Association between blood microbiome and type 2 diabetes mellitus: A nested case‐control study

Background Although recent studies have indicated that gut microbiome dysbiosis was significantly associated with the onset of type 2 diabetes mellitus (T2DM), information on the role of blood microbiome in T2DM development is scarce. Methods Fifty incident T2DM cases and 100 matched non‐T2DM controls were selected from a prospective cohort study of “135.” The composition of the blood microbiome was characterized using bacterial 16S ribosomal RNA (16S rRNA) gene sequencing from pre‐diagnostic blood sample. The amplicons were normalized, pooled, and sequenced on the Illumina MiSeq instrument using a MiSeq Reagent Kit PE300 v3 kit. Results Totally, 3 000 391 and 6 244 227 high‐quality sequences were obtained from T2DM patients and non‐T2DM controls, respectively. The mean diversity of the blood microbiome (Simpson, Chao1 and Shannon indices) was not different between two groups at baseline. At genus level, the Aquabacterium, Xanthomonas, and Pseudonocardia were presented with lower abundance, while Actinotalea, Alishewanella, Sediminibacterium, and Pseudoclavibacter were presented with higher abundance among T2DM cases compared to those in non‐T2DM controls. As the results shown, participants carried the genus Bacteroides in blood were significantly associated with a decreased risk for T2DM development, with 74% vs 88% (adjusted OR: 0.367, 95% CI: 0.151‐0.894). However, participants carried the genus Sediminibacterium have an increased risk for T2DM, with adjusted OR (95% CI) being 14.098 (1.358, 146.330). Conclusions Blood microbiome may play an etiology role in the development of T2DM. These findings would be useful to develop microbiota‐based strategies for T2DM prevention and control.

T2DM cases were defined as incident cases (for those with no previous history of diabetes and had never used diabetic medication at the time of the baseline survey), and a diagnosis of T2DM was confirmed by a physician, or the reported use of T2DM medication at the time of the follow-up survey.
In the present nested case-control study, 100 controls were matched to 50 T2DM cases with sex and age (5 years). Pregnant women and individuals with cancers, chronic viral hepatitis, renal failure, chronic enteritis, and diarrhea were excluded. Additionally, no subject (in both T2DM and control groups) had taken antibiotic, probiotic, or prebiotic products since two months before the sample collection. This study received approval from the Ethics Committee of SIPCDC (Suzhou Industrial Park Centers for Disease Control and Prevention) in accordance with the 1975 Declaration of Helsinki.
Informed consent was obtained from all individual participants.

| Sample collection and laboratory measurement
Blood samples were taken by venipuncture after at least 8-hour overnight fast. Tubes were centrifuged at 3000 g for 10 minutes at room temperature for separation. Plasma samples were frozen at −80°C for storage as quickly as possible. Fasting plasma glucose (FPG), triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL), alanine aminotransferase (ALT), and aspartate aminotransferase (AST) were measured using an autoanalyzer (Olympus AU640, Tokyo, Japan).

| DNA extraction, 16S rRNA gene amplification, and sequencing
Sample processing, DNA isolation, and PCR steps were conducted in a laminar air flow bench, illuminated with a UV lamp prior to use in order to avoid possible contaminants. Blood bacterial ge-

| Derivation of microbiome data and quality controls
Sequence reads processing was performed using QIIME (Quantitative

| Characteristics of the studied population
Demographic characteristics of cases (participants who developed T2DM) and controls (participants who did not develop T2DM) are shown in Table 1. T2DM cases and non-T2DM controls in the present study were similar with respect to the matching factors of age and gender. In addition, smoking, drinking, baseline FPG did not significantly differ between cases and controls. However, participants who developed T2DM were more likely to be with hypertension, increased levels of TC, TG, and LDL at baseline. As expected, the plasma level of HDL in control group was apparently lower than that in T2DM group at baseline.

| Sequencing data summarization
The total number of reads obtained from 150 participants was 9 968 238. After filtering and removing the chimeric sequences, we obtained 3 000 391 high-quality sequences from the participants developed T2DM (60 007/sample) and 6 244 227 sequences from non-T2DM controls (62 442/sample), respectively. As the results shown in Table 2 (Table S1).
As the results shown in Table 3, the mean diversity of the blood microbiome (Simpson, Chao1 and Shannon indices) was not different between the T2DM cases and non-T2DM controls at baseline.

| Associations of microbiome composition with diabetes
According to the limited criteria described in the "materials and methods," four phyla, 14 classes, 37 orders, 97 families, and 196 genera were included for further analysis in the present study. The results showed that no significant difference in relative abundance of bacterium was detected between two groups at phylum level. were much higher in T2DM cases than those in non-T2DM controls (Table S1-S5, Table 5).

TA B L E 4 Richness and diversity estimators in different groups
Because of low relative abundance of most pathogens in blood, we further examined the associations of blood microbiome with the onset of T2DM by characterizing the participants as carriers and non-carriers of the pathogens. As the results shown in Table 6 (Table 6).
However, the results from FDR correction indicated that no significant association was detected between blood microbiome and T2DM development (Table S5).

| D ISCUSS I ON
Recently, an increasing number of studies have reported that human blood contains an authentic microbiome. 18,[20][21][22] In the present study, our data showed that the peripheral blood collected from both cases and controls has a diverse bacterial microbiota, dominated by the phyla Proteobacteria, Bacteroidetes, Firmicutes, and Actinobacteria. This result is somewhat similar to the findings reported by previous studies. The results from a France study observed that the peripheral blood from healthy donors contains bacterial DNA mostly from the Proteobacteria phylum (between 80.4% and 87.4%), Actinobacteria phylum (between 6.7% and 10.0%), the Firmicutes (between 3.0% and 6.4%), and Bacteroidetes (between 2.5% and 3.4%) phyla. 16 To date, whether a diverse bacterial community is present in the blood of T2DM patients remains unclear. Larsen et al 23 reported that the relative abundance of Firmicutes was significantly higher, whereas the proportion of Bacteroidetes and Proteobacteria were much lower in fecal in diabetic persons than their non-diabetic counterparts. Additionally, the ratio of gut Firmicutes to gut Bacteroidetes (F/B) (the relative abundance in gut) was much higher in the patients with T2DM. 23 However, our present data did not observe that the composition of blood microbiome in T2DM patients is different with those in non-T2DM control at phyla level. Moreover, there was no significant difference in the value of F/B between T2DM participants and non-T2DM controls (data not shown). It is well known that the blood microbiome is derived primarily from the gut microbiome as a result of bacterial translocation. [24][25][26][27] However, as previously reported, the blood and gut microbiomes differ significantly from each other, indicating that the intestinal barrier, immune cells, and liver might play a role of filtering and affecting the bacterial translocation. [28][29][30][31][32] These results could partially explain the difference of the bacterial community diversity between the gut and blood. of the bacterial taxa in our blood samples had a low prevalence and abundance, and our study did not have sufficient power to carry out a systematic evaluation of this. Additionally, because the results from FDR correction showed that no significant association was detected between blood microbiome and T2DM development, more investigation with a large sample size is needed in future.

| CON CLUS IONS
In conclusion, our results suggested that blood microbiome may play an etiology role in the development of T2DM. In future, larger follow-up studies are warranted in order to further determine the relationships between blood microbiome and the risk of T2DM.

ACK N OWLED G M ENTS
The authors thank all individuals who took part in the study. We

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
JQ and YJ did the sequencing and data analyses. HZ, JQ, YJ, and CD were responsible for participants' recruitment, consent, and sample handling. JQ and CD did the statistical analyses and data interpretation. HZ and CD wrote the manuscript, and all authors reviewed the final version.