Abstract
Context and Objectives
Glucose metabolism becomes progressively impaired with older age. Fasting glucose and insulin resistance are risk factors for premature death and other adverse outcomes. We aimed to identifying plasma metabolites associated with altered glucose metabolism and insulin resistance in older community-dwelling adults.
Participants and Methods
A targeted metabolomics approach was used to identify plasma metabolites associated with impaired fasting plasma glucose, 2-hour plasma glucose on oral glucose tolerance testing, and homeostatic model assessment insulin resistance (HOMA-IR) in 472 participants who participated in the Baltimore Longitudinal Study of Aging, with a mean (SD) age of 70.7 (9.9) years.
Results
We measured 143 plasma metabolites. In ordinal logistic regression analyses, using a false discovery rate of 5% and adjusting for potential confounders, we found that alanine, glutamic acid, and proline were significantly associated with increased odds of abnormal fasting plasma glucose. Phosphatidylcholine (diacyl C34:4, alkyl-acyl C32:1, C32:2, C34:2, C34:3, and C36:3) was associated with decreased odds of abnormal fasting plasma glucose. Glutamic and acid phosphatidylcholine alkyl-acyl C34:2 were associated with increased and decreased odds of 2-hour plasma glucose, respectively. Glutamic acid was associated with increased odds of higher tertiles of HOMA-IR. Glycine; phosphatidylcholine (diacyl C32:0, alkyl-acyl C32:1, C32:2, C34:1, C34:2, C34:3, C36:2, C36:3, C40:5, C40:6, C42:3, C42:4, and C42:5); sphingomyelin C16:0, C24:1, and C26:1; and lysophosphatidylcholine C18:1 were associated with decreased odds of abnormal HOMA-IR.
Conclusions
Targeted metabolomics identified four plasma amino acids and 16 plasma lipid species, primarily containing polyunsaturated fatty acids, that were associated with abnormal glucose metabolism and insulin resistance in older adults.
The relationship of plasma metabolites with altered glucose metabolism was characterized in 472 older adults. Altered glucose metabolism was associated with 4 amino acids and 16 lipid species.
Diabetes is a major public health problem among older adults. In the United States, nearly one-third of adults older than 60 years have diabetes, of whom approximately half are undiagnosed (1). The prevalence of diabetes in older adults is more than twice that in middle-aged adults (1). The risk of altered glucose metabolism and insulin resistance increases with age (2) and is associated with lower muscle strength (3), mobility disability (4), impaired cognition (5), frailty (4), and increased mortality (6). Type 2 diabetes (T2D) is characterized by hyperglycemia due to reduced biological response to insulin signaling, especially in skeletal muscle, and later on impaired insulin secretion by pancreatic β-cells (7). Finding new lifestyle and pharmacological interventions that effectively prevent or delay the progression and consequences of diabetes is a public health priority. However, this research field is hampered by the little knowledge of the biological pathways and mechanisms leading to hyperglycemia and insulin resistance in older persons.
Epidemiological studies have implicated several metabolites, including branched-chain amino acids, aromatic amino acids, phosphatidylcholine, and lysophosphatidylcholine with an increased risk of hyperglycemia and T2D (8–10). As these studies have mainly focused on middle-aged adults, it is unclear whether the findings of these studies also apply to the older population. We used a targeted metabolomics approach to characterize the relationship of plasma metabolites with abnormal plasma fasting glucose, 2-hour glucose on oral glucose tolerance testing, and homeostatic model assessment insulin resistance (HOMA-IR) in a cohort of older community-dwelling adults.
Participants and Methods
Participants
Study participants were 472 adults, aged 50 to 97 years, who participated in the Baltimore Longitudinal Study of Aging (BLSA) and were seen between January 2006 and December 2008. The participants in this subcohort of the BLSA were selected based upon age (50 years and older) and with complete measures of fasting plasma glucose and glucose tolerance test. The BLSA is a prospective open cohort study of community-dwelling volunteers, largely from the Baltimore/Washington area. The study was established in 1958 and is described in detail elsewhere (11). BLSA participants return to the National Institute on Aging (NIA) Clinical Research Unit at Medstar Harbor Hospital in Baltimore, Maryland, at age-specific intervals for 2.5 days of medical, physiological, and psychological examinations. The BLSA has continuing approval from the Intramural Research Program of the NIA and the institutional review board of the National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina. The protocol for the current study was also approved by the institutional review board of the Johns Hopkins School of Medicine.
Laboratory methods
Blood was collected from participants who stayed overnight at the NIA Clinical Research Unit, Medstar Harbor Hospital in Baltimore, Maryland. Blood samples were drawn from the antecubital vein between 7:00 and 8:00 am after an overnight fast. Participants were not allowed to smoke, engage in physical activity, or take medications before the blood sample was collected. Blood samples were immediately stored at 4°C, centrifuged within 4 hours, and then immediately aliquoted and frozen at −80°C. The collection of EDTA plasma in the studies is consistent with guidelines for biomarker studies (12). Participants were sitting in a semi-reclining chair during glucose tolerance testing. Fasting plasma was collected at baseline, after which participants drank 75 g of glucose in a 300-mL solution (SunDex; Fisherbrand, Pittsburgh, PA), and blood samples were drawn 2 hours after oral administration. Plasma glucose concentration was measured by the glucose oxidase method (Beckman Instruments, Inc., Fullerton, CA). Plasma insulin was measured using an enzyme-linked immunosorbent assay (Mercodia, Uppsala, Sweden) with inter- and intra-assay coefficients of variation <4%. Insulin resistance was estimated by HOMA-IR, calculated using fasting glucose and insulin measurements (13). Concentrations of plasma triglycerides and total cholesterol were determined by an enzymatic method (ABA-200 ATC Biochromatic Analyzer; Abbott Laboratories, Irving, TX). The concentration of high-density lipoprotein (HDL) cholesterol was determined by a dextran sulfate–magnesium precipitation.
Plasma metabolites were measured using liquid chromatography (LC)/tandem mass spectrometry (MS) and the AbsoluteIDQ p180 kit (Biocrates Life Sciences AG, Innsbruck, Austria), a widely used metabolomics platform for clinical and epidemiological studies. Metabolites were extracted and concentrations measured using the manufacturer’s protocol for a 5500 QTrap (Sciex, Framingham, MA) mass spectrometer equipped with an electrospray ionization source, a Shimadzu CBM-20A command module and LC-20AD pump, and a Shimadzu SIL-20AC-HT autosampler (Shimadzu Corporation, Kyoto, Japan) and a CTO-10AC column oven heater, running with Analyst 1.5.2 software, as described elsewhere (14). Briefly, 10 μL of plasma was pipetted onto a 96-well kit (Biocrates Life Sciences AG). The samples were dried at room temperature for 30 minutes. Then, 50 µL of 5% PITC reagent was added and incubated for 20 minutes, and the plate was dried under nitrogen for 1 hour. Next, 300 µL of 5 mM ammonium acetate in methanol was added and incubated at room temperature on a shaker (450 rpm) for 30 minutes. The plate was centrifuged at 500 × g for 2 minutes and labeled; 50 µL of each sample was transferred to a deep 96-well LC plate, and 10 µL of each sample was transferred to the 96-well flow-injection analysis (FIA) plate. To the LC plate, 450 µL of 40% methanol (in HPLC-grade water) was added. To the FIA plate, 490 µL of FIA running solvent was added. Then, 10 µL was injected onto the Eclipse XDB C18, 3.5 μm, 3.0 × 100 mm, with a Phenomenex C18 Security Guard Cartridge, 3.0 mm ID. The mobile phase consisted of solvent A (water containing 0.2% formic acid) and solvent B (acetonitrile containing 0.2% formic acid), with the following gradient: 0 to 0.5 minutes, 0% B; 5.5 minutes, 95% B; 6.5 minutes, 95% B; 7.0 minutes, 0% B; and 9.5 minutes, 0% B. LC plate evaluation of the samples was carried out using the MetIDQ software. The FIA plate was run with a 20-µL injection directly onto the mass spectrometer at a flow of 30 µL/min with water/acetonitrile (1:1) containing 0.2% formic acid as the mobile phase, with the following flow rate program: 0 to 1.6 min, 30 µL/min; 2.4 minutes, 200 µL/min; 2.80 minutes, 200 µL/min; and 3.00 minutes, 30 µL/min. Concentrations were calculated using the Analyst/MetIDQ software and reported in µmol/L. The method measured 143 metabolites, including 25 amino acids, 11 biogenic amines, hexose, 10 sphingolipids, 7 acylcarnitines, and 89 glycerophospholipids (lyso-, diacyl-, and acyl-alkyl phosphatidylcholines). Glycerophospholipids are differentiated on the basis of ester and ether bonds in the glycerol moiety. Diacyl or “aa” indicates that fatty acids are bound with ester bonds at the sn-1 and sn-2 positions on the glycerol backbone. Acyl-alkyl or “ae” indicates that the fatty acid at the sn-1 position is bound with an ether bond. The total number of carbon atoms and double bonds in fatty acid chains is represented by “C x:y,” where x denotes the number of carbons and y denotes the number of double bonds. The MS spectra were evaluated using Analyst/MetIDQ (Biocrates Life Sciences AG) software. Human plasma samples spiked with standard metabolites were used to monitor the reproducibility of the assay. Plasma metabolites that were below the limit of quantification were excluded from the data analyses. The inter- and intra-assay coefficients of variation ranged from 5% to 15% for nearly all analytes.
Statistical methods
The analysis presented here was based on a cross-sectional design. Continuous variables were reported as mean ± SD. Normal fasting glucose, impaired fasting glucose, and diabetic fasting glucose were defined as fasting plasma glucose ≤99 mg/dL, 100 to 125 mg/dL, and >125 mg/dL, respectively (13). Normal glucose tolerance, impaired glucose tolerance, and diabetic glucose tolerance were defined as 2-hour plasma glucose of ≤139 mg/dL, 140 to 199 mg/dL, and ≥200 mg/dL, respectively (13). HOMA-IR was divided into tertiles (<1.23, 1.23 to 2.18, >2.18). Renal insufficiency was defined as estimated glomerular filtration rate of <60 mL/min/1.73 m2 using the Modification of Diet in Renal Disease equation of Levey and colleagues (15). Plasma metabolites and demographic, anthropometric, laboratory, and clinical characteristics were compared across the respective categories of fasting plasma glucose, 2-hour plasma glucose, and HOMA-IR using χ2 tests for categorical variables and Kruskal-Wallis tests for continuous variables. Ordinal multivariate logistic regression models were used to examine the relationship between plasma metabolites and fasting plasma glucose, 2-hour plasma glucose, and HOMA-IR, respectively, adjusting for age, sex, body mass index (BMI), hypertension, coronary artery disease, and chronic kidney disease. The linear relationship of the data across glucose metabolism categories in the ordinal logistic regression models was verified using the proportional odds assumption. To determine how much the results could be influenced by participants with established diabetes, additional sensitivity analyses were conducted in which participants with established diabetes were excluded from the analysis. A false discovery rate (FDR) approach was used to correct for multiple testing (16, 17). All associations reported as significant passed the FDR cutoff of 5%. The q value is the proportion of false discoveries and the minimum FDR at which a result can be called significant. An FDR-adjusted q value of 0.05 implies that 5% of significant tests will result in false positives. All analyses were conducted using SAS version 9.13 (SAS Institute, Cary, NC).
Results
The characteristics of adults by fasting plasma glucose concentrations are shown in Table 1. There was a higher proportion of women with elevated fasting plasma glucose. BMI and triglycerides were higher among adults with elevated fasting plasma glucose. Total cholesterol, low-density lipoprotein (LDL) cholesterol, and HDL cholesterol were lower among adults with elevated fasting plasma glucose. The prevalence of hypertension, diabetes, and chronic kidney disease was higher among adults with elevated fasting plasma glucose. The characteristics of adults by 2-hour plasma glucose concentrations on oral glucose tolerance testing are shown in Table 2.
Table 1.
Characteristics of Study Participants by Fasting Plasma Glucose Concentrations in the BLSA
| Characteristic |
Fasting Plasma Glucose (mg/dL)
|
P Value a | ||
|---|---|---|---|---|
| ≤99 (n = 312) | 100–125 (n = 138) | >125 (n = 22) | ||
| Age, y | 70 (15) | 72 (14) | 70 (8) | 0.26 |
| Male sex, % | 52.9 | 42.8 | 22.7 | 0.002 |
| BMI, kg/m2 | 25.9 (3.9) | 28.0 (3.9) | 28.9 (4.2) | <0.0001 |
| Current smoking, % | 1.6 | 2.2 | 4.6 | 0.36 |
| Systolic blood pressure, mm Hg | 118 (15) | 119 (14) | 120 (14) | 0.66 |
| Diastolic blood pressure, mm Hg | 66 (9) | 66 (8) | 68 (8) | 0.33 |
| Triglycerides, mg/dL | 92 (42) | 112 (53) | 113 (50) | 0.0001 |
| Total cholesterol, mg/dL | 192 (34) | 195 (42) | 157 (33) | <0.0001 |
| LDL cholesterol, mg/dL | 111 (31) | 116 (16) | 84 (33) | 0.0001 |
| HDL cholesterol, mg/dL | 62 (16) | 57 (16) | 50 (15) | <0.0001 |
| Hypertension, % | 47.1 | 62.3 | 81.8 | <0.0001 |
| Diabetes, % | 3.5 | 13.0 | 81.8 | <0.0001 |
| Coronary artery disease, % | 3.2 | 10.9 | 18.2 | <0.0001 |
| Stroke, % | 6.4 | 4.4 | 13.6 | 0.78 |
| Heart failure, % | 2.9 | 2.2 | 9.1 | 0.45 |
| Cancer, % | 6.7 | 10.1 | 4.6 | 0.52 |
| Chronic kidney disease, % | 30.8 | 40.6 | 54.6 | 0.005 |
Values are presented as mean (SD) or percentages.
χ 2 tests for categorical variables, Kruskal-Wallis tests for continuous variables.
Table 2.
Characteristics of Study Participants by Oral Glucose Tolerance Test in the BLSA
| Characteristic |
2-Hour Plasma Glucose (mg/dL)
|
P Value a | ||
|---|---|---|---|---|
| ≤139 (n = 302) | 140–199 (n = 99) | ≥200 (n = 44) | ||
| Age, y | 70 (10) | 72 (9) | 72 (9) | 0.05 |
| Male sex, % | 53.4 | 43.4 | 29.6 | 0.001 |
| BMI, kg/m2 | 26.1 (3.8) | 27.4 (4.4) | 28.5 (4.3) | 0.0001 |
| Current smoking, % | 0.7 | 5.0 | 2.3 | 0.06 |
| Systolic blood pressure, mm Hg | 117 (15) | 120 (13) | 119 (12) | 0.16 |
| Diastolic blood pressure, mm Hg | 66 (9) | 66 (9) | 65 (7) | 0.61 |
| Triglycerides, mg/dL | 95 (43) | 103 (47) | 118 (55) | 0.02 |
| Total cholesterol, mg/dL | 193 (35) | 195 (39) | 170 (41) | <0.0001 |
| LDL cholesterol, mg/dL | 113 (32) | 115 (34) | 95 (41) | 0.0001 |
| HDL cholesterol, mg/dL | 61 (16) | 59 (16) | 52 (16) | 0.0005 |
| Hypertension, % | 46.0 | 60.6 | 75.0 | <0.0001 |
| Diabetes, % | 2.7 | 2.0 | 63.6 | <0.0001 |
| Coronary artery disease, % | 3.3 | 7.1 | 22.7 | <0.0001 |
| Stroke, % | 6.0 | 7.1 | 4.6 | 0.92 |
| Heart failure, % | 2.0 | 3.0 | 4.6 | 0.27 |
| Cancer, % | 6.6 | 7.1 | 15.9 | 0.07 |
| Chronic kidney disease, % | 31.8 | 36.4 | 47.7 | 0.04 |
Values are presented as mean (SD) or percentages.
χ 2 tests for categorical variables, Kruskal-Wallis tests for continuous variables.
There was a higher proportion of women with elevated 2-hour plasma glucose. BMI and triglycerides were higher among adults with elevated 2-hour plasma glucose. Total cholesterol, LDL cholesterol, and HDL cholesterol were lower among adults with elevated 2-hour plasma glucose. The prevalence of hypertension, diabetes, coronary artery disease, and chronic kidney disease was higher among adults with elevated fasting plasma glucose. The characteristics of adults by tertiles of HOMA-IR are shown in Table 3. BMI was higher in adults with higher HOMA-IR. Plasma triglycerides were higher and total cholesterol, LDL cholesterol, and HDL cholesterol were lower among those with higher HOMA-IR. The prevalence of hypertension and diabetes was higher among adults with higher HOMA-IR. Of the 50 participants with established diabetes, the proportion who were taking sulfonylureas, thiazolidinediones, biguanide, biguanide combination treatment, insulin, and other classes of diabetes medications was 38%, 26%, 12%, 12%, 4%, and 6%, respectively.
Table 3.
Characteristics of Study Participants by Tertiles of HOMA-IR in the BLSA
| Characteristic |
HOMA-IR
|
P Value a | ||
|---|---|---|---|---|
| <1.23 (n = 155) | 1.23–2.18 (n = 155) | >2.18 (n = 155) | ||
| Age, y | 70 (11) | 71 (10) | 70 (9) | 0.57 |
| Male sex, % | 45.16 | 54.84 | 53.55 | 0.18 |
| BMI (kg/m2) | 24.5 (3.3) | 26.6 (3.4) | 29.0 (4.2) | <0.0001 |
| Current smoking, % | 1.3 | 3.2 | 1.3 | 0.36 |
| Systolic blood pressure, mm Hg | 116 (15) | 119 (16) | 121(12) | 0.005 |
| Diastolic blood pressure, mm Hg | 66 (8) | 65 (9) | 66 (8) | 0.23 |
| Triglycerides, mg/dL | 81 (31) | 95 (41) | 122 (56) | <0.0001 |
| Total cholesterol, mg/dL | 196 (35) | 186 (34) | 192 (42) | 0.04 |
| LDL cholesterol, mg/dL | 114 (32) | 106 (30) | 114 (38) | 0.08 |
| HDL cholesterol, mg/dL | 65 (17) | 60 (15) | 54 (15) | <0.0001 |
| Hypertension, % | 27.7 | 49.7 | 56.1 | <0.0001 |
| Diabetes, % | 4.5 | 7.1 | 18.7 | <0.0001 |
| Coronary artery disease, % | 3.9 | 5.2 | 9.7 | 0.08 |
| Stroke, % | 6.4 | 7.7 | 3.9 | 0.34 |
| Heart failure, % | 3.9 | 7.1 | 5.2 | 0.45 |
| Cancer, % | 5.8 | 10.3 | 7.1 | 0.31 |
| Chronic kidney disease, % | 25.2 | 24.5 | 27.1 | 0.86 |
Values are presented as mean (SD) or percentages.
χ 2 tests for categorical variables, Kruskal-Wallis tests for continuous variables.
The plasma concentrations of metabolites across the three categories of fasting plasma glucose, 2-hour plasma glucose, and HOMA-IR are shown in Supplemental Tables 1, 2, and 3, respectively.
Twenty-eight plasma metabolites were significantly associated with elevated fasting plasma glucose in the bivariate analyses adjusted by age and sex. Multivariable ordinal logistic regression models for the 28 plasma metabolites and elevated fasting plasma glucose are shown in Table 4, adjusted by age, sex, and BMI (model 1), and a final model also adjusted by hypertension, coronary artery disease, and chronic kidney disease (model 2). Alanine, glutamic acid, and proline were significantly associated with increased odds of abnormal fasting plasma glucose in the final multivariable ordinal logistic regression model. Phosphatidylcholine aa C34:4, ae C32:1, ae C32:2, ae C34:2, ae C34:3, and ae C36:3 were significantly associated with decreased odds of abnormal fasting plasma glucose in the final multivariable ordinal logistic regression model. When participants with established diabetes were excluded, phosphatidylcholine aa C34:4, ae C34:3, and ae C36:3; glutamic acid; and proline were still significantly associated with decreased odds of abnormal fasting plasma glucose in the final multivariable ordinal logistic regression model, whereas two metabolites were of marginal significance: phosphatidylcholine ae C34:2 (P = 0.07) and alanine (P = 0.056).
Table 4.
Multivariable Ordinal Logistic Regression Models for Plasma Metabolites and Other Risk Factors Associated With Abnormal Fasting Glucose in the Adults in the BLSA
| Metabolite |
Model 1
a |
Model 2
b |
||||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | q Value | OR | 95% CI | P Value | q Value | |
| Phosphatidylcholine aa C32:0 | 0.56 | 0.27–1.16 | 0.12 | 0.18 | 0.64 | 0.30–1.35 | 0.24 | 0.33 |
| Phosphatidylcholine aa C34:4 | 1.86 | 1.10–3.17 | 0.02 | 0.05 | 1.91 | 1.12–3.25 | 0.01 | 0.05 |
| Phosphatidylcholine aa C36:0 | 0.75 | 0.45–1.25 | 0.28 | 0.35 | 0.74 | 0.44–1.25 | 0.27 | 0.35 |
| Phosphatidylcholine aa C38:0 | 0.68 | 0.37–1.22 | 0.20 | 0.28 | 0.70 | 0.38–1.27 | 0.24 | 0.33 |
| Phosphatidylcholine aa C38:1 | 0.84 | 0.63–1.12 | 0.24 | 0.31 | 0.79 | 0.59–1.07 | 0.13 | 0.19 |
| Phosphatidylcholine aa C38:6 | 0.80 | 0.48–1.33 | 0.39 | 0.46 | 0.86 | 0.51–1.45 | 0.58 | 0.64 |
| Phosphatidylcholine aa C40:3 | 0.88 | 0.45–1.72 | 0.71 | 0.77 | 0.87 | 0.44–1.72 | 0.70 | 0.75 |
| Phosphatidylcholine aa C40:6 | 0.82 | 0.50–1.34 | 0.44 | 0.49 | 0.83 | 0.51–1.37 | 0.48 | 0.56 |
| Phosphatidylcholine ae C32:1 | 0.37 | 0.19–0.73 | 0.004 | 0.01 | 0.39 | 0.20–0.78 | 0.008 | 0.03 |
| Phosphatidylcholine ae C32:2 | 0.41 | 0.20–0.82 | 0.01 | 0.04 | 0.41 | 0.20–0.84 | 0.01 | 0.04 |
| Phosphatidylcholine ae C34:1 | 0.50 | 0.25–1.01 | 0.05 | 0.09 | 0.55 | 0.27–1.10 | 0.09 | 0.16 |
| Phosphatidylcholine ae C34:2 | 0.32 | 0.18–0.65 | 0.001 | 0.005 | 0.39 | 0.21–0.74 | 0.004 | 0.02 |
| Phosphatidylcholine ae C34:3 | 0.39 | 0.22–0.67 | 0.001 | 0.005 | 0.42 | 0.24–0.73 | 0.002 | 0.01 |
| Phosphatidylcholine ae C36:2 | 0.49 | 0.26–0.84 | 0.03 | 0.07 | 0.53 | 0.28–1.04 | 0.06 | 0.12 |
| Phosphatidylcholine ae C36:3 | 0.39 | 0.21–0.71 | 0.002 | 0.001 | 0.42 | 0.22–0.77 | 0.005 | 0.02 |
| Phosphatidylcholine ae C38:0 | 0.96 | 0.56–1.64 | 0.89 | 0.89 | 1.01 | 0.59–1.74 | 0.95 | 0.95 |
| Phosphatidylcholine ae C38:6 | 0.53 | 0.29–0.97 | 0.04 | 0.08 | 0.56 | 0.30–1.03 | 0.06 | 0.12 |
| Phosphatidylcholine ae C40:2 | 0.91 | 0.49–1.70 | 0.78 | 0.80 | 0.98 | 0.52–1.84 | 0.95 | 0.95 |
| Phosphatidylcholine ae C40:6 | 0.60 | 0.33–1.10 | 0.10 | 0.15 | 0.60 | 0.32–1.10 | 0.10 | 0.16 |
| Phosphatidylcholine ae C42:4 | 0.43 | 0.22–0.87 | 0.01 | 0.05 | 0.47 | 0.23–0.94 | 0.03 | 0.08 |
| Phosphatidylcholine ae C42:5 | 0.46 | 0.22–0.96 | 0.04 | 0.08 | 0.47 | 0.22–0.99 | 0.05 | 0.11 |
| Phosphatidylcholine ae C44:4 | 0.65 | 0.31–1.34 | 0.24 | 0.31 | 0.67 | 0.32–1.40 | 0.28 | 0.36 |
| Sphingomyelin C24:1 | 0.45 | 0.22–0.94 | 0.03 | 0.07 | 0.53 | 0.25–1.11 | 0.09 | 0.16 |
| Sphingomyelin C26:1 | 0.72 | 0.38–1.39 | 0.33 | 0.40 | 0.73 | 0.38–1.42 | 0.36 | 0.44 |
| Alanine | 2.82 | 1.53–5.17 | 0.001 | 0.005 | 2.72 | 1.47–5.05 | 0.001 | 0.01 |
| Glutamic acid | 1.81 | 1.313–2.50 | 0.0003 | 0.005 | 1.81 | 1.31–2.49 | 0.0003 | 0.004 |
| Proline | 3.04 | 1.593–5.80 | 0.001 | 0.005 | 2.99 | 1.55–5.78 | 0.001 | 0.01 |
| Valine | 2.40 | 0.99–5.84 | 0.05 | 0.09 | 2.67 | 1.08–6.61 | 0.03 | 0.08 |
Adjusted by age, sex, and BMI.
Adjusted by age, sex, BMI, hypertension, coronary artery disease, and chronic kidney disease.
Seventeen plasma metabolites were significantly associated with elevated 2-hour plasma glucose in the bivariate analyses adjusted by age and sex. Multivariable ordinal logistic regression models for the 17 plasma metabolites and elevated 2-hour plasma glucose are shown in Table 5, adjusted by age, sex, and BMI (model 1), and a final model also adjusted by hypertension, coronary artery disease, and chronic kidney disease (model 2). Phosphatidylcholine ae C34:2 and glutamic acid were significantly associated with 2-hour plasma glucose in the final multivariable ordinal logistic regression model. When participants with established diabetes were excluded, glutamic acid was still significantly associated with 2-hour plasma glucose in the final multivariable ordinal logistic regression model, whereas phosphatidylcholine ae C43:2 was of marginal significance (P = 0.10).
Table 5.
Multivariable Ordinal Logistic Regression Models for Plasma Metabolites and Other Risk Factors Associated With Abnormal Glucose Tolerance in the Adults in the BLSA
| Metabolite |
Model 1
a |
Model 2
b |
||||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | q Value | OR | 95% CI | P Value | q Value | |
| Phosphatidylcholine aa C36:0 | 0.71 | 0.41–1.21 | 0.21 | 0.24 | 0.72 | 0.42–1.25 | 0.25 | 0.30 |
| Phosphatidylcholine aa C38:0 | 0.63 | 0.34–1.17 | 0.14 | 0.19 | 0.69 | 0.37–1.29 | 0.25 | 0.30 |
| Phosphatidylcholine aa C38:6 | 0.76 | 0.45–1.30 | 0.32 | 0.34 | 0.83 | 0.48–1.43 | 0.51 | 0.54 |
| Phosphatidylcholine aa C40:1 | 0.35 | 0.13–0.93 | 0.03 | 0.06 | 0.38 | 0.14–1.01 | 0.05 | 0.09 |
| Phosphatidylcholine ae C32:1 | 0.49 | 0.24–0.97 | 0.04 | 0.06 | 0.56 | 0.28–1.12 | 0.10 | 0.16 |
| Phosphatidylcholine ae C32:2 | 0.38 | 0.18–0.78 | 0.009 | 0.03 | 0.42 | 0.20–0.86 | 0.01 | 0.06 |
| Phosphatidylcholine ae C34:2 | 0.35 | 0.18–0.66 | 0.001 | 0.01 | 0.40 | 0.21–0.77 | 0.007 | 0.04 |
| Phosphatidylcholine ae C34:3 | 0.44 | 0.25–0.76 | 0.004 | 0.02 | 0.49 | 0.28–0.86 | 0.01 | 0.06 |
| Phosphatidylcholine ae C36:3 | 0.46 | 0.25–0.85 | 0.01 | 0.04 | 0.51 | 0.27–0.96 | 0.03 | 0.08 |
| Phosphatidylcholine ae C38:0 | 0.87 | 0.50–1.15 | 0.62 | 0.62 | 0.94 | 0.53–1.64 | 0.83 | 0.83 |
| Phosphatidylcholine ae C38:6 | 0.55 | 0.29–1.02 | 0.06 | 0.08 | 0.60 | 0.32–1.12 | 0.11 | 0.17 |
| Phosphatidylcholine ae C44:4 | 0.61 | 0.29–1.28 | 0.19 | 0.23 | 0.67 | 0.31–1.42 | 0.29 | 0.33 |
| Sphingomyelin C24:1 | 0.48 | 0.22–1.03 | 0.06 | 0.08 | 0.60 | 0.27–1.31 | 0.20 | 0.28 |
| Lysophosphatidylcholine a C18:2 | 0.56 | 0.34–0.91 | 0.02 | 0.04 | 0.60 | 0.37–0.98 | 0.04 | 0.09 |
| Alanine | 2.04 | 1.11–3.76 | 0.02 | 0.04 | 2.01 | 1.08–3.72 | 0.02 | 0.08 |
| Glutamic acid | 1.58 | 1.15–2.17 | 0.005 | 0.02 | 1.60 | 1.17–2.02 | 0.003 | 0.03 |
| Glycine | 0.42 | 0.21–0.85 | 0.016 | 0.04 | 0.47 | 0.23–0.94 | 0.034 | 0.08 |
Adjusted by age, sex, and BMI.
Adjusted by age, sex, BMI, hypertension, coronary artery disease, and chronic kidney disease.
Twenty-five plasma metabolites were significantly associated with tertiles of HOMA-IR in the bivariate analyses adjusted by age and sex. Multivariable ordinal logistic regression models for the 25 plasma metabolites and HOMA-IR are shown in Table 6, adjusted by age, sex, and BMI (model 1), and a final model also adjusted by hypertension, coronary artery disease, and chronic kidney disease (model 2). Nineteen plasma metabolites were significantly associated with decreased odds of HOMA-IR in the final multivariable ordinal logistic regression model 2: phosphatidylcholine aa C32:0, ae C32:1, ae C32:2, ae C34:1, ae C34:2, ae C34:3, ae C36:2, ae C36:3, ae C40:5, ae C40:6, ae C42:3, ae C42:4, and ae C42:5; sphingomyelin C16:0, C24:1, and C26:1; glutamic acid; glycine; and lysophosphatidylcholine C18:1. When participants with established diabetes were excluded, phosphatidylcholine ae C32:2, ae C34:1, ae C34:2, ae C34:3, ae C40:5, ae C42:4, and ae C42:5; sphingomyelin C16:0, C24:1, and C26:1; glutamic acid; glycine; and lysophosphatidylcholine C18:1 were still significantly associated with decreased odds of HOMA-IR in in the final multivariable ordinal logistic regression model, and phosphatidylcholine aa C32:1 was of marginal significance (P = 0.07).
Table 6.
Multivariable Ordinal Logistic Regression Models for Plasma Metabolites and Other Risk Factors Associated With HOMA-IR in the Adults in the BLSA
| Metabolite |
Model 1
a |
Model 2
b |
||||||
|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | q Value | OR | 95% CI | P Value | q Value | |
| Phosphatidylcholine aa C32:0 | 0.40 | 0.20–0.77 | 0.007 | 0.01 | 0.44 | 0.22–0.86 | 0.01 | 0.02 |
| Phosphatidylcholine aa C38:4 | 1.65 | 0.87–3.12 | 0.119 | 0.12 | 1.56 | 0.82–2.96 | 0.17 | 0.17 |
| Phosphatidylcholine ae C32:1 | 0.41 | 0.22–0.74 | 0.004 | 0.01 | 0.41 | 0.22–0.76 | 0.005 | 0.01 |
| Phosphatidylcholine ae C32:2 | 0.36 | 0.20–0.67 | 0.001 | 0.007 | 0.35 | 0.19–0.65 | 0.001 | 0.01 |
| Phosphatidylcholine ae C34:1 | 0.39 | 0.21–0.72 | 0.003 | 0.01 | 0.42 | 0.22–0.79 | 0.007 | 0.01 |
| Phosphatidylcholine ae C34:2 | 0.38 | 0.22–0.66 | 0.001 | 0.007 | 0.42 | 0.24–0.74 | 0.003 | 0.01 |
| Phosphatidylcholine ae C34:3 | 0.41 | 0.25–0.67 | 0.0003 | 0.007 | 0.45 | 0.27–0.73 | 0.001 | 0.01 |
| Phosphatidylcholine ae C36:2 | 0.45 | 0.25–0.80 | 0.007 | 0.01 | 0.51 | 0.28–0.90 | 0.02 | 0.03 |
| Phosphatidylcholine ae C36:3 | 0.46 | 0.27–0.78 | 0.004 | 0.01 | 0.51 | 0.29–0.87 | 0.01 | 0.02 |
| Phosphatidylcholine ae C38:1 | 0.79 | 0.62–1.01 | 0.063 | 0.07 | 0.78 | 0.60–1.00 | 0.05 | 0.06 |
| Phosphatidylcholine ae C38:2 | 0.60 | 0.39–0.93 | 0.023 | 0.03 | 0.65 | 0.42–1.00 | 0.05 | 0.06 |
| Phosphatidylcholine ae C40:1 | 0.65 | 0.36–1.17 | 0.155 | 0.15 | 0.64 | 0.35–1.16 | 0.14 | 0.15 |
| Phosphatidylcholine ae C40:5 | 0.47 | 0.28–0.78 | 0.004 | 0.01 | 0.46 | 0.27–0.77 | 0.003 | 0.01 |
| Phosphatidylcholine ae C40:6 | 0.50 | 0.29–0.86 | 0.012 | 0.01 | 0.50 | 0.29–0.86 | 0.01 | 0.02 |
| Phosphatidylcholine ae C42:3 | 0.55 | 0.31–0.96 | 0.036 | 0.04 | 0.55 | 0.31–0.97 | 0.04 | 0.05 |
| Phosphatidylcholine ae C42:4 | 0.40 | 0.22–0.74 | 0.004 | 0.01 | 0.41 | 0.22–0.75 | 0.004 | 0.01 |
| Phosphatidylcholine ae C42:5 | 0.41 | 0.21–0.79 | 0.008 | 0.01 | 0.39 | 0.20–0.76 | 0.006 | 0.01 |
| Sphingomyelin C16:0 | 0.33 | 0.16–0.67 | 0.002 | 0.01 | 0.36 | 0.18–0.73 | 0.005 | 0.01 |
| Sphingomyelin C24:1 | 0.33 | 0.17–0.63 | 0.001 | 0.007 | 0.35 | 0.17–0.68 | 0.002 | 0.01 |
| Sphingomyelin C26:1 | 0.44 | 0.24–0.79 | 0.006 | 0.01 | 0.42 | 0.23–0.77 | 0.005 | 0.01 |
| Glutamic acid | 1.36 | 1.04–1.78 | 0.024 | 0.03 | 1.34 | 1.02–1.75 | 0.03 | 0.04 |
| Glycine | 0.41 | 0.23–0.75 | 0.003 | 0.01 | 0.42 | 0.23–0.75 | 0.004 | 0.01 |
| Lysophosphatidylcholine a C17:0 | 0.62 | 0.37–1.04 | 0.074 | 0.08 | 0.60 | 0.35–1.00 | 0.05 | 0.06 |
| Lysophosphatidylcholine a C18:1 | 0.51 | 0.30–0.85 | 0.011 | 0.01 | 0.52 | 0.31–0.88 | 0.02 | 0.03 |
| Lysophosphatidylcholine a C18:2 | 0.66 | 0.43–1.02 | 0.065 | 0.07 | 0.71 | 0.46–1.08 | 0.11 | 0.12 |
Adjusted by age, sex, and BMI.
Adjusted by age, sex, BMI, hypertension, coronary artery disease, and chronic kidney disease.
Six metabolites were associated with both 2-hour plasma glucose and HOMA-IR in the final models: glutamic acid and phosphatidylcholine (alkyl-acyl C32:1, C32:2, C34:2, C34:3, C36:3). There were two metabolites associated with all three outcomes in the final models (elevated fasting plasma glucose, 2-hour plasma glucose, and HOMA-IR): glutamic acid and phosphatidylcholine alkyl-acyl C34:2.
Discussion
The current study shows that abnormal fasting plasma glucose is associated with higher plasma alanine, glutamic acid, and proline and lower plasma phosphatidylcholine (diacyl C34:4, alkyl-acyl C32:1, C32:2, C34:2, C34:3, and C36:3). Two-hour plasma glucose was associated with higher plasma glutamic acid and lower plasma phosphatidylcholine alkyl-acyl C34:2. Insulin resistance was associated with plasma higher glutamic acid and lower plasma glycine; phosphatidylcholine (diacyl C32:0, alkyl-acyl C32:1, C32:2, C34:1, C34:2, C34:3, C36:2, C36:3, C40:5, C40:6, C42:3, C42:4, and C42:5); sphingomyelin C16:0, C24:1, and C26:1; and lysophosphatidylcholine C18:1. Metabolomics has been increasingly applied to identify new plasma biomarkers for abnormal glucose metabolism, insulin resistance, and risk of T2D mellitus (18). There is considerable heterogeneity in plasma metabolites that have been implicated with prediabetes and T2D mellitus due to differences in analytical technique, study design, and geographical location, lifestyle, diet, and age in various study populations (18). The potential role of plasma metabolites implicated in abnormal glucose metabolism and insulin resistance in the current study, consistencies, and differences with previous studies are discussed below.
Alanine plays a key role in gluconeogenesis by the liver as part of the glucose-alanine cycle (19, 20). Alanine is the primary amino acid that is released by skeletal muscle in the circulation, where it is taken up by the splanchnic bed for conversion to glucose (19, 20). The sources of alanine in skeletal muscle include branched-chain amino acids that are oxidized to produce alanine and glutamine (21). To complete the glucose-alanine cycle, glucose is taken up from plasma and can be converted to pyruvate and subsequently to alanine in a variety of tissues, including skeletal muscle (22). A previous study showed that elevated plasma alanine concentrations predicted incident T2D mellitus in a cohort of 9369 Finnish men (9).
In the current study, elevated glutamic acid was associated with abnormal fasting glucose, abnormal 2-hour glucose, and insulin resistance. A previous study of 45 specific metabolites measured using LC/tandem MS in the Framingham Heart Study and the Malmö Diet and Cancer Study showed that elevated plasma glutamic acid was associated with insulin resistance (23). In 182 nondiabetic adults, plasma glutamic acid was positively correlated with insulin resistance (24). Glutamic acid stimulates glucagon release from human pancreatic α-cells via ionotropic glutamate receptors (25). Glutamic acid also increases the transamination of pyruvate to alanine (26), which plays a role in gluconeogenesis as noted above. Elevated plasma proline was also associated with abnormal fasting glucose. Proline can be synthesized from either glutamic acid or ornithine.
The current study corroborates the association between low serum glycine and HOMA-IR, which has been previously described in a study of 73 functionally limited older adults, aged 70 to 85 years (27), and 83 healthy adults, aged 20 to 80 years, with normal glucose tolerance (28). Plasma glycine concentrations are reduced in patients with obesity (29) and T2D mellitus (30). Prospective studies show that plasma glycine is an independent predictor of incident T2D (10). Glycine participates in several different metabolic pathways. In humans, glycine is synthesized predominantly from glucose (via serine) and from dietary betaine. Glycine can also be synthesized from glyoxylate and threonine (31). There is no evidence for reduced glycine synthesis in patients with diabetes; thus, it has been hypothesized that glycine utilization is increased in diabetes, because patients with diabetes undergo excessive fatty acyl group formation. Glycine is an acceptor of acyl groups and buffers excess acyl groups by generating acyl-glycine derivatives (31).
It is notable that five of the six phosphatidylcholine species associated with fasting plasma glucose appeared to contain polyunsaturated fatty acyl groups (i.e., had two or more double bonds). The single plasma phosphatidylcholine associated with 2-hour plasma glucose was also a diene. Ten of the 17 glycerophospholipids associated with HOMA-IR appeared to contain a polyunsaturated fatty acid (i.e., contained at least two double bonds). Although there are several possible isomers for each lipid annotation, it is likely that in the phosphatidylcholines, the most dominant unsaturated fatty acyls are linoleic acid (C18:2) and oleic acid (C18:1) and saturated fatty acyl is palmitic acid (C16:0) because these three fatty acids are found in the highest concentrations in human plasma (32, 33). These findings suggest that phosphatidylcholine species with long chain length and unsaturated fatty acyls are protective against diabetes risk and insulin resistance. Lysophosphatidylcholine C18:1, which was associated with a lower risk of insulin resistance, is generated in human plasma via three pathways: (1) by phospholipase A2 on phosphatidylcholine that is membrane bound or on the polar surface of lipoproteins; (2) by the activity of endothelial lipase, including phospholipase A1, on HDLs (34); and (3) from phosphatidylcholine during the formation of cholesteryl esters. Lysophosphatidylcholine C18:1 can serve as a ligand for specific G protein–coupled signaling receptors and can also be converted to lysophosphatidic acid by autotaxin, an adipokine. Lysophosphatidic acid has specific receptors involved in growth, differentiation, and glucose metabolism (35). A previous study in the European Prospective Investigation into Cancer and Nutrition–Potsdam Study showed that alkyl-acyl phosphatidylcholines (C34:3, C40:6) and sphingomyelin C16:0 were independent predictors of incident T2D mellitus (9).
Acylcarnitines have been hypothesized to play a role in insulin resistance (36), and elevated plasma acylcarnitines have been associated with insulin sensitivity in obese adults with T2D (37). In the current study, we found no significant association between plasma acylcarnitines and fasting plasma glucose, 2-hour plasma glucose, and HOMA-IR. These findings are consistent with a study in which plasma acylcarnitine concentrations were not associated with HOMA-IR in 60 obese adults before or after a weight loss intervention (38).
The fasting glucose testing and oral glucose tolerance testing revealed that there was a considerable portion of adults in the BLSA who had undiagnosed diabetes. These results are consistent with national estimates that show that approximately one-fourth of adults in the United States have undiagnosed diabetes (39).
The strengths of this study are a well-characterized aging cohort with standardized collection of fasting plasma samples and measurements of plasma fasting glucose, oral glucose tolerance testing, and insulin measurements. The BLSA is a convenience sample of adults who tend to be healthier, with higher educational attainment and socioeconomic status than the general population. The findings from this study cannot necessarily be generalized to other study populations. It is possible that there may be residual confounding factors present after adjusting for the available covariates. A limitation of the study is that no data were collected on the dietary intake of the participants in this analysis. Although the study was focused on abnormal glucose metabolism in older adults, the cohort included a small number of participants who were younger than 65 years. The study may have been limited in power to detect associations between plasma metabolites with lower effect size with abnormal glucose metabolism. The cross-sectional design of the current study limits causal inferences about the relationship of plasma metabolites with altered glucose metabolism.
In conclusion, older adults with altered glucose metabolism have alterations in specific plasma amino acids and lipids. Future work is needed to gain insight into the relationship of these specific amino acid and lipid alterations with altered glucose metabolism over time, as well as how these metabolites could be related to aging phenotypes associated with altered glucose metabolism such as sarcopenia.
Supplementary Material
Acknowledgments
Financial Support: This study was supported by the National Institutes of Health grants R01 AG27012 (to R.D.S.), R01 EY024596 (to R.D.S.), R56 AG052973 (to R.D.S.), and P30 AG021334 (to P.Z.), Johns Hopkins University Older Americans Independence Center, and the Intramural Branch of the National Institute on Aging (to L.F.).
Author Contributions: All authors contributed to data analysis and interpretation and wrote the manuscript.
Disclosures: The authors have nothing to disclose.
Glossary
Abbreviations:
- BLSA
Baltimore Longitudinal Study of Aging
- BMI
body mass index
- FDR
false discovery rate
- FIA
flow-injection analysis
- HDL
high-density lipoprotein
- HOMA-IR
homeostatic model assessment insulin resistance
- LC
liquid chromatography
- LDL
low-density lipoprotein
- MS
mass spectrometry
- NIA
National Institute on Aging
- T2D
type 2 diabetes
References
- 1. Cowie CC, Rust KF, Ford ES, Eberhardt MS, Byrd-Holt DD, Li C, Williams DE, Gregg EW, Bainbridge KE, Saydah SH, Geiss LS. Full accounting of diabetes and pre-diabetes in the U.S. population in 1988–1994 and 2005–2006. Diabetes Care. 2008;32(2):287–294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Meigs JB, Muller DC, Nathan DM, Blake DR, Andres R; Baltimore Longitudinal Study of Aging . The natural history of progression from normal glucose tolerance to type 2 diabetes in the Baltimore Longitudinal Study of Aging. Diabetes. 2003;52(6):1475–1484. [DOI] [PubMed] [Google Scholar]
- 3. Kalyani RR, Metter EJ, Egan J, Golden SH, Ferrucci L. Hyperglycemia predicts persistently lower muscle strength with aging. Diabetes Care. 2014;38(1):82–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kalyani RR, Tian J, Xue QL, Walston J, Cappola AR, Fried LP, Brancati FL, Blaum CS. Hyperglycemia and incidence of frailty and lower extremity mobility limitations in older women. J Am Geriatr Soc. 2012;60(9):1701–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Overman MJ, Pendleton N, O’Neill TW, Bartfai G, Casanueva FF, Forti G, Rastrelli G, Giwercman A, Han TS, Huhtaniemi IT, Kula K, Lean MEJ, Punab M, Lee DM, Correa ES, Ahern T, Laurent MR, Verschueren SMP, Antonio L, Gielen E, Rutter MK, Vanderschueren D, Wu FCW, Tournoy J; EMAS Study Group . Glycemia but not the metabolic syndrome is associated with cognitive decline: findings from the European Male Ageing Study. Am J Geriatr Psychiatry. 2017;25(6):662–671. [DOI] [PubMed] [Google Scholar]
- 6. Metter EJ, Windham BG, Maggio M, Simonsick EM, Ling SM, Egan JM, Ferrucci L. Glucose and insulin measurements from the oral glucose tolerance test and mortality prediction. Diabetes Care. 2008;31(5):1026–1030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Helman A, Avrahami D, Klochendler A, Glaser B, Kaestner KH, Ben-Porath I, Dor Y. Effects of ageing and senescence on pancreatic β-cell function. Diabetes Obes Metab. 2016;18(Suppl 1):58–62. [DOI] [PubMed] [Google Scholar]
- 8. Wang TJ, Larson MG, Vasan RS, Cheng S, Rhee EP, McCabe E, Lewis GD, Fox CS, Jacques PF, Fernandez C, O’Donnell CJ, Carr SA, Mootha VK, Florez JC, Souza A, Melander O, Clish CB, Gerszten RE. Metabolite profiles and the risk of developing diabetes. Nat Med. 2011;17(4):448–453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Stancáková A, Civelek M, Saleem NK, Soininen P, Kangas AJ, Cederberg H, Paananen J, Pihlajamäki J, Bonnycastle LL, Morken MA, Boehnke M, Pajukanta P, Lusis AJ, Collins FS, Kuusisto J, Ala-Korpela M, Laakso M. Hyperglycemia and a common variant of GCKR are associated with the levels of eight amino acids in 9,369 Finnish men. Diabetes. 2012;61(7):1895–1902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Floegel A, Stefan N, Yu Z, Mühlenbruch K, Drogan D, Joost HG, Fritsche A, Häring HU, Hrabě de Angelis M, Peters A, Roden M, Prehn C, Wang-Sattler R, Illig T, Schulze MB, Adamski J, Boeing H, Pischon T. Identification of serum metabolites associated with risk of type 2 diabetes using a targeted metabolomic approach. Diabetes. 2012;62(2):639–648. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Shock NW, Greulich RC, Andres RA, Arenberg D, Casta PT, Lakatta EG, Tobin JD, et al. . Normal Human Aging: the Baltimore Longitudinal Study of Aging. Washington, DC: Government Printing Office; 1984. [Google Scholar]
- 12. Tuck MK, Chan DW, Chia D, Godwin AK, Grizzle WE, Krueger KE, Rom W, Sanda M, Sorbara L, Stass S, Wang W, Brenner DE. Standard operating procedures for serum and plasma collection: Early Detection Research Network Consensus Statement Standard Operating Procedure Integration Working Group. J Proteome Res. 2009;8(1):113–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Genuth S, Alberti KG, Bennett P, Buse J, Defronzo R, Kahn R, Kitzmiller J, Knowler WC, Lebovitz H, Lernmark A, Nathan D, Palmer J, Rizza R, Saudek C, Shaw J, Steffes M, Stern M, Tuomilehto J, Zimmet P; Expert Committee on the Diagnosis and Classification of Diabetes Mellitus . Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care. 2003;26(11):3160–3167. [DOI] [PubMed] [Google Scholar]
- 14. Schmerler D, Neugebauer S, Ludewig K, Bremer-Streck S, Brunkhorst FM, Kiehntopf M. Targeted metabolomics for discrimination of systemic inflammatory disorders in critically ill patients. J Lipid Res. 2012;53(7):1369–1375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D; Modification of Diet in Renal Disease Study Group . A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med. 1999;130(6):461–470. [DOI] [PubMed] [Google Scholar]
- 16. Storey JD. A direct approach to false discovery rates. J R Stat Soc B. 2002;64(3):479–498. [Google Scholar]
- 17. Storey JD, Taylor JE, Siegmund D. Strong control, conservative point estimation, and simultaneous conservative consistency of false discovery rates: a unified approach. J R Stat Soc B. 2004;66(1):187–205. [Google Scholar]
- 18. Liggi S, Griffin JL. Metabolomics applied to diabetes-lessons from human population studies. Int J Biochem Cell Biol. 2017;93:136–147. [DOI] [PubMed] [Google Scholar]
- 19. Felig P, Pozefsky T, Marliss E, Cahill GF Jr. Alanine: key role in gluconeogenesis. Science. 1970;167(3920):1003–1004. [DOI] [PubMed] [Google Scholar]
- 20. Felig P. The glucose-alanine cycle. Metabolism. 1973;22(2):179–207. [DOI] [PubMed] [Google Scholar]
- 21. Felig P, Wahren J, Sherwin R, Palaiologos G. Amino acid and protein metabolism in diabetes mellitus. Arch Intern Med. 1977;137(4):507–513. [PubMed] [Google Scholar]
- 22. Perriello G, Jorde R, Nurjhan N, Stumvoll M, Dailey G, Jenssen T, Bier DM, Gerich JE. Estimation of glucose-alanine-lactate-glutamine cycles in postabsorptive humans: role of skeletal muscle. Am J Physiol. 1995;269(3 Pt 1):E443–E450. [DOI] [PubMed] [Google Scholar]
- 23. Cheng S, Rhee EP, Larson MG, Lewis GD, McCabe EL, Shen D, Palma MJ, Roberts LD, Dejam A, Souza AL, Deik AA, Magnusson M, Fox CS, O’Donnell CJ, Vasan RS, Melander O, Clish CB, Gerszten RE, Wang TJ. Metabolite profiling identifies pathways associated with metabolic risk in humans. Circulation. 2012;125(18):2222–2231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Seibert R, Abbasi F, Hantash FM, Caulfield MP, Reaven G, Kim SH. Relationship between insulin resistance and amino acids in women and men. Physiol Rep. 2015;3(5):e12392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Cabrera O, Jacques-Silva MC, Speier S, Yang SN, Köhler M, Fachado A, Vieira E, Zierath JR, Kibbey R, Berman DM, Kenyon NS, Ricordi C, Caicedo A, Berggren PO. Glutamate is a positive autocrine signal for glucagon release. Cell Metab. 2008;7(6):545–554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Adeva M, González-Lucán M, Seco M, Donapetry C. Enzymes involved in l-lactate metabolism in humans. Mitochondrion. 2013;13(6):615–629. [DOI] [PubMed] [Google Scholar]
- 27. Lustgarten MS, Price LL, Phillips EM, Fielding RA. Serum glycine is associated with regional body fat and insulin resistance in functionally-limited older adults. PLoS One. 2013;8(12):e84034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Takashina C, Tsujino I, Watanabe T, Sakaue S, Ikeda D, Yamada A, Sato T, Ohira H, Otsuka Y, Oyama-Manabe N, Ito YM, Nishimura M. Associations among the plasma amino acid profile, obesity, and glucose metabolism in Japanese adults with normal glucose tolerance. Nutr Metab (Lond). 2016;13(1):5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Glynn EL, Piner LW, Huffman KM, Slentz CA, Elliot-Penry L, AbouAssi H, White PJ, Bain JR, Muehlbauer MJ, Ilkayeva OR, Stevens RD, Porter Starr KN, Bales CW, Volpi E, Brosnan MJ, Trimmer JK, Rolph TP, Newgard CB, Kraus WE. Impact of combined resistance and aerobic exercise training on branched-chain amino acid turnover, glycine metabolism and insulin sensitivity in overweight humans. Diabetologia. 2015;58(10):2324–2335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Drábková P, Šanderová J, Kovařík J, kanďár R. An assay of selected serum amino acids in patients with type 2 diabetes mellitus. Adv Clin Exp Med. 2015;24(3):447–451. [DOI] [PubMed] [Google Scholar]
- 31. Adeva-Andany M, Souto-Adeva G, Ameneiros-Rodríguez E, Fernández-Fernández C, Donapetry-García C, Domínguez-Montero A. Insulin resistance and glycine metabolism in humans. Amino Acids. 2018;50(1):11–27. [DOI] [PubMed] [Google Scholar]
- 32. Bradbury KE, Skeaff CM, Crowe FL, Green TJ, Hodson L. Serum fatty acid reference ranges: percentiles from a New Zealand national nutrition survey. Nutrients. 2011;3(1):152–163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Abdelmagid SA, Clarke SE, Nielsen DE, Badawi A, El-Sohemy A, Mutch DM, Ma DW. Comprehensive profiling of plasma fatty acid concentrations in young healthy Canadian adults. PLoS One. 2015;10(2):e0116195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Gauster M, Rechberger G, Sovic A, Hörl G, Steyrer E, Sattler W, Frank S. Endothelial lipase releases saturated and unsaturated fatty acids of high density lipoprotein phosphatidylcholine. J Lipid Res. 2005;46(7):1517–1525. [DOI] [PubMed] [Google Scholar]
- 35. Grzelczyk A, Gendaszewska-Darmach E. Novel bioactive glycerol-based lysophospholipids: new data—new insight into their function. Biochimie. 2013;95(4):667–679. [DOI] [PubMed] [Google Scholar]
- 36. Schooneman MG, Vaz FM, Houten SM, Soeters MR. Acylcarnitines: reflecting or inflicting insulin resistance? Diabetes. 2012;62(1):1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Mihalik SJ, Goodpaster BH, Kelley DE, Chace DH, Vockley J, Toledo FG, DeLany JP. Increased levels of plasma acylcarnitines in obesity and type 2 diabetes and identification of a marker of glucolipotoxicity. Obesity (Silver Spring). 2010;18(9):1695–1700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Schooneman MG, Napolitano A, Houten SM, Ambler GK, Murgatroyd PR, Miller SR, Hollak CE, Tan CY, Virtue S, Vidal-Puig A, Nunez DJ, Soeters MR. Assessment of plasma acylcarnitines before and after weight loss in obese subjects. Arch Biochem Biophys. 2016;606:73–80. [DOI] [PubMed] [Google Scholar]
- 39. Menke A, Casagrande S, Geiss L, Cowie CC. Prevalence of and trends in diabetes among adults in the United States, 1988–2012. JAMA. 2015;314(10):1021–1029. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.