Clinical Predictors of Response to Faecal Microbiota Transplantation in Patients with Active Ulcerative Colitis Free

Abstract

Background

Faecal microbiota transplantation [FMT] has been shown to be effective for induction of remission in patients with active ulcerative colitis [UC]. At present, the clinical factors impacting the response to FMT in UC remain unclear.

Methods

Patients with active UC treated with multisession FMT via colonoscopy at weeks 0, 2, 6, 10, 14, 18 and 22 were analysed. Response to FMT was defined as achievement of corticosteroid-free clinical remission at week 30. Patient and disease characteristics were evaluated to determine the predictors of response to FMT.

Results

Of 140 patients with active UC treated with FMT, 93 (mean age 34.96 ± 11.27 years, 62.36% males [n = 58], mean Mayo clinic score 8.07 ± 2.00) who completed the multisession FMT protocol were analysed. Fifty-seven [61.29%] patients achieved clinical remission. Younger age (odds ratio [OR] for age 0.93, 95% confidence interval [CI] 0.89–0.97, p = 0.001), moderate [Mayo clinic score 6–9] disease severity [OR 3.01, 95% CI 1.12–8.06, p = 0.025] and endoscopic Mayo score 2 [OR 5.55, 95% CI 2.18–14.06, p < 0.001] were significant predictors of remission on univariate analysis. Younger age, disease extent E2 and endoscopic Mayo score 2 [OR 0.925, 95% CI 0.88–0.97, p = 0.002; OR 2.89, 95% CI 1.01–8.25, p = 0.04; and OR 8.43, 95% CI 2.38–29.84, p = 0.001, respectively] were associated with clinical remission on multivariate logistic regression. A mathematical model [nomogram] was developed for estimating the probability of remission with the FMT protocol.

Conclusion

Younger age, disease extent E2 and endoscopic Mayo score 2 significantly predict achievement of clinical remission with FMT in active UC. The prediction model can help in selecting individuals for FMT. Validation in larger cohorts is needed.

1. Introduction

Our understanding of the role of the gut microbiome in health and disease is gradually expanding. The advancements in nucleotide sequencing methods have revealed a characteristic composition of gut microbiota in ulcerative colitis [UC], characterized by a reduction in Bacteroidetes, reduced diversity within Firmicutes and an increased proportion of Proteobacteria.¹ Faecal microbiota transplantation [FMT] is a novel approach to correct gut microbiome dysbiosis and restore microbial communities.² FMT has been found to be effective for induction of remission in patients with active UC in randomized controlled trials [RCTs].^3–7 Our group has also reported that multisession FMT via the colonoscopic route can induce clinical remission and aid in steroid withdrawal.⁸ The beneficial effect of FMT on the maintenance of both endoscopic and histological remissions has also been reported.⁹ Despite encouraging initial experience with FMT in UC, nearly half of the patients do not respond regardless of repeated interventions, the cause remaining unknown. Evaluating the clinical factors influencing the efficacy of FMT in UC is important to identify those patients who are likely to respond to this therapy.

2. Methods

2.1. Study design

This is a single-centre analysis of a prospectively maintained database of patients with active UC treated with FMT between September 2015 and December 2019 at Dayanand Medical College and Hospital, a tertiary care centre in northern India. The study was approved by the Institutional Review Board [IEC No. 2015-113] and was performed in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants. All authors had access to the study data and reviewed and approved the final manuscript.

2.2. Study population

Adult patients [aged 18–60 years] with steroid-dependent [defined as inability to reduce steroids below prednisolone 10 mg/day within 3 months of starting steroids or relapse within 3 months of stopping steroids]¹⁰ or chronic moderate–severe active UC [defined as Mayo clinic score≥ 6 and Mayo endoscopic subscore ≥ 2] were offered FMT as an add-on therapy. The standard of care treatment was continued, including oral 5-aminosalicylates [5-ASA; stable dose for 4 weeks], topical 5-ASA [for patients with E1 disease], azathioprine [on therapy for > 12 weeks and stable dose for 4 weeks, if not intolerant], and/or corticosteroids [on therapy for ≥ 4 weeks in a tapering dose schedule of prednisone 40 mg/day for 2–4 weeks followed by dose reduction of 5 mg/week to a daily dose of 20 mg/day and then slow taper by 2.5 mg/week until prednisone is discontinued]. Patients with previous exposure to biologics [infliximab, last dose at least 8 weeks before enrolment] were also included. Treatment modifications, including use of corticosteroids [initiation or increase in the dose] and biologics [infliximab], were permitted if the patient had worsening of disease activity [adjudged by an increase in the number of stools, presence of blood in stools and increase in faecal calprotectin]. Patients who had indeterminate colitis, a history of colorectal malignancy, pregnancy, past history of surgery, exposure to antibiotics or probiotics in the last 4 weeks, and evidence of infections such as Clostridium difficile, cytomegalovirus, parasites or extra-intestinal infections requiring antibiotics were excluded. Patients who did not complete the multisession colonoscopic FMT protocol were excluded from the analysis.

2.3. Intervention

A freshly collected donor stool sample [80 g, Bristol stool scale 3 or 4] was homogenized with preservative-free normal saline using a blender [Stomacher 400 circulator; Seward Ltd] until it reached a liquid consistency [200 mL]. Faecal slurry was infused via a colonoscope into the terminal ileum or caecum within 6 h of collection of stool and 1 h of preparation of the slurry. All the recipients received samples from a young healthy asymptomatic voluntary unrelated donor [male, aged 38 years, body mass index 23 kg/m²] who had no comorbidities or disorders known to be associated with changes in gut microbiota. The donor screening and testing was done as per recommendations of an international consensus conference on stool banking for FMT in clinical practice.¹¹ The donor stool was tested for enteric pathogens including parasites [Cryptosporidium, Entamoeba histolytica, Giardia, Cyclospora], bacteria [Salmonella, Vibrio, Plesiomonas, Shigella, Escherichia coli, Campylobacter, Yersinia and Clostridium difficile] and viruses [Adenovirus, Astrovirus, Norovirus, Rotavirus and Sapovirus]. Additionally, stool cultures for Salmonella, Shigella, Escherichia coli, Campylobacter and Yersinia were also done. The donor was negative for hepatitis A, B, C and E, human immunodeficiency virus, cytomegalovirus [IgM] and syphilis. Prospective screening for enteric infections was done every 12 weeks and any symptoms of infection between the last screening and time of donation were investigated. The stool sample was not accepted if the donor had taken antibiotics or probiotics in the previous 12 weeks. FMT sessions were scheduled at weeks 0, 2, 6, 10, 14, 18 and 22 via colonoscopy performed under conscious sedation. Antibiotics or loperamide were not used pre-FMT. The recipients were encouraged to retain stool slurry for as long as possible.

Disease activity including endoscopic severity [using Mayo clinic score] and response to therapy were analysed at each visit. All the FMT procedures were performed by a single endoscopist [R.M.] and the endoscopic severity scores were reviewed [from the recorded video-colonoscopy] by a second expert [A.S.] and verified.

2.4. Outcomes

Response to FMT was defined as achievement of corticosteroid-free clinical remission [Mayo clinic score ≤ 2, with each sub-score ≤ 1 and an endoscopic subscore of 0]¹² at week 30. Various factors considered to affect response to FMT were divided into three categories:

• patient characteristics [age and sex];

• disease characteristics (disease duration, disease extent and severity, exposure to anti-tumour necrosis factor [anti-TNF] agents/immunosuppressants and/or steroids, biomarkers for disease severity including C-reactive protein and faecal calprotectin); and

• FMT-related variables [faecal slurry retention time].

2.5. Adverse event recording

Adverse events such as fever, abdominal pain, nausea, vomiting, loss of appetite, worsening of diarrhoea, abdominal distension, perianal or rectal pain, flatulence, borborygmi, bloating, constipation, urinary tract infection, respiratory tract infection, etc., were recorded after each FMT. A serious adverse event was defined as any untoward medical occurrence after FMT, resulting in inpatient hospitalization or prolongation of hospitalization, was life-threatening or resulted in death of the patient. Pre-procedural/intra-procedural and immediate post-procedural adverse events were recorded before/during the procedure and after 48 h by a telephone call, respectively. Adverse events occurring subsequently in the course of time were recorded on follow-up visits in the outpatient department.

2.6. Statistical analysis

Summary statistics were calculated for both group of patients [i.e. those who achieved clinical remission with FMT and those who did not]. Cohen’s kappa was used for inter-rater reliability testing for endoscopic Mayo score. The non-parametric Wilcoxon rank-sum test and linear regression tests were used for continuous and categorical variables respectively. Using the standard α = 0.05 cut off, p < 0.05 was considered statistically significant. A multivariable logistic regression analysis was performed to determine the final predictors for response to FMT.

A mathematical model [nomogram] was created to predict the probability of achieving remission by calculating a point estimate of the response variables in the logistic regression model associated with remission. For building a nomogram, each predictor was scored on the basis of estimated logistic regression coefficients. The greatest impact predictor was determined and sequentially other predictors were scored in proportion to the points assigned to the greatest impact predictor.¹³ This translated complex mathematical models into a simple graph of scaled variables facilitating a quick approximation of event probability.¹⁴ All statistical analyses were performed using Stata 12 [Stata Statistical Software: Release 12; StataCorp].

3. Results

In total, 140 patients were treated with FMT for active UC between September 2015 and December 2019. Of these, 93 [66.42%] completed the multisession FMT protocol while 22 [15.71%] patients dropped out due to adverse events [fever: n = 7, worsening of diarrhoea: n = 6, post-procedure abdominal discomfort: n = 5 and flatulence: n = 4]. The remaining 25 [17.86%] patients were still continuing with FMT but were not included as they had not completed the multisession FMT protocol at the time of analysis. Only those patients who completed the predefined FMT protocol [n = 93] were included in the analysis. Most of the enrolled patients had moderate disease severity (Mayo clinic score 6–9; n = 71 [76.34%]). All [100%] patients were on concomitant 5-ASA [dose 3.6–4.8 g/day], 78.49% [n = 73] on corticosteroids [dose 10–40 mg/day, median 20 mg/day] and 51.61% [n = 48] on azathioprine [dose 1.5–2 mg/kg/day]. Nineteen [20.43%] patients had previous exposure to biologics [infliximab, n = 15; adalimumab biosimilar, n = 4]. None of the patients received multiple biologics before enrolment. The baseline characteristics of patients are summarized in Table 1.

Fifty-seven patients [61.29%] achieved steroid-free clinical remission [percentage of agreement for endoscopic Mayo score: 94.2%, Cohen’s k: 0.88]. All those patients who achieved clinical remission were able to taper and withdraw corticosteroids. Thirty-six [38.71%] patients did not achieve the end point of corticosteroid-free clinical remission. In these patients, however, the mean Mayo clinic score declined from 8.91 ± 1.96 to 5.81 ± 1.96 and ten [27.77%] patients had a decrease in the anatomical disease extent from pancolitis [E3] to left-sided colitis [E2] [n = 5] or left-sided colitis [E2] to proctitis [E1] [n = 5]. Nonresponders to the multisession FMT protocol were continued on corticosteroids, started on biologics or subjected to colectomy subsequently. None of the patients required rescue therapy with biologics until week 30.

3.1. Univariate and multivariate outcome analysis

On univariate analysis, the variables found to be associated with achieving remission with FMT included younger age (odds ratio [OR] for age 0.93, 95% confidence interval [CI] 0.89–0.97, p = 0.001), moderate [Mayo clinic score 6–9] disease severity [OR 3.01, 95% CI 1.12–8.06, p = 0.025] and endoscopic Mayo score 2 [OR 5.55, 95% CI 2.18–14.06, p < 0.001]. Because disease severity and endoscopic Mayo score have overlapping criteria, a test of their interaction was performed using a general linear model and no significant interaction was found [p = 0.266]. These two parameters were therefore retained. Disease extent E3, severe [Mayo clinic score ≥ 10] disease and endoscopic Mayo score 3 were associated with failure of FMT in inducing remission [OR 0.36, 95% CI 0.14–0.88, p = 0.024; OR 3.01, 95% CI 1.1262–8.0689, p = 0.025; and OR 5.55, 95% CI 2.1896–14.0675, p < 0.001 respectively] [Table 2]

Male gender [OR 1.60, 95% CI 0.67–3.77, p = 0.28], disease extent E2 [OR 2.10, 95% CI 0.89–4.96, p = 0.08], lower disease duration [OR for disease duration 0.92, 95% CI 0.84–1.01, p = 0.07] and no concomitant use of corticosteroids at the time of enrolment [OR 1.62, 95% CI 0.56–4.71, p = 0.36] were other variables associated with response to FMT, but these failed to achieve statistical significance.

For multivariate logistic regression, variables were selected via backward elimination with a p-value criterion of 0.1. The variables included in the prognostic model were age, disease duration, disease extent E2, moderate disease severity and endoscopic Mayo score 2. Three of these variables [age, disease extent E2 and endoscopic Mayo score 2] were identified as favourable predictors of response to FMT [OR 0.925, 95% CI 0.88–0.97, p = 0.002; OR 2.89, 95% CI 1.01–8.25, p = 0.04; and OR 8.43, 95% CI 2.38–29.84, p = 0.001, respectively] [Table 3].

3.2. Subgroup analysis of biologic-experienced and naïve patients

The predictors of response to FMT were also evaluated in the subgroups of biologic-experienced [n = 19] and biologic-naïve patients [n = 74]. In the biologic-experienced patients, only endoscopic Mayo score 2 at baseline predicted response to FMT [p = 0.027] [Supplementary Table 1]. On the other hand, younger age, shorter disease duration, moderate disease severity and endoscopic Mayo score 2 were associated with response to FMT in patients naïve to biologics [p = 0.003, 0.050, 0.014 and 0.001, respectively] [Supplementary Table 2]. The clinical remission rates were similar in patients with or without prior exposure to biologics (12/19 [63.15%] vs 45/74 [60.81%]; p = 0.85).

3.3. Development of nomogram

A mathematical predictive model [nomogram] was derived from all the independent variables included in the multivariate logistic regression model [age, disease duration, disease severity, disease extent and endoscopic Mayo score]. The nomogram was characterized by one scale corresponding to each variable: a score scale, a total score scale and a probability scale. The probability of remission was calculated by adding the score for all the determinants [Figure 1].

3.4. Safety and adverse events

The most common adverse events noted were abdominal discomfort, flatulence, borborygmi and low-grade fever (n = 28 [30.10%], n = 12 [12.90%], n = 10 [10.75%] and n = 8 [8.6%] respectively). Seven patients [7.52%] had worsening of diarrhoea, which was transient and self-limiting. No serious adverse events were noted.

4. Discussion

FMT is emerging as a novel therapy in management of UC.^3–9 However, the response rate varies considerably and the factors influencing efficacy of FMT in UC remain unclear. Although the gut microbiota is regarded a key player in dictating the biochemical/immune mechanisms in UC, analysing bacterial taxonomy alone is not an accurate tool to predict response as there is no uniform set of microorganisms explaining response to FMT in the existing literature. So far, the roles of patient- and disease-related parameters likely to influence response have not been reported. To our knowledge, this is the first study focused on describing the clinical predictors of response to FMT in UC.

Of 93 patients studied, 57 [61.29%] achieved steroid-free clinical remission. Of the various factors studied, younger age, disease extent E2 and endoscopic Mayo score 2 were associated with achievement of clinical remission on both univariate and multivariate analyses. A mathematical prediction model, the nomogram, was built using the independent factors predictive of response, to estimate the probability of remission with FMT.

A single session of FMT has been shown to be efficacious in patients with C. difficile infection.¹⁵ However, for UC, which involves complex pathological mechanisms, therapeutic microbial manipulation with a single FMT session is unlikely to have sustained benefits. Gut microbiota in chronic diseases is resilient and reverts back to its original composition; repeated interventions are therefore needed to induce a lasting change.^16,17 We therefore used a multisession colonoscopic protocol for infusion of FMT slurry.

Gut microbiota composition varies with age of the host. The molecular and functional changes associated with ageing impact microbial selection by the host and microbe–microbe interactions, as well as microbial evolution. Physiologically in humans, the microbiota in younger age has immune-modulatory functions, whereas old age-associated bacterial communities are enriched with pathobionts.¹⁸ The dysbiosis associated with UC overlaid on the physiological changes in microbiota is expected to influence the outcomes with microbiota-targeted therapies such as FMT. We demonstrated that younger age was associated with a higher probability of remission. As none of the patients enrolled in the current analysis were aged > 60 years, we cannot comment on the efficacy of FMT in the elderly. Better remission rates were also achieved in patients with shorter disease duration, implying microbiota modulation early in the disease course may be beneficial as a perturbation in the microbiome might be more easily restored early.

The endoscopic severity of mucosal lesions generally reflects the inflammatory disease activity.^19–21 Patients with higher Mayo endoscopy subscores and histologically active disease are less likely to achieve and sustain remission.^22–24 Our findings of E2 disease with endoscopic Mayo score 2 responding to FMT indicate that FMT could be an effective therapy only for moderately severe distal colitis and not for severe extensive colitis. This is in contrast to findings from previous RCTs that did not detect an interaction between disease distribution or severity and response to FMT.^3,5

Interestingly, no significant effect of sex, haemoglobin, inflammatory biomarkers, previous exposure to biologics, concomitant steroids/immunosuppressants or faecal slurry retention time was noted on outcomes with FMT. Bacterial engraftment is dependent on the composition of the pre-FMT microbial community.²⁵ Although faecal slurry retention time did not affect results with FMT, the authors believe that in addition to microbial taxonomy, adequate retention of faecal slurry is important for bacterial engraftment; however, this needs to be studied in larger structured trials.

Microbial diversity of the donor stool is the most important predictor of FMT outcome. In one of the earliest RCTs on FMT in UC, out of nine patients who entered remission, seven had received FMT from the same donor. Recipients of FMT from other donors did not respond.³ In the current study also, all patients received FMT from a single donor. High rates of achievement of clinical remission with a single donor, as noted in this study, are plausible but not typical. A multidonor infusion to ensure greater microbial diversity than from individual donors is an attractive alternative. However, multidonor intensive FMT did not significantly increase the response rate [27%] compared to an RCT using single-donor FMT [24%].^3,4 Our understanding of the gut-microbiome–host interactions is still in its infancy. Detailed characterization of donors and recipients with a multi-omics approach [metagenomics, metatranscriptomics and metabolomics] is needed to identify and correct the functional deficiencies. Such donor–recipient matching may potentially improve the efficacy of FMT.

The strength of our study is its large sample size. This allowed for development of a predictive model with good discriminative power at identifying individuals who are likely to respond to FMT. This may assist treating physicians in selecting appropriate patients for FMT. However, the predictive model requires further validation by prospective studies. The limitations of our study include that is from a single centre, lack of microbiota analysis, lack of randomization and risks of selection bias. Hence, validation in large, randomized datasets is warranted. Also it would be interesting to study the influence of microbial factors on the current predictive model.

The exact place of FMT in the therapeutic armamentarium of UC remains undefined. Selection of patients most likely to benefit from microbial manipulation will be a crucial determinant. We propose that younger age, moderately severe left-sided colitis [E2] with endoscopic Mayo score 2 employing a multisession colonoscopic FMT protocol can predict clinical remission in patients with active UC.

Conflict of Interest

The authors have no personal or financial conflicts of interest to declare.

Author Contributions

A.So.: Conception and design; supervision, analysis and interpretation of the data; drafting of the article; critical revision of the article for important intellectual content; final approval of the article. A.Si: conception; collection, drafting of the article, critical revision of the article for important intellectual content; final approval of the article. R.M.: critical revision of the article for important intellectual content; final approval of the article. V.M.: design, supervision, critical revision of the article for important intellectual content; final approval of the article. K.K.: final approval of the article. D.S.: collection, critical revision of the article for important intellectual content; final approval of the article. N.B.: statistical analysis, final approval of the article. K.D.: statistical analysis, final approval of the article.

Funding

None.

References