FormalPara Infographic
figure a
FormalPara Key Summary Points

Why carry out this study?

Recurrent Clostridioides difficile infection (CDI) imposes substantial clinical, economic, and HRQL burdens on patients, with subsequent recurrences intensifying these burdens, highlighting the need for effective treatments of first recurrence to enhance patient outcomes and alleviate healthcare system costs.

Fecal microbiota, live-jslm (REBYOTA, [RBL]), newly FDA-approved for preventing CDI recurrence following antibiotic treatment in adults, is the only microbiota-based live biotherapeutic product with a phase 3 trial that included patients at first recurrence, offering a unique opportunity to assess RBL as an effective treatment for patients at first recurrence.

There has been increased interest and research on treating patients at first recurrence of CDI to prevent further recurrences and our study adds important evidence to this emerging field of improving patient outcomes among this population.

What was learned from the study?

This post hoc analysis of the PUNCH CD3 trial found that, among patients at first CDI recurrence, RBL-treated patients had significantly higher probability of treatment success (81% vs. 60% [placebo]), and reported clinically meaningful improvements in HRQL compared to placebo, at week 8 of the double-blind period.

RBL is shown to be effective in preventing further CDI recurrences and improving HRQL among patients at first CDI recurrence, offering an effective treatment for these patients.

Digital Features

This article is published with digital features, including an infographic, to facilitate understanding of the article. To view digital features for this article, go to https://doi.org/10.6084/m9.figshare.24807243.

Introduction

Clostridioides difficile (C. diff) is an anaerobic spore-producing bacterium that causes an infection of the large intestine, with symptoms ranging from diarrhea to life-threatening colon damage [1]. C. diff infection (CDI) was associated with roughly 30,000 deaths annually and $6.3 billion in healthcare costs in 2016 in the USA [2,3,4,5]. Antibiotics alone, such as vancomycin and fidaxomicin, are the current standard-of-care for treatment of CDI [6, 7]. Although helpful in treating the initial phase of CDI, these antibiotics can disrupt intestinal flora resulting in further dysbiosis within the gut and increasing the risk of subsequent CDI recurrences [7]. Up to 35% of patients treated for a primary CDI episode experience at least one recurrence [8,9,10,11,12]. Moreover, the risk of additional recurrences increases after the first and can be as high as 65% for two or more recurrences [8].

Recurrent CDI (rCDI) is associated with substantial morbidity and mortality [13,14,15], and can greatly decrease a patient’s health-related quality of life (HRQL) [16]. Compared with first recurrence, two or more subsequent CDI recurrences are associated with increased emergency department and inpatient visits, resulting in substantially higher healthcare resource utilization and medical costs [17]. New effective treatments for first CDI recurrence are needed to improve patient outcomes of morbidity, mortality, and HRQL, which would potentially reduce healthcare resource utilization and associated costs.

There has been an increasing interest in exploring the utility of live biotherapeutic products (LBPs) for treating the early stage of CDI recurrence, particularly at first recurrence [18,19,20]. More data from phase 3 randomized, double-blind trials are needed to assess effective treatments for patients with first recurrence of CDI.

REBYOTA (fecal microbiota, live-jslm [RBL]), the first novel microbiota-based LBP, was recently approved by the US Food and Drug Administration (FDA) for the prevention of CDI recurrence following standard-of-care antibiotic treatment in adults with rCDI [21]. RBL was shown to significantly increase the treatment success rate of CDI prevention (70.6% vs. 57.5%) when compared with placebo in the phase 3 randomized, double-blinded placebo-controlled clinical trial, PUNCH CD3 [7]. RBL has exhibited long-term efficacy in maintaining CDI free in up to 92% for up to 24 months 22,23,24,25,26]. Moreover, treatment with RBL led to a significant improvement in patients’ HRQL, particularly in physical and mental health compared to placebo [27]. In post hoc analyses of PUNCH CD3, microbiome diversity and composition shifted significantly in patients with treatment response, with greater shifts among RBL-treated than placebo-treated responders [28].

Of the recently FDA-approved LBPs, RBL is the only one with a phase 3 trial (PUNCH CD3) that included patients at first CDI recurrence, thus providing a unique opportunity to rigorously assess treatment efficacy of RBL for this patient population. Here we investigated the efficacy of RBL in preventing future CDI recurrences and its impact on HRQL among patients at first recurrence using data from the PUNCH CD3 trial.

Methods

Data Source

De-identified, individual patient-level data from the PUNCH CD3 clinical trial were used in this study [7]. PUNCH CD3 (NCT03244644) is a randomized, double-blind, placebo-controlled phase 3 trial evaluating the safety and efficacy of RBL [7]. The trial included adult patients who were at least 18 years old, had one or more CDI recurrences, and completed at least one round of standard-of-care oral antibiotic therapy or had at least two episodes of hospitalization due to severe CDI within the last year. Patients were randomized in a 2:1 ratio to receive one dose of RBL or placebo (normal saline solution) and followed for 8 weeks. Additional PUNCH CD3 study details have been previously published [7].

Study Overview

A post hoc analysis was conducted among a subset of patients at first recurrence from the PUNCH CD3 modified intention-to-treat (mITT) population, which included patients who successfully received blinded treatments. The mITT population excluded patients who withdrew prior to treatment, for whom treatment was attempted but not completed, or who discontinued prior to evaluation of treatment failure/success for the primary endpoint if the reason for dropout was unrelated to CDI; RBL’s FDA approval was based on results from the mITT population. As-observed data from baseline and the double-blind period (i.e., weeks 1, 4, and 8) were used for the analyses. Patients who experienced a recurrence after blinded treatment had the option to receive open label RBL at the investigator’s discretion per study protocol. To maintain double-blindness, participants who received another dose of RBL before week 8 were excluded from the HRQL analyses as the open-label RBL treatment may bias the patient-reported HRQL.

All aspects of the study were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. As this analysis used de-identified clinical trial data, no institutional board review or patient consent was required for this study.

Outcomes

The outcomes of interest were treatment success and HRQL impacts at week 8 (the end of the double-blind period). Per trial protocol, treatment success at week 8 was defined by the absence of C. difficile associated diarrhea (without the need for retreatment as assessed by subject interview and physical exam). HRQL was measured during the double-blind period using the C. diff Health-related Quality-of-Life Questionnaire (Cdiff32) [29, 30], a validated, disease-specific survey that quantifies HRQL including a total score and three domain scores (physical, mental, and social). Cdiff32 comprises 32 self-administered questions about the impact of CDI, with each question offering five responses on a Likert-scale indicating how a person is affected by CDI-related concerns in the past 7 days (see Supplementary Material for the questionnaire). A 10-point shift is considered to be a clinically meaningful change [31]. The total and domain scores of Cdiff32 ranged from 0 (worst score) to 100 (best score) after aggregating and rescaling of responses to the questions. Cdiff32 scores were collected at baseline and at weeks 1, 4, and 8 during the double-blind trial period.

Statistical Analyses

All statistical analyses were conducted among patients at first recurrence within the mITT trial population. To avoid potential patient reporting bias, for the HRQL analysis, the sample set was further restricted to patients who completed Cdiff32 while remaining double-blind to the study intervention (i.e., participants who received an open-label dose of RBL before week 8 were excluded). Baseline demographic and disease characteristics were summarized using descriptive statistics by treatment arm for four sample sets: (1) the mITT trial population in the primary trial publication [7], (2) the subset of patients at first recurrence within the mITT trial population (used for efficacy analyses), (3) the mITT trial population with HRQL data available [27], and (4) the subset of patients at first recurrence with HRQL data (used for HRQL analyses). Mean and standard deviation (SD) were calculated for continuous variables and counts and percentages were calculated for categorical variables. Sample sets 2 and 4 were used for analyses in this study, while sample sets 1 and 3 were summarized as a reference for the full trial population.

The proportions of patients with treatment success at week 8 were summarized by treatment arm. As the current post hoc analysis included only the subset of patients at first recurrence, logistic regressions were conducted to estimate the efficacy of RBL vs. placebo accounting for baseline differences in known risk factors for recurrence. The dependent variable was treatment success at week 8 with treatment group (placebo vs. RBL) as the exposure variable of interest, controlling for age group (≥ 75 years vs. < 75 years), sex (male vs. female), prior proton pump inhibitor use, and prior treatment with fidaxomicin. The selection of these variables for inclusion in the model was based on prior literature, which found these to be risk factors for recurrence of CDI [32,33,34,35,36,37,38,39]. The odds ratios (OR) with 95% confidence interval (CI) were reported. The average predicted probabilities of treatment success for RBL and placebo were calculated using the least squares mean (LS-mean) method.

Impacts on HRQL were summarized via their absolute and change from baseline scores for the Cdiff32 total score and each domain score by treatment arm at weeks 1, 4, and 8 using mean and SD. Mean Cdiff32 scores over the 8-week double-blind period were plotted by treatment arm using trajectory graphs. Absolute Cdiff32 scores were compared between baseline and weeks 1, 4, and 8 using Wilcoxon rank-sum tests and changes from baseline to week 8 were compared between treatment arms with Wilcoxon signed-rank tests. Additional analyses were conducted to evaluate responses to individual Cdiff32 questions between RBL and placebo using Wilcoxon signed-rank tests. Statistical significance was based on a two-sided alpha level of 0.05.

Results

Patient Demographic and Baseline Characteristics

A total of 86 patients (53 [61.6%] RBL-treated and 33 [38.4%) placebo-treated) at first recurrence were included in this analysis, which constituted 32.8% of the total mITT population (N = 262) in the PUNCH CD3 trial. Patients in the RBL and placebo arms were on average 58.6 (SD 20.1) and 57.8 (15.9) years old, respectively, and 66.0% and 66.7% were female (Table 1). Most patients were treated with vancomycin at baseline (83.0% for RBL and 90.9% for placebo) and 15.1% (RBL) and 15.2% (placebo) had previously been hospitalized for CDI.

Table 1 Baseline characteristics, and observed treatment success rates at week 8, of patients with rCDI in the mITT and first recurrence populations of PUNCH CD3
Full size table

Among the 86 patients at first recurrence, 66 had Cdiff32 scores available at both baseline and week 8 and were included in the HRQL analyses. Patients in the RBL and placebo arms were on average 56.8 (SD 19.0) and 59.1 (16.7) years old, respectively, and 62.8% and 69.6% were female (Table 2). Most patients were treated with vancomycin at baseline (83.7% for RBL and 91.3% for placebo) and 16.3% (RBL) and 17.4% (placebo) had previously been hospitalized for CDI. The baseline characteristics for patients at first recurrence (efficacy and HRQL samples) were similar to those of patients in the overall mITT population in the PUNCH CD3 trial (Tables 1 and 2) [27, 40].

Table 2 Baseline characteristics of patients with rCDI in the mITT and first recurrence populations of PUNCH CD3 who had HRQL (Cdiff32) data at baseline and week 8
Full size table

Efficacy/Treatment Success at Week 8

At week 8, 62 of the 86 patients at first recurrence (42 in 53, 79.2%, RBL-treated, 20 in 33, 60.6%, placebo-treated; relative risk 0.53, p = 0.06) had no recurrence of CDI (i.e., treatment success). In the multivariable logistic regression analysis adjusting for patient characteristics at baseline, the RBL group had significantly lower odds of developing recurrence of CDI at week 8, and therefore a greater probability of treatment success for CDI prevention, compared to the placebo group (OR 0.35 [95% CI 0.13, 0.98]). The average probabilities of treatment success at week 8 were estimated at 81% (95% CI 68%, 90%) for RBL and 60% (42%, 75%) for placebo, representing a 21% absolute increase and a 35% relative increase in treatment success for RBL compared to placebo.

Health-related Quality of Life

Total Score

The majority of patients at first recurrence had Cdiff32 scores at both baseline and week 8: 43/53 (81%) of RBL-treated patients and 23/33 (70%) of placebo-treated patients. All patients had Cdiff32 scores at baseline; the mean Cdiff32 total scores were 40.9 (SD 16.1; median 45.3) for RBL and 46.4 (19.7; median 48.4) for placebo. The total scores improved significantly from baseline to weeks 1, 4, and 8 in both groups (all p < 0.01) (Table 3, Fig. 1a). Mean changes from baseline through week 8 were all significantly greater for RBL-treated than placebo-treated patients, with the largest difference at week 8 (35.2 [SD 19.7] points for RBL and 21.7 [23.4] for placebo; change score difference 13.5 [5.7], p < 0.05).

Table 3 Summary of HRQL (Cdiff32 score) changes from baseline to weeks 1, 4, and 8 among RBL- or placebo-treated patients in PUNCH CD3
Full size table
Fig. 1
figure 1

Cdiff32 total (a) and domain (b) scores during the double-blind treatment period, stratified by treatment arm (mITT population of PUNCH CD3, as-observed data) among all patients. The double-blinded period of the PUNCH CD3 trial spanned until week 8. Cdiff32 Clostridioides difficile Health-related Quality-of-Life Questionnaire, mITT modified intention-to-treat, RBL REBYOTA

Full size image

Physical Domain Score and Item-Level Analysis

Baseline Cdiff32 physical domain scores were 48.3 (SD 19.7) for RBL and 51.8 (22.8) for placebo; the median scores were 48.2 and 51.8, respectively. Cdiff32 physical scores improved significantly from baseline to weeks 1, 4, and 8 in both groups (all p < 0.01) (Table 3, Fig. 1b). RBL showed trends for greater improvement than placebo, but this difference in mean changes from baseline was only significant at week 1 (change score difference 11.5 [SD 5.2], p < 0.05) and not at other time points. The mean change at week 8 was 11.9 (SD 6.1, p = 0.07) points higher for RBL than for placebo (36.8 [20.3] vs. 24.9 [25.3], respectively).

RBL-treated patients, on average, reported larger improvements on individual item-level questions in the physical domain, with a few notable differences. Specifically, by week 8, responses on three questions showed significantly greater improvements for RBL-treated patients, namely questions “Because of your C. diff infection, have you had trouble sleeping?”, “Because of your C. diff infection have you been woken up from sleep?”, and “Have you avoided wearing some clothes (tight clothes, dress, light-colored clothes…)?”. Additionally, although not statistically significant, responses on one question showed a noteworthy 21.6-point higher improvement for RBL-treated patients, i.e., “Are you afraid that certain food will worsen your C. diff infection?”.

Mental Domain Score and Item-Level Analysis

Baseline Cdiff32 mental scores were 31.3 (SD 15.8) for RBL and 39.2 (19.6) for placebo; the median scores were 32.1 and 35.7, respectively. The Cdiff32 mental scores improved significantly from baseline to weeks 1, 4, and 8 in both groups (all p < 0.01) (Table 3, Fig. 1b). Mean changes from baseline through week 8 were all significantly greater for RBL compared with placebo, with the largest difference at week 8 (34.8 [SD 22.7] vs. 18.6 [23.3], respectively; change score difference 16.2 [6.0], p < 0.01).

For the majority of the item-level questions in the mental domain, RBL-treated patients reported significantly larger improvements than placebo by week 8, including “I feel that I am not in control of my C. diff infection”, “I feel my life is less enjoyable because of my C. diff infection”, “I feel much stressed because of my C. diff infection”, “I feel isolated from others because of my C. diff infection”, and “I worry about transmitting my C. diff infection to my family and/or friends”. Additionally, larger but not statistically significant improvements were observed on three questions, namely “I believe that any stress can worsen my C. diff infection”, “I feel irritable because of my C. diff infection,” and “Are you afraid that your C. diff infection could get worse in the future?”.

Social Domain Score and Item-Level Analysis

Baseline mean Cdiff32 social scores were 48.6 (SD 23.0) for RBL and 52.8 (24.5) for placebo; the median scores were 43.8 and 50.0, respectively. The Cdiff32 social domain scores improved significantly from baseline to weeks 1, 4, and 8 in both groups (all p < 0.01) (Table 3, Fig. 1b). The improvements were numerically higher for RBL than for placebo. At week 8, on average, RBL-treated patients scored 7.6 (SD 7.4, p = 0.45) higher score on the social domain than placebo did (28.8 [22.4] and 21.2 [31.4], respectively). At the item level, RBL-treated patients, on average, reported larger improvements on those in the social domain, although they were not significantly greater than those of placebo-treated patients.

Discussion

Recurrence of CDI is common and substantially impacts patient outcomes including increased morbidity and mortality [13,14,15], compromised HRQL [41,42,43], and high economic burden [44]. Because of the high and multifaceted burden imposed by rCDI on patients, there is an increased interest in treating patients at first recurrence to prevent further recurrences, improve their outcomes, and minimize the subsequent health and economic impacts. The efficacy and HRQL impact of RBL have been studied in phase 2 and phase 3 trials in 971 patients with rCDI and published elsewhere [22, 27]. Here we conducted post hoc analyses of the PUNCH CD3 clinical trial [7] on RBL’s efficacy and HRQL impact in patients at first recurrence. PUNCH CD3 is the only phase 3, randomized, double-blind trial of approved LBPs that includes patients at first recurrence. Our study adds important evidence to the emerging field of improving patient outcomes among the population of patients with rCDI at first recurrence.

From the efficacy perspective, this analysis found a significantly higher treatment success rate at week 8 in RBL-treated patients at first CDI recurrence than those receiving placebo (81% vs. 60%), after adjusting for baseline characteristics. Despite the reduced sample size in this subgroup analysis compared to the overall mITT population of the PUNCH CD3 trial, RBL demonstrated statistically significant improvements in efficacy, along with HRQL, over placebo, indicating RBL is an effective treatment for patients at first recurrence [7].

HRQL is particularly important for rCDI given its adverse effects on patients’ physical and psychological well-being [41, 45]. From the HRQL perspective, our analysis showed that RBL-treated patients at first recurrence benefited from significant larger improvements in HRQL, as measured by the Cdiff32, compared with those receiving placebo, for the total score and the mental domain. The differences in observed improvements for patients treated with RBL compared to placebo on the Cdiff32 total, physical, and mental domains exceeded the minimal clinically important difference (MCID) threshold (i.e., a 10-point change) [31]. Moreover, the differences in improvements were rapid (as early as at week 1) and continued throughout week 8. These rapid improvements align with the observed trend of change in gut microbiome composition in patients treated with RBL [46] and the HRQL analyses of the PUNCH CD3 trial population [27]. In addition, patients with a healthier gut microbiome have also been reported to have better HRQL [47].

The greater improvements in HRQL among RBL-treated patients at first recurrence were reflected in multiple aspects of daily life and emotional well-being as measured by the Cdiff32 questionnaire. Analysis of the individual Cdiff32 questions shed more light on these specific considerations. While improvements were observed in both treatment groups through the 8-week double-blind period, RBL-treated patients reported having fewer difficulties in “carrying out daily activities”, fewer concerns with “leaving the house” and in clothing choice, were less bothered by time one “spent on the toilet”, and stronger improvements in sleep. Particularly in the mental health domain, patients did not feel as isolated, depressed or stressed, and they felt more in control. At week 8, the improvements in mental health were more pronounced in RBL-treated patients as they reported worrying less about CDI and felt more in control and were more able to enjoy life. Overall, patients who received RBL appeared to feel more at ease starting in week 1. This sense continued to increase thereafter with a greater feeling of control of their CDI and their life and with a sense of return to normal life as measured by Cdiff32 over the double-blind trial period.

Our study contributes to a growing body of research focusing on early intervention, in particular, at first recurrence for rCDI. RBL, a newly approved therapy, has demonstrated greater treatment success and higher HRQL improvements in patients at first recurrence over placebo. The observed improvements in mental health may also suggest the potential restoration of gut microbiome diversity and inhibition of antibiotic-induced dysbiosis, resulting in patients experiencing better mental health. These patients may have subsequently benefited with a restoration of the gut–brain axis and rebalanced neurotransmitters for improved psychological well-being [48]. Indeed, in a separate exploratory analysis, where a subset of patients at first recurrence in PUNCH CD3 trial with microbiome data available, shifts in key microbiome taxa abundances were found between baseline and week 8 toward healthier microbiota, suggesting a link between observed improvements in HRQL and gut microbiota among patients at first recurrence. Further research is needed to quantify the association between gut microbiome and quality of life in patients with rCDI.

Since this is a post hoc subgroup analysis, patients in the first recurrence subset may no longer be randomized between RBL and placebo arm, which may introduce differences in patient characteristics at baseline between treatment arms. However, an examination of the baseline characteristics of patients at first recurrence showed consistency with the full mITT population of the PUNCH CD3 trial. Furthermore, the efficacy outcome was analyzed adjusting for possible baseline confounding factors. For the HRQL analysis, the proportion of patients with available Cdiff32 data at both baseline and week 8 was limited to a further reduced subset of patients (66 of the 86 patients at first recurrence). Patients who experienced a recurrence and received an open-label RBL dose at the investigator’s discretion prior to week 8 were excluded from the HRQL analysis to avoid potential bias. Because a greater proportion was excluded in the placebo group (30%) than the RBL group (19%) at week 8, it is more likely that we may have underestimated the HRQL benefit of RBL over placebo.

Conclusion

This study adds important evidence to the emerging research field of improving patient outcomes among patients with rCDI at first recurrence. RBL showed significantly higher treatment success and HRQL benefit when compared to placebo in patients at first CDI recurrence. Additionally, RBL demonstrated rapid (as early as at week 1) and continued improvements in HRQL as measured by Cdiff32 scores throughout the 8-week double-blinded clinical trial period. RBL provides a new and effective treatment option for patients at first recurrence of CDI.