Developing Microbiome Rehabilitation Biomarkers for Clostridium Difficle Infections: Continued Evaluation of a Prototype MICROBIOME HEALTH INDEX™ (MHI™)

Ken Blount PhD, Courtney Jones BS, Elena Deych MS, Bill Shannon PhD MBA

Background

• Dysbiosis, or disruption of a healthy microbiome, is strongly associated with Clostridium difficile infections (CDI). RBX2660 is a standardized, stabilized microbiota restoration therapy in clinical development for preventing recurrence of CDI (rCDI).
• In two recent Phase 2 clinical trials, RBX2660, a standardized, stabilized microbiota restoration therapeutic, was more efficacious than controls for preventing rCDI, and associated studies indicated that RBX2660 restored a healthier microbiome among responding participants.
• Since quantitative biomarkers for microbiome dysbiosis and/or restoration have not been established, we are evaluating a prototype unidimensional MICROBIOME HEALTH INDEX™ (MHI™).
• Herein we report a pooled MHI analysis of two Phase 2 trials of RBX2660 for preventing rCDI and evaluate the potential of MHI as a predictor of clinical efficacy.

Methods

• Included in this analysis are 127 RBX2660 product samples and 339 stool samples collected at indicated time points from a total of 176 participants with recurrent CDI who received at least one dose of RBX2660 as part of the PUNCH CD 2 Phase 2B trial (NCT02299570) or the PUNCH Open Label Phase 2 trial
  (NCT02589847).
• Success was defined as the absence of CDI at 8 weeks after the last treatment.

Microbiome Analysis

• PUNCH CD2 participant samples underwent 16S sequencing. Data from the PUNCH OL participant samples were generated using an ultra-shallow shotgun sequencing that generates taxonomic profiles with species-level resolution (CoreBiome, Minneapolis, MN).
• Relative taxonomic abundances at the class level were calculated from OTU data for each time, treatment, outcome group, and the mean and upper/lower confidence limits defined by by fitting to a Dirichlet-multinomial distribution using maximum likelihood estimation.

Conclusions

• MHI is agnostic from sequencing method and can effectively distinguish patients with dysbiosis from healthier patients.
• Retrospective MHI evaluation at 7 days post-treatment could establish a putative Microbiome Efficacy Endpoint that may predict 8-week clinical response.
• Future efforts will determine whether more specific taxonomic characterization below the class level and/or inclusion of diversity metrics provide a more precise index. Also, we will evaluate MHI prospectively in ongoing clinical trials as an exploratory endpoint.