Increased Intestinal Microbial Diversity following Fecal Microbiota Transplant for Active Crohn’s Disease

Abstract

Background

The microbiota in the lumen of patients with Crohn’s disease (CD) is characterized by reduced diversity, particularly Firmicutes and Bacteroidetes. It is unknown whether the introduction of the intestinal microbiota from healthy individuals could correct this dysbiosis and reverse mucosal inflammation. We investigated the response to fecal microbial transplantation (FMT) from healthy individuals to subjects with active CD

Methods

We performed a prospective open-label study (uncontrolled) of FMT from healthy donors to subjects with active CD. A single FMT was performed via colonoscopy. Recipients’ microbial diversity, mucosal T-cell phenotypes and clinical and inflammatory parameters were measured over 12 weeks, and safety over 26 weeks.

Results

Nineteen subjects were treated with FMT and completed the study follow-up. Fifty-eight percent (11/19) demonstrated a clinical response (HBI decrease > 3) following FMT. Fifteen subjects had sufficient pre/post fecal samples for analysis. A significant increase in microbial diversity occurred after FMT (p=0.02). This was greater in clinical responders than non-responders. Patients who experienced a clinical response demonstrated a significant shift in fecal microbial composition towards their donor’s profile as assessed by the Bray-Curtis index at 4 weeks (p=0.003). An increase in regulatory T-cells (CD4+CD25+CD127lo) was also noted in recipients’ lamina propria following FMT. No serious adverse events were noted over the 26-week study period.

Conclusion

In this open label study, FMT led to an expansion in microbial bacterial diversity in patients with active CD. FMT was overall safe, although the clinical response was variable. Determining donor microbial factors that influence clinical response is needed prior to randomized clinical trials of FMT in CD.

ClinicalTrials ID # NCT01847170

INTRODUCTION

Crohn’s disease (CD) is a chronic inflammatory condition best characterized as an abnormal immune response to an unknown environmental or microbiological stimulus in a genetically susceptible individual.^{1, 2} While no specific pathogen has been identified to cause CD, clear associations with the intestinal microbiome have been noted.^3–5 Most common are reduced diversity of bacterial phyla including Firmicutes and Bacteroidetes.⁶ A recent, large cohort study of treatment-naïve, pediatric CD subjects demonstrated that abundance of bacterial families such as Veillonellaceae, Enterobacteriaceae, Pasteurellaceae, Fusobacteriaceae along with decreases in Erysipelotrichales, Bacteroidales and Clostridiales correlated to disease activity.⁴ Beyond alterations in the composition of the microbiome, the function of the microbiome is notably different, with changes in oxidative stress pathways and carbohydrate metabolism.⁵

These associations have led investigators to target intestinal dysbiosis as a treatment for CD. The central hypothesis is that inflammation, and/or symptoms of CD, are related to dysbiosis and can be ameliorated through restoration of a healthy microbiome. Fecal Microbiota Transplant (FMT) is one method of introducing a diverse mix of bacterial families to the diseased intestinal tract that can restore intestinal diversity.^{7, 8} FMT has revolutionized the treatment for multiple recurrent Clostridium difficile infection with cure rates over 90%.⁹ Following FMT for recurrent C. difficile infection, recipients tend to take on their donor microbiota immediately after FMT, although later on distinctly diverge from the donor profile and restore diversity on an individual level.^{10, 11} Overall FMT appears safe in the short term for C. difficile, however its effects on the disease process in patients with IBD is more unpredictable.^{12, 13} Two RCTs in patients with ulcerative colitis have yielded mixed results with one trial finding a significant benefit following FMT, while another without statistical difference between FMT and placebo.^{14, 15} Unlike UC, CD is characterized by limited colonic distribution, a strong association with genes linked to bacterial sensing (such as NOD2 and CARD9)¹⁶, and subgroups that are particularly responsive to antibiotic therapy.¹⁷ Thus, the disease process in CD may be more responsive to alterations in microbial diversity. With that in mind we undertook a prospective, open-label study of active CD to assess the recipient microbial and clinical response to FMT.

METHODS

Patient selection

Eligible subjects required documentation of typical histopathology for CD of the colon, or colon and ileum, > 3 months duration of disease, and a Harvey Bradshaw Index (HBI) of ≥ 5. Subjects must have failed standard therapy with either mesalamine > 2 weeks, thiopurines > 3 weeks, anti-TNF antibody > 12 weeks or were steroid dependent. A 12-week washout period was required for patients exposed to cyclosporine, tacrolimus, infliximab, adalimumab, certolizumab pegol, or natalizumab. Antibiotics were held 60 days prior to FMT and probiotics 30 days prior to FMT. Steroids to tapered to 20mg of prednisone (or equivalent) per day at least 2 weeks prior to FMT

Exclusion criteria included age < 18, indeterminate colitis, isolated proctitis, isolated ileal or small bowel disease, fulminate colitis, infectious colitis, pregnant or nursing, ASA class > II, decompensated cirrhosis, diabetes, active malignancy, lupus, chronic kidney disease (GFP < 60ml/min), history of valvular heart disease (including rheumatic heart disease and endocarditis). Eligible patients were screened two weeks prior to FMT for C. difficile, stool culture, and ova and parasite stool analysis. A stable medication regimen was required for the two weeks prior to FMT.

Donor selection

Donors were healthy males without any relation to the patient. Non related donors were chosen given that siblings of patients with CD may have dysbiotic features such as reduced diversity.¹⁸ Donors were screened using the full length donor history questionnaire prepared by the AABB Donor History Task Force, and underwent blood testing for HIV, hepatitis A IgM, hepatitis B surface antigen and core antibody, hepatitis C antibody, CMV viral load, RPR, CBC, ALT, AST, alkaline phosphatase, and total bilirubin. If the donor passed initial screening, a stool sample was tested for Salmonella, Shigella, Yersinia, Campylobacter, Vibrio, Escherichia coli, C. difficile PCR, Giardia and Cryptosporum DFA, microscopic exam for O&P, Cyclospora, Microsporidium and Isospora. Each donor contributed stool for five aliquots of 50 grams of fecal microbiota material for a donor : recipient paring of 1:5. Four donors were chosen to limit that chance that an ineffective donor was chosen by chance and to provide variability in donor profile.

Donor stool processing

Donor stool was processed within hours of collection following the protocol by Hamilton et al.¹⁹ Briefly, 50g of donor stool was mixed with 250cc of sterile normal saline in a stainless steel blender for fifteen minutes. The resultant slurry was passed through sieves to remove persistent particulate matter. The material was then centrifuged for 15 minutes at 6000 rpms. The supernatant was discarded and the remaining precipitant was re-suspended in half the original volume with 10% glycerol and immediately frozen at −80°C. On the day of FMT, the donor sample was removed from the freezer and placed in a water bath for 4 hours. One hour prior to administration sterile normal saline was added to bring the total volume of donor material to 250cc.

Fecal Transplantation

All Patients underwent full bowel prep with magnesium citrate on the day before colonoscopy. No antibiotics were administered before FMT. Colonoscopy was performed to the terminal ileum and donor fecal material was administered starting in the terminal ileum and right colon and continued distally. Followed colonoscopy patients were administered 4mg of Imodium and positioned on their right side for 20 minutes.

Outcome measures

Clinical parameters including HBI, short Inflammatory Bowel Disease Questionnaire (sIBDQ), Crohn’s Disease Endoscopic Index of Severity (CDEIS), and blood work were collected for 12 weeks following FMT. A clinical response was defined as HBI change of > 3 without an increase in CD related medications at week 4. If a subject required an increase in CD related medications, the clinical parameters were held at the value prior to the start of therapy for statistical analysis. For missing data the prior value was carried forward. Changes in clinical parameters were measures by repeated measures ANOVA followed by Tukey’s multiple comparison post-hoc test, while changes in laboratory parameters and CDEIS were assessed with the Wilcoxon paired sign-rank test. Statistical analysis for clinical parameters was performed using Prism version 5. Three stool samples were collected: a sample prior to FMT, 4 weeks post FMT and 8 weeks post FMT. Mucosal biopsies were obtained at the time of FMT and 12 weeks post FMT. A 12 week stool sample was not collected as it was thought the prep for a colonoscopy might influence the microbiome.

Lamina Propria Cell Isolation

Lamina propria mononuclear cells (LPMCs) were isolated from freshly obtained colonic biopsies at week 0 (before FMT) and week 12. In brief, the colonic biopsies were washed in HBSS-calcium-magnesium free solution and then were incubated in HBSS containing 0.75 mM EDTA (Sigma-Aldrich) and 1 mM dithiothreitol (Sigma-Aldrich) at 37°C for about 30 min to remove the epithelium. The tissues were digested further in RPMI 1640 medium (Cellgro, Mediatech Inc) containing 400 U/ml collagenase IV (Sigma-Aldrich) and 0.01 mg/ml DNase I (Sigma-Aldrich) in a shaking incubator at 37°C. This step was repeated three to five times. LPMCs cells released from the tissues were purified by a 40 to 100% Percoll (GE Healthcare) gradient. LPMCs were cultured in complete RPMI 1640 medium containing 10% fetal bovine serum, 2 mM glutamine, 25mM HEPES, 100 U/ml penicillin, and 100 μg/ml streptomycin

Flow Cytometry

For cell surface staining, single-cell suspensions were prepared and labeled for 15 minutes at 4°C with optimal dilutions of each mAb. The following anti-human monoclonal antibodies were used; fluorescein isothiocyanate (FITC)-mouse anti-human CD39 (BU-61; Ancell Corp. Bayport MN, USA); phycoerythrin (PE)- mouse anti-human CD73 (AD2; BD Pharmingen, San Jose, CA, USA); Allophycocyanin (APC)- mouse anti-human CD25 (BC96; E-bioscience, San Diego, CA, USA); Pacific Blue (PB)-mouse anti-human CD127 (eBioRDR5; E-bioscience, San Diego, CA, USA); phycoerythrin-Cychrome 7 (PE-Cy7)-mouse anti-human CD4 (RPA-T4; E-bioscience, San Diego, CA, USA), An isotype-matched control mAb (murine IgG1) was used to determine nonspecific staining for gating. Expression of cell-surface or intracellular markers was assessed using a flow cytometer (LSRII; Becton Dickinson, Mountain View, CA). After gating on live cells determined by scatter characteristics, data were analyzed using FlowJo software (Tree Star, Ashland, OR). The number of CD73+ cells in the CD4+ population was expressed as a percentage based on flow cytometry cell numbers.

Whole-genome shotgun sequencing

Whole-genome shotgun sequencing libraries were prepared as follows. DNA extractions from stool were carried out using the QIAamp DNA Stool Mini Kit (QIAGEN, Inc., Valencia, CA, USA). Metagenomic DNA samples were quantified by Quant-iT PicoGreen dsDNA Assay (Life Technologies) and normalized to a concentration of 50 pg μL⁻¹. Illumina sequencing libraries were prepared from 100–250 pg DNA using the Nextera XT DNA Library Preparation kit (Illumina) according to the manufacturer’s recommended protocol, with reaction volumes scaled accordingly. Insert sizes and concentrations for each pooled library were determined using an Agilent Bioanalyzer DNA 1000 kit (Agilent Technologies). Metagenomic libraries were sequenced on the Illumina HiSeq 2500 platform, targeting ~5.0 Gb of sequence per sample with 101 bp, paired-end reads.

Quantification of microbial taxa in WGS data

Reads were quality controlled using the KneadData pipeline (https://bitbucket.org/biobakery/kneaddata). The pipeline consists of trimming low-quality bases, dropping reads below 60 nucleotides, and filtering out potential human contamination. Quality controlled samples were profiled taxonomically using MetaPhlAn 2.0²⁰, following Bowtie 2–2.1.0²¹ alignment to the MetaPhlAn 2.0 unique marker database (http://huttenhower.sph.harvard.edu/metaphlan2).

Principal Coordinate Analysis plots

Principal Coordinate Analysis plots were generated using t-Distributed Stochastic Neighbor Embedding (t-SNE, R package Rtsne)²² using Bray-Curtis dissimilarity, BC_ij = Σ_S|x_si − x_sj|/Σ_S|x_si + x_sj|, where x_si denotes the abundance of species s in sample i, as the distance measure. When comparing recipient and donor samples in downstream analyses, 1 − BC was used as a measure of similarity.

Microbial diversity

Microbial richness (alpha diversity) was measured using Shannon’s diversity index, H = −Σp_i log₂ p_i on species level data. Difference between responders and non-responders was tested using Student’s t-test.

Functional quantification of WGS data

Samples were profiled functionally using HUMAnN2 (http://huttenhower.sph.harvard.edu/humann2)²³ and the resulting gene family specific abundance measurements were mapped to gene onthology (GO) terms.^{24, 25} We focused on the “Biological Process” hierarchy within GO, allowing protein annotations to propagate upward through the child-parent relationships among GO terms. Following previous work,^{26, 27} we isolated a subset of “informative” GO terms, defined as terms associated with >k proteins for which no descendant term was associated with >k proteins (here, k = 2,000, which equates to ~1 of every 5,000 UniRef50 protein families). This procedure yielded a comprehensive but manageable set of 247 non-redundant GO Biological Process terms for subsequent analysis.

Taxonomic and functional differences in responders and non-responders

Microbial differences between responders and non-responders were tested using MaAsLin,⁵ a linear modelling system adapted for microbial community data (http://huttenhower.sph.harvard.edu/maaslin, default parameters were used). In brief, MaAsLin applies a generalized linear model using each clade as an independent target variable after outlier removal (Grubbs test), a variance stabilizing arcsin square root transform for binomial data, and false discovery rate multiple hypothesis correction. For each taxon, mixed effect model with relative abundance as a target variable was fitted. Binary predictors for treatment response (responder/non-responder) and treatment status (before/after) were used as fixed effects. Donor ID and subject ID were used as random effects (strRandomCovariates argument in MaAsLin R package) to account for differences between the donors and subjects, respectively.

RESULTS

Demographic information for all 19 patients is shown in Table 1. All 19 patients successfully completed FMT without any adverse events during the procedure. One patient, with severe disease prior to FMT, proceeded to colectomy 8 weeks after FMT. Otherwise all patients completed the 12-week study period. Overall, 11/19 (58%) subjects exhibited a clinical response (decrease in HBI >3, without escalation of CD related medications) by week 4 after the FMT and were categorized as ‘Responders’. Of the ‘Non-responders’, seven required an increase in CD related medications over the study period. Of the 19 subjects, 15 had sufficient pre and post FMT fecal samples for further exploratory analysis.

Table 1.

Baseline characteristics of study participants

Characteristic	n (%)*
Age (SD)	36 (12.3)
Male	12 (63)
Female	7 (37)
Age as diagnosis (Montreal Classification)
A1 (age < 16)	7 (37)
A2 (age between 17 and 40)	10 (53)
A3 (age > 40)	2 (11)
Location (Montreal Classification)
L2 (colonic disease)	7 (37)
L3 (ileocolonic disease)	12 (63)
Behavior (Montreal Classification)
B1 (inflammatory)	16 (84)
B2 (stricturing)	2 (11)
B3 (penetrating)	1 (1)
Peri-anal disease	6 (32)
Mean duration of disease in years (SD)	12.5 (10.6)
Medication at the time of FMT
Mesalamines	3 (16)
Steroids	8 (42)
Immunomodulators	2 (11)
Prior exposure to anti-TNF	10 (52)
Prior IBD surgery	5 (25)
Smoking status
Current	2 (11)
Ex-smoker	2 (11)
Never smoker	15 (79)

^*Percent unless otherwise noted

FMT: fecal microbial transplantation

TNF: Tumor Necrosis Factor

Global Shifts in Microbial Similarity and Diversity

Principal Coordinate Analysis based on Bray-Curtis dissimilarity showed no systematic trend in patients with active Crohn’s disease when comparing to healthy donors (Figure 1a). Donor phyla profiles are noted in supplemental Figure 1. Bacteroidetes and Firmicutes dominated the microbiome as is expected for health individuals. One donor had approximately a 10% relative abundance of Verrucomicrobia, a known normal variant of the intestinal microbiome. For most subjects, including many responders, the samples before and after FMT clustered together, while some responders showed substantial shift towards their donor (Figure 1b). However, species-level similarity between donor and recipient after transplantation was significantly greater among clinical responders when compared to non-responders, both at week 4 and 8 following FMT (p < 0.003 and 0.008 respectively, Figure 2). We next analyzed if the incorporation of donor species into recipients’ microbiota altered its species level diversity, measured by Shannon diversity. Bacterial diversity (alpha diversity) was significantly greater amongst all subjects within 8 weeks following FMT (p < 0.02), and the increase in diversity was greater for clinical responders versus non-responders after FMT (p < 0.03) (Figure 3). These data suggest a single colonoscopic FMT can alter the fecal bacterial composition of patients with active CD in the short-term, and these alterations enhance the diversity of the intestinal microbiota.

Figure 1.

Principal Coordinate Analysis plots of all analyzed stool samples based on Bray-Curtis dissimilarity and colored by A. clinical response and B. donor. All available data is shown, although some subjects are missing a pre or post FMT sample.

Figure 2.

Similarity (1 – Bray-Curtis dissimilarity) between the stool samples of responders and their donors at 4 weeks and 8 weeks. Similarity increased for responders while no such shift was observed in non-responders. All available data is shown however comparisons only reflect subjects with sufficient paired samples.

Figure 3.

Changes in microbial diversity A. An increase in diversity among all subjects was noted after FMT. B. Responders demonstrated more substantial increase in alpha diversity compared to non-responders. Note: Both post-FMT samples (weeks 4 and 8) were used in this analysis. While all data is shown, only subjects with a pre and post sample were included in analysis.

Shifts in Specific Bacterial Families and metabolic pathways

Specific bacterial families previously associated with CD were assessed in subjects before and after FMT.⁴ At baseline Veillonellaceae was present in all subjects, Enterobacteriaceae was found in 93% of subjects, Pasteurellaceae in 60% and Fusobacteriaceae in 27%. Among donors, these four bacterial families were less abundant compared to the CD patients at baseline (mean±s.d. abundance in donors/CD patients: Veillonellaceae 2.2e-3±4.2e-3/0.034±0.056, Enterobacteriaceae 7.3e-3±8.7e-3/0.022±0.027, Pasteurellaeae 6.7e-5±1.3e-4/8.9e-4±1.2e-3, Fusobacteriaceae 3.4e-5±6.8e-5/2.4e-3±8.1e-3). However, we did not observe significant change in the abundance of these families after FMT. Recipient family and genus level changes pre and post FMT are presented in supplemental figures 2 and 3 respectively. When a more extensive list of bacterial species was analyzed in pre- and post- FMT fecal samples, significant changes were noted (p < 0.05, false discovery rate (Q) < 0.25, Supplemental table 1).

Multiple metabolic pathways were increased in responders following FMT compared to non-responders (Supplemental Figure 4). These pathways included serine and glutamine metabolic pathways, folic acid metabolic pathways, and lipid A biosynthetic pathways. The majority of pathways increased in responders were related to increased energy metabolism or components needed for bacterial cell surface or cell walls. The accession number for the microbial data reported in this paper can be found at NCBI BioProject: PRJNA290380.

Mucosal T-cell Response

Since microbial peptides influence T-lymphocyte phenotypes in the intestinal mucosa, we next sought to determine of there were alterations in mucosal immunology post-FMT.²⁸ Regulatory T cells (CD4+CD25+ and CD4+Foxp3+) in the peripheral blood exhibit reduced function or number in IBD compared to healthy controls.^{29, 30} However, Tregs defined as such often have T effector cells included in the population and thus CD127lo has been proposed as an additional marker of Tregs.^31–33 Based on this prior work, we examined mucosal CD4+CD25+CD127lo T-cells before and after FMT. Compared to baseline, there was a significant increase in the proportion of mucosal CD4+CD25+CD127lo T-cells by flow cytometry at 12 weeks after FMT (mean 5% versus 2% at baseline, p < 0.05, Figure 4A). Clinical responders did not have a greater proportion of CD4+CD25+CD127lo T-cells at week 12 when compared to non-responders (mean 5% versus 5.3%). Effector T-cells CD4(+)CD39(+)CD161(+) cells, which are associated with Th17 cell differentiation³⁴ were not decreased 12 weeks after FMT (Figure 4B).

Figure 4.

Specific T-cell populations. A. proportion of mucosal CD4+CD25+CD127lo T-cells by flow cytometry at 12 weeks after FMT (mean 5% versus 2% at baseline, p=0.05), B. CD4(+)CD39(+)CD161(+) cells were not decreased 12 weeks after FMT.

Clinical response

Clinical indices of disease activity (HBI and sIBDQ) over time are displayed in Figure 5A & B. Overall, there was a significant trend in HBI over time towards an improved HBI (p < 0.0001). Tukey’s multiple comparison test demonstrated significant differences comparing all pre to post FMT time points for HBI score (Figure 5A) and quality of life (sIBDQ, Figure 5B). While clinical indices improved, no changes were noted in CDEIS or CRP scores 12 weeks after FMT (Figure 5C&D). At week 4, 58% (11/19) had a clinical response (HBI decrease >= 3) and 53% of subjects were in clinical remission (HBI < 5). Seventy-three percent (8/11) of subjects who had a clinical response at week 4 had a sustained clinical response at week 8, and 55% (6/11) had a sustained response at week 12. No clinical characteristics appeared to have any impact on clinical response (gender, disease distribution, disease duration, or prior medication use), although all subgroup analysis were limited by small size (data not shown).

Figure 5.

Clinical disease indices following FMT: A. Harvey Bradshaw Index (HBI) before and after FMT and B. sIBDQ before and after FMT. Repeated measures ANOVA demonstrates a significant change over time (p < 0.0001) for both HBI (A) and sIBDQ (B), while no significant change was noted in CDEIS (C) or CRP (D).

Safety

Overall FMT was well tolerated in the setting of active Crohn’s disease (Supplemental Table 2). The most common reported adverse event was GI related, typically mild abdominal cramping or constipation. As reported elsewhere, low grade and subjective fevers were noted following FMT. One patient developed hives the week following FMT, which required treatment with oral steroids. Another patient had no clinical response to FMT and proceeded to surgery 8 weeks after FMT.

DISCUSSION

This study reports, for the first time, the impact of colonoscopic FMT on microbial and immunological parameters in patients with active Crohn’s disease. The data suggests that recipients’ microbial diversity can be enhanced by this technique, and this development is associated with a clinical response in this uncontrolled observational trial. FMT was safe for the majority of participants over the study period. However some patients experienced a clinical worsening or non-response after FMT necessitating an escalation of CD related medications and in one case surgery. As all patients were experiencing a CD flare at baseline, we were unable to determine if those who did not respond simply had progression of their disease or were made worse by FMT. It is notable that the subject who proceeded to surgery had medical refractory disease prior to FMT.

In general, participants had a symptomatic improvement following FMT as measured by clinical indices, although inflammatory parameters were unchanged. There are multiple reasons for this observation. First, given the small number of participants and lack of control group, this could be due to placebo effect or natural variation of disease. The latter is less likely as HBI scores remained stable over a two week period prior to FMT. Dosing of FMT may also play a role. Higher or more frequent doses of donor microbiota may be needed to improve engraftment and impact inflammation. Similarly, it may be that certain bacterial phyla or families are important to replete and should be enriched pre-FMT to obtain an optimal response. Lastly altering the intestinal microbiota may only decrease symptoms of inflammation without affecting the underlying inflammation. Many of the symptoms of CD, such as abdominal cramping, gas, diarrhea, may due to dysbiosis as a result of inflammation. Thus intervening on the dysbiosis may help symptoms without modulating inflammation. While our study has suggested a signal for FMT in active CD, more work needs to be done to elucidate the mechanisms, optimal dosing, and delivery of microbiota before being studied in large clinical trials.

Data for FMT in IBD is limited and contradictory with two randomized clinical trials for FMT in UC demonstrating mixed results.^{14, 15} Two prior observational trials found modest efficacy for FMT in CD.^{35, 36} Both studies were performed with nasogastric administration of donor fecal material and had limited description of the host’s microbial response; although Suskind et al noted that subjects who did not engraft, failed to have a clinical benefit.³⁵ While the host response to FMT was variable, our data suggest that restoring colonic diversity is associated with a clinical improvement in CD. Subjects with active CD at baseline had a dramatic variability in colonic microbiome likely reflecting the dysbiosis noted in active CD. Additionally, we note multiple metabolic changes associated with clinical response, that are likely reflective of improved microbial diversity.

Prior studies have noted specific families associated with CD. Enterobacteriaceae are gram negative bacteria that commonly include pathogenic species such as Salmonella, Shigella, Proteus, and E. coli. Members of this family have been linked to colonic inflammation, specifically increased IL-1b via NLRP3 inflammasome.^{37, 38} In one study, 90% of subjects with Crohn’s disease demonstrated intra-cellular E coli.³⁹ In our cohort, abundance of Enterobacteriaceae was often found in a flare state. While Fusobacteriaceae has been reported as correlating with increased disease severity in patients with pouchitis,⁴⁰ subjects in our cohort had minimal presence of this family and no correlation to disease state. An increase in Pasteurellaceae (predominately Haemophilus parainfleunzae) in mucosal biopsy specimens is correlated with disease activity in Crohn’s disease.⁴ In our cohort, the presence of Pasteurellaceae was noted more often in a flaring state. While previous observational studies have demonstrated that these families are associated with dysbiosis in CD, we were unable to associate changes with these bacterial families to clinical response. It remains unclear what the significance of these associations are, but it is unlikely that a single bacterial family will emerge as a strong predictor of disease state or response to intervention such as FMT.

Analysis of mucosal T-cells determined that there was an increase in Tregs (CD4+CD25+CD127lo), while little change was noted on Th17 phenotype cells (CD4+CD39+CD161+) after FMT. These changes may have been induced by alterations noted in the microbiome, but in the absence of a control group such hypotheses cannot be confirmed from this study.

Colonoscopic FMT for active Crohn’s disease was overall safe. Adverse events were similar in to other studies of FMT with subjective fevers and chills as well as abdominal cramping and change in bowel pattern being common. One patient developed hives following FMT, underscoring the importance of biologically active components in donor stool. Only one subject had a clear worsening of symptoms following FMT. Two weeks post FMT, this subject’s HBI increased from 9 to 20 and the patient proceeded to surgery within 2 months. This patient had medically refractory disease prior to FMT. IBD flares following FMT have been reported in up to a quarter of IBD patients post FMT for C. difficile, although it is unknown if this is related to the FMT, the infection, the procedure, or natural progression of disease.^{12, 13} However given this finding it is important to counsel patients that there may be a chance that FMT could lead to a CD flare.

Our study is limited by the lack of a control group and small sample size. However, the study was exploratory in nature and as such, was not designed to assess if FMT is an effective treatment for CD. The clinical response following FMT may have been related to natural variation of disease or placebo effect. While the association of increased similarity of microbial profiles is provocative, it is important to note that we cannot infer causation. Subjects who improved may naturally develop a more diverse microbiota. Additionally we were unable to completely account for diet and non-IBD medication use over the study period, which may have an effect on the microbiome. While our microbial analysis was robust, it was based off of stool samples, which may not be as accurate as mucosal biopsies. We chose not to administer antibiotics prior to FMT as this strategy is not proven in IBD. While this may have decreased microbiota engraftment, pre-FMT antibiotics could also affected the underlying CD making it difficult to assess if antibiotics or FMT caused changes post-FMT. Lastly, seven subjects required an increase in CD related medications, which could have been responsible for changes in the microbiome. All of these subjects were classified as non-responders and thus increased microbial diversity as a result of CD related medications would likely bias our results towards the null hypothesis.

We demonstrate that FMT for CD is safe and provides symptomatic improvement in select patients potentially through increasing the overall diversity of colonic microbiome. Clinical responders were more likely to have a post-FMT microbiome that was similar to their donor as opposed to non-responders, who did not shift towards their donor. While eventually randomized controlled trials of FMT will need to be done to assess the clinical efficacy of FMT in CD, our results suggest that immediate future studies should focus on identifying components of donor microbiota that result in a decrease in potentially pathogenic bacterial families and identifying a link between bacterial communities and intestinal inflammation. In conclusion, FMT is safe in active CD and may induce clinical remission in select patients though increased diversity in the microbiome.