Washed microbiota transplantation reduces glycemic variability in unstable diabetes

Abstract

Background: Dysbiosis of gut microbiota is causally linked to impaired host glucose metabolism. We aimed to study effects of the new method of fecal microbiota transplantation, washed microbiota transplantation (WMT), on reducing glycemic variability (GV) in unstable diabetes.

Methods: Fourteen eligible patients received three allogenic WMTs and were followed up at 1 week, 1 month, and 3 months. Primary outcomes were daily insulin dose, glucose excursions during meal tests, and GV indices calculated from continuous monitoring or self-monitoring glucose values. Secondary outcomes were multiomics data, including 16S rRNA gene sequencing, metagenomics, and metabolomics to explore underlying mechanisms.

Results: Daily insulin dose and glucose excursions markedly dropped, whereas GV indices significantly improved up to 1 month. WMT increased gut microbial alpha diversity, beta diversity, and network complexity. Taxonomic changes featured lower abundance of genera Bacteroides and Escherichia-Shigella, and higher abundance of genus Prevotella. Metagenomics functional annotations revealed enrichment of distinct microbial metabolic pathways, including methane biosynthesis, citrate cycle, amino acid degradation, and butyrate production. Derived metabolites correlated significantly with improved GV indices. WMT did not change circulating inflammatory cytokines, enteroendocrine hormones, or C-peptide.

Conclusions: WMT showed strong ameliorating effect on GV, raising the possibility of targeting gut microbiota as an effective regimen to reduce GV in diabetes.

Keywords: fecal microbiota transplantation; glycemic variability; hypoglycemia; microbiome; unstable diabetes.

FIGURE 1

(A) Flow diagram of study design and participating patients. (B) Changes of daily insulin dose. Data of the same subject at different follow‐up time points are connected by gray lines. False discovery rate (represented as q‐value) was calculated using repeated‐measures one‐way analysis of variance corrected by Benjamini and Hochberg method for multiple comparisons, n = 14. (C) Postprandial glucose excursions expressed as percentage of baseline during steamed bun meal test (SBMT). All data are represented as mean ± SEM, *q < 0.05 by mixed‐effects model for repeated measures corrected by Benjamini and Hochberg method, n = 9–12. (D–H) Changes of glycemic variability (GV) indices at 1 week (T1W) calculated using continuous glucose monitor (CGM) data: (D) MAGE, (E) SDBG, (F) BG > 11.1 mmol/L, (G) LAGE, and (H) MBG. ^† p < .05 by paired t test (two‐tailed), n = 14. (I–J) Changes of GV indices at T1W, 1 month (T1M), and 3 months (T3M) calculated using self‐monitoring of blood glucose data: (I) TIR and (J) hypoglycemic episodes. ^$ q < 0.05 as stated in (B); ^# q < 0.05 by repeated‐measures Friedman test corrected by Benjamini and Hochberg method, n = 14. (K) Representative CGM 24‐h glucose profiles demonstrating improved glycemic stability in two patients by WMT. Values of some GV indices were the same for different participants. Thus, both points and connecting lines overlapped with each other, exhibiting fewer than 14 points. LAGE, largest amplitude of glycemic excursions; MAGE, mean amplitude of glycemic excursion; MBG, mean blood glucose; SDBG, SD of blood glucose; TIR, time in range; WMT, washed microbiota transplantation.

FIGURE 2

Washed microbiota transplantation (WMT) alters gut microbiota composition in recipient patients as revealed by 16S rRNA gene sequencing. (A) Comparison of alpha diversity indices, including Observed Species, Faith's phylogenetic diversity, and ACE index, between T0 and follow‐up time points. Box plots are presented with means and whiskers denoting minimum and maximum values. *q < 0.05, **q < 0.01 by ordinary one‐way analysis of variance with Benjamini and Hochberg false discovery rate correction for multiple comparisons, n = 12–14. (B) Comparison of beta diversity represented as PCoA that was plotted based on Unweighted Unifrac distances. Associated statistical analyzes were conducted using analysis of molecular variance, ^† p < .05, ^††† p < .001 when compared to T0. (C) Linear discriminate analysis coupled with effect size measurements of the signature differential taxa. LDA score of 4 was used as cutoff. (D) Cooccurrence network of gut microbiota showing interactions among genera. Each node represents a genus, nodes of the same color belong to one phylum, node size represents taxonomic abundance. Lines connecting nodes represent Spearman coefficients. Line thickness represents the interaction strength. Red and green lines indicate positive and negative interactions, respectively. Only significantly correlated genera with p < .05 are adopted to construct the network. (E) Network diameter refers to the length of the longest of all the computed shortest paths between any two nodes. Graph density refers to the ratio of the number of actual edges over that of all possible edges. Average degree refers to average number of edges per node. LDA, linear discriminant analysis; PC1, principal coordinate 1; PC2, principal coordinate 2; PCoA, principal coordinates analysis.

FIGURE 3

Washed microbiota transplantation (WMT) promotes divergent functional shifts in gut microbiome as revealed by metagenomics. (A) PCoA analysis of changes in the abundance of KEGG genes and associated box plots presented with means and whiskers denoting minimum and maximum values of Bray‐Curtis distance. *q < 0.05 as stated in Figure 2(A), n = 7–10. (B) Box plots with means showing relative changes of representing KEGG orthology at 1 week (T1W) and 1 month (T1M) in comparison to T0. Whiskers denote minimum and maximum values. Data were log transformed. *p < .05 and **p < .01 by permutation test, n = 7–9. (C) Enriched KEGG pathways using differential KOs. The length of bar was determined by the number of KOs mapping to each pathway. (D) Illustration of distinct metabolic pathways regulated by WMT. Differential KOs rendering each identified pathway are positioned above the reactions they catalyze. Red and blue indicate up‐ and downregulated KOs, respectively. Dashed lines indicate a series of reactions involving multiple steps. ABC, ATP‐binding cassette; KEGG, Kyoto encyclopedia of genes and genomes; KO, KEGG orthology; PC1, principal coordinate 1; PC2, principal coordinate 2; PCoA, principal coordinates analysis; TCA, tricarboxylic acid.

FIGURE 4

Washed microbiota transplantation (WMT) alters profiles of fecal metabolites. (A) Heatmap showing changes of fecal metabolites determined by targeted metabolomics. Colors changing from blue to red indicate higher abundance. p values of comparisons between T0 and each post‐WMT time point for each metabolite were calculated via either unpaired t test (two‐tailed) or Mann–Whitney test, depending on whether sample data passed normality test, n = 10–12. (B) Heatmap of Spearman's correlation coefficients between fecal metabolites and glycemic variability (GV) indices calculated using self‐monitoring of blood glucose data. Red and blue color indicates positive and negative coefficient, respectively. *p < .05, **p < .01, and ***p < .001. AUC, area under the curve; CV, coefficient of variation; HbA1c, glycated hemoglobin; LAGE, largest amplitude of glycemic excursions; MBG, mean blood glucose; PPGE, postprandial glucose excursion; SBMT, steamed bun meal test; TIR, time in range.

FIGURE 5

Washed microbiota transplantation (WMT) alters profiles of serum metabolites. (A) Heatmap showing changes of serum metabolites determined by non‐targeted metabolomics, n = 11–14. (B) Heatmap of Spearman's correlation coefficients between serum metabolites and glycemic variability (GV) indices. (C) The 8 positively correlated metabolites between fecal and serum compartments. Color range varies from light red (weaker correlation) to dark red (stronger correlation). Statistical methods were the same as stated in Figure 4. AUC, area under the curve; BG, blood glucose; CGM, continuous glucose monitor; CV, coefficient of variation; HbA1c, glycated hemoglobin; LAGE, largest amplitude of glycemic excursions; MBG, mean blood glucose; PPGE, postprandial glucose excursion; SBMT, steamed bun meal test; SMBG, self‐monitoring of blood glucose; TIR, time in range.

FIGURE 6

Schematic summary of proposed mechanisms underlying WMT‐induced therapeutic effects (feature figure). Up‐ and downregulated items by WMT are shown in red and blue colors, respectively. The top portion shows representative differential taxa, for which the letters “p,” “c,” “o,” “f,” “g,” and “s” represent phylum, class, order, family, genus, and species, respectively. The middle portion shows major differential pathways and related fecal metabolites. The bottom portion shows representative serum differential metabolites. All the as‐shown metabolites have been demonstrated to be associated with reduced GV. Because only the abundance of fecal L‐tyrosine correlated with that in serum, here we propose gut microbial metabolism of L‐tyrosine may contribute to improved systemic glycemic control. Solid open arrow indicates a confirmed causal relationship; dashed open arrow indicates this interaction still needs to be validated; the solid line with unfilled arrowhead indicates translocation. 2‐HPAA, ortho‐hydroxyphenylacetic acid; 2‐OG, oxoglutaric acid; AABA, L‐alpha‐aminobutyric acid; GV, glycemic variability; HPHPA, 3‐3‐hydroxyphenyl‐3‐hydroxypropanoic acid; HPLA, hydroxyphenyllactic acid; MA, maleic acid; MMA, methylmalonic acid; WMT, washed microbiota transplantation.