Gut Microbiome-Based Diagnostic Model to Predict Diabetes Mellitus

Figures & data

Table 1. Baseline characteristics of study population

Variable	2 Diabetes(n = 44)	Control(n = 47)	P value
Age, yr	58 ± 8	59 ± 12	0.556
Male	24(54.55%)	28(59.57%)	0.393
Body mass index, kg/m^2	26.06 ± 3.639	24.55 ± 3.159	0.61
Smoking	17(38.64%)	16(34.04%)	0.406
Drinking	12(27.27%)	13(27.66%)	0.577
Basic diseases
Coronary heart disease	35(79.55%)	37(78.72%)	0.565
Essential hypertension	31(70.45%)	36(76.60%)	0.335
Long term oral/subcutaneous injection medication
Metformin	40(90.9%)	/	/
Glucosidase inhibitors	11(25%)	/	/
DDP-4 inhibitors	2(4.55%)	/	/
Insulin	16(36.36%)	/	/
Aspirin	33(75%)	34(72.34%)	0.481
Clopidogrel	30(68.18%)	25(53.19%)	0.106
Stain	32(72.72%)	26(55.32%)	0.065
β-block	27(61.36%)	28(59.57%)	0.516
Calcium Channel Blockers	12(27.27%)	7(14.89%)	0.116
ACEI or ARB^a	18(40.91%)	18(38.30%)	0.484
Laboratory results
White blood cells, ×10^9/L	6.76 ± 1.9	6.67 ± 1.98	0.838
Neutrophilicgranulocyte,%	59.91 ± 11.36	57.49 ± 13.39	0.354
Hemoglobin, ×10^12/L	136.50 ± 21.34	138.55 ± 13.47	0.587
Platelets, ×10^9/L	235.34 ± 65.98	221.85 ± 53.13	0.284
blood urea nitrogen,mmol/L	5.37 ± 1.91	5.05 ± 1.32	0.366
Creatinine,umol/L	62.00(46.34–72.00)	64.00(44.20–72.50)	0.948
Uric acid,umol/L	314.45 ± 91.63	320.28 ± 77.47	0.746
FBG^b,mmol/L	6.59(5.77–9.43)	4.7(4.27–5.395)	<0.001
Total cholesterol, mmol/L	3.64 ± 0.90	3.69 ± 1.13	0.849
Triglyceride, mmol/L	1.72(1.11–2.39)	1.38(1.01–2.02)	0.184
LDL^c,mmol/L	2.20 ± 0.77	2.20 ± 0.92	0.972
HDL^d,mmol/L	1.01 ± 0.27	1.15 ± 0.27	0.13
Total bilirubin,umol/L	10.1(7.4–14.4)	10.6(8.3–14.8)	0.553

Data are presented as median (interquartile range), mean ± SD, or number (%).

ACEI or ARB^a: angiotensin converting enzyme ingibitors or angiotensin receptor blocker.

FBG^b: fasting blood glucose; LDL^c: low density lipoprotein; HDL^d: high density lipoprotein.

Figure 1. Increased fecal microbial diversity in patients with DM vs healthy controls.(a)Rarefaction Curve.(b) Rank abundance curve.(c) Venn diagram displaying the overlaps between groups showing that 3362 of the total richness of 23,839 OTUs were shared between the CAD patients and the healthy controls. OTUs: operational taxonomy units.(d) Circle packing chart

Figure 2. Stacked histogram of species composition in the top 10 of each group. (a) Phylum level. (b) Family level. (c) Genus level. (d) Species level

Figure 3. Comparison of fecal microbial diversity, as estimated by the Chao1 index and Shannon index (a), Faith-pd index and Observed_species index (b)

Figure 4. Beta diversity was calculated using unweighted UniFrac [left] or brau_curtis [right] by PCoA and NMDS, indicating a symmetrical distribution of fecal microbial community among all the samples

Figure 5. Phylogenetic profifiles and differences of gut microbes between patients and healthy controls. [A] LEfSe method identifified the most differentially abundant taxons between the patients and healthy controls. [B] Heatmap of correlation between differential bacteria and clinical indicators. #: Tentative names in greenenes database

Figure 6. Important biomarkers. [A] The top 12 bacteria belong to the genus level. [B] ROCs curve with AUC for the diagnostic performance of the gut microbial model. [C] Comparison of the POD of gut microbiome. [Ruminococcus] Temporary name

Table 2. Candidate variables for clinical model development

Variables	AUC	P values	95% CI
FBG	0.839	0.000	0.742–0.937
HDL	0.337	0.016	0.212–0.462
EH	0.658	0.019	0.533–0.783
BMI	0.613	0.095	0.481–0.744
CAD	0.599	0.140	0.471–0.728
Ne	0.589	0.189	0.458–0.720
SMOKING	0.552	0.437	0.421–0.684
LDL	0.451	0.469	0.318–0.584
TG	0.546	0.496	0.414–0.678
TC	0.457	0.527	0.326–0.589
WBC	0.463	0.580	0.330–0.595
Cr	0.471	0.663	0.336–0.605
UA	0.474	0.698	0.340–0.608
SEX	0.475	0.706	0.342–0.607
TBIL	0.477	0.737	0.343–0.612
AGE	0.480	0.766	0.345–0.614
PLT	0.519	0.774	0.383–0.656
HBL	0.491	0.898	0.356–0.627
DRINGKING	0.503	0.962	0.371–0.635

Figure 7. Instructions for using the nomogram

Figure 8. [A] Calibration curve. [B] The AUCs for the diagnostic performances of the clinical model, microbiome model and the combined model