Patient Characteristics
From February 2016 to September 2017, 154 UC patients were assessed for eligibility, and ultimately 15 subjects were recruited and randomized. Of these, 7 individuals were randomly assigned to the FMT and 8 to the placebo arm. Three subjects (1 in the FMT and 2 in the placebo group) did not meet endoscopic criteria for inclusion (Mayo score ≥1) and were excluded from the study (Fig. 1). The remaining 12 subjects (6 in each group) received at least one dose of study treatment. While all 6 subjects allocated to the FMT arm completed all treatments and follow-up assessments, 1 patient in the placebo group dropped out at 6 weeks due to worsening disease. The two study groups were exhibited comparable baseline demographic and clinical characteristics (Table 1).
Table 1. Baseline Subject Demographics
|
|
Group
|
Variable
|
|
Active
|
Placebo
|
N
|
|
6
|
6
|
Age
|
Mean(SD)
|
41 (15)
|
52 (15)
|
Sex
|
#(%) Male
|
4 (67%)
|
3 (50%)
|
Duration UC
|
Mean(SD) yrs
|
8.9 (9.1)
|
9.8 (10.6)
|
BMI*
|
Mean(SD)
|
24 (3)
|
29 (4)
|
Treatment with Biologic
|
#(%) yes
|
2 (33%)
|
1 (17%)
|
Fecal calprotectin
|
Mean(SD)
|
573 (659)
|
408 (277)
|
Total Mayo score
|
Mean(SD)
|
6.3 (2.0)
|
6.7 (1.2)
|
Mayo Symptom subscore
|
Mean(SD)
|
4.8 (1.5)
|
4.3 (1.0)
|
Mayo Endoscopic subscore
|
Mean(SD)
|
1.5 (0.8)
|
2.3 (0.5)
|
Endoscopic UCEIS^ score
|
Mean(SD)
|
7.0 (1.8)
|
8.5 (1.8)
|
Histologic Severity Score**
|
Mean(SD)
|
3.4 (1.2)
|
4.3 (2.0)
|
IBDQ^^ total score
|
Mean(SD)
|
142.8 (16.8)
|
120.2 (25.1)
|
IBDQ bowel system subscore
|
Mean(SD)
|
4.2 (0.7)
|
4.3 (0.9)
|
Baseline clinical characteristics of subjects randomized to both the active FMT and placebo study arms. *Body Mass Index ^Ulcerative Colitis Endoscopic Index of Severity *Geboes Score ^^Inflammatory Bowel Questionnaire
Fig 1. CONSORT diagram showing the flow of subjects through the study. Following randomization, but prior to administration of designated intervention, 1 subject in the treatment group and 2 subjects in the placebo group had no evidence of disease upon endoscopic evaluation and were excluded from the remainder of the study.
Safety Evaluation
Among study subjects that received at least one dose of active or placebo therapy, adverse events possibly or probably related to FMT were few (4 total) and were equally distributed between groups (2/6 vs 2/6; p = 1.00) (Table 2). The only serious adverse event was a worsening of disease activity, which occurred in one subject from each group. Both of these subjects required escalation of therapy (prednisone taper) during the treatment period (at 6 and 4 weeks following initial FMT, respectively). Mild adverse events included nausea (36 hours prior to colonoscopic delivery of placebo material) and fever (24 hours following FMT). Notably, this febrile patient also reported fever 24 hours prior to the initial FMT procedure, making causality uncertain. No infectious complications occurred.
Frozen FMT capsules were distributed to subjects in 4-week allotments (28 pills). cFMT capsules were maintained in home freezers for a total of 84 doses per subject. Medication adherence logs revealed strong adherence with <1% of missed doses across both arms (9/1,008); however, many subjects expressed frustration regarding the strict study guidelines imposed to ensure capsule temperature stability, particularly travel restrictions.
Table 2. Adverse Events by Treatment Assignment
Adverse events
|
FMT (n = 6)
|
Placebo (n = 6)
|
p value
|
AE possibly or probably related to FMT or sham FMT, n (%)
|
2/6 (33)
|
2/6 (33)
|
1.0
|
AE Type and severity, n (%)
|
|
|
nausea, mild
|
0
|
1 (50)
|
1.0
|
fever, mild
|
1 (50)
|
0
|
1.0
|
worsening disease, severe
|
1 (50)
|
1 (50)
|
1.0
|
Adverse events by treatment group that were possibly or probably related to FMT.
Comparisons were made by Fisher’s exact test.
Clinical and Histologic Outcomes
At baseline, no significant differences in histologic or endoscopic scoring were detected between the two groups (Table 1). At 12-week follow-up, the mean endoscopic UCEIS score decreased from 7.0 ± 1.8 to 6.2 ± 2.3 in the FMT group and from 8.0 (1.4) to 7.6 (1.8) in the placebo group (p = 0.60). The mean histologic Geboes score decreased from 3.4 ± 1.2 to 2.3 ± 2.4 in FMT-treated subjects, and from 4.0 ± 2.1 to 3.8 ± 2.0 in the placebo group (p = 0.28). Fecal calprotectin levels decreased from 573 ± 659 to 298 ± 428 in FMT-treated subjects, and from 413 ± 309 to 369 ± 309 in the placebo group. The difference in change of fecal calprotectin between the two groups approached statistic significance (p = 0.08). Alternatively, mean CRP levels increased in both groups, from 4.3 ±7.3 to 10.12 ±10.43 in the FMT group and from 8.1 ± 10.1 to 10.65 ±11.31 in the placebo group with no difference in the proportion of subjects with levels >0 mcg/g noted between groups (p = 1.0). While this study was not powered to predict a clinical response, per subject and per groupvalues for markers of clinical and physiologic disease activity are presented for informative purposes (Tables 3 and 4). In total, two of six (2/6) subjects (subjects E, and W) in the FMT group achieved clinical remission versus none (0/6) in the placebo group (RR = infinity; CI: 0.38-infinity; p = 0.45) and three of six (3/6) subjects (E, W, P) in the FMT group met the study definition of clinical response versus one (1/6) in the placebo group (subject B) (RR = 3.00; CI: 0.42-21.20, p = 0.55). It is worth noting that the FMT subject (subject P) who met the definition of clinical response but not remission, required steroid therapy in the middle of the intervention period (week 6), making any attributions of clinical improvement to FMT difficult. FMT subjects achieving clinical remission (E, W) were considered “FMT responders” for the purposes of additional exploratory analyses, including both immunologic and microbiota-based investigations. Representative photomicrographs of biopsy samples from an FMT-responder and non-responders obtained before and after treatment are shown in Figure 3.
Placebo subjects exhibited inconsistent changes over time, with no clear improvements in symptomatology or clinical evidence of disease. Four of six (66%) placebo subjects required escalation or adjustments in their pharmacologic treatment regimens: one (subject I) at week 4 (this subject dropped out of the trial at this point) and the other 3 study subjects at the end of the treatment period (weeks 12 and 13).
Table 3. Changes in clinical, endoscopic, and histologic evidence of disease by subject
|
Study Code
|
Age
|
Sex
|
Extent of Disease
|
Duration of Disease (yrs)
|
Maintenance Therapy
|
BMI
|
Change in Total Mayo Score
|
Change in Endoscopic UCEIS Score
|
Change in Endoscopic Mayo Score
|
Change in Histologic Score
|
Change in Fecal Calprotectin (mcg/g)
|
Change in Total IBDQ Score
|
FMT
|
E
|
46
|
F
|
pan-colitis
|
5.5
|
Mesalamine
|
20.9
|
-7
|
-2
|
0
|
-3
|
-285
|
92
|
W
|
35
|
M
|
pan-colitis
|
7.5
|
Vedolizumab
|
27.8
|
-3
|
-1
|
0
|
-1.3
|
-189
|
47
|
F
|
20
|
M
|
pan-colitis
|
3.8
|
Mesalamine
|
25
|
-1
|
-1
|
0
|
-2
|
-164
|
16
|
A
|
65
|
F
|
L-Sided
|
26.2
|
Mesalamine
|
20.9
|
3
|
0
|
1
|
-1.2
|
? => 375
|
8
|
N
|
44
|
M
|
pan-colitis
|
0.2
|
Sulfasalazine
|
25.6
|
1
|
-3
|
0
|
0.9
|
>1000 => >1000
|
59
|
P
|
38
|
M
|
pan-colitis
|
10.2
|
Mercaptopurine
|
25.2
|
-3
|
2
|
-1
|
0
|
-42
|
32
|
Placebo
|
B
|
68
|
M
|
pan-colitis
|
4.4
|
Mesalamine
|
28.8
|
-2
|
-1
|
0
|
0.8
|
-132
|
21
|
G
|
58
|
M
|
L-Sided
|
27.8
|
Mesalamine
|
26.9
|
1
|
0
|
0
|
-2
|
4
|
32
|
Y
|
65
|
M
|
L-Sided
|
0.4
|
Mesalamine
|
36.15
|
0
|
0
|
0
|
0
|
74
|
4
|
V
|
47
|
F
|
pan-colitis
|
8.8
|
Adalimumab
|
29.2
|
-1
|
0
|
0
|
0
|
27
|
35
|
T
|
31
|
F
|
pan-colitis
|
0.8
|
Mesalamine
|
29.1
|
0
|
-1
|
0
|
0
|
-137
|
32
|
I
|
40
|
F
|
pan-colitis
|
16.3
|
Mesalamine
|
25
|
DROPPED OUT DUE TO WORSENING DISEASE ACTIVITY
|
^Ulcerative Colitis Endoscopic Index of Severity *Geboes Score; IBDQ, Inflammatory Bowel Disease Questionnaire; L-Sides, left-sided disease; BMI, Body Mass Index; yrs, years; wks, weeks
Fig 2. Longitudenal markers of clinical disease and inflammation. Each line represents a single subject over time. Modified Mayo Score includes subject-reported rectal bleeding, stool frequency, and physician global assessment. IBDQ, Inflammatory Bowel Disease Questionairre; CRP, serum C-Reactive Protein (mg/L); Calprotectin (mcg/g) and Lactoferrin (positive/negative) were measured in stool.
Table 4. Changes in clinical, endoscopic, and histologic evidence of disease by group
Variable
|
Group
|
|
Screen or Procedure
|
12 week
|
[12 wk] - [Bl]
|
P value*
|
CRP
|
Active
|
#(%)>0
|
2 (33%)
|
4 (67%)
|
2 (33%)
|
0.16
|
|
Placebo
|
#(%)>0
|
2 (40%)
|
3 (60%)
|
1 (20%)
|
0.32
|
|
Active – Placebo
|
|
|
|
1.00
|
Fecal calprotectin
|
Active
|
Mean(SD)
|
573 (659)
|
298 (428)
|
-275 (246)
|
0.07
|
|
Placebo
|
Mean(SD)
|
413 (309)
|
369 (309)
|
-44 (90)
|
0.34
|
|
Active – Placebo
|
|
|
-231 (185)
|
0.08
|
Fecal lactoferrin
|
Active
|
#(%) positive
|
7 (100%)
|
4 (67%)
|
0 (0%)
|
--
|
|
Placebo
|
#(%) positive
|
6 (75%)
|
5 (100%)
|
0 (0%)
|
--
|
|
Active – Placebo
|
|
|
|
--
|
Endoscopic UCEIS score
|
Active
|
Mean(SD)
|
7.0 (1.8)
|
6.2 (2.3)
|
-0.8 (1.7)
|
0.29
|
|
Placebo
|
Mean(SD)
|
8.0 (1.4)
|
7.6 (1.8)
|
-0.4 (0.5)
|
0.18
|
|
Active – Placebo
|
|
|
-0.4 (1.3)
|
0.60
|
Endoscopic Mayo score
|
Active
|
Mean(SD)
|
1.5 (0.8)
|
1.5 (0.5)
|
0 (0.6)
|
1.00
|
|
Placebo
|
Mean(SD)
|
2.2 (0.4)
|
2.2 (0.4)
|
0 (0)
|
--
|
|
Active – Placebo
|
|
|
0 (0.5)
|
1.00
|
Mayo symptom score
|
Active
|
Mean(SD)
|
5.0 (1.5)
|
3.5 (3.2)
|
-1.5 (3.4)
|
0.33
|
|
Placebo
|
Mean(SD)
|
4.2 (1.1)
|
4.0 (1.4)
|
-0.2 (1.1)
|
0.70
|
|
Active – Placebo
|
|
|
-1.3 (2.6)
|
0.44
|
Histology (Geboes Score)
|
Active
|
Mean(SD)
|
3.4 (1.2)
|
2.3 (2.2)
|
-1.1 (1.4)
|
0.11
|
|
Placebo
|
Mean(SD)
|
4.0 (2.1)
|
3.8 (2.0)
|
-0.2 (1.0)
|
0.63
|
|
Active – Placebo
|
|
|
-0.9 (1.2)
|
0.28
|
IBDQ bowel system
|
Active
|
Mean(SD)
|
4.2 (0.7)
|
5.2 (1.4)
|
1.0 (1.6)
|
0.19
|
|
Placebo
|
Mean(SD)
|
4.1 (0.8)
|
4.8 (1.2)
|
0.6 (0.7)
|
0.11
|
|
Active – Placebo
|
|
|
0.3 (1.3)
|
0.67
|
IBDQ emotional health
|
Active
|
Mean(SD)
|
4.4 (0.9)
|
5.3 (0.8)
|
0.9 (1.6)
|
0.23
|
|
Placebo
|
Mean(SD)
|
4.9 (1.1)
|
5.3 (1.0)
|
0.4 (0.4)
|
0.09
|
|
Active – Placebo
|
|
|
0.5 (1.2)
|
0.53
|
IBDQ systemic systems
|
Active
|
Mean(SD)
|
4.4 (1.2)
|
4.9 (1.0)
|
0.5 (2.0)
|
0.53
|
|
Placebo
|
Mean(SD)
|
4.2 (1.1)
|
4.6 (0.9)
|
0.5 (0.9)
|
0.31
|
|
Active – Placebo
|
|
|
0.0 (1.6)
|
0.96
|
IBDQ social function
|
Active
|
Mean(SD)
|
5.0 (0.5)
|
5.7 (1.6)
|
0.6 (1.3)
|
0.27
|
|
Placebo
|
Mean(SD)
|
5.1 (1.2)
|
5.9 (1.0)
|
0.8 (0.5)
|
0.02
|
|
Active – Placebo
|
|
|
-0.2 (1.0)
|
0.79
|
IBDQ total score
|
Active
|
Mean(SD)
|
142.8 (16.8)
|
169.0 (34.0)
|
26.2 (48.4)
|
0.24
|
|
Placebo
|
Mean(SD)
|
146.4 (26.1)
|
163.4 (27.2)
|
17.0 (14.4)
|
0.06
|
|
Active – Placebo
|
|
|
9.2 (37.3)
|
0.69
|
CRP, serum C-reactive Protein (mg/L). Endoscopic UCEIS, Ulcerative Colitis Endoscopic index of Severy; Mayo symptom score includes subject-reported rectal bleeding, stool frequency, and physician global assessment. IBDQ, Inflammatory Bowel Disease Questionairre; Calprotectin (mcg/g) and Lactoferrin (pos/neg) were measured in stool. *For within-group comparisons, paired t tests were used for continuous variables and McNemar’s chi square test was used for categorical variables. Between-group comparisons were conducted using two-sample t tests for continuous variables and Fisher’s exact test for categorical variables.
Fig 3. Histologic, endoscopic and clinical parameters of a representative FMT responder (E), non-responder (N), and placebo subject (Y) before and after treatment. Hematoxylin-eosin staining of intestinal mucosa highlight acute and chronic changes and are accompanied by Geboes score (0, structural change only; 1, chronic inflammation; 2, lamina propria neutrophils; 3, neutrophils in epithelium; 4, crypt destruction; and 5, erosions or ulcers), 2x, insets at 20x, scale bar, 50 micrometers; UCEIS, Ulcerative Colitis Endoscopic Index of Severity; fecal calprotectin (mcg/g), and IBDQ, inflammatory bowel disease questionnaire (scale ranging from 32 (worst) to 224 (best))
Longitudinal Phenotyping of Peripheral Blood T-cells
Baseline T cell populations of interest were first compared between UC subjects and healthy controls. The frequency of total lymphocytes obtained following PBMC separation, as well as the CD4:CD8 ratio, were similar between groups. T regulatory cell frequencies were also similar (mean of 3.12% ± 0.41 in UC patients vs. a mean of 3.42% ± 0.54 in controls) with comparable proportions positive for IL-17A and IL-10. No T regulatory cells were IFNg+. The frequency of mucosal-associated invariant T (MAIT) cells was decreased in UC patients (0.62% ± 0.15 vs. 1.67% ± 0.46) and IL-17A positivity occurred almost exclusively in UC-derived MAIT cell populations (3.42 ± 1.27 vs. 0.1759 ± 0.09). Alternatively, IFNg secretion was increased in MAIT cells from healthy controls (46.97 ± 7.15 vs. 24.16 ± 6.01), (Fig. 4).
T cell populations were examined before FMT, and then at weeks 4, 8, and 12 during the cFMT maintenance period. By week 4, the frequency of total PBMC lymphocytes and CD4:CD8 T cell ratios increased in the majority of FMT non-responders, while variable changes were observed placebo subjects. One notable exception was subject F who showed a prominent decrease in CD4:CD8 ratio through week 8, after which a reverse dynamic to baseline occurred. Longitudinal frequencies of T regulatory and MAIT cell populations remained relatively constant across groups. By week 4, IL-17A+ MAIT and IFNg+ MAIT cells decreased in FMT responders, remained suppressed through week 8, and then returned to baseline by week 12. The number of subjects is insufficient to evaluate the statistical significance of these observed changes.
Fig 4. Longitudinal T cell profiling of subjects by flow cytometry and intracellular cytokine staining. a Representative gating scheme showing MAIT cell identification and their cytokine production; b Frequency of total lymphocytes within peripheral blood mononuclear cell isolations and their CD4:CD8 T cell ratios. c,d Comparison of T regulatory and MAIT cell frequencies between UC patients and healthy controls (HC) with longitudinal frequencies and INFγ+, IL-17A+, and IL-10+ proportions displayed by treatment group and clinical response (black, placebo; red, non-responser; green, responser). Each line is an individual subject. Controls are displayed with placebo results (C). Between-group comparisons were conducted using two-sample t tests and p values < 0.05 considered significant.
Intestinal Microbiota Analysis by 16S Sequencing
Relative abundances
Across all time points, stool samples were dominated at the phylum level by Firmicutes, and Bacteroidetes, which accounted for 88.90% of all sequence reads. Bacteria present at lower proportions included Proteobacteria, and Actinobacteria, accounting for 6.9% and 4.0% of total reads, respectively. At the genus level, samples were dominated by Clostridiales and Bacteroidales, with a lower proportion of Burkholderiales, Bifidobacteriales, Selenomonadlaes, Enterobacteriales, Lactobacillales observed at various time points. A strong antibiotic effect was observed following a 7-day course of Metronidazole and Ciprofloxacin in all subjects. Changes included a decrease in gram negative and anaerobic bacteria of the Firmicutes and Bacteroidetes phyla and an increase in gram positive Actinobacteria, including from the genus Bifidobacteriales, and Lactobaccilales Fig. 5. This effect was associated with a decrease in alpha (Shannon) diversity and increased divergence (Jensen-Shannon) from baseline. While these changes were mitigated by the cessation of antimicrobials, neither group returned to their own baseline by 18-week follow-up (Figs 6 and 7).
Fig. 5. Relative abundance of fecal microbiota in subjects with ulcerative colitis (UC) before and after treatment at the phylum and genus levels. Different colors represent different bacterial species, each bar represents one patient sample. a, b. most abundant taxa by phylum and genus level, respectively. Arrow denotes day of FMT (*or placebo); c 23 most abundant taxa of donors and subject at species level, arranged by subject, treatment group, and day.
Alpha and Beta Diversity
No difference in alpha or beta diversity was observed between treatment groups at baseline (Fig. 6). FMT did not increase alpha (Shannon) diversity in recipients but did lead to community-level changes in the gut microbiota creating measurable similarity (beta diversity, Jensen-Shannon divergence index) between FMT subjects and their donor. This convergence, which we termed ‘Donor Divergence Index’, remained statistically significant through 8 weeks of dosing (p < 0.01) and although losing significance (p= 0.16), could still be detected at week 20, > 8 weeks following cessation of oral cFMT therapy (Fig. 7).
Fig. 6 Alpha diversity measured by Shannon index. a alpha diversity in placebo subjects grouped by week; b alpha diversity in FMT subjects grouped by week. Comparisons between groups made by Student’s t-test and p values of <0.05 were considered significant.
Fig. 7 Beta diversity measured by Jenson-Shannon diversity. a Beta diversity comparing subjects to their own baseline overtime; b Beta diversity comparing subjects to donors. Aggregate data is presented by treatment and clinical response (black, placebo; red, non-responser; green, responser). Comparisons between groups made by Student’s t-test and p values of <0.05 were considered significant; c Principle component analysis of donors and study subjects. Each dot represents one sample, and subjects are colored by the treatment group (yellow shades represent placebo, blue and red represent primary donor).