The Commensal Microbe Veillonella as a Marker for Response to an FGF19 Analog in NASH

Abstract

Background and aims: The composition of the human gut microbiota is linked to health and disease, and knowledge of the impact of therapeutics on the microbiota is essential to decipher their biological roles and to gain new mechanistic insights. Here we report the effect of aldafermin, an analog of the gut hormone FGF19, versus placebo on the gut microbiota in a prospective, phase 2 study in patients with NASH.

Approach and results: A total of 176 patients with biopsy-confirmed nonalcoholic steatohepatitis (NASH) (nonalcoholic fatty liver disease activity score ≥ 4), fibrosis (F1-F3 by NASH Clinical Research Network criteria), and elevated liver fat content (≥ 8% by magnetic resonance imaging-proton density fat fraction) received 0.3 mg (n = 23), 1 mg (n = 49), 3 mg (n = 49), and 6 mg (n = 28) aldafermin or placebo (n = 27) for 12 weeks. Stool samples were collected on day 1 and week 12 and profiled using 16S ribosomal RNA gene sequencing; 122 patients had paired stool microbiome profiles at both day 1 and week 12. Overall, the state of the gut microbial community was distinctly stable in patients treated with aldafermin, with all major phyla and genera unaltered during therapy. Patients treated with aldafermin showed a significant, dose-dependent enrichment in the rare genus Veillonella, a commensal microbe known to have lactate-degrading and performance-enhancing properties, which correlated with changes in serum bile acid profile.

Conclusions: Veillonella may be a bile acid-sensitive bacteria whose enrichment is enabled by aldafermin-mediated suppression of bile acid synthesis and, in particular, decreases in toxic bile acids. This study provides an integrated analysis of gut microbiome, serum bile acid metabolome, imaging, and histological measurements in clinical trials testing aldafermin for NASH. Our results provide a better understanding of the intricacies of microbiome-host interactions (clinicaltrials.gov trial No. NCT02443116).

© 2020 by the American Association for the Study of Liver Diseases.

Figures

Fig. 1

Stable gut microbiome with aldafermin therapy in patients with NASH. (A) Study design. Cohort 1 was a placebo‐controlled, double‐blind study comparing aldafermin 3 mg and 6 mg versus placebo for 12 weeks; cohort 2 was a dose‐expansion study evaluating aldafermin 0.3 mg, 1 mg, and 3 mg for 12 weeks; and cohort 3 further expanded the assessment of aldafermin 1 mg in additional patients for 12 weeks. Overall, a total of 176 patients with NASH received 0.3 mg (n = 23), 1 mg (n = 49), 3 mg (n = 49), or 6 mg (n = 28) aldafermin or placebo (n = 27) for 12 weeks in this phase 2 trial of aldafermin. The 16S rRNA sequencing of fecal samples collected at baseline (day 1) and week 12 (end of treatment) was performed. Patients had liver biopsy at baseline and underwent MRI and laboratory tests at baseline and week 12. (B) Gut microbial richness and evenness measured by alpha diversity (P values by Mann‐Whitney U test with FDR corrections using Benjamini‐Hochberg method). (C) Gut microbial beta biodiversity. Beta diversity was evaluated using UniFrac‐based analysis. In the principal coordinates analysis, no clustering was observed at the PC1 versus PC2 plot, indicating stable phylogenetic composition of the samples. All patients with paired stool samples at both day 1 and week 12 were included in the analysis: placebo (n = 18), aldafermin 0.3 mg (n = 17), 1 mg (n = 30), 3 mg (n = 36), and 6 mg (n = 21). Abbreviations: D1, day 1; W12, week 12.

Fig. 2

Phylogenetic abundance at the phylum level. (A) Relative abundance of top phyla by visits in patients treated with placebo or aldafermin. Taxonomic compositions of the most prevalent phyla are displayed, with ranks ordered from left to right by their decreasing abundance. Only the top 10 phylotypes are shown for clarity (P > 0.05 for all comparisons by Kruskal‐Wallis test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). (B) Relative abundance of top phyla by treatment groups (P > 0.05 for all comparisons by Mann‐Whitney U test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). The boxes represent the interquartile range (IQR), from the first and third quartiles, and the inside line represents the median. The whiskers denote the lowest and highest values within 1.5 IQR from the first and third quartiles. The circles represent outliers beyond the whiskers. All patients with paired stool samples at both day 1 and week 12 were included in the analysis: placebo (n = 18), aldafermin 0.3 mg (n = 17), 1 mg (n = 30), 3 mg (n = 36), and 6 mg (n = 21).

Fig. 3

Phylogenetic abundance at the genus level. (A) Relative abundance of top genera by visits in patients treated with placebo or aldafermin. The phylotypes with median relative abundances greater than 0.01% of total abundance in either the placebo or aldafermin groups were included for analysis. Only the 10 most abundant genera in each group are shown for clarity (P > 0.05 for all comparisons by Kruskal‐Wallis test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). (B) Relative abundance of top genera by treatment groups (P > 0.05 for all comparisons by Mann‐Whitney U test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). The boxes represent the IQR, from the first and third quartiles, and the inside line represents the median. The whiskers denote the lowest and highest values within 1.5 IQR from the first and third quartiles. The circles represent outliers beyond the whiskers. All patients with paired stool samples at both day 1 and week 12 were included in the analysis: placebo (n = 18), aldafermin 0.3 mg (n = 17), 1 mg (n = 30), 3 mg (n = 36), and 6 mg (n = 21).

Fig. 4

Aldafermin enriches the rare genus Veillonella. (A) Box and whisker plot of relative abundance of Veillonella. Stool samples from patients treated with placebo, aldafermin 0.3 mg, aldafermin 1 mg, aldafermin 3 mg, and aldafermin 6 mg were assessed by 16S rRNA gene sequencing. Dose‐dependent increases in Veillonella abundance were observed in the aldafermin groups, but not in the placebo group (***P < 0.001, **P < 0.01 vs. placebo by Kruskal‐Wallis test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). (B) Proportion of subjects with fecal samples that were positive for Veillonella at week 12. Presence of Veillonella in stool samples was considered positive (**P < 0.01, *P < 0.05 vs. placebo by Fisher’s exact test). (C) Aldafermin did not affect other microbes that are of oral origin or typically associated with Veillonella, nor did it affect microbes that ferment other substrates or have an ethanol‐producing property. Shown are box and whisker plots of relative abundance in placebo and aldafermin groups. All patients with paired stool samples at both day 1 and week 12 were included in the analysis: placebo (n = 18), aldafermin 0.3 mg (n = 17), 1 mg (n = 30), 3 mg (n = 36), and 6 mg (n = 21). Abbreviations: D1, day 1; W12, week 12.

Fig. 5

Enrichment of Veillonella inversely correlates with toxic, hydrophobic bile acids. (A) Box and whisker plots of serum bile acids from placebo and aldafermin‐treated patients by GC‐MS. Shown are concentrations of GCA, GCDCA, GDCA, and DCA at baseline (day 1) and end‐of‐treatment (week 12) (***P < 0.001, **P < 0.01, *P < 0.05 vs. placebo): placebo (n = 27), aldafermin 0.3 mg (n = 23), 1 mg (n = 49), 3 mg (n = 49), and 6 mg (n = 28). (B) Correlation between Veillonella OTU reads and concentrations of GCA, GCDCA, GDCA, and DCA at week 12. Spearman’s correlation coefficients were calculated; rho and P values are shown on the graphs.

Fig. 6

Correlation between gut microbiome and LFC measured by MRI‐PDFF. (A) Patients were grouped into the top quartile (Q4) versus the bottom quartile (Q1) according to baseline LFC. Differences in alpha diversity were observed between Q1 and Q4 subjects using the observed OTUs but not the Shannon index. (B) Weighted UniFrac analysis showed no separation between low liver fat (Q1) and high liver fat (Q4) populations at baseline, whereas unweighted UniFrac showed a trend of clustering. (C) No significant difference at the phylum and genus level between low liver fat (Q1) and high liver fat (Q4) populations. (D) Relative abundance of top phyla in subjects with ≥ 70% reduction in LFC (i.e., super‐responders). At the phylum level, Euryarchaeota significantly differed between aldafermin‐treated subjects who had ≥ 70% reduction in LFC and the placebo group (*P < 0.05 by Mann‐Whitney U test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). (E) Relative abundance of top genera in subjects with ≥ 70% reduction in LFC (i.e., super‐responders). Genera are displayed with ranks ordered from left to right by their decreasing abundance (left panel) or significance (right panel). Only the top 10 phylotypes are shown for clarity. At the genus level, although the top 10 genera are similar between groups, the minor genus Haemophilus is more abundant in subjects with ≥ 70% reduction in LFC than placebo‐treated subjects (*P < 0.05 by Mann‐Whitney U test with Benjamini‐Hochberg FDR‐adjusted multiple test correction). The boxes represent the IQR, from the first and third quartiles, and the inside line represents the median. The whiskers denote the lowest and highest values within 1.5 IQR from the first and third quartiles. The circles represent outliers beyond the whiskers. All patients with paired stool samples at both day 1 and week 12 were included in the analysis: placebo (n = 18), aldafermin 0.3 mg (n = 17), 1 mg (n = 30), 3 mg (n = 36), and 6 mg (n = 21).