Experimental human colonisation with non-toxigenic Clostridioides difficile: a placebo-controlled randomised clinical trial Authors: A. D. O. Hensen, C. Harmanus, P. H. Verbeek-Menken, J. P. R. Koopman, O. A. C. Lamers, G. V. T. Roozen, J. J. Janse, M. Balke-Buijs, M. Y. E. C. van der Stoep, P. Meij, I. M. van Amerongen-Westra, P. Schipper, C. Crul, L. Pattacini, K. Rox, F. Farowski, A. Tsakmaklis, M. J. G. T. Vehreschild, E. J. Kuijper, W. K. Smits, M. Roestenberg Published: 2026 ABSTRACT Clostridioides difficile infections remain a major global healthcare burden, underscoring the need for novel therapies. Human colonisation models provide mechanistic insight into C. difficile colonisation and facilitate identification of novel intervention targets. We conducted a placebo-controlled, randomised clinical trial (NCT05693077) administering non-toxigenic C. difficile (NTCD) capsules to healthy participants to assess safety and colonisation as primary endpoints, and microbiota susceptibility as a secondary endpoint. A total of 69 healthy participants (18-45 years), not previously colonised with C. difficile and without recent antibiotic use, were enrolled following a health assessment. NTCD capsules administered for five consecutive days at low or high dose, was safe with no dose-response relationship in colonisation outcomes. Vancomycin pretreatment induced colonisation success: with 5% colonisation without, 32% after one day, and 84% after five days vancomycin pretreatment. Some participants that cleared vancomycin rapidly acquired non-challenge C. difficile strains prior to NTCD challenge. Microbiota profiling (using shotgun metagenomics) revealed reduced α-diversity and pronounced community restructuring. These findings highlight the impact of antibiotic-mediated microbiota disruption, the widespread environmental presence of C. difficile, and the feasibility of meaningful microbiota assessment in small-scale intervention trials, thereby providing a robust tool to investigate this globally impactful infection. Introduction Clostridioides difficile remains the leading cause of healthcare-associated diarrhoea worldwide, accounting for an estimated 15,600 deaths and 284,000 disability adjusted life years (DALYs) in 20211. C. difficile infection (CDI) is a toxin-mediated disease, and risk factors for its acquisition include antibiotic use, increased age, immunosuppression and prolonged hospitalisation2,3. Of concern, the incidence of community-acquired CDI is increasing4, potentially driven by the widespread dissemination of C. difficile in animals, the environment, and food5. Consequently, CDI constitutes a major global health challenge, with profound clinical, societal, and economic implications6. This challenge is further exacerbated by the limitations of conventional antibiotic treatment for CDI, which is associated with high relapse rates7 and the growing threat of antimicrobial resistance6. As such, there is an urgent need for novel, more effective treatment strategies. Microbial restoration therapies, such as faecal microbiota transplantation (FMT), have shown promise in providing high-level sustained prevention of recurrent CDI8–10. However, the complex interactions between C. difficile and the human microbiome remain poorly understood. Because of the uniqueness and complexity of the human microbiome, animal models have limited value for the development and testing of novel therapies targeting CDI. A comprehensive understanding of the specific components of the microbiome that contribute to colonisation resistance as well as resistance to infection would be crucial to the development of novel (preventive) treatment strategies. Here, we develop a human model of C. difficile colonisation to investigate colonisation resistance conferred by the host microbiome. The model uses a non-toxigenic C. difficile (NTCD) strain that lacks the pathogenicity locus (PaLoc) necessary for toxin production and therefore cannot induce C. difficile disease. This study thus focuses on establishing a C.difficile-controlled human colonisation model, which could also serve as a foundation for developing a C.difficile-controlled human infection model (CHIM) in the future. Supported by encouraging preliminary safety data from a first-in-human study of NTCD as a preventive therapeutic11,12, we aimed to develop a colonisation model with controlled dosing to enable investigation of host susceptibility and resistance markers. For this purpose, we designed a clinical trial in which NTCD spores (from strain L-NTCD03) are administered through capsules13 in an outpatient setting, followed by weekly safety assessments and high-frequency sampling for colonisation assessment and microbiota changes. Results We enrolled 69 healthy participants between September 2023 and April 2024 in an adaptive dose-design, randomised double-blind controlled clinical trial consisting of three consecutive stages (Fig.1). In stage 2, 22 participants were included instead of the intended 23. Overall loss to follow-up was minimal (N = 2). The median age of the participants was 22 years, with a male-to-female ratio of 33:36. No significant differences in participant characteristics were observed between the study stages (Supplementary Table 1). The study design is schematically depicted in Fig. 2.Fig. 1CONSORT diagram.Both lost to follow-ups were due to non-study related issues and did not complete follow-up until D28. Abbreviations used: pos positive, SSCY Salmonella, Shigella, Yersinia or Campylobacter spp.; PCR polymerase chain reaction, HH member household member.Fig. 2Schematic overview of study design.Schematic overview of study design; in stage 2, instead of the intended 23 participants (104 L-NTCD03 (N = 10), 107 L-NTCD03 (N = 10) and placebo (N = 3)), 22 participants were included and randomised (104 L-NTCD03 (N = 10), 107 L-NTCD03 (N = 10) and placebo (N = 2)). CONSORT diagram. Both lost to follow-ups were due to non-study related issues and did not complete follow-up until D28. Abbreviations used: pos positive, SSCY Salmonella, Shigella, Yersinia or Campylobacter spp.; PCR polymerase chain reaction, HH member household member. Schematic overview of study design. Schematic overview of study design; in stage 2, instead of the intended 23 participants (104 L-NTCD03 (N = 10), 107 L-NTCD03 (N = 10) and placebo (N = 3)), 22 participants were included and randomised (104 L-NTCD03 (N = 10), 107 L-NTCD03 (N = 10) and placebo (N = 2)). Primary outcomes: safety and C. difficile colonisation Exposure of healthy participants to five consecutive days of a low (104 colony-forming units (CFU)) or high dose (107 CFU) of L-NTCD03 capsules was safe and well-tolerated with no study-related serious adverse events (SAEs). In stage 3, one participant who had been pretreated with vancomycin for five days and was briefly colonised with L-NTCD03 (two positive faecal samples on D7 and D9), developed acute appendicitis on D14 (10 days after the end of dosing) requiring appendectomy. This event was considered unrelated to the study procedures. A total of 39 gastrointestinal (GI) adverse events (AEs) (Table 1) were reported as related to L-NTCD03 exposure in 60 participants, of which 87% (34/39) were reported as mild, 13% (5/39) moderate, and none severe (Supplementary Table 2). GI AEs were comparable between the high-dose and low-dose L-NTCD03 groups, with 19 GI AEs reported in the low-dose group and 20 GI AEs reported in the high-dose group (severity of low-dose: 74% (14/19) mild, 26% (5/19) moderate and high-dose: 100% (20/20) mild, Supplementary Table 2). In the placebo group (N = 9) a total of four GI AEs were reported, of which 75% (3/4) were considered mild and 25% (1/4) severe (abdominal pain spontaneous resolution within two days). Oral vancomycin pretreatment (one day in stage 2 (N = 22) and five days in stage 3 (N = 23)) was associated with the following solicited events: abdominal pain 9% (4/45), diarrhoea 7% (3/45), change in bowel habit 4% (2/45), nausea 4% (2/45), dyspepsia 2% (1/45) and flatulence 2% (1/45) (Supplementary Table 2).Table 1Overview of AEs related to NTCD spore ingestionAEs related to L-NTCD03 spores N (%)Stage 1Stage 2Stage 3TotalPlaceboLow doseHigh doseLow doseHigh doseLow doseHigh doseN = 10N = 10N = 10N = 10N = 10N = 10N = 60N = 9SolicitedAbdominal pain2 (20)3 (30)3 (30)2 (20)1 (10)2 (20)13 (22)2 (22)Change in bowel habit1 (10)-2 (20)-1 (10)1 (10)5 (8)-Diarrhoea--1 (10)-1 (10)2 (20)4 (7)-Dyspepsia-1 (10)--1 (10)2 (20)4 (7)-Flatulence1 (10)2 (20)--1 (10)1 (10)5 (8)1 (11)Nausea-1 (10)2 (20)1 (10)1 (10)-5 (8)-Constipation-------1 (11)UnsolicitedLow back pain-1 (10)----1 (2)-Abdominal distension-1 (10)----1 (2)-Burning sensation perianal----1 (10)-1 (2)-AEs are presented as the number and proportion (N(%)) of participants with a certain adverse event related to NTCD spore ingestion or placebo. Table 1Overview of AEs related to NTCD spore ingestionAEs related to L-NTCD03 spores N (%)Stage 1Stage 2Stage 3TotalPlaceboLow doseHigh doseLow doseHigh doseLow doseHigh doseN = 10N = 10N = 10N = 10N = 10N = 10N = 60N = 9SolicitedAbdominal pain2 (20)3 (30)3 (30)2 (20)1 (10)2 (20)13 (22)2 (22)Change in bowel habit1 (10)-2 (20)-1 (10)1 (10)5 (8)-Diarrhoea--1 (10)-1 (10)2 (20)4 (7)-Dyspepsia-1 (10)--1 (10)2 (20)4 (7)-Flatulence1 (10)2 (20)--1 (10)1 (10)5 (8)1 (11)Nausea-1 (10)2 (20)1 (10)1 (10)-5 (8)-Constipation-------1 (11)UnsolicitedLow back pain-1 (10)----1 (2)-Abdominal distension-1 (10)----1 (2)-Burning sensation perianal----1 (10)-1 (2)-AEs are presented as the number and proportion (N(%)) of participants with a certain adverse event related to NTCD spore ingestion or placebo. Overview of AEs related to NTCD spore ingestion AEs are presented as the number and proportion (N(%)) of participants with a certain adverse event related to NTCD spore ingestion or placebo. We found no clear dose-response relationship in colonisation rates (defined by having at least two C. difficile-positive faecal samples between three and 14 days after the last L-NTCD03 exposure) after exposure to low-dose or high-dose L-NTCD03 spores for five consecutive days (low-dose: 40% (12/30) colonised, high-dose: 39% (11/28) colonised). Rather, the duration of antibiotic pretreatment appeared to drive colonisation efficiency (Table 2). For this reason, the low-dose and high-dose L-NTCD03 groups were analysed together to compare colonisation rates between the different stages. Exposure to five days of L-NTCD03 spores alone resulted in 5% (1/20) colonisation with C. difficile (stage 1), which increased to 32% (6/19) following a one-day vancomycin pretreatment (stage 2) and 84% (16/19) following a five-day vancomycin pretreatment (stage 3) (Table 2). Persistent colonisation (defined as a positive gluD qPCR or positive culture for more than seven days, starting three days after the last exposure day (D7)) was observed in 78% (18/23) of all C. difficile-colonised participants (0% (0/1) in stage 1, 83% (5/6) in stage 2 and 81% (13/16) in stage 3). At the three-month follow-up timepoint (D84), 6% (4/67) of the participants still had a faecal sample positive for C. difficile (two in stage 2 and two in stage 3, Fig. 3). All four colonised participants showed spontaneous decolonisation within 10 months from study start (follow-up at 7 months: 2/4 decolonised; follow-up at 10 months: 4/4 decolonised).Fig. 3Colonisation patterns of colonised participants from the high- and low-dose L-NTCD03 groups.The figure includes only participants (N = 23) who became colonised after exposure to high- or low-dose L-NTCD03 (so does not include the one colonised placebo participant). Participant numbers are plotted on the y-axis and days since start of L-NTCD03 dosing on the x-axis. Each coloured box represents a faecal sample that tested positive for C. difficile. Blue boxes indicate samples positive for the challenge strain (L-NTCD03) and orange boxes indicate samples positive for a non-challenge C. difficile strain. Numbers within the orange boxes denote the corresponding C. difficile ribotypes (RT): 1 = RT043, 2 = RT001, 3 = RT056, 4 = RT010, 5 = RT216, 6 = RT070, 7 = RT295, 8 = RT258. Ribotyping was restricted to culture-positive samples; no ribotype data were available for culture-negative samples. Source data are provided as a Source Data file.Table 2Colonisation rate per study stageIntervention groupsColonisationproportion (number)Stage 1No vancomycin (N = 24)Stage 21D vancomycin (N = 21)Stage 35D vancomycin (N = 22)Low L-NTCD03 dose0% (0/10)40% (4/10)80% (8/10)High L-NTCD03 dose10% (1/10)22% (2/9)88% (8/9)Total attack rate5% (1/20)32% (6/19)84% (16/19)Placebo0% (0/4)0% (0/2)33% (1/3)Percentage of colonisation with in brackets the number of C. difficile colonised individuals per group (defined as having a positive faecal sample for C. difficile (gluD qPCR or culture) on at least two different timepoints between 3 and 14 days after the last L-NTCD03 exposure day). Bold values represent the combined colonisation (attack) rate of the high- and low-dose L-NTCD03 groups within each stage. Stage 1 participants received a low (104 CFU) or high dose (107 CFU) of L-NTCD03 spores or placebo for 5D, without vancomycin pretreatment. Stage 2 participants first received 1D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D, stage 3 participants first received 5D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D. Colonisation patterns of colonised participants from the high- and low-dose L-NTCD03 groups. The figure includes only participants (N = 23) who became colonised after exposure to high- or low-dose L-NTCD03 (so does not include the one colonised placebo participant). Participant numbers are plotted on the y-axis and days since start of L-NTCD03 dosing on the x-axis. Each coloured box represents a faecal sample that tested positive for C. difficile. Blue boxes indicate samples positive for the challenge strain (L-NTCD03) and orange boxes indicate samples positive for a non-challenge C. difficile strain. Numbers within the orange boxes denote the corresponding C. difficile ribotypes (RT): 1 = RT043, 2 = RT001, 3 = RT056, 4 = RT010, 5 = RT216, 6 = RT070, 7 = RT295, 8 = RT258. Ribotyping was restricted to culture-positive samples; no ribotype data were available for culture-negative samples. Source data are provided as a Source Data file. Table 2Colonisation rate per study stageIntervention groupsColonisationproportion (number)Stage 1No vancomycin (N = 24)Stage 21D vancomycin (N = 21)Stage 35D vancomycin (N = 22)Low L-NTCD03 dose0% (0/10)40% (4/10)80% (8/10)High L-NTCD03 dose10% (1/10)22% (2/9)88% (8/9)Total attack rate5% (1/20)32% (6/19)84% (16/19)Placebo0% (0/4)0% (0/2)33% (1/3)Percentage of colonisation with in brackets the number of C. difficile colonised individuals per group (defined as having a positive faecal sample for C. difficile (gluD qPCR or culture) on at least two different timepoints between 3 and 14 days after the last L-NTCD03 exposure day). Bold values represent the combined colonisation (attack) rate of the high- and low-dose L-NTCD03 groups within each stage. Stage 1 participants received a low (104 CFU) or high dose (107 CFU) of L-NTCD03 spores or placebo for 5D, without vancomycin pretreatment. Stage 2 participants first received 1D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D, stage 3 participants first received 5D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D. Colonisation rate per study stage Percentage of colonisation with in brackets the number of C. difficile colonised individuals per group (defined as having a positive faecal sample for C. difficile (gluD qPCR or culture) on at least two different timepoints between 3 and 14 days after the last L-NTCD03 exposure day). Bold values represent the combined colonisation (attack) rate of the high- and low-dose L-NTCD03 groups within each stage. Stage 1 participants received a low (104 CFU) or high dose (107 CFU) of L-NTCD03 spores or placebo for 5D, without vancomycin pretreatment. Stage 2 participants first received 1D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D, stage 3 participants first received 5D of vancomycin pretreatment (250 mg four times daily) followed after 7D by a low or high dose of L-NTCD03 spores or placebo for 5D. To assess the identity of the C. difficile strains recovered from the study participants, we performed capillary PCR ribotyping (Fig. 3). We found that the majority of the colonised participants (70%, 16/23) were colonised with the challenge strain L-NTCD03 (PCR ribotype 416). Among participants who received placebo (N = 9), one participant, pretreated with five days of vancomycin, was briefly colonised with L-NTCD03 (based on two positive faecal samples on D7 and D9), indicating transmission between participants (with no shared households) in the outpatient setup. Remarkably, 30% (7/23) of the colonised participants tested positive for C. difficile strains with a PCR ribotype inconsistent with the challenge strain (hereafter termed: non-challenge strains). Non-challenge strains were identified in seven participants; one participant of stage 2 and six participants of stage 3. Most participants who tested positive for non-challenge strains (71%, 5/7) already had a positive faecal sample at D0 (the baseline sample, collected seven days after the last vancomycin dosing and before the first L-NTCD03 exposure). Subsequently, of those colonised with non-challenge strains, 29% (2/7) failed to get colonised with L-NTCD03 despite repeat-dosing, whereas 71% (5/7) had intermittently positive samples for both a non-challenge strain and L-NTCD03. Notably, based on Ct values measured by gluD qPCR over time, bacterial loads were comparable between L-NTCD03 colonised participants and those colonised with non-challenge strains or a mixture of a non-challenge strain and L-NTCD03 (Supplementary Fig. 2). The toxigenic status of the isolated non-challenge strains was assessed using a tcdB qPCR on total DNA extracted from faeces and a toxin gene multiplex PCR on DNA of the cultured isolate (Supplementary Table 3). Among the seven participants colonised with a non-challenge strain, eight non-challenge isolates were recovered, of which 88% (7/8) were found to be toxigenic (tcdA + /tcdB + /cdt-) (Supplementary Table 3a). Two of the faecal samples from which toxigenic strains (belonging to PCR ribotypes 216 and 295, both in stage 3) were isolated, tested positive for free toxins using the ViDAS® C. difficile TOXIN A & B assay (Supplementary Table 3b). The two participants colonised with these ribotypes reported mild to moderate GI symptoms (one with diarrhoea and subfebrile temperature, and the other with nausea, dyspepsia, flatulence and abdominal pain) that were transient and self-limiting, with spontaneous decolonisation. Additionally, in participants who did not meet the definition of colonisation, toxigenic ribotypes were detected in 3/13 (23%) in stage 2 and 1/3 (33%) in stage 3. In addition, we quantified faecal vancomycin levels during its wash-out period (D-5, D-3 and D0) in stage 3 participants to assess the association of faecal vancomycin clearance with C. difficile colonisation outcomes following five days of vancomycin pretreatment (D-11 to D-7). For 77% (17/22) of stage 3 participants, faecal vancomycin concentrations exceeded the limit of detection at D-5 and/or D-3, with marked variability in clearance between participants (Supplementary Fig. 3). In all participants, vancomycin was fully cleared by D0 (before first L-NTCD03 exposure), seven days after the last vancomycin dosing (D-7). Exploratory subgroup analysis stratified by C. difficile colonisation status (Fig. 4a) suggested higher faecal vancomycin concentrations at D-5 and D-3 in colonised participants compared with non-colonised participants, driven by a subset of participants with slow clearance; though these differences were not statistically significant, likely reflecting the small sample sizes (two-tailed Mann-Whitney U test: D-5, U = 19, p = 0.07; D-3, U = 36, p = 0.60). Stratification by C. difficile status before the first L-NTCD03 exposure (D0) (Fig. 4b) indicated lower faecal vancomycin concentrations at D-5 and D-3 in participants who were already C. difficile-positive (e.g. for a non-challenge strain) at D0, compared with those who were C. difficile-negative at D0, with a statistically significant difference observed at D-5 (two-tailed Mann-Whitney U test: D-5, U = 19, p = 0.03; D-3, U = 32.5 p = 0.19).Fig. 4Subgroup analysis of faecal vancomycin concentrations in stage 3 participants.Subgroup analysis of faecal vancomycin concentrations in stage 3 participants stratified by C. difficile colonisation status (a) and C. difficile positivity prior to L-NTCD03 exposure on D0 (b). a Concentrations measured on day −5 (D-5), day −3 (D-3) and day 0 (D0) in colonised (filled diamonds, N = 17) and non-colonised (open diamonds, N = 5) stage 3 participants. Two-tailed Mann–Whitney U test: D-5, U = 19, p = 0.07; D-3, U = 36, p = 0.60; b concentrations measured on day −5 (D-5), day −3 (D-3) and day 0 (D0) in C. difficile-positive (filled circles, N = 6) and C. difficile-negative (open circles, N = 16) stage 3 participants prior to L-NTCD03 exposure on D0. Two-tailed Mann–Whitney U test: D-5, U = 19, p = 0.03; D-3, U = 32.5 p = 0.19. Stage 3 participants were pretreated with vancomycin for five days (250 mg, four times daily) from day −11 (D-11) to day −7 (D-7). Each symbol represents an individual measurement; horizontal bars indicate median values. The lower limit of detection (LLOD) was 0.01 ug/mL; values below the LLOD were assigned 0.005 ug/mL (half of the LLOD). Source data are provided as a Source Data file. Subgroup analysis of faecal vancomycin concentrations in stage 3 participants. Subgroup analysis of faecal vancomycin concentrations in stage 3 participants stratified by C. difficile colonisation status (a) and C. difficile positivity prior to L-NTCD03 exposure on D0 (b). a Concentrations measured on day −5 (D-5), day −3 (D-3) and day 0 (D0) in colonised (filled diamonds, N = 17) and non-colonised (open diamonds, N = 5) stage 3 participants. Two-tailed Mann–Whitney U test: D-5, U = 19, p = 0.07; D-3, U = 36, p = 0.60; b concentrations measured on day −5 (D-5), day −3 (D-3) and day 0 (D0) in C. difficile-positive (filled circles, N = 6) and C. difficile-negative (open circles, N = 16) stage 3 participants prior to L-NTCD03 exposure on D0. Two-tailed Mann–Whitney U test: D-5, U = 19, p = 0.03; D-3, U = 32.5 p = 0.19. Stage 3 participants were pretreated with vancomycin for five days (250 mg, four times daily) from day −11 (D-11) to day −7 (D-7). Each symbol represents an individual measurement; horizontal bars indicate median values. The lower limit of detection (LLOD) was 0.01 ug/mL; values below the LLOD were assigned 0.005 ug/mL (half of the LLOD). Source data are provided as a Source Data file. Secondary outcome: microbiota analyses As the duration of vancomycin pretreatment was the main factor determining colonisation success, differences in microbiome composition between one-day (1D) and five-day (5D) vancomycin pretreatment were evaluated using shotgun metagenomics. Both 1D and 5D vancomycin pretreatment resulted in a marked reduction in alpha diversity. In the 1D group, the Shannon index decreased (D-7 to D0) from 3.39 ± 0.62 to 2.78 ± 0.46 (p < 0.001), while in the 5D group it decreased (D-11 to D0) from 3.39 ± 0.65 to 2.30 ± 0.36 (p < 0.001). To account for baseline differences, we performed an Analysis of Covariance (ANCOVA). The analysis confirmed that the duration of vancomycin pretreatment had a significant effect on the alpha diversity at D0 (p = 0.001). The baseline-adjusted mean Shannon index was significantly lower in the 5D group (2.30 ± 0.42) compared to the 1D group (2.75 ± 0.41). In addition to the loss of diversity, we observed a pronounced shift in microbiome composition in both groups. Principal Coordinates Analysis (PCoA) of unweighted UniFrac distances showed distinct clustering by time point and vancomycin pretreatment duration (Fig. 5a). PERMANOVA revealed a significant overall effect of duration of vancomycin pretreatment and time (R² = 0.35, p < 0.001). Within-group analyses confirmed significant temporal changes in both 1D (R² = 0.19, p < 0.001) and 5D (R² = 0.35, p < 0.001) vancomycin pretreated groups. Particularly, for the 5D vancomycin group, microbiome composition showed progressive recovery during the vancomycin wash-out period (D-5, D-3, D0), yet composition still differed significantly between baseline and both the 1D and 5D vancomycin pretreated group at D0 (first day of L-NTCD03 exposure) (R² = 0.10, p < 0.001). These results demonstrate that even a single day of vancomycin exposure leads to a marked change in microbiome composition that affects colonisation resistance towards C. difficile.Fig. 5Bacterial beta-diversity and colonisation-associated taxa.Bacterial beta-diversity in relation to vancomycin pretreatment duration and sample timepoints (a) and LEfSe (Linear discriminant analysis Effect Size) analysis identifying taxa associated with colonisation (b). a Principal Coordinate Analysis (PCoA) of unweighted UniFrac distances, with each dot representing a sample. Open circles represent 1D vancomycin pretreatment, and closed circles represent 5D. Sample timepoints are colour-coded as follows: red, baseline (stage 2: D-7; stage 3: D-11); green, D-5 (stage 3), blue, D-3 (stage 3); and purple D0 (stages 2 and 3). Dashed lines (---) show 95% confidence intervals assuming a normal distribution, and solid lines (──) indicate 95% confidence ellipses. Differences in community composition were tested by PERMANOVA (permutational multivariate analysis of variance, 999 permutations; two-sided by construction, as the permutation-based pseudo-F test is non-directional). For each test below, the reported p-value corresponds to the minimum achievable with 999 permutations (p = 1/1000). Overall effect of vancomycin pretreatment duration and timepoint: R² = 0.35, p = 0.001. PERMANOVA within treatment groups: Vancomycin 1D, R² = 0.19, p = 0.001; Vancomycin 5D, R² = 0.35, p = 0.001. PERMANOVA at D0 comparing Vancomycin 1D vs. 5D, R² = 0.10, p = 0.001. Source data are provided as a Source Data file. b Microbiome composition at D0, following vancomycin pretreatment (1D or 5D), influences colonisation with C. difficile. Differentially abundant taxa were identified using LEfSe, which applies a non-parametric factorial Kruskal–Wallis rank-sum test (two-sided) followed by pairwise Wilcoxon rank-sum tests (two-sided) and Linear Discriminant Analysis (LDA). No adjustment for multiple comparisons was applied. Features with Kruskal–Wallis p < 0.05 and LDA score ≥ 2 are shown; bars indicate effect sizes (LDA scores). Source data are provided as a Source Data file. The full list of all taxa LEfSe attempted to test are provided in Supplementary Data file 1. Bacterial beta-diversity and colonisation-associated taxa. Bacterial beta-diversity in relation to vancomycin pretreatment duration and sample timepoints (a) and LEfSe (Linear discriminant analysis Effect Size) analysis identifying taxa associated with colonisation (b). a Principal Coordinate Analysis (PCoA) of unweighted UniFrac distances, with each dot representing a sample. Open circles represent 1D vancomycin pretreatment, and closed circles represent 5D. Sample timepoints are colour-coded as follows: red, baseline (stage 2: D-7; stage 3: D-11); green, D-5 (stage 3), blue, D-3 (stage 3); and purple D0 (stages 2 and 3). Dashed lines (---) show 95% confidence intervals assuming a normal distribution, and solid lines (──) indicate 95% confidence ellipses. Differences in community composition were tested by PERMANOVA (permutational multivariate analysis of variance, 999 permutations; two-sided by construction, as the permutation-based pseudo-F test is non-directional). For each test below, the reported p-value corresponds to the minimum achievable with 999 permutations (p = 1/1000). Overall effect of vancomycin pretreatment duration and timepoint: R² = 0.35, p = 0.001. PERMANOVA within treatment groups: Vancomycin 1D, R² = 0.19, p = 0.001; Vancomycin 5D, R² = 0.35, p = 0.001. PERMANOVA at D0 comparing Vancomycin 1D vs. 5D, R² = 0.10, p = 0.001. Source data are provided as a Source Data file. b Microbiome composition at D0, following vancomycin pretreatment (1D or 5D), influences colonisation with C. difficile. Differentially abundant taxa were identified using LEfSe, which applies a non-parametric factorial Kruskal–Wallis rank-sum test (two-sided) followed by pairwise Wilcoxon rank-sum tests (two-sided) and Linear Discriminant Analysis (LDA). No adjustment for multiple comparisons was applied. Features with Kruskal–Wallis p < 0.05 and LDA score ≥ 2 are shown; bars indicate effect sizes (LDA scores). Source data are provided as a Source Data file. The full list of all taxa LEfSe attempted to test are provided in Supplementary Data file 1. To identify taxa at D0 that influenced colonisation outcome regardless of vancomycin pretreatment (Fig. 5b), Linear Discriminant Analysis Effect Size (LEfSe) was used. Similar results were obtained when analysing only the participants who received 5D vancomycin pretreatment (Supplementary Fig. 1). Using an LDA score threshold of 2 and a significance level of p < 0.05, several taxa were found to be associated with colonisation outcome. Notably, the presence of Enterococcus sp. and Enterobacterales were associated with colonisation, whereas the genera Phocaeicola, Alistipes, Roseburia, Eubacterium, Dorea, and the species Clostridium scindens were associated with a non-colonised status, consistent with literature14–18. This demonstrates that despite a modest number of volunteers and large microbiome changes resulting from antibiotic pretreatment, meaningful results can be obtained from controlled colonisation experiments. Discussion In this study, we have demonstrated that administration of a low or high dose of L-NTCD03 spores over five consecutive days to healthy adult participants was safe and well tolerated, resulting in 5% colonisation without vancomycin pretreatment, 32% colonisation following a one-day vancomycin pretreatment and 84% colonisation following a five-day vancomycin pretreatment. No dose-response relationship was observed between L-NTCD03 inoculum dosing and colonisation rate; only vancomycin pretreatment was a primary determinant of colonisation success. An interesting observation was the increased susceptibility of healthy participants to colonisation with non-challenge C. difficile strains following vancomycin pretreatment, even after just one day of exposure. This could be explained by two, not mutually exclusive, mechanisms: vancomycin-induced disruption of the gut microbiota may 1) increase susceptibility to colonisation by environmental C. difficile strains, and/or 2) promote germination and outgrowth of (intracellular) C. difficile spores19 present below the limit of detection in the gastrointestinal tract. Although pre-existing colonisation below the limits of detection of our assays cannot formally be excluded, we deem the first scenario more likely as all participants tested negative for C. difficile by PCR and enrichment culture at screening and asymptomatic colonisation in healthy adults is reported to be 4–15%20. Measurements of faecal vancomycin levels revealed substantial interindividual variability in intestinal vancomycin clearance. Exploratory subgroup analyses indicated that participants who cleared vancomycin more rapidly are more likely to have a positive faecal sample for a non-challenge strain on D0 (Fig. 4b), which may hinder subsequent colonisation by the challenge C. difficile strain (Fig. 3). We hypothesise that vancomycin clearance rate determines colonisation outcomes by shaping the window between microbiota disruption and residual inhibitory vancomycin levels. Rapid clearance may create a transient period in which vancomycin concentrations fall below inhibitory levels while the microbiota remains disrupted, allowing non-challenge C. difficile strains to establish colonisation, whereas slower clearance maintains inhibitory levels and delays colonisation until vancomycin is washed out (after seven days) and L-NTCD03 challenge is initiated. These findings underscore the importance of administering the challenge strain during the susceptible vancomycin washout period (ideally immediately after discontinuation of vancomycin) in controlled human C. difficile challenge models to prevent acquisition of non-challenge strains as vancomycin levels decline. Remarkably, most of the non-challenge C. difficile strains were toxigenic (tcdA + /tcdB + /cdt-) and represented ribotypes that are also commonly reported in the Dutch national human CDI surveillance programme (0.19–3.8% of typed isolates in the period 2023–2025)21,22 and are also commonly found in Europe23,24. Free toxin was detected in faecal samples from two participants harbouring strains belonging to two different PCR ribotypes (PCR ribotype 216 and 295) (Supplementary Table 3). The two participants colonised with these strains experienced only mild to moderate gastrointestinal symptoms, which were self-limiting, highlighting the mild and transient nature of toxigenic C. difficile infection in healthy young individuals. This observation lends confidence to the feasibility of a controlled human infection model (CHIM) in which healthy participants are exposed to toxigenic C. difficile (TCD) with the goal of inducing mild clinical disease. Although a TCD CHIM has not yet been developed, insights from our NTCD human colonisation model could critically inform its design, providing a valuable and essential platform to accelerate the development of novel (toxin-directed) medicinal products and vaccines. Our findings suggest that dissemination of C. difficile in the environment allows for rapid colonisation of individuals treated with antibiotics. Moreover, it also raises the possibility that mild diarrhoea after antibiotic treatment, even in young healthy individuals, may be attributable to transient C. difficile rather than antibiotic side effects. Though greater awareness could potentially improve diagnosis and guide treatment decisions, and strengthen surveillance programmes, further studies are needed to confirm and strengthen our findings. Lastly, our results may help guide efforts to optimise the duration of antibiotic therapy in populations at high-risk of CDI. We observed that colonisation frequency increased substantially with longer antibiotic pretreatment (one day versus five days). In line with this, microbiome restoration occurred more rapidly following one day of antibiotic pretreatment compared with five days. These findings suggest that antibiotic treatment duration may be a critical and potentially modifiable risk factor for environmental C. difficile acquisition. The NTCD colonisation model can provide a unique and controlled platform to systematically investigate antibiotic-specific effects on colonisation risk. By comparing different antibiotic regimens or treatment durations, followed by (standardised) NTCD challenge, it becomes possible to quantify colonisation susceptibility alongside antibiotic clearance kinetics and microbiome restoration dynamics. Such studies could inform antimicrobial stewardship strategies and refine risk stratification for C. difficile acquisition in clinical practice. Comparison of our study with the pioneering first-in-human NTCD trial11 reveals notable methodological differences. Our study employed capsule-based NTCD spore administration (rather than a liquid formulation) to ensure controlled dosing, which was produced according to GMP standards13. The dosing regimens differed for both NTCD spores and vancomycin pretreatment. In our study, participants received once-daily 10⁴ or 10⁷ CFU NTCD spores for five days, rather than single-dose regimens, twice-daily dosing for five days, or once-daily dosing for 14 days. Vancomycin pretreatment also differed, consisting of either one or five days of 250 mg four times daily given seven days before NTCD exposure instead of five days of 125 mg four times daily administered one day before NTCD exposure. Most importantly, we incorporated frequent faecal sampling to provide high-resolution insights into C. difficile colonisation dynamics, extended the follow-up period to day 84 (rather than day 28) and included genotypic characterisation by PCR ribotyping and faecal vancomycin analyses alongside microbiota profiling to identify susceptibility markers. Despite these methodological differences, we similarly observed no dose-response relationship and stable C. difficile colonisation only after vancomycin pretreatment. Interestingly, placebo recipients in both studies became colonised with the challenge strain following vancomycin pretreatment, despite a different setting (inpatient versus outpatient), underscoring the high transmissibility of C. difficile among susceptible individuals. However, Villano et al. reported a lower susceptibility to non-challenge C. difficile strains after a five-day vancomycin pretreatment (6%, 2/36), compared with our study (32%, 7/22) (two-sided Fisher’s exact test, p = 0.02). This discrepancy may reflect differences in study setting (outpatient versus inpatient), influencing exposure to environmental C. difficile strains. Other contributing factors could include increased detection of C. difficile due to more frequent faecal sampling, higher assay sensitivity (qPCR plus enrichment culture instead of a regular culture), more extensive strain characterisation, the higher vancomycin dose used in our study (250 mg four times daily versus 125 mg) and the longer wash-out period between vancomycin dosing and NTCD challenge exposure. We acknowledge that the young, healthy study population in our trial does not reflect a typical (elderly) population at high risk for CDI, and therefore, the findings from this study may not be generalisable to all patient populations. For example, compared with younger adults, elderly individuals generally exhibit reduced gut microbiota diversity, lower abundance of beneficial commensal bacteria and a relative increase in opportunist organisms25. However, the ethical frameworks for intervention studies with healthy volunteers dictate that risks of (severe) adverse events should be minimised, while still allowing assessment of inter-individual variability in colonisation resistance and susceptibility markers. Given the novelty of controlled human colonisation models for C. difficile, it is good practice to choose a low-risk clinical trial design. Moreover, the rising incidence of community-acquired CDI (CA-CDI) has shifted disease demographics toward younger individuals26. For instance, in a recent population-based study, the median age of patients with CA-CDI was 50, compared to 72 years for patients with hospital-acquired CDI27. Nonetheless, the excellent safety profile of L-NTCD03 in the current study suggests that administration in an elderly population, within the context of a controlled human challenge study, may be feasible, provided that the indication for and timing of antibiotic pretreatment and L-NTCD03 administration are carefully considered. This includes evaluation of the need for vancomycin pretreatment and its associated safety considerations before extrapolating the current protocol to the elderly population. The favourable safety profile of L-NTCD03 also suggests a potential, early-stage avenue for exploring L-NTCD03 as a preventive therapeutic, although this application remains preliminary. To conclude, in our NTCD human colonisation model, which combines controlled capsule dosing, frequent faecal sampling, in-depth C. difficile characterisation, and microbiota analysis, we show that vancomycin-induced disruption of the microbiota is a primary driver of C. difficile colonisation. Moreover, we reveal the pervasive environmental dissemination of C. difficile and highlight the clinical relevance of these findings, providing robust tools to enhance the prevention and management of this globally impactful infection. Methods Study design and recruitment From September 2023 to February 2025, an adaptive dose-design, randomised, double-blind placebo-controlled clinical trial was conducted at Leiden University Medical Centre (LUMC), Leiden, The Netherlands. A total of 69 healthy participants aged 18-45 years, who were not colonised with C. difficile before the study and did not use antibiotics (or other microbiota-influencing products) within three months before inclusion, were enrolled after a health assessment (Fig. 1, please refer to Supplementary Table 5 for the full inclusion and exclusion criteria). The health assessment, performed between three months and one week before study start, included a medical history review, physical examination, general laboratory tests (including haematology and biochemistry) and faecal sample analysis to rule out C. difficile as well as Salmonella, Shigella, Campylobacter, Yersinia, Aeromonas and Plesiomonas shigelloides species28. The gender of participants was self-reported. Participants who developed diarrhoea or used antibiotics or other microbiota- or immune-altering medications between screening and study start were excluded. Female participants were counselled to use adequate contraception throughout the study and were tested for pregnancy with a urine pregnancy test before the first L-NTCD03/placebo exposure (D0). All participants needed to provide written informed consent. The study consisted of three consecutive intervention stages (Fig. 2); the dosing schedule of the second and third stages was based upon the (colonisation) results of the previous stages, trying to reach at least a 60% C. difficile colonisation rate. In stage 1, 24 participants were included and randomised into three groups: ten received a low dose (104 CFU) of L-NTCD03 spores for five days, ten received a high dose (107 CFU) of L-NTCD03 spores for five days, and four received a placebo for five days. In stage 2, 22 participants (instead of the intended 23 participants, see Fig.1 for details) were included and randomised into three groups. All participants first received one day of vancomycin pretreatment (250 mg four times daily), followed after a seven-day antibiotic wash-out period by five days of a low dose of L-NTCD03 spores (n = 10), a high dose of L-NTCD03 spores (n = 10) or placebo (n = 2). In stage 3, 23 participants were included and randomised, first receiving an extended vancomycin pretreatment of five days (250 mg four times daily), followed by a seven-day interval, by five days of a low dose of L-NTCD03 spores (n = 10), a high dose (n = 10) or placebo (n = 3). The participants visited the trial centre on the first day of vancomycin pretreatment (stage 2 and 3) and daily during the days of spores or placebo ingestion (D0-D4), with faeces collection before ingestion (i.e., the faeces sample of D0 is the baseline sample). Both participants and investigators were blinded to the intervention. After the last exposure day (D4), the participants visited the trial centre until D28 three times per week for faecal sample collection (for C. difficile diagnostics and microbiota analyses), once a week for AE collection (on D7, D14, D21 and D28) and two times in total for a safety blood test (on D14 and D28). After three months (D84), there was a final follow-up visit for faecal sample collection; participants who still tested positive for C. difficile at that time were asked to return for follow-up until decolonisation was confirmed. Rescue antibiotics were available in case of unexpected AEs. Asymptomatic C. difficile colonisation was not treated with antibiotics in line with international clinical guidelines29,30. AEs were recorded by participants in a diary and graded in four categories (mild, moderate, severe and serious) that were prespecified per protocol (Supplementary Table 6). The C. difficile capsule jars were prepared by laboratory staff, independent of the clinical investigators. Randomisation was done by an independent data manager. The vancomycin pretreatment in stages 2 and 3 consisted of open-label oral vancomycin (Vancomycine Strides 250 mg hard capsules; vancomycine hydrochloride; Strides Pharma). Oral vancomycin was chosen as antibiotic pretreatment based on its safety profile and prior successful experience in an experimental NTCD model11. Primary outcomes were safety and C. difficile colonisation (aiming for ≥ 60% attack rate). C. difficile colonisation was defined as having a positive faecal sample for C. difficile (gluD qPCR or culture) on at least two different time points between three and 14 days after the last day of L-NTCD03 exposure. Secondary outcomes were microbiota markers associated with successful C. difficile colonisation. Data capture was done using an electronic case report form (Castor CDMS v.2023.4.x.x.). The trial protocol was approved by the local ethics committee of Leiden, The Hague and Delft (METC LDD, file number NL83949.058.23) and registered at ClinicalTrials.gov (NCT05693077). NTCD spore production The production of C. difficile spore capsules for experimental human administration has been described previously13. The manufacturing process aligns with Good Manufacturing Practices (GMP) guidelines. Production was performed at the Leiden University Centre for Infectious Diseases (LUCID) Department of the LUMC, and the produced challenge material was released by a LUMC Qualified Person (QP). The NTCD strain used in this trial is L-NTCD03 (PCR ribotype 416, ST 39, clade 4, complete genome GenBank NZ_OX637968.1), which has been extensively characterised31. L-NTCD03 spores were administered in direct-release liquid-filled capsules (Lonza Vcaps® Plus Capsules size 0, Lonza, Colmar, France) for oral administration. Two different doses of spore capsules were produced: 104 and 107 CFU spores, formulated in buffered 70% glycerol in direct-release liquid-filled capsules. Placebo capsules were produced, consisting of identical capsules filled with 500ul buffered 70% glycerol. Production of the L-NTCD03 and placebo capsules underwent strict quality control (QC), independently reviewed by a QC officer, before release by a QP. Exposure to the capsules was done in the trial centre by directly observed treatment (DOT). C. difficile diagnostics C. difficile diagnostics, consisting of gluD qPCR and stool culture, were performed at the LUMC (Leiden, the Netherlands) on faecal samples collected during screening and from D0 onwards (Fig. 2). The gluD qPCR utilises primers specific to the glutamate dehydrogenase (gluD) gene (908CLD_gluDs: 5’- GTCTTGGATGGTTGATGAGTAC-3’; 909CLD_gluDas: 5’-TTCCTAATTTAGCAGCAGCTTC-3’) to amplify a 158-bp fragment, which is detected using a gluD-specific probe (699CLD_GLUD-TQ-FAM: 5’- FAM-AAGCCAGTTGAATTTGGTGG-BHQ1-3’) in a real-time qPCR assay32. Stool culturing was performed after enrichment of stool in CDEB MOD (C. difficile enrichment modified broth, Mediaproducts BV, Groningen, The Netherlands) for five days, followed by anaerobic plating with and without an ethanol shock (1:1) on selective CLO medium (Clostridioides difficile selective agar plates, bioMérieux, Marcy-l'Étoile, France) and incubation for five days. If C. difficile was cultured, a single colony from the culture plate was subjected to capillary PCR ribotyping33 and the pattern was matched to a library of reference patterns using BioNumerics software version 7.6 (bioMéreux, Marcy-l'Étoile, France). Selected non-challenge C. difficile strain samples were analysed for toxigenic status using tcdB qPCR on total DNA extracted from faeces, as well as a toxin gene multiplex PCR on DNA of the cultured isolate and free toxin detection in faeces using the ViDAS® C. difficile TOXIN A & B assay. Microbiome analyses Stool samples were stored at −80 °C and shipped on dry ice to the external service provider (Biomarker Technologies GmbH, Münster, Germany), where DNA extraction using the Tiangen Magnetic Soil and Stool DNA Kit (TIANGEN Biotech, Beijing, China) and shotgun metagenomic sequencing on a BGI platform (2×150 bp paired-end reads) were performed. The targeted sequencing depth was 5 Gb per sample, with an actual average output of 18.5 M paired reads (approximately 5.5 Gb per sample) after quality filtering. To ensure consistent processing, all samples were subjected to additional quality filtering and host read removal using KneadData (v0.12.0)34. Trimming was performed with Trimmomatic (v0.39; SLIDINGWINDOW:4:20 MINLEN:50), and host reads were removed using Bowtie2 (v2.4.2) against the hg37_v0.1 human reference genome. Taxonomic profiling was conducted using MetaPhlAn4 (v4.1.1, mpa_vJun23_CHOCOPhlAnSGB_202403)35. Vancomycin measurements Vancomycin concentrations in stool samples were determined using high performance liquid chromatograph-mass spectrometry/mass spectrometry (HPLC-MS/MS). In brief, 50 mg of a stool sample was suspended in 15 ml water supplemented with 0.1 % formic acid. Samples were vortexed and then subjected to ultra sonification at 35 kHz for 30 min. Next, 50 µl of a sample was extracted with 50 µl of extraction solvent composed of 20 µl ACN, 18.5 µl methanol, 5 µl glipizide (as 1 µg/ml in methanol as internal standard) and 10 µl water containing 10% formic acid. Samples were extracted at 800 rpm on an Eppendorf® MixMate vortex mixer for 15 min and centrifuged for 20 min at 4 °C at 2272 x g. Next, supernatants were transferred to new Greiner 96well-V-bottom plates. Plates were sealed and subjected to analysis. Likewise, a calibration curve including quality control samples was prepared from matrix stool samples. The calibration curve range was selected based on the expected concentration range. HPLC-MS/MS analysis was conducted on an AB Sciex QTrap6500plus mass spectrometer coupled to an Agilent 1290 Infinity II HPLC system. Liquid chromatography conditions were as follows: column: Agilent Zorbax Eclipse Plus C18, 50 × 2.1 mm, 1.8 µm; temperature: 30 °C; injection volume: 5 µL per sample; flow rate: 700 µL/min. Solvents: A: H2O + 0.1% formic acid; solvent B: 95% MeCN/5% H2O + 0.1% formic acid. Gradient was as follows: 99% A at 0 min until 0.1 min. Gradient from 99% A to 0% A until 4 min, 0 % A until 5 min. Mass transitions can be found in Supplementary Table 4. Peaks of samples were quantified using the calibration curve using Multiquant 3.0 software (AB Sciex). The accuracy of the calibration curve was determined using QCs independently prepared on different days. Statistical analysis The intention-to-treat (ITT) population includes all participants who received L-NTCD03 spores or placebo, and is used for assessment of safety outcomes. The per-protocol (PP) population consists of participants who received L-NTCD03 spores or placebo according to the protocol and completed follow-up through day 28; this population is used to assess colonisation frequency and microbiota correlates of colonisation. Baseline characteristics of participants are reported as both means with standard deviations and medians with ranges for continuous variables, and as frequencies with corresponding percentages for categorical variables. AEs are analysed by calculating the proportion of participants in each group who reported mild, moderate or severe adverse events, reported as frequencies with percentages. Colonisation frequency is reported as the total number and proportion of participants complying with the definition of colonisation or persistent colonisation (defined as having a positive gluD qPCR or culture for more than seven days, starting three days after the last exposure day (D7)). Above statistical analyses were performed using SPSS (v.29.0). Statistical analyses of microbiota composition were performed in R for Statistical Computing (version 4.5.1, R Foundation for Statistical Computing, Vienna, Austria)36 with metagenomic data imported using the phyloseq package (v1.52.0)37. Microbiome composition was visualised via Principal Coordinates Analysis (PCoA) based on UniFrac distance matrices. Differences in community structure were assessed using PERMANOVA. Differentially abundant taxa were identified using LEfSe (Linear Discriminant Analysis Effect Size)38. Statistical analyses were performed without imputation of missing values; for paired comparisons, pairs with a missing value at either timepoint were excluded (pairwise deletion). One faecal sample could not be obtained because no material was available; the corresponding cell is flagged as missing in the supplementary data. Faecal vancomycin concentrations and exploratory subgroup differences were analysed and visualised using GraphPad Prism (v.10.2.3). Between-group comparisons were performed using a two-tailed Mann–Whitney U test. These exploratory analyses were not adjusted for multiple comparisons due to their hypothesis-generating nature and limited sample size. Figures were produced in Microsoft Excel (v.2408), PowerPoint (v.2408), R (v. 4.5.1) (for microbiota data) and GraphPad Prism (v.10.2.3). Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1-7 Reporting Summary Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Data 1-7 Reporting Summary Transparent Peer Review file Source data Source Data Source Data