Associations of serum indolepropionic acid, a gut microbiota metabolite, with type 2 diabetes and low-grade inflammation in high-risk individuals

1Well-established lifestyle, and metabolic and genetic factors are currently used for stratifying people at high risk of developing type 2 diabetes (T2D). Even though physical inactivity, overweight, and obesity are generally accepted major risk factors contributing to the T2D incidence, the quality of the diet seems also to have a role.

2We recently performed a non-targeted metabolite profiling study in pre-selected participants with impaired glucose tolerance (IGT) from the Finnish Diabetes Prevention Study (DPS) and reported that serum indolepropionic acid (IPA) was associated with a lower likelihood of developing T2D. Furthermore, we replicated this association in two other independent cohorts. In DPS, IPA was the only metabolite linked with preservation of β-cell function in those who did not develop T2D.

3IPA is a specific microbial product from dietary tryptophan absorbed from the gut into the blood stream and is also found in cerebrospinal fluid,. In animal models, IPA exerts antioxidant and anti-inflammatory effects and possibly ameliorates glucose metabolism,.

4Because of the putative link of gut microbiota and T2D, we aimed at getting a more accurate picture of the interplay between IPA, T2D, glucose metabolism, inflammation, and diet. Therefore, we established a targeted quantitative method using liquid chromatography with triple quadrupole mass spectrometric detection (HPLC-QQQ-MS/MS) to measure the precise concentrations of IPA in serum samples from the DPS study.

5The DPS was a randomized, controlled, multicenter study carried out in Finland between the years 1993 and 2001 (ClinicalTrials.gov NCT00518167). A total of 522 individuals with IGT were randomly allocated into either a lifestyle intervention or control group. After a mean 4-year intervention (active study) period, the post-intervention follow-up was carried out with annual examinations. The DPS study design and methods have been reported in detail elsewhere and are described in the Supplementary Information (SI) material.

6The present study included all the participants who had fasting serum samples available for IPA analysis from the one-year follow-up. Altogether, IPA was measured in serum of 415 participants (n = 209 lifestyle and n = 206 control groups, respectively).

7Glucose and insulin levels were determined as previously described and as surrogate index of the first/early-phase insulin secretion we used the disposition index30 (DI30) (details in SI material). High sensitive C-reactive protein (hsCRP) was measured in fasting serum at IPA sampling (1-year follow-up) and yearly during the mean 4-year intervention (active study) period using an IMMULITE® 2000 Systems Analyzer (Siemens Healthcare Diagnostics, Inc. Tarrytown, NY). Genotyping of TCF7L2 rs7903146 and rs12255372 was performed as reported.

8IPA was quantified by HPLC-QQQ-MS/MS using reversed-phase separation technique. Commercial IPA and IPA-d2 were used as a standard and internal standard, respectively. Details on materials, sample preparation, HPLC-QQQ-MS/MS system, and method validation are described in the SI material.

9The data were analyzed using IBM SPSS Statistics 23 software (IBM, Inc., Armonk, NY). After data normalization, Cox proportional hazards regression models assessed the association of IPA with the risk of incident T2D during a mean follow-up of 7 years (range 1–14 years) after IPA sampling. In addition, analysis of variance models adjusted for study group tested the associations of IPA with TCF7L2 genotypes and with insulin secretion (DI30) during the long-term follow-up. For testing correlations, we applied Pearson’s correlation test. A value of P < 0.05 was considered significant.

10After quality control (detailed in SI material), 403 samples (202 lifestyle intervention and 201 control) were included in the final analyses (Table 1). IPA concentration was not different between the study groups (P = 0.14, Table 1). When participants with incident T2D at IPA sampling (1-year examination) were excluded (n = 4, intervention and n = 13, control), the results remained the same (P = 0.12).

11During a mean follow-up of 7 years since IPA sampling, the number of diabetes cases was 95 in the control and 71 in the intervention groups. Participants who progressed from IGT to diabetes compared with those who did not had reduced levels of IPA at 1-year follow-up (169 [104–264] vs. 207 [118–304] ng/ml, respectively; P = 0.05).

12We observed that higher IPA concentrations were inversely associated with the incidence of diabetes during the mean 7-year follow-up (odds ratio [confidence interval]: 0.86 [0.73–0.99], P = 0.04). A 1 SD increase in IPA was associated with a 14% decrease in the risk of developing diabetes. However, the association lost its significance when body mass index (BMI) (P = 0.15), fasting (P = 0.07), or 2 h (P = 0.26) plasma glucose at IPA sampling were also taken into account. In a model where DPS study group, age and sex were included the association of IPA with diabetes incidence was boderline (P = 0.09).

13The IPA concentrations tended to be directly associated with insulin secretion (DI30) during the mean 7-year follow-up (β = 0.10, P = 0.06; Fig. 1a). Models adjusted for sex, age, and DPS group retrieved similar results (P = 0.07 for the effect of IPA on DI30 during the follow-up).

14Because of the strong relationship of TCF7L2 genotype with T2D and insulin secretion,, we tested whether specific related genotypes could interfere in the relationship of IPA with insulin secretion. Overall, these genotypes did not influence IPA concentrations (STable 1) or its association with DI30 (P > 0.30 for each variant at each respective model).

15Our previous results suggested a correlation between IPA and dietary fiber, which was confirmed in the current study (Fig. 1b. r = 0.24, P = 1 × 10−6). There was only a mild correlation between saturated fat intake and IPA, which was no longer significant after controlling for fiber intake (STable 2).

16We found a negative correlation of IPA and serum hsCRP levels (r = − 0.22, P = 0.0001), even after controlling for study group (P = 0.0002) or BMI (P = 0.001). One-year serum hsCRP was inversely associated with DI30 during the long-term follow-up independently of study group (β = − 0.14, P = 0.01), but not after controlling for BMI (P = 0.30).

17Serum hsCRP also correlated negatively with fiber intake (r = − 0.22, P = 6.6 × 10−5) after controlling for study group (P = 6.6 × 10−5) or BMI (P = 0.001). When controlled for fiber intake, the correlation between serum IPA and hsCRP concentrations remained significant (P = 0.003). We also observed an impact of serum IPA at 1-year study on the average of circulating levels of hsCRP during the 4 years of the study (β = − 0.19, P = 0.001), independently of the study group (Fig. 1c).

18We established the quantitative HPLC-QQQ-MS/MS method for measuring serum IPA in the Finnish DPS. We showed a relationship of IPA with the incidence of T2D using the original design of the DPS and demonstrated a trend for an association with insulin secretion during a longer follow-up time of 7 years. Moreover, we demonstrated that higher serum IPA was associated with lower low-grade inflammation and higher dietary fiber intake. In our study, the predictive value of IPA on T2D incidence weakened after controlling for confounding factors. This suggests that healthy lifestyle changes resulting in higher fiber intake that protected from T2D may modify IPA concentrations and consequently diminish any associations in the whole cohort.

19IPA levels correlated negatively with hsCRP concentrations, which has been previously linked with an increased risk of T2D in the DPS population. The gut microbiota seem to have a role in T2D. Therefore, the beneficial effect of increasing dietary fiber and concomitant weight loss on gut microbiota could be linked to the production of IPA, which by enhancing intestinal barrier integrity and lowering inflammation–, ultimately leads to improved insulin secretion as seen in our study, thereby lowering the risk of T2D. In addition, due to IPA modulation of incretin hormones this could lead to enhanced insulin secretion.

20It has been suggested that higher IPA could ameliorate inflammation and cell oxidative damage,, thereby resulting on better insulin secretion due to preservation of β-cells, and consequently lowering the risk of T2D. Accordingly, lower hsCRP concentrations at IPA sampling was associated with a better insulin secretion during the follow-up years. However, it is known that weight loss has an impact on ameliorating beta-cell function and lowering inflammation,,. Therefore, it is not surprising that in our study obesity modified the association of insulin secretion with both IPA and inflammation.

21Importantly, neither the strongest common T2D-associated variant of TCF7L2 rs7903146 nor the previously reported T2D-associated variant rs7903146 in DPS modified the association of IPA levels and insulin secretion, confirming that the effect of TCF7L2 is probably not mediated by IPA.

22Strengths of the present study include the well-characterized and homogenous study population and yearly measurements of insulin secretion estimates during a long follow-up. Moreover, we developed a method for quantification of IPA in serum using all samples available from the DPS. Our study has limitations. The surrogate for insulin secretion (DI30) was based on indexes that were not measured by either the hyperinsulinemic-euglycemic clamp or the intravenous glucose tolerance test (IVGTT). Instead, we used an IVGTT for validation.

23In conclusion, we propose that the putative beneficial effects of IPA on lowering T2D risk relate to the interplay between high dietary fiber intake and decreased inflammation, or by the direct effect of IPA on β-cell function. Overall, our study further highlights the importance of the gut microbiota as a mediator for the development of metabolic disorders like T2D.