OBJECTIVE

People with type 1 diabetes are at risk for developing micro- and macrovascular complications. Little is known about the gut microbiome in long-standing type 1 diabetes. We explored differences in the gut microbiome of participants with type 1 diabetes compared with healthy control subjects and associated the gut microbiome with diabetes-related complications.

RESEARCH DESIGN AND METHODS

Microbiome data of 238 participants with type 1 diabetes with an average disease duration of 28 ± 15 years were compared with 2,937 age-, sex-, and BMI-matched individuals. Clinical characteristics and fecal samples were collected, and metagenomic shotgun sequencing was performed. Microbial taxonomy was associated with type 1 diabetes–related characteristics and vascular complications.

RESULTS

No significant difference in the α-diversity of the gut microbiome was found between participants with type 1 diabetes and healthy control subjects. However, 43 bacterial taxa were significantly depleted in type 1 diabetes, while 37 bacterial taxa were significantly enriched. HbA1c and disease duration explained a significant part of the variation in the gut microbiome (R2 > 0.008, false discovery rate [FDR] <0.05), and HbA1c was significantly associated with the abundance of several microbial species. Additionally, both micro- and macrovascular complications explained a significant part of the variation in the gut microbiome (R2 > 0.0075, FDR < 0.05). Nephropathy was strongly associated with several microbial species. Macrovascular complications displayed similar associations with nephropathy.

CONCLUSIONS

Our data show that the gut microbiome is altered in people with (long-standing) type 1 diabetes and is associated with glycemic control and diabetes-related complications. As a result of the cross-sectional design, the causality of these relationships remains to be determined.

Chronic hyperglycemia in type 1 diabetes drives the development of various microvascular complications, including retinopathy, neuropathy, and nephropathy (1), and accelerates macrovascular complications consisting of coronary artery disease, peripheral arterial disease, and stroke (2).

Several studies have suggested that the gut microbiome plays a role in the development of immune-mediated diseases, including type 1 diabetes (35). These include many cross-sectional studies in humans (4,6) and some studies showing that microbial changes are present before the onset of an immune-mediated disease (7). The gut microbiome and immune system are tightly connected, and their interaction is homeostatic. The microorganisms in the gut are a major source of antigen variation to which the immune system should control its response (8). The myelopoiesis and the epigenome of immune cells are affected by signals from the gut microbiome. Dysbiosis in the gut microbiome is associated with systemic effects on the immune system and can contribute to the development of autoimmune diseases (8,9).

Some reports have also suggested an association between the gut microbiome and the development of vascular complications (10). Alterations in gut microbial composition result in changes in the metabolites produced in the gut that are suggested to be partly responsible for the induction of chronic inflammation and to thereby promote cardiovascular disease (CVD) (10,11).

Only a few studies have investigated the role of the gut microbiome in long-standing type 1 diabetes. One study reported differences in the gut microbiome of 53 participants with type 1 diabetes with an average disease duration of 9 years compared with healthy control subjects (12). However, studies focusing on differences in the gut microbial composition in long-standing type 1 diabetes in larger cohorts and on the potential associations with diabetes-associated complications are lacking. Here, we characterized the gut microbiome using whole-genome shotgun sequencing in a cohort of patients with long-standing type 1 diabetes with a large range of glucose control, diabetes duration, and complications. We compared these results to a large, matched, control group to identify whether the microbiome composition is associated with type 1 diabetes–related traits.

Cohorts and Metadata Collection

The study included two different cohorts: 240 participants with type 1 diabetes recruited in the Radboud University Medical Center, the Netherlands, and 2,937 age-, sex-, and BMI-matched healthy participants from the Lifelines Dutch Microbiome Project cohort.

Participants with type 1 diabetes were included from the outpatient diabetes clinic of the Radboud University Medical Center, the Netherlands. All participants were diagnosed with type 1 diabetes based on accepted clinical criteria, and data include age of onset, severity of symptoms, glucose variability, and absolute insulin dependency, sometimes in combination with documented positivity for autoantibodies (mainly anti-GAD). Exclusion criteria were age <18 years, pregnancy, and the use of antibiotics in the 4 weeks before inclusion. Tests were rescheduled in case of fever in the week before inclusion. This cohort is part of the Human Functional Genomics Project (13).

Besides general characteristics, in-depth information related to type 1 diabetes was collected, including HbA1c (mean HbA1c was used, based on 35 measurements on average), disease duration, and use of medication. In addition, we determined the presence of diabetes-associated complications, including retinopathy, neuropathy, nephropathy, myocardial infarction, stroke, and peripheral arterial disease (14,15). One participant had celiac disease and was on a long-term gluten-free diet. We did not observe any significant differences in α-diversity, β-diversity, or relative abundances of individual taxa or pathways between the participant and the remainder of the cohort, and he was therefore not excluded. As metformin is known to change the composition of the gut microbiome, we removed 2 metformin users from the cohort, resulting in 238 participants with type 1 diabetes being included in our analysis.

Lifelines is a multidisciplinary prospective population-based cohort study with a three-generation design that is examining the health and health-related behaviors of 167,729 people living in the north of the Netherlands (16). During the first Lifelines data collection visit, all participants were invited to participate in a parallel project—The Dutch Microbiome Project (DMP)—on a voluntary basis. The DMP’s goal is to evaluate the impact of different exposures and lifestyles on gut microbiota composition. A total of 10,000 participants were recruited in the DMP, and 8,208 were retained for downstream analysis after stringent quality control. DMP participants who did not self-report any serious or chronic diseases (any cancer, any form of diabetes, celiac disease, inflammatory bowel disease [Crohn disease or ulcerative colitis] and irritable bowel syndrome [diagnosed based on Rome III questionnaires]) were age-, sex-, and BMI-matched to the participants with type 1 diabetes using the MatchIt version (v)3.0.2 package for R (17). The ratio of control subjects to case subjects was selected as the maximal number of control subjects for which no significant difference in age, sex, or BMI could be identified between case subjects and control subjects using a Mann-Whitney U test, which resulted in 2,937 control subjects.

Determination of Gut Microbiome

Fecal samples from the two cohorts were collected and processed using a uniform workflow, as described previously (6). Briefly, fecal samples were collected by participants, frozen within 15 min after production, transported on dry ice, and stored at −80°C. Microbial DNA extraction was performed using QIAamp Fast DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions, and DNA library preparation and sequencing were performed at Novogene, Beijing, China. Library preparation of samples with a total DNA yield of <200 ng (as determined by Qubit 4 Fluorometer) was performed using NEBNext Ultra DNA Library Prep Kit for Illumina, while libraries for other samples were prepared using NEBNext Ultra II DNA Library Prep Kit for Illumina. Sequencing was performed using the Illumina HiSEq 2000 platform to generate ∼8 gigabase (Gb) of 150 base pair paired-end reads per sample (mean 7.9 Gb, SD 1.2 Gb).

Metagenomic sequencing data were analyzed using standardized methods we have used previously (6). In brief, quality control was performed by removing low-quality reads (Phred quality ≤30), adapters, and human DNA (reference Genome Reference Consortium Human build [GRCh] 37/hg19) using KneadData tools v0.5.1. Relative abundances of microbial taxa were determined using MetaPhlAn2 v2.7.2 (18), and the functional potential of the microbiome was profiled using the HUMAnN2 pipeline v0.11.1 (19) integrated with the UniRef90 v0.1.1 database of proteins (20), the ChocoPhlAn pangenome database, and the DIAMOND alignment tool v0.8.22 (21). In this study, we focused on bacterial and archaea taxa and pathways present in at least 1% of samples with a relative abundance >0.001%. We removed redundant pathways by discarding one pathway among highly correlated (Spearman R2 > 0.95) pathways.

Case-Control Analysis of Microbiome Signatures in Type 1 Diabetes

Microbiome α-diversity was calculated on the level of microbial species and genera using the Shannon diversity index, while β-diversity was calculated as the Bray-Curtis dissimilarity on the level of species. Principal coordinate analysis was used to project the Bray-Curtis dissimilarity matrix into a limited number of dimensions (n = 5) for exploratory data analysis, visualization of data, and clustering. Comparisons between centroids of groups (participants with type 1 diabetes vs. healthy control subjects) were performed on principal coordinates using the Mann-Whitney U test.

To quantify the proportion of microbiome variance explained by the disease and participant phenotypes, including age, sex, BMI, and medication use, we performed permutational multivariate ANOVA using distance matrices (ADONIS) analysis on the Bray-Curtis dissimilarity matrix using 10,000 permutations.

To determine microbiome signatures of type 1 diabetes, we compared microbiome diversity and the relative abundances of bacterial taxa and pathways between participants with type 1 diabetes and matched healthy control subjects by constructing Gaussian family multivariate generalized linear models, with age, sex, BMI, smoking, sequencing read depth, and proton pump inhibitor (PPI) use included as covariates to correct for the potential confounding effects of these variables. Relative abundances of microbiome taxa and pathways were normalized by centered log-ratio transformation. As none of the participants reported use of antibiotics up to 3 months before the fecal sampling, antibiotic use was not considered in the analysis. Results with a false discovery rate (FDR) of <0.05 were deemed statistically significant.

Analysis of Microbiome Signatures of Type 1 Diabetes Phenotypes

Analysis of the microbiome signatures of phenotypes related to type 1 diabetes was performed within the type 1 diabetes cohort using methods analogous to the case–control analysis described above in Case-Control Analysis of Microbiome Signatures in Type 1 Diabetes. We considered the following phenotypes in the study: HbA1c (mmol/mol), insulin sensitivity (derived from insulin need in insulin units/kg/day), duration of diabetes (years), glycemic load (defined as HbA1c ∗ disease duration), microvascular complications (including nephropathy, neuropathy, and retinopathy), and macrovascular complications (including peripheral arterial disease and myocardial infarction). Of these, HbA1c and insulin sensitivity were found to have highly skewed distributions and were log-transformed before analysis to normalize the distribution. All analysis included corrections for age, sex, BMI, read depth, smoking, and PPI use.

Statistical Tools and Software

Calculations were performed using R 3.6.0 for Windows. Generalized multivariate models were implemented using the MaAsLin2 (v2.1.0) package for R and the glm function from the R base package. ADONIS analysis and calculations of α- and β-diversity were implemented using the vegan (v2.5.6) package for R. The Benjamini-Hochberg correction was used to control for multiple testing.

Baseline Characteristics

Characteristics of both cohorts are summarized in Supplementary Table 1. People with type 1 diabetes and healthy control subjects were well matched with respect to age, sex, and BMI. Participants with type 1 diabetes had an average age of 52 ± 16 years, an average HbA1c of 8.0% (63.6 mmol/mol), and an average disease duration of 28 ± 15 years. There were 40 participants (16.7%) with a history of macrovascular complications (myocardial infraction, stroke, or peripheral arterial disease), and 161 participants (67.3%) affected by one or more microvascular complications (retinopathy, neuropathy, or nephropathy).

Patients With Type 1 Diabetes Have Increased α-Diversity of Microbial Pathways

Data analysis revealed no differences in α-diversity of microbial taxa between participants with type 1 diabetes and healthy control subjects (Fig. 1). There was a small but significant (FDR = 0.001) increase in α-diversity of microbial biochemical pathways in participants with type 1 diabetes (Fig. 1G).

Figure 1

Principal component analysis (PCoA) of microbial species showing PC1/PC2 (A) and PC2/PC3 (B) of participants with type 1 diabetes (orange) and healthy control subjects (blue). PCoA of microbial pathways of PC1/PC2 (C) and PC2/PC3 (D) of participants with type 1 diabetes (orange) and healthy control subjects (blue). Shannon diversity of species (E), genera (F), and pathways (G).

Figure 1

Principal component analysis (PCoA) of microbial species showing PC1/PC2 (A) and PC2/PC3 (B) of participants with type 1 diabetes (orange) and healthy control subjects (blue). PCoA of microbial pathways of PC1/PC2 (C) and PC2/PC3 (D) of participants with type 1 diabetes (orange) and healthy control subjects (blue). Shannon diversity of species (E), genera (F), and pathways (G).

Close modal

Gut Microbiome Composition of Patients With Type 1 Diabetes Is Significantly Different

Next, we compared the microbial species and pathways between participants with type 1 diabetes and healthy control subjects. Significant differences in various species and pathways were observed between both groups. A total of 37 bacterial species were significantly enriched in participants with type 1 diabetes, whereas 43 bacterial taxa were significantly depleted (Supplementary Table 2). Depleted species included Alistipes putredinis (FDR = 1.1 ∗ 10−11), Prevotella copri (FDR = 7.7 ∗ 10−10), and Bifidobacterium longum (FDR = 3.5 ∗ 10–5) (Fig. 2A–C). Enriched species in participants with type 1 diabetes included multiple species from families Ruminococcaceae (FDR = 2.5 ∗ 10−9), Clostridiaceae (FDR = 8.3 ∗ 10−8) (Fig. 2D), Clostridiales (FDR = 3.2 ∗ 10−5), and genus Oscillibacter (FDR = 0.01). Interestingly, enriched species consisted mainly of opportunistic pathogens (Clostridiales, Oscillibacter), whereas commensal species were significantly depleted in participants with type 1 diabetes (Supplementary Table 2). Regarding microbial pathways, 28 pathways were significantly depleted in participants with type 1 diabetes, whereas 11 were enriched (Supplementary Table 3). Depleted pathways included the pyrimidine deoxyribonucleotides de novo biosynthesis III pathway (PWY.6545) (FDR = 2.3 ∗ 10−8) and the adenosine deoxyribonucleotides de novo biosynthesis II pathway (PWY.7220) (FDR = 1.0 ∗ 10–4) (Fig. 2E and G). Enriched pathways included the l-isoleucine biosynthesis II pathway (PWY.5101) (FDR = 1.0 ∗ 10–4) and the stachyose degradation pathway (PWY.6527) (FDR = 4.2 ∗ 10−4).

Figure 2

Relative abundance of three depleted species (Alistipes putredinis [A], Prevotella copri [B], and Bifidobacterium longum [C]), one enriched taxa (Clostridiaceae [D]), and four depleted pathways (EH). For all species/pathways, the FDR < 0.005. HC, healthy cohort.

Figure 2

Relative abundance of three depleted species (Alistipes putredinis [A], Prevotella copri [B], and Bifidobacterium longum [C]), one enriched taxa (Clostridiaceae [D]), and four depleted pathways (EH). For all species/pathways, the FDR < 0.005. HC, healthy cohort.

Close modal

Diabetes-Related Characteristics Explain a Significant Amount of Gut Microbiome Variance

To define the impact of diabetes-associated characteristics on the gut microbiome composition, microbiome variance per phenotype was determined. This approach was used to determine what factors would explain the variance in the gut microbiome between participants with type 1 diabetes and healthy control subjects and within participants with type 1 diabetes. In addition to looking at the individual species level, we also looked at enriched microbial pathways. When participants with type 1 diabetes were compared with healthy control subjects, beyond having type 1 diabetes, HbA1c explained a significant part of the variance in species (explained variance R2 > 0.004, FDR = 0.0017) and pathways (explained variance R2 > 0.003, FDR = 0.0014) (Fig. 3A and B). The diabetes-associated characteristics that impacted the microbiome variance included duration of diabetes, age of onset, HbA1c levels/glycemic load, and medication use. Age explained the largest amount of variation in microbiome composition (explained variance R2 = 0.013, FDR = 0.0066) (Fig. 3C), followed by the use of PPIs or platelet aggregation inhibitors (explained variance R2 = 0.011, FDR = 0.0092 and R2 = 0.011, FDR = 0.0092, respectively) and HbA1c (explained variance R2 = 0.010, FDR = 0.0092) (Fig. 3C). Within the cohort with type 1 diabetes, none of the diabetes-related clinical characteristics impacted the microbial pathway variation significantly (Fig. 3D).

Figure 3

Variance per phenotype of participants with type 1 diabetes compared with healthy controls for both species (A) and pathways (B). Variance per phenotype within the type 1 diabetes cohort for species (C) and pathways (D). Orange bars are significant (FDR < 0.005). IU, insulin units; MED, medication; NSAIDs, nonsteroidal anti-inflammatory drugs; VitD, vitamin D.

Figure 3

Variance per phenotype of participants with type 1 diabetes compared with healthy controls for both species (A) and pathways (B). Variance per phenotype within the type 1 diabetes cohort for species (C) and pathways (D). Orange bars are significant (FDR < 0.005). IU, insulin units; MED, medication; NSAIDs, nonsteroidal anti-inflammatory drugs; VitD, vitamin D.

Close modal

Diabetes-Related Characteristics Show Significant Associations With Microbial Species and Pathways

Since we observed a significant effect of diabetes-related characteristics on the microbiome variance, we set out to determine species-specific associations. Glycemic control, as reflected by HbA1c, showed a significant association with several microbial species, with the strongest positive association found with Dorea formicigenerans (FDR = 0.028) and the strongest negative associations found with genus Fecalibacterium (P = 0.0016), which was dominated by Fecalibacterium prausnitzii (P = 0.0016). Disease duration and HbA1c showed positive associations with bacteria from family Ruminococcaceae (P < 0.05), and disease duration showed a negative association with Roseburia intestinalis (P = 0.01). HbA1c also showed significant negative and positive associations with several microbial pathways. The highest negative FDRs were found for the galacturonate degradation pathway I (FDR = 0.008) and the super pathway of l-serine and glycine biosynthesis I (FDR = 0.008). Again, disease duration and glycemic load displayed similar associations, showing significant positive associations with both the flavin biosynthesis I bacteria and plants pathway and the super pathway of pyrimidine ribonucleotides de novo biosynthesis pathway.

Microbial Species and Pathways Are Significantly Associated With Diabetes-Related Complications

The presence of micro- and macrovascular complications explained a significant amount of the variance in gut microbiome composition on species level (R2 > 0.0075, FDR < 0.05) (Fig. 3). Overall, the different microvascular complications showed similar associations, whereas different macrovascular complications displayed different associations.

Diabetic nephropathy, retinopathy, and neuropathy were strongly associated with several microbial species (Fig. 4C), with the most significant associations found for diabetic nephropathy, which was positively associated with seven Clostridium species (FDR < 0.05), genus Bacteroides (FDR = 0.042), and family Bacteroidaceae (FDR = 0.03). Retinopathy showed a positive association with 15 species, mainly Clostridium species, and a negative association with 7 species, including R. intestinalis (P = 0.015). Diabetic neuropathy was positively associated with 14 species, of which the strongest association was with Anaerostipes (P = 0.0007), but there was also an association with two Ruminococcaceae species (P < 0.05). Diabetic nephropathy was also associated with a variety of microbial pathways, whereas fewer associations were found for other complications. The strongest positive associations were found for the folate transformation II pathway (PWY.3841) (FDR = 0.001), the formyl tetrahydrofolate biosynthesis pathway (FDR = 0.0022), and the super pathway of pyridoxal 5 phosphate biosynthesis and salvage (PWY.0.845; FDR = 0.0039).

Figure 4

Associations between type 1 diabetes–related characteristics and the most significant microbial taxa (A) and pathways (B). Associations between type 1 diabetes–related complications and the most significant microbial taxa (C) and pathways (D). Blue indicates positive associations. Yellow indicates negative associations. Colored squares indicate P < 0.05. – indicates an FDR < 0.05. IU, insulin units.

Figure 4

Associations between type 1 diabetes–related characteristics and the most significant microbial taxa (A) and pathways (B). Associations between type 1 diabetes–related complications and the most significant microbial taxa (C) and pathways (D). Blue indicates positive associations. Yellow indicates negative associations. Colored squares indicate P < 0.05. – indicates an FDR < 0.05. IU, insulin units.

Close modal

Presence of peripheral arterial disease and myocardial infarction were strongly associated with the same microbial species, but peripheral arterial disease was associated with more species (Fig. 4C). Although not FDR significant, these complications were positively associated with Slackia (P < 0.05), order Clostridiales, and five Clostridium species. One negative association was found for both macrovascular peripheral arterial disease and myocardial infarction, which was with Bifidobacterium adolescentis (P < 0.05). Macrovascular complications were positively associated with seven pathways, including the pyridoxal 5 phosphate biosynthesis I pathway (Fig. 4D). Macrovascular complications were negatively associated with three pathways, including the super pathway of pyridoxal 5 phosphate biosynthesis and salvage (PWY.0.845; P = 0.025) (Fig. 4D).

The current study identified clear differences in gut microbiome composition between participants with long-standing type 1 diabetes and healthy control subjects, with a significant part of the variation seemingly related to glycemic control. Gut microbiome composition was also associated with the presence of micro- and macrovascular complications, particularly diabetic nephropathy, independent of age, sex, BMI, PPI use, and smoking. Overall, our study provides evidence for a difference in gut microbiome composition in participants with long-standing type 1 diabetes compared with healthy control subjects that associates with glycemic control and disease-related complications. Given the cross-sectional design, this study cannot determine cause-and-effect relationships.

Previous research on the role of the gut microbiome in participants with type 1 diabetes primarily focused on the pathogenesis of the disease and found reduced diversity in gut microbiome composition in children with type 1 diabetes and children at risk for developing type 1 diabetes (22). These data are suggestive for the involvement of decreased microbial diversity in the autoimmune process that drives type 1 diabetes development. Interestingly, we did not observe significant differences in microbial diversity between participants with long-standing type 1 diabetes and healthy control subjects, a finding consistent with two previous studies in participants with long-standing type 1 diabetes (12,23). These results suggest that a recovery in diversity occurs over time.

We did identify differences in gut microbial species between participants with type 1 diabetes and healthy control subjects. Previous studies using 16S sequencing revealed conflicting results on this topic, finding both enrichment and depletion of Bifidobacteria in long-standing type 1 diabetes (12,24). Alongside a smaller cohort size, these differences between previous studies could be explained by the average disease duration of 10 years and the younger age of the participants (12,24). In our study, where participants had an average diabetes duration of 28 years, we observed a reduction in Bifidobacteria. Bifidobacteria are producers of the short-chain fatty acid butyrate, which has anti-inflammatory effects through suppression of nuclear factor-κB signaling, induces T regulatory cells in the gut mucosa, and is important in tight-junction expression (2527). The reduction in Bifidobacteria may be related to the more proinflammatory status in type 1 diabetes driving complications. Alongside a reduction in Bifidobacteria, we also observed a reduction in the commensal species D. longicatena. This species is highly prevalent in healthy individuals and has therefore been associated with a healthy gut microbiome (28). Interestingly, a lower abundance of D. longicatena in individuals with type 1 diabetes was previously observed to result in an increased risk of developing heart failure (29). Besides a lower abundance of several species, we found increased levels of Clostridiales and Oscillibacter in type 1 diabetes. These species are opportunistic pathogens that have previously been linked to type 2 diabetes and obesity in humans and mice (3032). Overall, our data suggest that the differences in the gut microbiome could be caused by a decrease in species linked to a healthy gut microbiome and an increase in species associated with disease.

Using the diabetes-associated characteristics collected for our study cohort, we tried to identify the driving force behind the changes in the microbiome composition in participants with type 1 diabetes. Hyperglycemia is an important risk factor in the development of several diabetes-related complications (33), and poor glycemic control is associated with systemic low-grade inflammation (33,34). HbA1c explained a significant amount of the variation found in the gut microbiome of participants with type 1 diabetes. At the species level, we found a significant positive association between HbA1c and D. formicigenerans. This species has previously been linked to the presences of glycoside hydrolases involved in mucin degradation and can thereby affect gut permeability (35). The positive association between HbA1c and D. formicigenerans may impact gut permeability in participants with poor glycemic control, and increased gut permeability can contribute to a higher systemic inflammatory status (11).

Disease duration and poor glycemic control were also positively correlated with Ruminococcaceae, which were previously found to be correlated with clinical characteristics such as HbA1c, blood pressure, and lysophosphatidylcholine (LPC)/phosphatidylcholine (PC) metabolites in patients with type 2 diabetes (36). The association with LPC/PC metabolites was found to be strongly age dependent (36). Since we found an association with disease duration, we suggest that Ruminococcaceae, and thereby LPC/PC metabolites, increase over time and might contribute to the development of CVD.

Our results suggest a link between the level of glycemic control and gut microbiome composition but cannot provide insight into causality. We hypothesize that poor glycemic control could result in an altered gut microbiome composition that may subsequently contribute to the development of systemic inflammation and an increased risk of complications. Conversely, the altered gut microbiome composition may affect glycemic control and thereby drive the risk of developing complications. Previous research in people with prediabetes reported clear associations between the composition of the microbiome and postprandial glucose excursions in response to carbohydrate ingestion (37). Further research would be needed to determine the causal relation between gut microbiome composition and diabetes-related characteristics.

The presence of diabetes results in an increased risk of developing micro- and macrovascular complications. CVD, which is linked to chronic inflammation (38), is the leading cause of death in participants with type 1 diabetes, and recent studies suggest that alterations in the gut microbiome could contribute to chronic inflammation and CVD risk (10). This is in line with our results, also revealing an association between complications and gut microbiome composition. The strongest associations were found between microvascular complications and the gut microbiome, particularly with nephropathy, which were mainly with family Clostridioforme (C. bolteae and C. clostridioforme) and Bacteroides, known opportunistic pathogens (39,40). There were fewer associations between other microvascular complications and microbial species. Nephropathy is considered as one of the most severe complications associated with diabetes. Most participants with type 1 diabetes had several complications, so although the number of participants with nephropathy was not high, many had additional complications. This may partially explain the more pronounced association. Furthermore, previous literature has shown an important role for the gut-kidney axis in the development of kidney disease (41), which may explain the stronger associations between nephropathy and the gut microbiome in our study.

We observed that the presence of microvascular complications was associated with increased levels in Clostridia, Dialister, and Anaerotruncus and decreases in butyrate producers from genera Roseburia and Alistipes. Butyrate producers are crucial for colon health due to their role in mucin production and their anti-inflammatory properties (42). We found similar associations between Clostridioforme (C. boltee and C. clostridioforme) and Bacteroides in participants with macrovascular complications. Furthermore, we found an association between micro- and macrovascular complications and Ruminococcus. Higher levels of Ruminococcus have previously been found in diabetes-prone rats. A common metabolite of Ruminococcus is butyrate, which is important for colon health (42). In a healthy gut, besides mucin-producing species, mucin-degrading species, such as Prevotella, are also important for mucin production (43). We observed that Prevotella were reduced in participants with type 1 diabetes, which could contribute to a reduction in mucin production, thereby increasing gut permeability.

Microbial species also produce metabolites, with one interesting metabolite being trimethylamine-N-oxide (TMAO). TMAO’s precursor (trimethylamine) is mainly produced by Firmicutes, and trimethylamine is oxidized to TMAO by the hepatic enzyme flavin monooxygenase 3 (FMO3). Plasma levels of TMAO have been shown to be predictive for the development of CVD and atherosclerosis (44). P. copri, which was increased in participants with type 1 diabetes in our cohort, has been linked to increased TMAO production (45). Furthermore, several Ruminococcaceae, part of the phylum Firmicutes, show a positive association with microvascular neuropathy and macrovascular peripheral arterial disease in our data. Overall, these data support the hypothesis that alterations in the gut microbiome can increase the risk of developing micro- and macrovascular complications via microbially produced metabolites such as butyrate and TMAO.

Our study has limitations. The cross-sectional design makes it impossible to determine causality. In addition, while our cohort consisted of a reasonable number of participants, it is relatively small for this type of study. The correlation structure in our data (HbA1c is correlated with complications and complications are correlated with other complications) and the limited number of cases of individual complications limited our power to detect specific complication-microbiome associations. Yet, our cohort included carefully selected people with type 1 diabetes with a wide range of disease duration, glycemic control (HbA1c), and presence of complications, which renders our findings generalizable to the majority of the population with established type 1 diabetes.

The study also has strengths, including the detailed microbiome analyses, the large matched control population, and the large amount of clinical information obtained for the cohort with type 1 diabetes.

In conclusion, the composition of the gut microbiome in participants with type 1 diabetes differs significantly from the microbiome of healthy control subjects, while diversity is not affected. Changes in the gut microbiome are associated with glycemic control and disease-related complications. Altogether, these results suggest that the gut microbiome is not only important in the context of type 1 diabetes development but may also be involved in the development of diabetes-associated complications.

This article contains supplementary material online at https://doi.org/10.2337/figshare.20032256.

J.I.P.v.H., and R.G. share first authorship. L.A.B.J., and C.J.T. share senior authorship.

This article is featured in a podcast available at diabetesjournals.org/journals/pages/diabetes-core-update-podcasts.

Acknowledgments. The authors thank all participants for their contribution. The authors thank Anouk Janssen, Department of Internal Medicine, Radboudumc, and Martin Jaeger, Department of Internal Medicine, Radboudumc, for the sample collection and Anneke Hijmans and Kiki Schraa, Department of Internal Medicine, Radboudumc, for processing the samples. The authors would like to thank the Center for Information Technology of the University of Groningen (Rijksuniversiteit Groningen [RUG]) for their support and for providing access to the Peregrine high-performance computing cluster, the Genomics Coordination Center (University Medical Center Groningen and RUG) for their support and for providing access to Calculon and Boxy high-performance computing clusters, and the MOLGENIS (Molecular Genetics Information System) team for data management and analysis support. The authors also thank K. McIntyre, Department of Genetics, University of Groningen, for English and content editing.

Funding. Sequencing of the cohort was funded by a grant from CardioVasculair Onderzoek Nederland (CVON 2012-03) to J.F. and A.Z.; J.I.P.v.H., R.S., L.A.B.J., C.J.T., R.G., and R.K.W. are supported by the collaborative TIMID project (LSHM18057-SGF) financed by the public-private-partnership allowance made available by Top Sector Life Sciences & Health to Samenwerkende Gezondheidsfondsen (SGF) to stimulate public–private partnerships and cofinanced by health foundations that are part of the SGF. R.K.W. is supported by the Seerave Foundation and the Dutch Digestive Foundation (16–14). A.Z. is supported by European Research Council (ERC) Starting Grant 715772, Netherlands Organization for Scientific Research (NWO) VIDI grant 016.178.056, CVON grant 2018–27, and NWO Gravitation grant ExposomeNL 024.004.017. J.F. is supported by the Dutch Heart Foundation IN-CONTROL (CVON2018–27), the ERC Consolidator grant (grant agreement no. 101001678), NWO-VICI grant VI.C.202.022, and the Netherlands Organ-on-Chip Initiative, an NWO Gravitation project (024.003.001) funded by the Ministry of Education, Culture and Science of the government of the Netherlands.

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

Author Contributions. J.I.P.v.H., and R.G. wrote the first draft of the manuscript. R.G. analyzed the data. R.S. reviewed the first draft of the manuscript. R.S., R.K.W., H.J.M.H., L.A.B.J., and C.J.T. conceived and planned the study. J.F., A.Z., and H.J.M.H. contributed to the bioinformatical framework and the DArm Gezondheid (DAG) 3 cohort. All authors provided critical feedback and helped to shape the research and the manuscript. L.A.B.J. and C.J.T. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Nathan
DM
,
Genuth
S
,
Lachin
J
, et al.
Diabetes Control and Complications Trial Research Group
.
The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus
.
N Engl J Med
1993
;
329
:
977
986
2.
Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Research Group
.
Risk factors for cardiovascular disease in type 1 diabetes
.
Diabetes
2016
;
65
:
1370
1379
3.
Zhou
H
,
Sun
L
,
Zhang
S
,
Zhao
X
,
Gang
X
,
Wang
G
.
Evaluating the causal role of gut microbiota in type 1 diabetes and its possible pathogenic mechanisms
.
Front Endocrinol (Lausanne)
2020
;
11
:
125
4.
Vatanen
T
,
Franzosa
EA
,
Schwager
R
, et al
.
The human gut microbiome in early-onset type 1 diabetes from the TEDDY study
.
Nature
2018
;
562
:
589
594
5.
Han
H
,
Li
Y
,
Fang
J
, et al
.
Gut microbiota and type 1 diabetes
.
Int J Mol Sci
2018
;
19
:
E995
6.
Brand
EC
,
Klaassen
MAY
,
Gacesa
R
, et al.;
Dutch TWIN-IBD consortium and the Dutch Initiative on Crohn and Colitis
.
Healthy cotwins share gut microbiome signatures with their inflammatory bowel disease twins and unrelated patients
.
Gastroenterology
2021
;
160
:
1970
1985
7.
Fuhri Snethlage
CM
,
Nieuwdorp
M
,
van Raalte
DH
,
Rampanelli
E
,
Verchere
BC
,
Hanssen
NMJ
.
Auto-immunity and the gut microbiome in type 1 diabetes: lessons from rodent and human studies
.
Best Pract Res Clin Endocrinol Metab
2021
;
35
:
101544
8.
Brown
EM
,
Kenny
DJ
,
Xavier
RJ
.
Gut microbiota regulation of T cells during inflammation and autoimmunity
.
Annu Rev Immunol
2019
;
37
:
599
624
9.
Murri
M
,
Leiva
I
,
Gomez-Zumaquero
JM
, et al
.
Gut microbiota in children with type 1 diabetes differs from that in healthy children: a case-control study
.
BMC Med
2013
;
11
:
46
10.
Kasselman
LJ
,
Vernice
NA
,
DeLeon
J
,
Reiss
AB
.
The gut microbiome and elevated cardiovascular risk in obesity and autoimmunity
.
Atherosclerosis
2018
;
271
:
203
213
11.
Cani
PD
,
Bibiloni
R
,
Knauf
C
, et al
.
Changes in gut microbiota control metabolic endotoxemia-induced inflammation in high-fat diet-induced obesity and diabetes in mice
.
Diabetes
2008
;
57
:
1470
1481
12.
de Groot
PF
,
Belzer
C
,
Aydin
Ö
, et al
.
Distinct fecal and oral microbiota composition in human type 1 diabetes, an observational study
.
PLoS One
2017
;
12
:
e0188475
13.
Netea
MG
,
Joosten
LA
,
Li
Y
, et al
.
Understanding human immune function using the resources from the Human Functional Genomics Project
.
Nat Med
2016
;
22
:
831
833
14.
Draznin
B
,
Aroda
VR
,
Bakris
G
, et al.;
American Diabetes Association Professional Practice Committee
.
11. Chronic kidney disease and risk management: Standards of Medical Care in Diabetes—2022
.
Diabetes Care
2022
;
45
(
Suppl. 1
):
S175
S184
15.
Draznin
B
,
Aroda
VR
,
Bakris
G
, et al.;
American Diabetes Association Professional Practice Committee
.
12. Retinopathy, neuropathy, and foot care: Standards of Medical Care in Diabetes—2022
.
Diabetes Care
2022
;
45
(
Suppl. 1
):
S185
S194
16.
Scholtens
S
,
Smidt
N
,
Swertz
MA
, et al
.
Cohort profile: LifeLines, a three-generation cohort study and biobank
.
Int J Epidemiol
2015
;
44
:
1172
1180
17.
Ho
D
,
Imai
K
,
King
G
,
Stuart
EA
.
MatchIt: nonparametric preprocessing for parametric causal inference
.
J Stat Softw
2011
;
42
:
1
28
18.
Truong
DT
,
Franzosa
EA
,
Tickle
TL
, et al
.
MetaPhlAn2 for enhanced metagenomic taxonomic profiling
.
Nat Methods
2015
;
12
:
902
903
19.
Franzosa
EA
,
McIver
LJ
,
Rahnavard
G
, et al
.
Species-level functional profiling of metagenomes and metatranscriptomes
.
Nat Methods
2018
;
15
:
962
968
20.
Suzek
BE
,
Wang
Y
,
Huang
H
,
McGarvey
PB
;
UniProt Consortium
.
UniRef clusters: a comprehensive and scalable alternative for improving sequence similarity searches
.
Bioinformatics
2015
;
31
:
926
932
21.
Buchfink
B
,
Xie
C
,
Huson
DH
.
Fast and sensitive protein alignment using DIAMOND
.
Nat Methods
2015
;
12
:
59
60
22.
Kostic
AD
,
Gevers
D
,
Siljander
H
, et al.;
DIABIMMUNE Study Group
.
The dynamics of the human infant gut microbiome in development and in progression toward type 1 diabetes
.
Cell Host Microbe
2015
;
17
:
260
273
23.
Stewart
CJ
,
Nelson
A
,
Campbell
MD
, et al
.
Gut microbiota of Type 1 diabetes patients with good glycaemic control and high physical fitness is similar to people without diabetes: an observational study
.
Diabet Med
2017
;
34
:
127
134
24.
Higuchi
BS
,
Rodrigues
N
,
Gonzaga
MI
, et al
.
Intestinal dysbiosis in autoimmune diabetes is correlated with poor glycemic control and increased interleukin-6: a pilot study
.
Front Immunol
2018
;
9
:
1689
25.
Arpaia
N
,
Campbell
C
,
Fan
X
, et al
.
Metabolites produced by commensal bacteria promote peripheral regulatory T-cell generation
.
Nature
2013
;
504
:
451
455
26.
Furusawa
Y
,
Obata
Y
,
Fukuda
S
, et al
.
Commensal microbe-derived butyrate induces the differentiation of colonic regulatory T cells
.
Nature
2013
;
504
:
446
450
27.
Canani
RB
,
Costanzo
MD
,
Leone
L
,
Pedata
M
,
Meli
R
,
Calignano
A
.
Potential beneficial effects of butyrate in intestinal and extraintestinal diseases
.
World J Gastroenterol
2011
;
17
:
1519
1528
28.
Tap
J
,
Mondot
S
,
Levenez
F
, et al
.
Towards the human intestinal microbiota phylogenetic core
.
Environ Microbiol
2009
;
11
:
2574
2584
29.
Kamo
T
,
Akazawa
H
,
Suda
W
, et al
.
Dysbiosis and compositional alterations with aging in the gut microbiota of patients with heart failure
.
PLoS One
2017
;
12
:
e0174099
30.
Blandino
G
,
Inturri
R
,
Lazzara
F
,
Di Rosa
M
,
Malaguarnera
L
.
Impact of gut microbiota on diabetes mellitus
.
Diabetes Metab
2016
;
42
:
303
315
31.
Jung
MJ
,
Lee
J
,
Shin
NR
, et al
.
Chronic repression of mTOR complex 2 induces changes in the gut microbiota of diet-induced obese mice
.
Sci Rep
2016
;
6
:
30887
32.
Lam
YY
,
Ha
CW
,
Campbell
CR
, et al
.
Increased gut permeability and microbiota change associate with mesenteric fat inflammation and metabolic dysfunction in diet-induced obese mice
.
PLoS One
2012
;
7
:
e34233
33.
Flynn
MC
,
Kraakman
MJ
,
Tikellis
C
, et al
.
Transient intermittent hyperglycemia accelerates atherosclerosis by promoting myelopoiesis
.
Circ Res
2020
;
127
:
877
892
34.
Tsalamandris
S
,
Antonopoulos
AS
,
Oikonomou
E
, et al
.
The role of inflammation in diabetes: current concepts and future perspectives
.
Eur Cardiol
2019
;
14
:
50
59
35.
Vacca
M
,
Celano
G
,
Calabrese
FM
,
Portincasa
P
,
Gobbetti
M
,
De Angelis
M
.
The controversial role of human gut Lachnospiraceae
.
Microorganisms
2020
;
8
:
E573
36.
Zhao
L
,
Lou
H
,
Peng
Y
,
Chen
S
,
Zhang
Y
,
Li
X
.
Comprehensive relationships between gut microbiome and faecal metabolome in individuals with type 2 diabetes and its complications
.
Endocrine
2019
;
66
:
526
537
37.
Zeevi
D
,
Korem
T
,
Zmora
N
, et al
.
Personalized nutrition by prediction of glycemic responses
.
Cell
2015
;
163
:
1079
1094
38.
de Ferranti
SD
,
de Boer
IH
,
Fonseca
V
, et al
.
Type 1 diabetes mellitus and cardiovascular disease: a scientific statement from the American Heart Association and American Diabetes Association
.
Circulation
2014
;
130
:
1110
1130
39.
Dehoux
P
,
Marvaud
JC
,
Abouelleil
A
,
Earl
AM
,
Lambert
T
,
Dauga
C
.
Comparative genomics of Clostridium bolteae and Clostridium clostridioforme reveals species-specific genomic properties and numerous putative antibiotic resistance determinants
.
BMC Genomics
2016
;
17
:
819
40.
Zafar
H
,
Saier
MH
Jr
.
Gut Bacteroides species in health and disease
.
Gut Microbes
2021
;
13
:
1
20
41.
Chen
YY
,
Chen
DQ
,
Chen
L
, et al
.
Microbiome-metabolome reveals the contribution of gut-kidney axis on kidney disease
.
J Transl Med
2019
;
17
:
5
42.
Roesch
LF
,
Lorca
GL
,
Casella
G
, et al
.
Culture-independent identification of gut bacteria correlated with the onset of diabetes in a rat model
.
ISME J
2009
;
3
:
536
548
43.
Tailford
LE
,
Crost
EH
,
Kavanaugh
D
,
Juge
N
.
Mucin glycan foraging in the human gut microbiome
.
Front Genet
2015
;
6
:
81
44.
Wilson
A
,
McLean
C
,
Kim
RB
.
Trimethylamine-N-oxide: a link between the gut microbiome, bile acid metabolism, and atherosclerosis
.
Curr Opin Lipidol
2016
;
27
:
148
154
45.
Brusca
SB
,
Abramson
SB
,
Scher
JU
.
Microbiome and mucosal inflammation as extra-articular triggers for rheumatoid arthritis and autoimmunity
.
Curr Opin Rheumatol
2014
;
26
:
101
107
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.