Fecal microbiota transplantation promotes reduction of antimicrobial resistance by strain replacement

Abstract

Multidrug-resistant organism (MDRO) colonization is a fundamental challenge in antimicrobial resistance. Limited studies have shown that fecal microbiota transplantation (FMT) can reduce MDRO colonization, but its mechanisms are poorly understood. We conducted a randomized, controlled trial of FMT for MDRO decolonization in renal transplant recipients called PREMIX (NCT02922816). Eleven participants were enrolled and randomized 1:1 to FMT or an observation period followed by delayed FMT if stool cultures were MDRO positive at day 36. Participants who were MDRO positive after one FMT were treated with a second FMT. At last visit, eight of nine patients who completed all treatments were MDRO culture negative. FMT-treated participants had longer time to recurrent MDRO infection versus PREMIX-eligible controls who were not treated with FMT. Key taxa (Akkermansia muciniphila, Alistipes putredinis, Phocaeicola dorei, Phascolarctobacterium faecium, Alistipes species, Mesosutterella massiliensis, Barnesiella intestinihominis, and Faecalibacterium prausnitzii) from the single feces donor used in the study that engrafted in recipients and metabolites such as short-chain fatty acids and bile acids in FMT-responding participants uncovered leads for rational microbiome therapeutic and diagnostic development. Metagenomic analyses revealed a previously unobserved mechanism of MDRO eradication by conspecific strain competition in an FMT-treated subset. Susceptible Enterobacterales strains that replaced baseline extended-spectrum β-lactamase-producing strains were not detectable in donor microbiota manufactured as FMT doses but in one case were detectable in the recipient before FMT. These data suggest that FMT may provide a path to exploit strain competition to reduce MDRO colonization.

Fig. 1:. Multi-drug resistant organism decolonization and recurrent infection among PREMIX clinical trial participants.

(A) PREMIX clinical trial schema. Blue bottles depict polyethylene glycol bowel preparation (prep), brown bottles depict FMT enema. Visits were conducted in cycles with visits on Days 1, 15, and 36, with additional stool sample collection on Day 2. Participants were randomized 1:1 to observation after bowel prep or bowel prep plus FMT. All participants with MDRO-positive stool culture(s) at Day 36 proceeded to an FMT cycle for up to 2 FMTs. (B) Results for (MDRO) selective stool cultures from PREMIX participants completing at least one visit cycle (N=11). Selective stool cultures were performed to isolate carbapenem-resistant Enterobacterales (CRE), extended-spectrum beta-lactamase producing Enterobacterales (ESBL), vancomycin-resistant Enterococcus spp (VRE), and multi-drug resistant Pseudomonas aeruginosa. (C) Kaplan-Meier plot showing shorter time to MDRO negative stool culture in the FMT group compared to observation followed by FMT, p = 0.046 by log rank test. (D) Kaplan-Meier plot showing renal transplant recipients with MDRO infection not enrolled in PREMIX (N=16) had shorter time to recurrent MDRO infection compared to PREMIX participants (N=11) at 180 days of follow up, p=0.037. Panels A and B created with biorender.com.

Fig. 2:. Microbiome dynamics after FMT for MDRO decolonization.

(A) Genus-level composition for participants (n=11) and FMT dose lots (n=6) from the stool donor (SD01). Gray dotted lines indicate first post-FMT samples (C1D02 & C2D02). Pathogenic genera are shaded orange, donor metagenome-assembled genome (MAGs) shaded purple. Non-pathogen genera with relative abundance <3% collapsed as “Other.” FMT participants with negative MDRO stool cultures at last visit shown with white facets, participants with positive MDRO stool cultures with gray facets, and FMT doses in a brown facet. (B) Inverse Simpson index increased after first FMT (Cycle 1, n=10, p = 0.03) but not Observation (n=5, p = 0.21). (C.) Normalized coverage (Reads per Kilobase per Genome Equivalent, RPKG) of antimicrobial resistance (AR) genes decreased after FMT to FMT dose levels. (D) Principal coordinates analysis (PCoA) of Bray-Curtis dissimilarity between donor and participant stool metagenomes shows trajectory of two baseline clusters grouped by partition around medoids (PAM) unsupervised clustering. Participants PM05, PM06, and PM09 were grouped as a cluster with more extreme dissimilarity to the donor compared to others. (E) PCoA of Bray-Curtis dissimilarity between samples from PREMIX, the Bar-Yoseph study of FMT for MDRO decolonization, 4 studies of FMT for RCDI, and the HMP HHS cohort suggest that most PREMIX participants had relatively mild microbiome disruption compared to RCDI patients.

Fig. 3:. Participants grouped by baseline dissimilarity from the stool donor had distinct metabolic shifts and donor taxa engraftment patterns after FMT.

(A) Log2-transformed mean peak intensity differences of annotated stool metabolites show greater changes from screening to last visit in glycocholate, isovalerate, and butyrate in participants with more extreme dissimilarity (n=3) vs less extreme (n=7). Boxes show first and third quartiles, whiskers illustrate values extending from the boxes to 1.5x the interquartile range. Metabolite data was not available for PM13. Representative engraftment of donor MAGs estimated by breadth of coverage with competitively mapped metagenome reads from participants classified by baseline donor dissimilarity as less extreme (B; PM01) and more extreme (C; PM06). Engraftment maps show fewer shared donor taxa at screening in the more extreme group and greater number of taxa that engrafted after FMT.

Fig. 4:. Gene alignment and metagenomic coverage of ESBL strain replaced by non-ESBL strain after FMT.

(A) Comparison of aligned, clustered representative genes from bacterial isolate draft genomes in participants that demonstrate replacement of high-identity ESBL strains (yellow) by a conspecific non-ESBL strain (blue). (B) Representative ESBL and non-ESBL isolate genome coverage breadth trends show detection of non-ESBL strain in stool sample metagenomes before FMT in PM01 and higher breadth of non-ESBL strains after FMT in PM01, PM03, and PM04. Breadth of 50% for specific detection of related strains is shown as a red dotted line.

Fig. 5:. Co-culturing experiments confirm genome- and metagenome-predicted competition of antibiotic susceptible and antibiotic resistant strains after FMT.

Lawns of baseline and replacing strains overlayed with suspensions of replacing and baseline strains, respectively, from PREMIX participants with conspecific strain replacement. Zone of clearance halos indicated by white arrows confirm competition by replacing colicinogenic strains and non-colicinogenic baseline strains (PM01, PM04) predicted by metagenome coverage breadth statistics (Fig. 4B) and genomic detection of colicin genes (Supplemental File S1.xlsx). No clearance was seen in baseline or replacing strains from other participants (PM03, PM07) in which colicin genes were not detected. In PM01, PM03, and PM04 the replacing strain was more susceptible to antibiotics than the baseline strain.