Unexpected Results From a Phase 2 Trial of a Microbiome Therapeutic for Clostridioides difficile Infection: Lessons for the Future

(See the Major Article by McGovern et al on pages 2132–40.)

“Failure is only the opportunity more intelligently to begin again.”

—Henry Ford

Infection with toxigenic Clostridioides difficile in the setting of antibiotic administration has become the exemplar of a disease where the indigenous microbiota plays a central role in pathogenesis [1]. Alteration of the structure and function of the gut bacterial community renders the gastrointestinal tract conducive to germination of ingested spores and the vegetative outgrowth of the pathogen. While we are still unraveling the precise molecular mechanisms by which the indigenous microbiota normally maintains colonization resistance against C. difficile there has been a push to target the gut microbiota for therapy of C. difficile infection (CDI) and other enteric infections [2]. In particular, the restoration of the normal structure and function of the microbiota to prevent or break the cycle of C. difficile recurrence is an area of intense study. The use of fecal transplantation for antibiotic-associated colitis predates the recognition of C. difficile as the primary pathogen for this clinical syndrome [3]. The efficacy of this treatment has resulted in the inclusion of fecal microbiota transplantation (FMT) as an alternate therapy for recurrent CDI (rCDI) in recently revised treatment guidelines [4].

The success of FMT for rCDI has spurred studies in the basic science of the microbiome (see [5] for my use of terms) related to C. difficile infection and also in the development of novel therapies. The development of therapeutics has followed several paths. Fecal banks have been developed where screened donor feces are collected and prepared for distribution, generally in frozen form. The advantage of this approach is that it eliminates the time, staff, protocols, and resources needed by each facility to prepare material for FMT [6]. In general, the donor feces are minimally processed, mimicking as closely as practically possible the use of fresh donor feces. However, this lack of processing has led to inadvertent transfer of pathogens to the FMT recipients [7]. The emergent coronavirus disease 2019 (COVID-19) outbreak has led the Food and Drug Administration (FDA) to issue additional guidance on the use of therapeutic FMT, for all indications, not just for rCDI [8]. This guidance followed the recognition that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can be detected in the feces of infected individuals.

Several groups have extended FMT by processing the donor feces prior to formulation in a form for delivery to patients. The article by McGovern and colleagues [9] published in this issue of Clinical Infectious Diseases reports on the results of a phase 2 trial of a microbiome therapeutic derived from donor feces. This trial was based on the promising results from a phase 1b trial that demonstrated that the investigational drug SER-109 was associated with a 96.7% efficacy of preventing relapse of CDI in patients who had previously had 3 or more episodes of rCDI [10]. SER-109 is derived from donor feces that have been treated using a proprietary method that involves ethanol treatment to eliminate potential pathogens and allow enrichment of spore-forming bacteria present. Previous work has demonstrated that members of the spore-forming fraction of feces (mainly members of the Firmicutes) were sufficient to restore colonization resistance and break the cycle of rCDI. On the basis of the results of the phase 1b trial, a phase 2 multicenter trial was undertaken to reduce rCDI in patients who had at least 3 episodes of CDI in the previous 9 months. A total of 89 patients were enrolled, with 59 receiving a fixed dose of SER-109 (equivalent to 1 × 10⁸ spores) and 30 receiving placebo.

Despite the remarkable results of the noncontrolled phase 1b trial, the phase 2 trial failed to demonstrate efficacy of SER-109 over placebo. Overall, rCDI rates were 44.1% in the SER-109 arm compared with 53.3% in the placebo arm, but this did not reach statistical significance (relative risk [RR], 1.2; 95% confidence interval [CI], .8–1.9). A planned subgroup analysis based on age did show a significant reduction in rCDI in subjects older than age 65 years (RR, 1.8; 95% CI, 1.1–2.8). However, given the failure of the primary endpoint, a great deal of concern rippled through the microbiome research community and adversely affected some corporate ventures hoping to capitalize on the great interest in the microbiome as a novel therapeutic, preventative, and diagnostic target.

In the current article, the investigators report on the failed phase 2 study, and to their credit, uncovered potential underlying reasons for the failure of SER-109 to reduce rCDI in the research cohort. The authors came to 2 main conclusions. First, while the phase 1b study used a range of doses, a single dose was used for the phase 2 trail. Post hoc analysis combining the results of both trials suggests that the dose of 1 × 10⁸ spores used in the phase 2 trail was too low. Analysis of the microbiota in SER-109 recipients indicated that successful treatment was associated with successful, early engraftment of the bacteria present in SER-109 in the recipient. The combined analysis indicated that higher doses were more likely to result in early engraftment, and it was suggested that the fixed dose used in the phase 2 trial was too low as a result. Second, of the patients who were enrolled in the study, which required a positive stool test for C. difficile, over 80% had the diagnosis made on the basis of a polymerase chain reaction (PCR)–based assay and only 19.1% of patients were enrolled based on a test that tested for the presence of C. difficile toxin in feces. It should be noted that the relative utility of nucleic acid amplification testing (NAAT) versus toxin-based testing for diagnosis of clinical CDI has been debated [11], and both approaches are currently included in the current clinical practice guidelines for CDI [4]. Some investigators suggest that a direct toxin test has greater specificity and positive-predictive value for true CDI as a positive PCR test may only indicate asymptomatic colonization or transient shedding of spores [12]. Regardless of the sensitivity and specificity of NAAT versus toxin testing for clinical CDI, for the purposes of the SER-109 clinical trial, if the use of the PCR assay resulted in the inclusion of patients who had symptoms based on something other than actual CDI—for example, postinfectious irritable bowel syndrome [13]—this could explain the nearly 50% response in recipients of placebo. The investigators have taken these results into consideration for their current phase 3 trial, which uses a higher dose of SER-109 and requires diagnosis of rCDI on the basis of a toxin-based test. It will be interesting to see if these study modifications will lead to more promising results at the conclusion of this ongoing trial.

The authors of the study should be commended for conducting a thorough “postmortem” on a study that failed to reach the planned endpoints. The results of their analysis, including the potential confounding effects of PCR versus toxin-based methods used to diagnose CDI, should inform others conducting clinical studies and trials of this important nosocomial infection. There is possibly another lesson that can be taken from failure of this phase 2 trial. The current study followed a “standard” randomized, placebo-controlled design, which is regarded as the gold standard for clinical studies. However, there has been a recent surge in interest for employing adaptive study designs in clinical trials. A detailed discussion of adaptive study designs is beyond the scope of this commentary, but these study designs have been proposed to limit the time and expense of large randomized controlled trials while still providing data that can be used to evaluate and develop promising new therapies [14]. Given the observed effect of dose in the SER-109 trial, this may have been detected using an adaptive dose-finding or adaptive group scheme. Adaptive study designs are increasingly being employed in the cardiovascular and cancer research areas [15] and the FDA recently published guidance for industry on the use of adaptive designs for clinical trials of drugs and biologics [16]. Perhaps the greater use of adaptive study designs for microbiome therapeutics will represent an important advance that will speed the development of novel therapies based on an understanding of the complex relationship that we have with our indigenous microbiota.

Notes

Financial support. Work in the Young Laboratory is funded by the National Institutes of Health (grant number AI124255).

Potential conflicts of interest. The author has served as a consultant to Vedanta Biosciences, Bio-K+ Internation, and Pantheryx. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.