Phage-microbe dynamics after sterile faecal filtrate transplantation in individuals with metabolic syndrome: a double-blind, randomised, placebo-controlled clinical trial assessing efficacy and safety

Abstract

Bacteriophages (phages) are bacterial viruses that have been shown to shape microbial communities. Previous studies have shown that faecal virome transplantation can decrease weight gain and normalize blood glucose tolerance in diet-induced obese mice. Therefore, we performed a double-blind, randomised, placebo-controlled pilot study in which 24 individuals with metabolic syndrome were randomised to a faecal filtrate transplantation (FFT) from a lean healthy donor (n = 12) or placebo (n = 12). The primary outcome, change in glucose metabolism, and secondary outcomes, safety and longitudinal changes within the intestinal bacteriome and phageome, were assessed from baseline up to 28 days. All 24 included subjects completed the study and are included in the analyses. While the overall changes in glucose metabolism are not significantly different between both groups, the FFT is well-tolerated and without any serious adverse events. The phage virion composition is significantly altered two days after FFT as compared to placebo, which coincides with more virulent phage-microbe interactions. In conclusion, we provide evidence that gut phages can be safely administered to transiently alter the gut microbiota of recipients.

Fig. 1. Study overview and outcomes on glucose metabolism.

a Overview of the study. b Glucose excursions during the oral glucose tolerance test. One person who was randomised to the FFT group had progressed to type 2 diabetes, which was not apparent at the time of screening. c Total area under the curve (AUC) for glucose, and d for C-peptide did not significantly differ between the groups. Within both groups there was a small increase in glucose AUC between day 0 and 28, which was nominal significant within the FFT group, although this significance disappeared after correction for multiple testing. e Insulin resistance (HOMA-IR) measures did not significantly differ between the groups, but significantly increased from day 0 to day 28 in both groups. f Glucose variability, expressed as time between 3.9-10 mmol/L glucose, improved only within the FFT group between day 0 and day 28, which was nominal significant, but disappeared after correcting for multiple testing. Sample size for both FFT and placebo group is n = 12 subjects in all plots. Box plots show the median (middle line), 25th, and 75th percentile (box), with the 25th percentile minus and the 75th percentile plus 1.5 times the interquartile range (whiskers), and outliers (single points). P-values according to paired two-sided Wilcoxon signed-rank test. Source data are provided in the Source Data file.

Fig. 2. Overview of faecal samples sequenced, richness and diversity.

a Overview of the faecal samples used for the bulk metagenomic sequencing (for bacteriome and phageome) and the metagenomic sequencing of the viral-like particles (VLP). b The richness (number of observed species) in the bacteriome, phage virions (VLP) and bulk-derived phageome (WGS) from baseline until follow-up at day 28. Though there were no significant differences between the placebo and faecal filtrate group, the richness in the bacteriome reduced slightly after both interventions. A similar trend was observed in the phageome (mostly prophages present in bacterial hosts), while the richness in the VLP fraction tended to increase slightly at day 2 for both interventions. c The alpha-diversity (Shannon index) of the bacteriome, phage virions (VLP) and bulk-derived phageome (WGS) from baseline until follow-up at day 28. Again, no significant differences were found between the interventions. Similar to the richness, the diversity of the bacteriome and phageome slightly decreased directly after the interventions. For the free phages, the diversity decreased slightly in the placebo group, but not in the faecal filtrate group. Sample size for both FFT and placebo group is n = 12 subjects in all plots. Error bars represent the standard error of the mean. Source data are provided in the Source Data file.

Fig. 3. Donor phage and new phage abundance in recipients.

a The percentage of phages that were shared between the donor and recipient within the phageome, and (b) within the phage virions after the faecal filtrate transplantation. There was a slight, non-significant increase in the relative abundance of (pro)phages shared with the donor after the intervention, while the relative abundance of phage virions that were shared with the donor slightly decreased. c The percentage of new phages that were present after the intervention within the bulk-derived phageome, and (d) within the phage virions. In both, the relative abundance of new phages increased over time and although not significant, this increase was slightly higher in the FFT group. Sample size for both FFT and placebo group is n = 12 subjects in all plots. Error bars represent the standard error of the mean. Source data are provided in the Source Data file.

Fig. 4. Compositional differences between the FFT and placebo group.

a Principal response curve showing how the FFT group (n = 12) differs from the placebo (n = 12, set to zero) in the bacteriome, bulk-derived phageome (WGS), and phage virion (VLP) composition. The coefficient is the canonical coefficient of treatment and significance in dispersion over time and at each separate time-point was calculated with permutation tests, corrected for multiple testing by the Benjamini-Hochberg method. b Principal component analysis of VLP composition after centred log-ratio transformation. Large points show the mean of each group (both n = 12). c Log fold change for all 216 viral populations (VP) indicated as differentially abundant by ANCOM-BC analysis. For legibility, VP names are not shown, these can be found in Supplementary Data 1. d Bacterial host species of which the phages are enriched among differentially abundant VPs. The first column shows the number of differentially abundant VPs, the second the total number of VPs linked to a given host in the dataset, and padj shows the level of significance after testing for enrichment with a hypergeometric test, adjusted for multiple testing by the Benjamini-Hochberg method. e Splits up the data on the first column of D by participant group. Source data are provided in the Source Data file.

Fig. 5. Correlations between viral populations and their host bacteria.

a Correlations plots showing the change in relative abundance between day 0 and day 2 of viral populations (VP) versus host bacteria. Each datapoint represents the interactions between the VPs and metagenome-assembled genomes (MAGs) of a particular species within a given sample. Linkages between VPs and MAGs were based on either CRISPR spacer hits or the presence of the VP as a prophage in the MAG. The Spearman correlation coefficient showed that phage-bacterium interactions in FFT samples tended toward antagonism, while those in placebo samples were protagonistic. b Same as a, but showing the change in relative abundance between day 2 and day 28. The Spearman correlation coefficient showed that phage-bacterium interactions in placebo samples were protagonistic. c Same as a and b, but showing the change in relative abundance between day 0 and day 28. There was no overall correlation between changes in abundance of VPs and host MAGs. Shaded areas indicate 95% confidence interval. Significance according to two-sided Spearman’s rank correlation coefficient. Source data are provided in the Source Data file.