Nine Amino Acids Are Associated With Decreased Insulin Secretion and Elevated Glucose Levels in a 7.4-Year Follow-up Study of 5,181 Finnish Men

Type 2 diabetes is often preceded by a long period of prediabetes, characterized by insulin resistance and impaired insulin secretion. Conversion to diabetes happens when insulin secretion from the pancreas is no longer able to compensate for insulin resistance in peripheral insulin-sensitive tissues (1).

Several metabolites, including especially branched-chain amino acids (BCAAs) and aromatic amino acids, have been reported to be associated with the risk of type 2 diabetes in previous studies (2–5). However, most of these studies have been cross-sectional, and none of these studies has investigated the association of amino acids (AAs) with changes in insulin secretion in a longitudinal setting.

We investigated the associations of twenty AAs with insulin secretion, insulin resistance, and glycemia in a large Finnish prospective population-based METabolic Syndrome In Men (METSIM) cohort.

The METSIM study comprises 10,197 Finnish men, aged from 45 to 73 years, randomly selected from the population register of Kuopio town, Eastern Finland. The study design and laboratory methods have previously been described (6,7). Glucose tolerance was evaluated with a 2-h glucose tolerance test (OGTT) (75 g glucose) including three time points (glucose and insulin levels measured at 0, 30, and 120 min) according to the American Diabetes Association criteria (8). Of 5,181 men without diabetes at entry included in the present analysis, 4,851 participated in the follow-up study (mean follow-up time 7.4 ± 2.9 years). This subset of 5,181 men had clinical and laboratory characteristics (Supplementary Table 1) similar to those of the entire METSIM population without diabetes, and therefore this subsample can be considered to present the entire METSIM cohort (7). A total of 522 participants developed incident type 2 diabetes. The study was approved by the Ethics Committee of the University of Kuopio and Kuopio University Hospital. All study participants gave written informed consent.

Glucose and insulin areas under the curve (AUCs) in an OGTT were calculated by the trapezoidal method. The Matsuda insulin sensitivity index (Matsuda ISI) was calculated as previously published (9). Insulin secretion index (InsAUC_0–30/GluAUC_0–30) was calculated based on an OGTT as follows: (insulin at 0 min + insulin at 30 min)/ (glucose at 0 min + glucose at 30 min). The selection of Matsuda ISI (among 6 insulin sensitivity indices compared with the M value from euglycemic clamp) and InsAUC_0–30/GluAUC_0–30 (among 11 insulin secretion indices compared with insulin secretion during a frequently sampled intravenous glucose tolerance test) was based on our previous validation study (7). Disposition index (DI), a measure of insulin secretion adjusted for prevailing insulin sensitivity, was calculated as Matsuda ISI × (InsAUC_0–30/GluAUC_0–30) (7).

Metabolites were measured as part of Metabolon's untargeted Discovery HD4 platform using ultra-high-performance liquid chromatography–tandem mass spectroscopy. All methods used a Waters ACQUITY UPLC and a Thermo Scientific Q Exactive high-resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. Raw data were extracted and quality control processed using Metabolon’s hardware and software. Peaks were quantified using AUC. For studies spanning multiple days, a data normalization step was performed to correct variation resulting from instrument interday tuning differences. Compounds were identified by comparison with library entries of purified standards or recurrent unknown entities. The metabolite levels were rescaled to have a median equal to 1.

The metabolomics analyses were performed in three batches. Batch 1 included 999 samples with 717 metabolites identified, batch 2 included 1,231 samples with 778 metabolites identified, and batch 3 included 3,000 samples with 843 metabolites identified. Twenty AAs from a total of 857 metabolites were included in current statistical analysis. The percentage of missing values for AAs was from 0 to 0.2%.

We conducted statistical analyses using IBM SPSS Statistics, version 25. We log-transformed all continuous traits with the exception of age and follow-up time to correct for their skewed distribution. Intercorrelations between the AAs were calculated by Pearson correlations (Supplementary Fig. 1). We examined the association of AAs with Matsuda ISI, DI, and glucose levels with linear regression and presented the results as standardized regression coefficients (β and SE). We applied Cox regression to associate the levels of AAs with incident type 2 diabetes. P < 5.8 × 10⁻⁵ (corrected for 857 metabolites) was considered statistically significant and P < 0.05 nominally significant.

In our cross-sectional study all AAs, except for arginine, histidine, and threonine, were significantly (P < 5.8 × 10⁻⁵) associated with insulin sensitivity (Matsuda ISI) (Table 1). Glycine, serine, glutamine, and asparagine were associated with improved insulin sensitivity, whereas all other AAs were associated with decreased insulin sensitivity. Glutamate, tyrosine, isoleucine, and alanine had the largest effect on the reduction of insulin sensitivity. In a prospective study, the effect sizes of AAs on Matsuda ISI were substantially smaller than those in the baseline study. Only 6 of 20 AAs showed significant changes in Matsuda ISI during the follow-up. Glutamate, tyrosine, leucine, aspartate, and phenylalanine were associated with a decrease in Matsuda ISI, and glycine was associated with an increase in Matsuda ISI.

In a cross-sectional study, 14 AAs were significantly associated with a decrease in insulin secretion (DI); the most significant reductions in the DI were observed for glutamate, tyrosine, alanine, and isoleucine (Table 2). Glycine, glutamine, serine, and asparagine were associated with a significant increase in the DI. Nine of the 14 AAs that were significantly associated with the DI in the cross-sectional study remained significant also in the prospective study. The largest effects on the DI were observed for leucine, isoleucine, tyrosine, and glutamate.

In the cross-sectional study, the effect sizes (β) of the AAs were significantly larger for insulin resistance (except for glycine and serine, which increased insulin sensitivity) than for reduction in insulin secretion (Supplementary Table 1 and Supplementary Fig. 2). By contrast, in the prospective study effect sizes for the reduction of insulin secretion were numerically larger compared with effect sizes for insulin resistance and showed a statistically significant difference for tryptophan, alanine, isoleucine, leucine, and valine among the nine AAs, which were associated with reduced insulin secretion during the follow-up.

All AAs that were significantly associated with reduced insulin secretion (phenylalanine, tryptophan, tyrosine, alanine, leucine, isoleucine, valine, aspartate, and glutamate) significantly increased fasting or 2-h glucose levels. All these AAs were also associated with higher risk of incident type 2 diabetes (N = 522), whereas glycine and glutamine were associated with a lower risk of type 2 diabetes (Table 3). In the follow-up study, tryptophan, alanine, isoleucine, and valine were associated with reduced insulin secretion and increased fasting and 2-h glucose levels but not with impairment in insulin sensitivity.

We investigated the association of all twenty AAs with insulin secretion, insulin sensitivity, and fasting and 2-h glucose levels in a large randomly selected population-based METSIM cohort. Our 7.4-year follow-up study reports novel findings: 1) several AAs were associated with both impaired insulin secretion and impaired insulin sensitivity and 2) only the AAs that were associated with impaired insulin secretion prospectively were associated with an increase in glucose levels.

The novel finding in our METSIM 7.4-year follow-up study was that nine AAs (phenylalanine, tryptophan, tyrosine, alanine, isoleucine, leucine, valine, aspartate, and glutamate) were significantly associated with reduced insulin secretion, an important contributor in the conversion to diabetes. Among these nine AAs, five (phenylalanine, tyrosine, alanine, aspartate, and glutamate) were significantly associated with decreases in both insulin secretion and insulin sensitivity and five with incident type 2 diabetes (tyrosine, alanine, isoleucine, aspartate, and glutamate).

In our study, 17 of 20 AAs were associated with insulin resistance in cross-sectional analyses, in agreement with the results of earlier studies, but only 6 of them were associated with insulin resistance in our follow-up study. Our results agree with previous findings showing significant associations of branched-chain AAs (BCAAs) isoleucine, valine, and leucine with insulin resistance (10–13). However, in our study, BCAAs were also associated with reduced insulin secretion, suggesting that elevated levels of BCAAs may over time result in a decrease in insulin secretion. Impairment in BCAA catabolism has been suggested to result in the accumulation of potentially toxic intermediates that contribute to β-cell mitochondrial dysfunction and eventually to the apoptosis of β-cells (14).

We showed that phenylalanine and tyrosine were significantly associated with decreased insulin secretion and elevated levels of fasting and 2-h glucose in our follow-up study. We also confirmed the results of previous studies, which have shown that phenylalanine and tyrosine are significantly associated with insulin resistance and diabetes (15,16).

Tryptophan was associated with impaired insulin secretion in our prospective study. Tryptophan is an essential AA metabolized predominantly (∼95%) by the kynurenine pathway. Kynurenines are involved in inflammation, immune response, and excitatory neurotransmission (17). Previous studies have suggested that several metabolites of the kynurenine pathway are diabetogenic in humans (18). Tryptophan metabolites inhibit both proinsulin synthesis and glucose- and leucine-induced insulin release from rat pancreatic islets (19).

Glutamate and aspartate were significantly associated with decreased insulin secretion and increased insulin resistance in our follow-up study. A previous study has reported increased levels of glutamate in insulin resistance (20), but previous studies have not investigated the association of glutamate with insulin secretion in a prospective setting. In transgenic mice with β-cell–specific glutamate dehydrogenase deletion, glucose-stimulated insulin secretion was reduced by 37%, demonstrating the essential role of glutamate in the regulation of insulin secretion (21). The association of aspartate with increased risk of type 2 diabetes is a novel finding.

Alanine, one of the most abundant AAs in the circulation (22), was associated in our follow-up study with impaired insulin secretion and type 2 diabetes but not with insulin resistance. Glucagon is an important regulator of AA metabolism, and hyperglucagonemia is associated with increased levels of AAs, including alanine, aspartate, and glutamate (23). Unfortunately, we did not measure glucagon levels in our study.

Glycine was significantly associated with increases in insulin sensitivity and insulin secretion as previously published (24). Glycine stimulates insulin secretion by acting on glycine receptors and N-methyl-d-aspartate receptors in β-cells (24). Glycine also decreased fasting glucose nominally and 2-h glucose significantly, as previously reported (25), in agreement with its beneficial effects on insulin secretion and insulin sensitivity.

We summarize our results of effects of nine AAs (isoleucine, leucine, valine, phenylalanine, tyrosine, tryptophan, alanine, aspartate, and glutamate) on insulin sensitivity and insulin secretion in Fig. 1. The early effect of AAs on glucose metabolism seems to be an increase in insulin resistance given that the effect sizes of AAs (negative β) on insulin resistance were considerably larger than those on insulin secretion in a cross-sectional analysis. However, in the follow-up study the effects sizes of AAs (negative β) were consistently larger on reduction of insulin secretion than on insulin sensitivity. This suggests that AAs may have adverse effects on insulin secretion and consecutively on the risk of hyperglycemia and type 2 diabetes. Our study is, however, a prospective population-based study and cannot prove causality.

The strengths of this study are a large and homogeneous study population and the validation of the methods used as surrogate markers for insulin sensitivity and secretion. We used a very conservative threshold for statistical significance to increase credibility of our conclusions. The limitation of the study is that only middle-aged and elderly Finnish men were included in the study, and therefore we do not know whether the results are valid for women, all age-groups, and other ethnic and racial groups. We are not aware of any unselected population sample where insulin and glucose measurements at 0, 30, and 120 min in an OGTT and metabolomics including all 20 AAs had been performed. Therefore, we could not replicate our findings in other populations.

In conclusion, we demonstrate that nine AAs were significantly associated with reduced insulin secretion and elevated glucose levels in a prospective population-based study of 5,181 Finnish men. Further studies are, however, needed to investigate the role of insulin secretion in diabetogenic effects of AAs.

Funding. The research leading to these results has received support from the Innovative Medicines Initiative Joint Undertaking under European Medical Information Framework grant agreement no. 115372 (to M.L. and U.S.). The METSIM study was supported by grants from Academy of Finland (321428), Sigrid Juselius Foundation, Finnish Foundation for Cardiovascular Research, Kuopio University Hospital, and Centre of Excellence of Cardiovascular and Metabolic Diseases, supported by the Academy of Finland (to M.L.).

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

Author Contributions. J.V. conceived the study, performed metabolomics and genetic data analyses, and wrote and revised the manuscript. A.S. performed metabolomics data analyses and revised the manuscript. U.S. and J.K. contributed to the discussion and revised the manuscript. M.L. conceived the study, wrote and reviewed the manuscript, and supervised the entire study. M.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.