Long-Term Bacterial and Fungal Dynamics following Oral Lyophilized Fecal Microbiota Transplantation in Clostridioides difficile Infection

Clostridioides difficile infection (CDI) is a substantial health concern worldwide, complicated by patterns of increasing antibiotic resistance that may impact primary treatment. Orally administered fecal microbiota transplantation (FMT) is efficacious in the management of recurrent CDI, with specific bacterial species known to influence clinical outcomes.

R ecurrence following Clostridioides difficile infection (CDI) is common following antibiotic therapy, usually within 30 days of treatment cessation (1). Fecal microbiota transplantation (FMT) is now recognized as the preferred treatment in recurrent CDI. It alleviates the microbial imbalance seen in the disease and reduces the risk of recurrences compared with antibiotic therapy (2,3). Lyophilization uses a freeze-drying technique that allows encapsulation of FMT and dose standardization to 10 10 or more bacteria per capsule, along with maintenance of bacterial viability for greater than 7 months following production (4,5). Oral lyophilized FMT is effective in the treatment of recurrent CDI with engraftment of donor microbial species in the recipient (4). While existing evidence suggests particular bacterial species, such as those involved in carbohydrate fermentation to short-chain fatty acids, are important in mediating FMT outcomes (6)(7)(8)(9), information on the role of the mycobiome in disease development and FMT therapeutic outcomes remains limited. Furthermore, little is known about persistent microbial engraftment beyond 12 weeks and whether reduction in donor microbial contribution in the recipient over time will lead to recurrence of disease.
The growing resistance of C. difficile to antibiotics is also a significant health concern (10) since currently recommended therapeutic guidelines for primary CDI involve antibiotic therapy. The role of FMT in the treatment of primary CDI is not established; however, early evidence suggests comparable efficacy with antibiotics (11)(12)(13), indicating that it could serve as an alternate therapeutic strategy.
Here, we report the outcomes of a prospectively enrolled "real world" cohort of consecutive patients with primary and recurrent CDI treated with oral lyophilized FMT and investigate how resultant changes in the bacterial and fungal communities can impact therapeutic outcomes.

RESULTS
Orally administered lyophilized FMT is safe and durably effective for treating CDI. A total of 37 patients received oral FMT for CDI (15 primary CDI, 22 recurrent CDI) with a median follow up of 17 weeks (range, 4 to 26). The mean 6 standard deviation (SD) age was 42.3 6 19.8 years. Twenty (54%) patients were male. Patient and disease characteristics are outlined in Table 1. In total, 33 (89.2%) patients had sustained clinical and biochemical cure to the end of follow up, of whom 3 (9.1%) required a second course of oral FMT. Sustained CDI cure rates were similar between primary and recurrent disease ( Minor gastrointestinal adverse events occurred in 45% of patients, including nausea, abdominal discomfort, constipation, and diarrhea; all events were self-limiting. One patient had fevers following treatment that resolved without intervention. There were no serious adverse events attributable to oral FMT therapy. No patients required colectomy during the follow-up period. Bacterial community in responders to FMT appear to be more susceptible to manipulation. Twenty-three patients (62.2% of patient cohort) and 4 donors (100% of donors) provided a total of 166 fecal samples. Patients (18 responders and 5 nonresponders) were sampled from baseline up to week 26 following FMT therapy when possible. Baseline samples from two patients could not be obtained. Analysis of the bacterial community of responders and nonresponders to FMT as well as their donors identified key differences in the effect of FMT on these groups. In patients who responded to FMT, there was a significantly lower baseline alpha diversity than in donors and a significant increase in alpha diversity measures to levels observed in donors following FMT ( Fig. 1A; see Fig. S2A and B in the supplemental material). These findings were not replicated in the nonresponders ( Fig. 1B; Fig. S2C and D), with a sustained difference in species richness between donors and recipients throughout the examined time points (Fig. S2C). There was a sustained but incomplete shift in bacterial composition in responders toward the donors, with all post-FMT samples being significantly different from baseline samples (P , 0.005 for all, permutational multivariate analysis of variance [PERMANOVA]) as well as donor samples (P , 0.006 for all, PERMANOVA) (Fig. 1C). This again was not replicated in nonresponders, with one patient even showing a shift away from donors (Fig. 1D). The most prominent discriminatory factor between responders and nonresponders was the significant changes in relative abundances of dominant operational taxonomic units (OTUs) toward donor levels in responders (Fig. 1E), which was not observed in nonresponders ( Fig. 1E and F). This result corresponded to the depletion of Enterobacteriaceae and enrichment of Faecalibacterium sp.
(linear discriminant analysis [LDA] score of .4 and P , 0.05) (Fig. 1E). A possible explanation for the resistance to beneficial microbiome manipulation in nonresponders was their higher levels of Ruminococcaceae at baseline and robust enrichment of Bifidobacterium sp. (see Fig. S3 in the supplemental material).
Fungal richness and Penicillium sp. were associated with FMT failure. Differences in the mycobiome between responders and nonresponders were examined. While no difference was observed in b-diversity (composition) metrics (see Fig. S4 in the supplemental material), nonresponders at baseline had a significantly higher relative abundance of one Penicillium taxon than responders (LDA score of 3.68 and P , 0.05) ( Fig. 2A). No other taxon initially identified as different using linear discriminant analysis effect size (LEfSe) survived sensitivity analysis (see Fig. S5 in the supplemental material). There was a borderline nonsignificant (P = 0.072) difference in baseline fungal species richness in responders compared with that of nonresponders (Fig. 2B), with higher richness in responders that decreased with FMT (P = 0.051). Other alpha diversity metrics were not significantly different between responders and nonresponders (see Fig. S6 in the supplemental material).
Donor microbiome engraftment persisted up to 6 months following FMT. Given the increased alpha diversity and compositional shifts in FMT recipients toward the donors, the levels and persistence of donor engraftment after FMT therapy were studied. The following two strategies were adopted: the first was a one-to-one strategy where the specific donors were matched to their recipient, and the second was an allto-one strategy where we did not differentiate between donors. As expected, there was a significantly higher donor contribution following FMT across both strategies ( Fig. 3A; see Fig. S7A in the supplemental material). While there was initial engraftment in both responders and nonresponders, it was more robust in responders and there was a reduction in donor species persistence seen in nonresponders from week 4 to 12 ( Fig. 3B; Fig. S7B). However, this decrease in persistence may be the result of the lower number of samples in nonresponders and not a biological effect. Notably, high levels of donor contribution were seen up to 26 weeks (P , 0.001) following treatment in responders ( Fig. 3; Fig. S7), suggesting stable long-term engraftment of bacteria even with single dose of oral lyophilized FMT. In contrast, fungal source tracking from donor to patient showed no significant increases from baseline to post-FMT (see Fig. S8 in the supplemental material). Donor microbiome and FMT efficacy. FMT was derived from four individual unrelated and unmatched donors ( Table 2). Across the whole cohort (n = 37), each donor provided FMT for at least 1 treatment failure. The most regular donor who provided 57% of the treatments had 2 episodes of treatment failure, of which both were in the setting of further antibiotic therapy. Given this result, donor microbial features associated with therapy outcomes were assessed. Species evenness and Shannon's diversity were similar across donors and were relatively stable across time ( Fig. 1A; Fig. S2B). Some variability in species richness was observed at an inter-but not intradonor level (Fig. S2A). These differences were not associated with treatment response. Donors clustered independently according to their composition (Fig. 4A) but did not appear to influence therapy outcome. Despite this finding, a common feature across donor FMTs contributing to treatment failure was significantly higher levels of taxa from Bacteroides sp. (98.81% sequence similarity to Bacteroides vulgatus) and Clostridium XIVa (LDA score of .3.5 and P , 0.05) (Fig. 4B), with the former likely reflecting lower levels of Firmicutes sp. in these samples.
Primary CDI showed more pronounced microbiome shifts than recurrent CDI. Primary and recurrent CDI had similar clearance rates post-FMT. The baseline microbiomes of these patients and the effects of FMT were compared, with the analysis . Both patients with primary CDI and recurrent CDI responded similarly, with increases in bacterial alpha diversity measures and a shift in beta diversity toward the donor following FMT (Fig. 5; Fig. S9). However, the changes in bacterial community measures in primary CDI appeared to be more robust than those in recurrent CDI (P , 0.015 for all, PERMANOVA) (Fig. 5). Next, features that could discriminate between primary and recurrent CDI were assessed, and the relative abundance of a Veillonella taxon was identified as a strong marker of primary CDI (LDA score of .4 and P , 0.05) (Fig. 5C). This taxon was also found to be a marker in responders to FMT therapy (Fig. 1E). In contrast, on stratification of samples based on recurrent or primary CDI, no significant differences were observed across all mycobiome metrics (data not shown).  Contribution was determined using SourceTracker with donor samples assigned as specific sources to their matched patient samples (sinks). Patient samples at recurrence were labeled in black and with patient number. Significance was tested using ANOVA with Tukey's multiple-comparison tests. Only baseline samples of responders (P-W0-Resp) were significantly different from other groups (denoted by asterisks above groups).*, P , 0.05; **, P , 0.01; ***, P , 0.001; ****, P , 0.0001.

DISCUSSION
In this prospective real-world cohort of consecutive patients with CDI, oral lyophilized FMT was safe and highly effective in treating both recurrent and primary CDI, with prolonged bacterial engraftment in patients who responded to therapy.
Our study showed that bacterial engraftment is successful in responders with a single treatment of orally administered lyophilized FMT, with microbial diversity increasing and composition shifting toward the donor profiles. Furthermore, microbial  Only responders were included in this analysis. Severe C. difficile infections or those with persisting disease despite antibiotics were excluded due to low numbers leaving primary (-P) and recurrent (-R) infections. (A) Shannon's diversity (H¨) index across different sample groups. Significance was tested using ANOVA with Tukey's multiple-comparison test, and only P-W0-P was found to be statistically significantly different from other post-FMT groups. (B) Principal-coordinate analysis of Bray-Curtis resemblance matrix generated from square-root-transformed relative abundances of bacterial OTUs. All patient subgroups (P-) were significantly different from the donors (DON) when tested using pairwise PERMANOVA (P , 0.046 for all). P-W0-P was consistently significantly different from all other post-FMT sample groups in primary CDI (P , 0.015 for all). This result could not be replicated in the patients with recurrent CDI. (C) Heatmap of mean relative abundance of bacterial OTUs found to be significantly different between patients with primary and recurrent CDI at baseline. Tests were performed using LEfSe, and a strict cutoff LDA score of .4 and P value of ,0.05 were applied. changes persisted out to 26 weeks following therapy. We did not see a significant shift in recipient composition in patients who did not respond to therapy. FMT success was characterized by a decrease in Enterobacteriaceae and an increase in Faecalibacterium sp. These findings are consistent with previous studies suggesting a beneficial role for short-chain fatty acid-producing species in the context of FMT (7).
In our cohort, the mycobiome signatures identified were less robust than those of bacterial signatures. We did however find that the recipient fungal richness and the presence of Penicillium sp. were associated with reduced therapeutic outcome. Penicillium sp. has been recognized as a prominent fungal element associated with CDI that may increase intestinal dysbiosis (14). This has been suggested to occur through the antibacterial compounds these species produce, which may affect the capacity of the microbiota to recover (15). Only one study has examined the impact of the mycobiome on therapeutic outcomes in CDI patients following FMT, which was administered via a nonoral route. In a previously published CDI cohort by Zuo et al., Candida albicans was shown to have a negative impact on patient response to naso-duodenal FMT therapy, and Penicillium species was a favorable finding (16). While we did not see a significant association between Candida species and a lack of response, we identified two novel and robust coexclusion relationships with Dorea longicatena and Faecalibacillus intestinalis, bacterial taxa that were enriched in donors and responders following therapy. Furthermore, the presence of one Penicillium taxon was associated with the lack of response to therapy. There are a few potential explanations for this discrepancy. The previous study assessed an Asian cohort of CDI, where a less severe CDI phenotype is sometimes seen (17). Dietary and geographical environmental differences might also explain the discordant Penicillium findings since these fungi are commonly found in foods.
It has been hypothesized that the lyophilization process can damage certain bacterial species and that an upper gastrointestinal (GI) route of administration can further reduce beneficial bacteria entering the colon (18). Whether the lack of shift in the nonresponder recipient microbiome is the cause for cases of treatment failure and a more intensive regime may improve clinical success rate are yet to be determined. In our study, one patient who had no clinical response to therapy and had low donor contribution at week 1 had clinical success with further oral FMT therapy from the same donor, supporting the notion of using repeated oral FMT in those CDI patients who do not have an initial or sustained clinical response to therapy.
Limitations of our study include the relatively small patient numbers from an uncontrolled cohort in a single expert center. While our findings are encouraging, a randomized controlled clinical trial is required to confirm the efficacy of oral FMT in primary CDI. A strength of the study was the frequent sample collection the cohort underwent, enabling longitudinal characterization of microbial dynamics, particularly fungal analysis out to 26 weeks, which has not been done in previous CDI cohorts. One notable characteristic of our patient cohorts is the younger age compared with those reported in the literature, and we believe this characteristic may reflect the mild-moderate disease included in this study.
In conclusion, our data confirmed that orally administered lyophilized FMT is effective in treating recurrent CDI and suggested that this form of therapy is safe and effective for primary CDI. Microbial engraftment was sustained in responders out to 6 months following therapy, with specific bacterial changes found to be associated with treatment outcomes. Novel coexclusion relationships were identified between Candida sp. and specific bacterial species associated with treatment efficacy, supporting the clinical relevance of transkingdom dynamics in CDI.

MATERIALS AND METHODS
Study design. Consecutive adult patients with CDI treated with oral lyophilized FMT were prospectively enrolled in this "real world" cohort between 2015 and 2018 at a single center in Sydney, Australia. CDI was diagnosed by clinical symptoms and confirmed by detection of fecal CDI toxin and/or culture. Patients were treated with a single administration of 6 capsules (each 0.35-g lyophilized stools). No patients received FMT through another route. Treatment episodes were classified as either primary CDI for an initial episode of infection or as recurrent CDI for persistent or repeated disease following an Oral Fecal Microbiota Transplantation for C. difficile appropriate course of antibiotics. Patients were followed weekly for 6 weeks and then monthly for 6 months to assess for symptom response, recurrence, and adverse events. Fecal samples were collected at baseline as well as weeks 1, 2, 3, 4, 6, 12, and 26 following FMT therapy for CDI toxin and culture. Sustained CDI cure was defined as a resolution of diarrhea in addition to loss of CDI toxin and/or culture with no symptom recurrence during the follow-up period. The study was approved by the Centre for Digestive Diseases Human Research Ethics Committee (CDD18/CO5). Written informed consent was obtained from all recruited study subjects.
Fecal donors. Fresh stool was obtained from four individual donors over a 2-year period. Donors were unmatched and unrelated and were screened according to previously published protocols (19).
Lyophilized FMT production. Donated stool was processed within 4 h of collection. Donor stool was homogenized with a cryoprotectant (trehalose and cysteine) and stored at 280°C for up to 2 weeks before being lyophilized. The freeze-dried product from each individual donor was then double encapsulated and stored at 280°C until dispensation. Each capsule contained 0.35 g of lyophilized stool.
Sample collection and DNA extraction. Fecal samples were collected from individual donors and study participants for gastrointestinal microbial community profiling. Twenty-three patients (62.2% of patient cohort) and 4 donors (100% of donors) provided a total of 166 fecal samples from baseline as well as weeks 1, 2, 3, 4, 6, 12, and 26 following FMT therapy. All samples were homogenized and then stored at 280°C until nucleic acid extraction. DNA was extracted using the QIAamp PowerFecal DNA kit (Qiagen, Chadstone, VIC, Australia) according to the manufacturer's instructions.
16S rRNA amplicon sequencing. The V4 region of the 16S rRNA gene was amplified using the Kapa HiFi HotStart ReadyMix (95°C for 3 min; 25 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s; followed by a final step of 72°C for 5 min) and the Earth microbiome primers (515F and 806R) (20) as previously described (21). Indices and Illumina sequencing adapters were attached using the Nextera XT index kit, and sequencing was performed with Illumina MiSeq 2 Â 250-bp chemistry at the Ramaciotti Centre for Genomics. Three negative-control samples (extraction kit reagents) were included as part of the sequencing run. Raw reads were analyzed using mothur v1.42.3, (22,23) with Silva SEED 16S rRNA reference (v123) alignment, clustering with opticlust method, and classification using RDP v16 (3%). The resulting data matrix was subsampled (read depth, 25,287 clean reads/sample) and used for statistical analysis.
ITS region amplicon sequencing. The fungal internal transcribed spacer (ITS) region was amplified using the primers fITS7 and ITS4 (24,25). Raw reads were analyzed using mothur v1.42.3. Reads were clustered by abundance (method, agc) and classified using the UNITE v6 database (cutoff, 0.05). The same three extraction controls as above were used for the fungal ITS sequencing, and minimal amplification was observed (see Fig. S1 in the supplemental material). The resulting data matrix was used for statistical analysis (mean read depth, 47,906 clean reads/sample).
Statistical analysis. Calculation of alpha diversity measures, correlation of resemblance matrices (RELATE), principal-coordinate analysis (PCoA), analysis of similarities (ANOSIM), and permutational multivariate ANOVA (PERMANOVA) were performed using Primer-E v6. Per taxon analyses were conducted using LEfSe (26). Source tracking of microbial taxa was performed using SourceTracker (27) within the Metagenomics for Environmental Microbiology Galaxy framework (28). Procrustes and protest analyses were performed using the R package "vegan," and nonparametric correlation analyses were performed using the framework outlined in Reshef et al. (29). All additional statistical analyses were performed using GraphPad Prism v8.
Data availability. The data sets generated during the current study are available in the European Nucleotide Archive repository under the accession numbers PRJEB37800 (16S) and PRJEB37810 (ITS).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.

ACKNOWLEDGMENTS
This project was supported through a grant from the Gastroenterological Society of Australia. Crestovo provided support for manufacturing equipment required to produce the FMT. N.O.K. is supported by a Scientia Fellowship from the University of