Resetting the Microbial Landscape: Donor Microbiome Engraftment in Patients Treated with RBX2660 for Multi-Recurrent Clostridium difficile Infection

Arnab Ray, MD, Courtney Jones, BS, Bill Shannon, PhD, MBA, Sharina Carter, BS

Background

• Recurrent Clostridium difficile infections (rCDI) are associated with an altered microbiome composition and diversity compared to healthy patients.
• RBX2660 is a standardized microbiota-based drug manufactured from live human microbes, designed to rehabilitate a patient’s intestinal microbiota.
• This analysis evaluated the extent to which RBX2660 defines the microbiome for patients treated in PUNCH CD 2—a randomized, double-blind, placebo-controlled study evaluating RBX2660 for the prevention of rCDI.

Methods

• Patients in the 3-arm PUNCH CD 2 trial were randomized 1:1:1 to receive either: 2 doses of RBX2660 (Group A); 2 doses of placebo (Group B); or 1 dose of RBX2660 followed by 1 dose of placebo (Group C). Doses were administered by enema 7 days apart.
• Each RBX2660 dose originated from a single donor-traceable lot manufactured from one of 19 possible donors. Note that all donors underwent pre-enrollment pathogen screening and post-donation blood testing.
• Success was defined as the absence of Clostridium difficile -associated diarrhea at 8 weeks following completion of the last treatment. Patients were classified as a treatment failure if all four(4) of the following criteria were met: recurrence of diarrhea less than 8 weeks after administration of the last assigned study enema, a positive laboratory diagnosis of C. difficile as conducted and reported by the study investigator, a need for retreatment for CDI, and no other cause for CDI symptoms.
• Patients were requested to submit stool samples at baseline and at 7, 30, and 60 days after treatment.
• Longitudinal 16s rRNA analysis using the Illumina MiSeq platform was performed on stool samples collected from patients in Groups A and C, and on samples from the RBX2660 donor lots utilized in the study.
• The variable region V4 was targeted to identify the operational taxonomic units (OTUs) in each sample.
• Microbiome data for all patients at study entry were combined as a treatment-naïve baseline.
• OTU data for patients who responded to treatment in Groups A and C were combined as a single “success” cohort at each time point for microbiome comparisons
• Relative abundance data from patient samples were grouped longitudinally and compared to donors’ microbiome data using a Bray- Curtis dissimilarity calculation with non-metric multi-dimensional scaling. Mean group microbiomes and relative taxonomic abundance at the class level were determined using the Dirichlet-Multinomial distribution and compared with a generalized Wald-type test.
• Differentiation between sample communities was visualized using a Kullback-Leibler (KL) divergence analysis model (BioRankings, St. Louis, MO), which calculates a mean log fold change between two communities.

PUNCH CD 2 Clinical Trial Results¹

• A single dose of RBX2660 was significantly better than placebo (67%, Group C vs 46% Group B, P=0.48) (Figure 2).
• A second dose of RBX2660 one week later (Group A) did not provide additional clinical benefit.
• Groups A and C combined were superior to placebo (P=0.046)

Results

• Microbiome analysis was conducted on 42 patients in Groups A and C who had completed longitudinal collection (n=113 stool samples) and were associated with 19 RBX2660 donor lots (n=136 product samples).
• The 19 RBX2660 donor lots had similar taxonomic distributions, with a group mean that was highly divergent from and significantly different from baseline microbiomes (Figures 3, 4, and 5).
• After treatment, the group means of patients’ microbiomes progressed to the mean donor microbiome over time, with the largest shift observed 7 days after treatment (Figure 3). The shift of group means from baseline was observed at all time points (Figure 5).
• The relative abundances of key taxonomic classes became more related to the donor microbiome with increasing time post-treatment (Figure 4), most notably Bacteroidia, Clostridia, Gammaproteobacteria, and Bacilli, but did not completely converge with the donor microbiome during the first 60 days.
• Regardless of donor lot received, the mean patient microbiome did not completely converge to the mean donor microbiome by day 60 post-treatment (p>0.05 for all time points).
• A single dose of RBX2660 was sufficient to induce the maximal microbiome change (Group C), since a second dose of RBX2660 after 7 days (Group A) did not induce additional microbiome changes (Figure 6).

Conclusion

RBX2660 donor microbiomes contribute to, but do not fully define, a patient’s post-treatment microbiome. One dose of RBX2660 is sufficient for microbiome change associated with successful outcome. Further work is needed to define specific taxa and strains that directly engraft from RBX2660 to patient and to evaluate whether early engraftment correlates with eventual treatment response.

References

1. Dubberke ER et al. Efficacy and Safety of RBX2660 for the Prevention of Recurrent Clostridium difficile Infection: Results of the PUNCH CD 2 Trial Infectious Diseases Week (ID Week) October 26-30, 2016, New Orleans, LA.