, differing in body mass index (BMI): Normal weight (NW), n = 33, BMI = 22.23 ± 1.60; overweight, n = 42, BMI = 25.9 ± 1.07; and obesity, n = 22, BMI = 31.23 ± 2.31). All T2DM cases were diagnosed within 12 mo at the First People's Hospital of Yunnan Province, Kunming, China. Serum amino acids were analyzed using ultra-performance liquid chromatography/triple stage quadrupole mass spectrometry.

Fasting serum AAAs, BCAAs, glutamate, and alanine levels were significantly greater and high-density lipoprotein levels were significantly lower in obese T2DM patients than in NW-T2DM patients, especially among male patients. Arginine, histidine, leucine, methionine, and lysine were greater in male patients than in female patients. Moreover, histidine, alanine, glutamate, lysine, valine, methionine, leucine, isoleucine, tyrosine, phenylalanine, and tryptophan were significantly correlated with abdominal adiposity, body weight and BMI, respectively.

Heterogeneously elevated amino acids, especially BCAAs/AAAs, are found in new-onset T2DM patients in differing BMI categories, which may indicate a potentially skewed prediction of T2DM development by BCAA/AAAs, i.e., more accurate and reliable prediction in obese and male individuals than in NW individuals and females. This skewness may limit the universal application of this T2DM predictor.

This study has limitations. The lack of BMI-matched healthy control individuals makes it impossible to determine whether a potential low grade hyperaminoacidemia/BCAA/AAA elevation exists in normal-weight T2DM patients. The moderate sample size and lack of lifestyle and genetic data of this cross-sectional study could not discern the causality and/or mechanisms of the heterogeneity. Further studies should include both metabolically healthy and metabolically unhealthy individuals in differing BMI categories to overcome the abovementioned limitations.