The intestinal microbiota and metabolites in patients with anorexia nervosa

Abstract

Brain-gut microbiota interactions are intensively studied in connection with various neurological and psychiatric diseases. While anorexia nervosa (AN) pathophysiology is not entirely clear, it is presumably linked to microbiome dysbiosis. We aimed to elucidate the gut microbiota contribution in AN disease pathophysiology. We analyzed the composition and diversity of the gut microbiome of patients with AN (bacteriome and mycobiome) from stool samples before and after renourishment, and compared them to healthy controls. Further, levels of assorted neurotransmitters and short-chain fatty acids (SCFA) were analyzed in stool samples by MS and NMR, respectively. Biochemical, anthropometric, and psychometric profiles were assessed. The bacterial alpha-diversity parameter analyses revealed only increased Chao 1 index in patients with AN before the realimentation, reflecting their interindividual variation. Subsequently, core microbiota depletion signs were observed in patients with AN. Overrepresented OTUs (operation taxonomic units) in patients with AN taxonomically belonged to Alistipes, Clostridiales, Christensenellaceae, and Ruminococcaceae. Underrepresented OTUs in patients with AN were Faecalibacterium, Agathobacter, Bacteroides, Blautia, and Lachnospira. Patients exhibited greater interindividual variation in the gut bacteriome, as well as in metagenome content compared to controls, suggesting altered bacteriome functions. Patients had decreased levels of serotonin, GABA, dopamine, butyrate, and acetate in their stool samples compared to controls. Mycobiome analysis did not reveal significant differences in alpha diversity and fungal profile composition between patients with AN and healthy controls, nor any correlation of the fungal composition with the bacterial profile. Our results show the changed profile of the gut microbiome and its metabolites in patients with severe AN. Although therapeutic partial renourishment led to increased body mass index and improved psychometric parameters, SCFA, and neurotransmitter profiles, as well as microbial community compositions, did not change substantially during the hospitalization period, which can be potentially caused by only partial weight recovery.

Keywords: BMI; EDE-Q; Microbiome; SCFA; bacteriome; dysbiosis; gut-brain-microbiota axis; mycobiome; neurotransmitter; renourishment.

Figures

Figure 1.

Gut bacteriome alpha diversity variation between control individuals vs. AN1 vs. AN2 assessed based on A) Observed OTUs number, B) Total OTU richness predicted by Chao 1 index, and C) Shannon index. Significant differences between categories (p < 0.05 according to Tukey post-hoc tests) are indicated by different letters above the bars

Figure 2.

Proportions of dominating bacterial classes (represented by >1% of reads) in the three studied groups

Figure 2.

Proportions of dominating bacterial classes (represented by >1% of reads) in the three studied groups

Figure 3.

PCoA showing variation in bacterial microbiota composition between controls vs. AN1 vs. AN2. Compositional variation was assessed based on A) relative abundance-based (Bray-Curtis) and B) prevalence-based (Jaccard) dissimilarities

Figure 3.

PCoA showing variation in bacterial microbiota composition between controls vs. AN1 vs. AN2. Compositional variation was assessed based on A) relative abundance-based (Bray-Curtis) and B) prevalence-based (Jaccard) dissimilarities

Figure 4.

Bacterial core microbiota in the studied groups. A) Proportion of reads corresponding to core bacterial OTUs (i.e. detected in >90% samples) in each studied group and B) heatmap showing their prevalences

Figure 4.

Bacterial core microbiota in the studied groups. A) Proportion of reads corresponding to core bacterial OTUs (i.e. detected in >90% samples) in each studied group and B) heatmap showing their prevalences

Figure 5.

Relative abundances of bacterial OTUs (squared-root transformed) that varied, according to DESeq2 analyses (FDR

Figure 6.

A) PCoA showing variation in relative abundances of predicted metabolic pathways between controls vs. AN1 vs. AN2, and B) Relative abundances of bacterial metagenomic pathways (squared-root transformed) that varied, according to DESeq2 analyses (FDR

Figure 6.

A) PCoA showing variation in relative abundances of predicted metabolic pathways between controls vs. AN1 vs. AN2, and B) Relative abundances of bacterial metagenomic pathways (squared-root transformed) that varied, according to DESeq2 analyses (FDR

Figure 7.

Fungal microbiota variation in the three studied groups. A) Dominating fungal class proportions (represented by >1% of reads); PCoA for B) Bray-Curtis and C) Jaccard dissimilarities between samples

Figure 7.

Fungal microbiota variation in the three studied groups. A) Dominating fungal class proportions (represented by >1% of reads); PCoA for B) Bray-Curtis and C) Jaccard dissimilarities between samples

Figure 8.

Significant associations between bacterial OTU abundances and concentrations of SCFAs or neurotransmitters. Predictions and 95% confidence intervals for negative binomial generalized linear mixed models (GLMMs) are shown

Figure 8.

Significant associations between bacterial OTU abundances and concentrations of SCFAs or neurotransmitters. Predictions and 95% confidence intervals for negative binomial generalized linear mixed models (GLMMs) are shown