Metabolite Signature of Albuminuria Involves Amino Acid Pathways in 8661 Finnish Men Without Diabetes

Abstract

Objective

To investigate the metabolite signature of albuminuria in individuals without diabetes or chronic kidney disease to identify possible mechanisms that result in increased albuminuria and elevated risk of type 2 diabetes (T2D).

Research Design and Methods

The study cohort was a population-based Metabolic Syndrome In Men (METSIM) study including 8861 middle-aged and elderly Finnish men without diabetes or chronic kidney disease at baseline. A total of 5504 men participated in a 7.5-year follow-up study, and 5181 of them had metabolomics data measured by Metabolon’s ultrahigh performance liquid chromatography-tandem mass spectroscopy.

Results

We found 32 metabolites significantly (P < 5.8 × 10^-5) and positively associated with the urinary albumin excretion (UAE) rate. These metabolites were especially downstream metabolites in the amino acid metabolism pathways (threonine, phenylalanine, leucine, arginine). In our 7.5-year follow-up study, UAE was significantly associated with a 19% increase (hazard ratio 1.19; 95% confidence interval, 1.13–1.25) in the risk of T2D after the adjustment for confounding factors. Conversion to diabetes was more strongly associated with a decrease in insulin secretion than a decrease in insulin sensitivity.

Conclusions

Metabolic signature of UAE included multiple metabolites, especially from the amino acid metabolism pathways known to be associated with low-grade inflammation, and accumulation of reactive oxygen species that play an important role in the pathogenesis of UAE. These metabolites were primarily associated with an increase in UAE and were secondarily associated with a decrease in insulin secretion and insulin sensitivity, resulting in an increased risk of incident T2D.

Increased urinary albumin excretion (UAE) is a marker of nephropathy and reflects endothelial dysfunction and kidney damage. Multiple pathways are affected in albuminuria, but the mechanisms leading to an increase in UAE are incompletely understood (1). Urinary albumin excretion is an early marker for an increased risk of cardiovascular disease, kidney disease, total mortality, and metabolic syndrome (2–5). Urinary albumin excretion is already increased in prediabetes and in patients with newly diagnosed type 2 diabetes (T2D) (6).

Urinary albumin excretion is influenced by both genetic and environmental factors (7). Over 60 independent associations of genetic variants with the urine albumin creatine ratio, a proxy marker of UAE, have been reported. These variants are related to multiple cardiovascular traits (8). Among these traits elevated total triglyceride levels, low-density lipoprotein cholesterol (LDL-C) levels, waist-to-hip ratio adjusted for body mass index (BMI), and hypertension were causally associated with the albumin creatine ratio in a recent Mendelian randomization study (9).

High-throughput metabolomics studies covering hundreds to thousands of metabolites have revolutionized biomarker research (10). Previous studies have characterized the metabolite signature of albuminuria and proteinuria in people with chronic kidney disease or T2D (11–18) but there are no large-scale, population-based studies investigating the metabolite signature of albuminuria in individuals without diabetes. Finding metabolic markers of albuminuria early in people at high risk of T2D is of great interest because they could predict the development of renal complications.

Individuals with prediabetes, having elevated fasting glucose, impaired glucose tolerance, or both, have an increased prevalence of albuminuria compared with control subjects (19). Several studies have shown that hyperinsulinemia and insulin resistance are associated with microalbuminuria in individuals without diabetes (20, 21) and that albuminuria is also associated with increased risk of T2D, independent of blood pressure level (22, 23). However, no evidence of a causal relationship between fasting insulin or fasting glucose with the urine albumin creatine ratio has been found in a Mendelian randomization study (9). Additionally, microvascular (endothelial) dysfunction has been proposed as a link between UAE and incident T2D (24).The aim of our study is to explore the metabolite signature of albuminuria in individuals without diabetes or chronic kidney disease to identify possible mechanisms that result in increased albuminuria and elevated risk of T2D.

Study population

The Metabolic Syndrome In Men (METSIM) study includes 10 197 Finnish men, aged 45–73 years at baseline, randomly selected from the population register of the town of Kuopio, Eastern Finland (25). We applied identical protocols and similar clinical and laboratory measurements at the cross-sectional and follow-up studies of the METSIM cohort. We measured height to the nearest 0.5 cm and weight using a calibrated digital scale (Seca 877; Seca, Hamburg, Germany) to the nearest 0.1 kg. We calculated BMI as the weight in kilograms divided by the height in meters squared. We evaluated glucose tolerance with a 2-hour glucose tolerance test (75 g glucose), and measured glucose and insulin levels at 0, 30, and 120 minutes. Glucose tolerance status was based on the American Diabetes Association criteria at both baseline and follow-up studies (26). Normal glucose tolerance (N = 3012) was defined by fasting glucose <5.6 mmol/l and 2-hour glucose (<7.8 mmol/l). Prediabetes was defined as an increase in fasting glucose (5.6–6.9 mmol/l, n = 4299) or 2-hour glucose (7.8–11.0 mmol/l, N = 310), or both (n = 1041). Altogether, 1412 participants had T2D at the baseline study, and they were excluded from all statistical analyses. A total of 8662 men without diabetes and who had UAE measurement at baseline (age 57 ± 7 years, BMI 26.8 ± 3.8 kg/m², mean ± standard deviation) were included in the current study. Their fasting plasma glucose was 5.7 ± 0.5 mmol/l and 2-hour glucose 6.0 ± 1.7 mmol/l. A total of 5504 men who had follow-up data including UAE participated in the follow-up study (mean follow-up time 7.5 ± 2.9 years). We have previously shown that clinical and laboratory characteristics of this subset of men did not differ from those of the entire METSIM population without diabetes and, therefore, we consider that this subsample represents the entire METSIM cohort (27). During the follow-up study, 1020 participants developed incident T2D. The Ethics Committee of the University of Kuopio and Kuopio University Hospital approved the study. All study participants gave written informed consent.

Laboratory measurements

We measured plasma glucose using an enzymatic hexokinase photometric assay (Konelab Systems Reagents; Thermo Fisher Scientific, Vantaa, Finland), and insulin by immunoassay (ADVIA Centaur Insulin IRI no. 02230141; Siemens Medical Solutions Diagnostics, Tarrytown, New York). Albumin was determined by Immunoturbidimetric method (Konelab Albumin/Microalbuminuria system reagents, REF no 981660, Thermo Electron Corp, Vantaa, Finland) from the first urine sample in the morning (µg/minute). Total triglycerides were measured using enzymatic colorimetric tests (Konelab Systems Reagents; Thermo Fisher Scientific, Vantaa, Finland), and LDL-C by enzymatic colorimetric tests (Konelab Systems Reagents; Thermo Fisher Scientific, Vantaa, Finland). Serum concentrations of high-sensitivity C-reactive protein (hs-CRP) were assayed using kinetic immunoturbidimetry (near infrared particle immunoassay, IMMAGE Immunochemistry System; Beckman Coulter, Fullerton, California) and plasma interleukin 1 receptor antagonist using a photometric immunoassay (enzyme-linked immunosorbent assay [ELISA]) method (Quantikine DRA00 Human IL-1RA; R&D Systems, Inc., Minneapolis, Minnesota).

We measured metabolites as part of Metabolon’s untargeted Discovery HD4 platform using ultrahigh-performance liquid chromatography–tandem mass spectroscopy, as previously described in detail (28). Metabolites were measured from 5181 participants in 3 batches: batch 1 included 999 samples with 717 metabolites, batch 2 included 1231 samples with 778 metabolites, and batch 3 included 3000 samples with 843 metabolites. All metabolites having >50% missing values, including mainly those with data available exclusively from a single batch, were omitted from statistical analyses.

Calculations

We calculated the Matsuda insulin sensitivity index (Matsuda ISI), as previously published (29), and insulin secretion index (InsAUC_0–30/GluAUC_0–30) as follows: (insulin at 0 minutes + insulin at 30 minutes)/(glucose at 0 minutes + glucose at 30 minutes). We have previously validated the Matsuda ISI as the best index for the insulin sensitivity index, as compared with the M value of the euglycemic hyperinsulinemic clamp, and InsAUC_0–30/GluAUC_0–30 as the best marker of insulin secretion, as compared with insulin secretion during a frequently sampled intravenous glucose tolerance test (30). We calculated the disposition index (DI), a measure of insulin secretion adjusted for prevailing insulin sensitivity, as Matsuda ISI × (InsAUC_0–30/GluAUC_0–30) (30).

Statistical analysis

We conducted statistical analyses using IBMSPSS Statistics, version 25. We log-transformed all continuous traits except for age and follow-up time to correct for their skewed distribution. We applied Cox regression to associate the levels of metabolites with incident T2D and presented the results as a hazard ratio (HR) and 95% confidence interval (CI). We tested Cox proportionality assumption for the metabolites using survival and survminer packages in R and found that a fitted Cox regression model adequately described the data. P < 5.8 x 10^–5 (corrected for 857 metabolites, Bonferroni correction) was considered statistically significant. We examined the association of the metabolites with Matsuda ISI, DI, and glucose levels, with linear regression adjusted for batch effect and follow-up time and present the results as standardized regression coefficients (β and standard error [SE]).

Results

UAE and its association with clinical and laboratory parameters

From 8662 participants of the METSIM study without diabetes at baseline, 7831 had UAE <20 µg/minute (7.4 ± 3.8 µg/minute, mean ± standard deviation), 761 had UAE 20–200 µg/minute (47.4 ± 37.1 µg/minute), and 70 had UAE >200 µg/minute (611.6 ± 551.5 µg/minute). Urinary albumin excretion was associated significantly with obesity parameters of BMI, waist circumference, waist-to-hip ratio, and fat mass (Table 1). We also found significant associations of UAE with systolic and diastolic blood pressure, total triglycerides, and inflammatory markers hs-CRP and IL-1RA, but only a weak nonsignificant association with age and LDL cholesterol.

Table 1.

Associations of urinary albumin excretion with obesity markers, blood pressure, and laboratory measurements in the METSIM study

Variable	Beta	SE	P-value
Age, years	0.024	0.217	2.75E-02
BMI, kg/m2	0.102	0.002	2.18E-21
Waist, cm	0.102	0.001	1.39E-21
Waist-to-hip ratio	0.089	0.001	1.00E-16
Fat mass, %	0.054	0.004	4.11E-07
Systolic BP, mmHg	0.169	0.002	2.91E-56
Diastolic BP, mmHg	0.138	0.001	6.70E-38
Total triglycerides, mmol/l	0.080	0.006	1.05E-13
LDL cholesterol, mmol/l	-0.024	0.004	2.39E-02
hs-CRP, ml/l	0.061	0.014	1.49E-08
IL-1RA, pg/ml	0.057	0.007	9.49E-08

Abbreviations: BMI, body mass index; BP, blood pressure; LDL, low-density lipoprotein; hs-CRP, high-sensitivity C-reactive protein; IL-1RA, interleukin 1 receptor antagonist; SE, standard error.

Metabolic signature of UAE

Table 2 shows metabolites are significantly associated with the increase in UAE. Almost all metabolites that were significantly and positively associated with UAE were also significantly and inversely associated with the Matsuda ISI and the DI. β-coefficients were substantially larger for the Matsuda ISI and the DI than for UAE. In general, metabolites were more strongly associated with the Matsuda ISI than with DI. Formiminoglutamate, argininate, 2-oxoarginine, kynurenate, 1-stearoyl-2-oleyl-GPE (18:0/18:1), N-acetyltryptophan, palmitoyl-linoleoyl-glycerol (16:0/18:2) [1], N-acetylkynurenine, and 3-(4-hydroxyphenyl)lactate had the strongest associations (β < -0.250) with the Matsuda ISI, and stearidonate (18:4n3) with DI (β < -0.230). As summarized in Fig. 1, many metabolites significantly associated with an increase in UAE were downstream metabolites from the amino acid pathways (tryptophan, phenylalanine, and tyrosine pathways).

Table 2.

Metabolites significantly associated with urine albumin excretion, Matsuda ISI, and Disposition Index at baseline of the METSIM study

Metabolite	Subclass	Direct parent	UAE		Matsuda ISI		DI
			Beta	P	Beta	P	Beta	P
AMINO ACIDS
Tryptophan pathway
Xanthurenate	–	–	0.089	1.6E-10	-0.248	4.1E-73	-0.106	3.2E-14
Indolelactate	–	–	0.078	2.5E-08	-0.159	1.6E-30	-0.024	8.6E-02
Kynurenate	–	–	0.072	2.2E-07	-0.283	1.2E-95	-0.139	1.4E-23
N-acetyltryptophan	–	–	0.061	1.1E-05	-0.284	1.3E-96	-0.107	1.6E-14
N-acetylkynurenine [2]	–	–	0.057	4.7E-05	-0.274	2.1E-89	-0.099	1.3E-12
Histidine pathway
Formiminoglutamate	–	–	0.080	1.5E-05	-0.365	1.9E-93	-0.166	1.3E-19
1-methylhistidine	–	–	0.065	2.8E-06	-0.103	1.3E-13	-0.026	6.3E-02
Leucine pathway
3-hydroxy-3-methylglutarate	–	–	0.078	2.5E-08	-0.118	2.0E-17	-0.092	4.2E-11
Phenylalanine pathway
N-acetylphenylalanine	–	–	0.077	3.7E-08	-0.226	1.1E-60	-0.101	3.1E-13
Phenylpyruvate	–	–	0.069	7.4E-07	-0.180	1.0E-38	-0.057	3.8E-05
Phenyllactate	–	–	0.062	8.7E-06	-0.164	2.4E-32	-0.058	3.3E-05
Tyrosine pathway
3-(4-hydroxyphenyl)lactate	–	–	0.056	5.6E-05	-0.317	8.4E-121	-0.141	2.1E-24
Arginine pathway
Argininate	–	–	0.075	4.8E-05	-0.295	2.9E-60	-0.077	2.8E-05
Methionine pathway
N-formylmethionine	–	–	0.061	1.2E-05	-0.073	1.3E-07	-0.019	1.8E-01
NUCLEOSIDES
Orotidine	–	Pyrimidine nucleosides	0.077	1.4E-06	-0.147	1.4E-20	-0.047	3.5E-03
N1-methylinosine	–	Purine nucleosides	0.073	1.8E-07	-0.184	2.0E-40	-0.077	3.0E-08
N6-carbamoylthreonyladenosine	–	Purine nucleosides	0.060	1.7E-05	-0.138	2.2E-26	-0.061	1.1E-05
N2,N2-dimethylguanosine	–	Purine nucleosides	0.060	1.7E-05	-0.202	1.4E-48	-0.090	8.3E-11
PHENYL SULFATES
4-ethylphenylsulfate	Aryl sulfates	Phenyl sulfates	0.075	6.8E-08	-0.042	2.6E-03	0.052	1.7E-04
DICARBOXILIC ACIDS
Maleate	–	–	0.074	9.5E-08	-0.138	2.9E-23	-0.076	5.3E-08
KETO ACIDS
2-oxoarginine	–	Short-chain keto acids	0.074	5.6E-05	-0.460	4.5E-154	-0.131	7.6E-13
LIPIDS
Glycerolipids
Palmitoyl-linoleoyl-glycerol (16:0/18:2) [1]	DAG	1,2-DAG	0.058	3.1E-05	-0.256	7.0E-78	-0.061	1.2E-05
Glycerophospholipids
1-stearoyl-2-oleoyl-GPE (18:0/18:1)	GPE	PE	0.069	8.3E-07	-0.309	9.3E-115	-0.093	2.5E-11
Omega fatty acids
Stearidonate (18:4n3)	PUFA	Long-chain PUFA	0.072	2.8E-07	-0.100	5.3E-13	-0.232	6.6E-64
Linolenate [alpha or gamma; (18:3n3 or 6)]	PUFA	Long chain PUFA	0.060	1.6E-05	-0.047	6.8E-04	-0.169	2.9E-34
Cyclohexanols
Myo-inositol	Alcohols and polyols	Cyclohexanols	0.068	1.3E-06	-0.021	1.4E-01	-0.017	2.10E-01
Acyl-carnitines
Tiglylcarnitine (C5:1-DC)	Fatty acid esters	Acyl carnitine	0.064	4.45-06	-1.0E-4	9.9E-01	-0.056	6.6E-05
Hydroxy acids
Malate	–	Beta-hydroxy acids	0.063	6.4E-06	-0.067	1.4E-06	-0.088	3.0E-10
Vitamins and co-factors
Pantothenate	Alcohols and polyols	Secondary alcohols	0.063	6.8E-06	-0.171	3.4E-35	-0.111	1.3E-15
NON-METAL OXOANIONIC COMPOUNDS
Sulfate	Nonmetal sulfates	Non metal sulfates	0.068	9.7E-07	-0.086	4.7E-10	-0.012	3.7E-01
ORGANIC SULFURIC ACIDS
3-methyl catechol sulfate [1]	Aryl sulfates	Phenyl sulfate	0.058	3.5E-05	-0.031	2.6E-02	0.058	2.8E-05

Beta, SE, and P-values are based on linear regression. Statistically significant P-value <5.8E-05.

Abbreviations: DAG, diacylgycerol; DI, disposition index; GPE, glycerophosphoethanolamine; Matsuda ISI, Matsuda index; METSIM, Metabolic Syndrome In Men; PE, phosphatidylethanolamine; PUFA, polyunsaturated fatty acids; SE, standard error; UAE, urinary albumin excretion.

Figure 1.

Metabolites associated with urinary albumin excretion in the tryptophan, phenylalanine, and tyrosine pathways. Blue color in boxes indicates statistical significant association with albuminuria. Abbreviation: TCA, tricarboxylic acid.

Association of UAE with the risk of T2D in the follow-up study

Given the fact that almost all metabolites associated with UAE were also significantly and inversely associated with the Matsuda ISI and the DI, we wanted to investigate the association of UAE with the risk of T2D. A total of 1020 out of 8661 participants having all data available developed incident T2D during a 7.5-year follow-up study. Urinary albumin excretion was significantly associated with the risk of T2D (Table 3). Unadjusted HR and its 95% CI were 1.19 (1.13–1.25), P < 1.1E-10; after adjustment for age, BMI, smoking, and physical activity, P < 8.0E-05; and after being additionally adjusted for insulin sensitivity, P = 0.0005, and insulin secretion, P = 0.0111. Urinary albumin excretion was significantly associated with the risk of T2D both in participants with normoglycemia and with prediabetes at baseline.

Table 3.

Association of urinary albumin excretion with incident type 2 diabetes

UAE at Baseline	HR (95% CI)	P	P a	P b	P c
All individuals (n = 8661)	1.19 (1.13–1.25)	1.1E-10	8.0E-05	0.0005	0.011
Normal glucose tolerance (n = 2965)	1.25 (1.06–1.48)	0.008	0.017	0.032	0.027
Prediabetes (n = 5649)	1.14 (1.08–1.21)	6.0E-06	0.013	0.014	0.070

Hazard’s ratio (95% CI) and P-value based on Cox-regression analyses. Abbreviations: CI, confidence intervals; HR, hazards ratio; UAE, urinary albumin excretion.

P ^a adjusted for age, BMI, smoking, and physical activity at baseline.

P ^b was additionally adjusted for the Matsuda ISI.

P ^c was adjusted for the DI.

Associations of UAE with glucose levels, insulin sensitivity, and insulin secretion

Urinary albumin excretion was associated with an increase in both fasting (β = 0.047, P = 5.1E-04) and 2-hour glucose (β = 0.060, P = 7.0E-06) during the follow-up study (Table 4). After adjustment for confounding factors, the associations weakened and were P = 0.033 for fasting and P = 0.003 for 2-hour glucose. Urinary albumin excretion was inversely associated with the Matsuda ISI (β = -0.077, P = 8.7E-09) and remained significant after the adjustment for age, BMI, smoking, and physical activity (P = 0.002). Urinary albumin excretion was inversely associated with DI at follow-up (β-0.040, P = 0.003), but after the adjustment for confounding factors it lost its statistical significance.

Table 4.

Association of urinary albumin excretion at baseline with glucose levels, Matsuda ISI, and disposition index at follow-up (N = 5491)

Variable	β	SE	P	P a
Fasting glucose	0.047	0.002	5.1E-04	0.033
2-hour glucose	0.060	0.005	7.0E-06	0.003
Matsuda ISI	-0.077	0.012	8.7E-09	0.002
Disposition index	-0.040	0.008	0.003	0.142

β, SE, and P-values were based on unadjusted linear regression analyses. P^a adjusted for age, BMI, smoking, and physical activity. Abbreviations: Matsuda ISI, Matsuda index; SE, standard error.

Next we analyzed the changes of the Matsuda ISI, the DI, and glucose levels in the quintiles of UAE at baseline and follow-up visits. We found that in individuals with normal glucose tolerance the Matsuda ISI decreased significantly from the lowest UAE quintile to the highest quintile, but no significant decrease was observed in the DI or glucose levels (31). In individuals with prediabetes (ie, impaired fasting glucose or impaired glucose tolerance or both) there was a significant reduction in the Matsuda ISI and the DI, and an increase in fasting and 2-hour glucose from the lowest UAE quintile to the highest UAE quintile (31). Quite similar results were obtained at the follow-up visit (31).

Figure 2 shows the percentage changes in the Matsuda ISI, the DI, and glucose levels at baseline from the lowest to the highest quintile of UAE. In participants with the normal glucose tolerance, the Matsuda ISI decreased by 12.5%, whereas the DI was decreased by only 1.8%, and fasting and 2-hour glucose levels increased by 0.6% and 1.5%, respectively. In individuals with prediabetes, the change in the Matsuda ISI was of the same magnitude as for individuals with normal glucose tolerance (-11.6%), but a decrease in the DI (-6.8%) and an increase in 2-hour glucose (6.0%) were substantially larger than in individuals with normal glucose tolerance. Fasting glucose level was increased only by 1.0%. At the follow-up visit in the normal glucose tolerance category the Matsuda ISI was decreased by 8.0%, the DI by 6.6%, and fasting glucose and 2-hour glucose was increased by 0.4% and 4.3%, respectively. In individuals with prediabetes, the change in the Matsuda ISI was -18.6% and in the DI was -7.8%, and an increase in fasting glucose was 1.0% and in 2-hour glucose was 8.8%.

Figure 2.

Percentage changes in Matsuda ISI, DI, and glucose levels (the highest UAE quintile vs the lowest quintile) at baseline (A) and follow-up (B) in participants with normal glucose tolerance and with prediabetes of the METSIM study. Abbreviations: 2hPG, 2 hour plasma glucose; DI, disposition index; FPG, fasting plasma glucose; Matsuda ISI, Matsuda index.

Discussion

The METSIM study is the first large, randomly selected population-based cohort to investigate the metabolite signature of UAE. We identified several novel metabolites associated with UAE, and these metabolites also had significant associations with impaired insulin secretion and insulin sensitivity as pathophysiological mechanisms leading to T2D.

The underlying etiology and pathophysiology of albuminuria are complex and diverse, and poorly understood. Different regulators and signaling pathways are involved, including endothelial growth factor, caveolin-1, CRP, nuclear factor-κB, and mitochondrial reactive oxygen species (32). The association of albuminuria with the risk of T2D has been repeatedly reported (4, 22) but causal mechanisms for this relationship remain unclear.

Risk factors for albuminuria and T2D overlap. A Mendelian randomization study has reported causal relationships between obesity markers (BMI, waist-to-hip ratio), blood pressure, total triglycerides, LDL-C, and the urine albumin creatine ratio, but not with fasting insulin or fasting glucose (9). We also found significant associations of UAE with obesity parameters, blood pressure, and total triglycerides, but the association with LDL-C was not statistically significant. Similarly, a previous Mendelian randomization study has shown that BMI, waist-to-hip ratio, and LDL-C level were causally related to the risk of T2D, whereas total triglyceride levels and blood pressure were not (33). Given the fact that a Mendelian randomization study did not confirm that insulin level, a proxy marker for insulin resistance, is causally related to the urine albumin creatine ratio in a previous study (9), insulin resistance is likely to be secondary to other mechanisms leading to albuminuria.

We analyzed the associations of UAE with metabolites to identify possible pathways and biomarkers for albuminuria. Several metabolites were significantly associated with UAE, and their β values were comparable to clinical (obesity, blood pressure) and laboratory (total triglycerides, inflammatory markers) traits in the METSIM cohort. Many of them were downstream metabolites of the amino acid pathways, especially of those of an essential amino acid tryptophan.

Kidneys play an important role in tryptophan metabolism, and chronic kidney disease upregulates the tryptophan pathway in which tryptophan 2,3-dioxygenase and indoleamine 2,3-dioxygenase 1 overexpression result in kynurenine biosynthesis (34), which stimulates inflammation, leukocyte activation, and cytokine production (35). Tryptophan is metabolized predominantly by the kynurenine pathway, and elevations of many tryptophan pathway metabolites have been reported in patients with chronic kidney disease (11, 12), in patients with T2D, and albuminuria (13). Kynurenine and its downstream metabolites, xanthurenate, kynurenate (14), and tryptophan downstream metabolite indolelactate (15), have been associated with proteinuria. Similarly, our study indicated that several metabolites of the tryptophan pathway were significantly associated with UAE, including xanthurenate, indolelactate, kynurenate, N-acetyltryptophan, and N-acytylkynurenine. These metabolites were also associated with decreased insulin secretion and insulin sensitivity even more significantly than with UAE, except for indolelactate, which was not associated significantly with insulin secretion. Kynurenines are involved in inflammation, immune response, and excitatory neurotransmission (36), and several metabolites of this pathway are diabetogenic in humans (37). Tryptophan metabolites inhibit both proinsulin synthesis and glucose- and leucine-induced insulin release from rat pancreatic islets (38). Accordingly, the metabolites of the kynurenine pathway were strongly associated with decreases in insulin sensitivity and insulin secretion.

We found that N-acetylphenylalanine, a downstream metabolite in the phenylalanine pathway, was associated with UAE, as previously reported (39). Other metabolites of the phenylalanine pathway, gamma-glutamylphenylalanine, phenyllactate, phenylpyruvate, and 3-(4-hydoxyphenyl)lactate), a metabolite of the tyrosine pathway, were significantly associated with increased UAE. In a deficiency of phenylalanine hydroxylase phenylalanine is converted into phenylpyruvate instead of tyrosine. Phenylketones are associated with albuminuria and proteinuria (40). The novel finding in our study is that these phenylalanine metabolites were also associated with decreased insulin secretion and insulin sensitivity.

Metabolites from the leucine (3-hydroxy-3-methylglutarate), glutamate (formiminoglutamate), and arginine (argininate, 2-oxoarginine) pathways were also strongly associated with UAE and with decreases in insulin secretion and insulin sensitivity in our study. As far as we know, these findings are novel. Stearidonate (18:4n3), maleate, pantothenate, linolenate, and palmitoyl-linoleoyl-glycerol (16:0/18:2) were also significantly associated with UAE, and decreases in insulin secretion and insulin sensitivity. Sulfates (sulfuric acid) are products of metabolism of sulfur containing amino acids, including methionine, cysteine, and taurine, and their intracellular accumulation trigger production of inflammatory cytokines and reactive oxygen species production (41). Interestingly, the administration of phenyl sulfate, a gut microbiota-derived metabolite, induces albuminuria and podocyte damage in experimental models of diabetes (42).

Urinary albumin excretion was associated with an increased risk of T2D by 19% in METSIM participants without diabetes at baseline. This increase was statistically significant after the adjustment for age, BMI, smoking, and physical activity. Further adjustment for the DI weakened the P-value more than the adjustment for the Matsuda ISI, suggesting that impaired insulin secretion was more important than insulin resistance for the conversion to T2D. Our finding that impaired insulin secretion is more important than insulin resistance as a mechanism leading to T2D in individuals having high UAE is a novel finding. This agrees with a previous Mendelian randomization study showing that insulin secretion has a causal role in the etiology of T2D (43) and that conversion to diabetes does not happen without a defect in insulin secretion (44). Urinary albumin excretion was significantly associated with decreases both in the the Matsuda ISI and the DI at baseline and did not change substantially at the follow-up. Increases in fasting glucose levels were relatively small in both groups, and 2-hour glucose increased only in individuals with prediabetes.

Our study showed that obesity, high blood pressure, waist-to-hip ratio, waist circumference, total triglyceride, and inflammatory markers were associated with UAE. A previous Mendelian randomization study has demonstrated that from these risk factors, obesity, high blood pressure, and triglyceride were causally related to the urine albumin creatine ratio, whereas high insulin level (insulin resistance) failed to demonstrate causality (9). Given the fact that this Mendelian randomization study did not support high insulin levels as a causal factor for the development of albuminuria, adverse changes in metabolites in our study are likely to be secondary to increased albuminuria. Several downstream metabolites, especially from the tryptophan and phenylalanine pathways, result in low-grade inflammation and the accumulation of reactive oxygen species that play an important role in the pathogenesis of UAE. We demonstrated that these metabolites were associated not only with UAE but even more strongly with insulin sensitivity and insulin secretion, and thus on the risk of T2D. Our study suggests that several adverse metabolic changes resulting in an increase of UAE may secondarily induce insulin resistance and impaired insulin secretion. We have previously shown that 9 amino acids, phenylalanine, tryptophan, tyrosine, alanine, isoleucine, leucine, valine, aspartate, and glutamate were significantly associated with decreases in insulin secretion and the elevation of fasting or 2-hour glucose levels (27), demonstrating that amino acid pathways play an important role in increasing the risk of T2D.

The strength of our study is that the METSIM study is a large randomly selected population cohort, and that we measured albuminuria by UAE and not by the urine albumin creatinine ratio, which is a less accurate method of assessing albuminuria (45). Additionally, we used a very conservative threshold for statistical significance in our analyses of metabolites to increase credibility of our conclusions. Additionally, we applied validated measures of insulin secretion and insulin resistance (30). The limitations of the study are that only middle-aged and elderly men were included, and that our study included only Finns; therefore, our findings need to be replicated in women and other populations. Finally, our study is an association study that does not allow to make causal conclusions from our results.

In summary, our study suggests that multiple metabolites, especially from the amino acid metabolism pathways, contribute to the metabolite signature of UAE, and that these metabolites are associated with decreases both in insulin secretion and insulin sensitivity, resulting in an increased risk of T2D.

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability

Restrictions apply to the availability of data generated or analyzed during this study to preserve the confidentiality of the participants. The corresponding author will, on request, detail the restrictions and any conditions under which access to some data may be provided.