Assessing the Relationship between the Gut Microbiota and Inflammatory Bowel Disease Therapeutics: A Systematic Review

Current inflammatory bowel disease (IBD) treatments including non-biological, biological, and nutritional therapies aim to achieve remission and mucosal healing. Treatment efficacy, however, is highly variable, and there is growing evidence that the gut microbiota influences therapeutic efficacy. The aim of this study was to conduct a systematic review and meta-analysis to define changes in the gut microbiota following IBD treatment and to identify microbial predictors of treatment response. A systematic search using MEDLINE/Embase and PubMed was performed in July 2022. The review was conducted based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines. Studies were included if they reported longitudinal microbiota analysis (>2 weeks) using next-generation sequencing or high-throughput sequencing of faecal/mucosal samples from IBD patients commencing treatment. Meta-analysis on alpha-diversity changes following infliximab treatment was conducted. Thirty-nine studies met the inclusion criteria, and four studies were included in the meta-analysis. An increase in alpha diversity was observed following treatment with 5-aminosalicylates, corticosteroids, and biological therapies in most studies. Characteristic signatures involving the enrichment of short-chain-fatty-acid-producing bacteria including Faecalibacterium prausnitzii and a reduction of pathogenic bacteria including various Proteobacteria were demonstrated following treatment with specific signatures identified based on treatment outcome. The meta-analysis demonstrated a statistically significant increase in bacterial richness following infliximab treatment (standardised mean difference −1.16 (−1.50, −0.83), p < 0.00001). Conclusion: Distinct microbial signatures are seen following treatment and are associated with treatment response. The interrogation of large longitudinal studies is needed to establish the link between the gut microbiota and IBD therapeutic outcomes.


Introduction
Inflammatory bowel diseases (IBDs), of which the main subtypes are ulcerative colitis (UC) and Crohn's disease (CD), are chronic immune-mediated inflammatory conditions that require lifelong treatment [1]. The natural history of IBD is that of quiescent disease interspersed with flares in disease activity with significant socio-economic implications [2,3].
Randomised controlled trials (RCTs), cohort studies, and observational studies reporting on microbiota changes in patient faecal samples and/or mucosal biopsies were evaluated. Eligible studies were included in this systematic review if they met all the inclusion criteria. Studies that reported duplicated results or where data could not be extracted were excluded.
Studies involving adults or children (>5 years old) previously diagnosed with IBD using clinical, endoscopic, and/or pathologic features and using a commonly accepted method to assess disease progression and outcomes were eligible for inclusion. Studies were included if subjects on IBD therapy were followed up for at least two weeks with microbial analysis of faecal or biopsy samples performed at baseline and the endpoint using next-generation or high-throughput sequencing. Microbial analysis reporting was required to provide information on the presence or abundance of microbial taxa.
Studies were excluded if they performed solely culture-based analysis, did not report on patterns of individual bacterial taxa differences, or if the microbial analysis did not in-clude information on the baseline status prior to therapy. Additionally, case reports, studies of fewer than ten patients, studies focusing on children <5 years old, conference articles, review articles, animal studies, studies that did not provide information on antibiotic usage, or studies reported in languages other than English were excluded.

Outcome Assessment
The primary outcomes were (i) changes in the gut microbiome of IBD patients following commencement of treatment with biological therapies, 5-ASA, corticosteroids, antibiotics, or nutritional therapies and (ii) the identification of microbial predictors of treatment response.

Data Extraction and Analysis
The Newcastle and Ottawa scale (NOS) was used to assess the quality of the studies [44] Cohort studies were scored across 3 categories: selection (4 questions) and comparability (1 question) of study groups and ascertainment of the outcome of interest (3 questions) with all questions with a score of 1, except for the comparability of study groups, in which separate points were awarded for adjusting for confounders (maximum of 2 points). An aggregate score of 8 or greater was suggestive of a high-quality study. Different scales were used to assess cohort studies and RCTs.

Data Synthesis and Statistical Methods
Information from the included studies was extracted into a Microsoft Excel spreadsheet. Where available, the following clinical data were extracted for each trial: study type, country of origin, diagnosis, age, duration of disease, intervention and its dosage, duration of intervention, sample types, length of follow-up, method of microbiota composition analysis, and microbiota changes. Studies were eligible for the meta-analysis if more than two studies shared the same outcome and had reported usable data in a compatible metric. Revman5 (Review Manager (RevMan) (Computer program), Version 5.4.1, The Cochrane Collaboration, 2020) was used to conduct the meta-analysis. A meta-analysis of four studies involving 82 patients looking at the mean differences in alpha diversity measured by the Shannon index at baseline and after exposure to biologic treatment with anti-TNFalpha (infliximab) was undertaken. Shannon index values were analysed using an inverse variance model with a 95% confidence interval. Values are reported as standardised mean differences (SMDs). p-values were two-tailed, and statistical significance was set at p < 0.05. The I2 test was used to quantify heterogeneity estimates, i.e., <30%, 30-59%, 60-75%, and >75% defined the thresholds for low, moderate, substantial, and considerable heterogeneity, respectively. When statistically significant heterogeneity was not present, pooled estimates (95% CI) were calculated using the fixed-effects model. If significant heterogeneity was present, pooled estimates (95% CI) were calculated using the random-effects model [45][46][47].

Search Results
The search strategy identified 10,008 references that were imported for screening, and 1266 duplicates were removed. Three reviewers (C.M., G.L., and T.S.) independently performed an initial screen of the titles and abstracts based on the eligibility criteria, with 8479 articles excluded at this stage. Then, 263 studies were assessed for full-text eligibility. Of these, 205 were excluded, and of these, 95 were conference proceedings, 19 full texts that were not retrievable, 16 were review articles/study protocols, 20 reported on the microbiota without using high-throughput sequencing, 12 had insufficient microbial data, 11 lacked longitudinal data, 16 had insufficient data on intervention, 3 had the wrong intervention, 2 had the wrong outcomes, 1 included fewer than six patients, 1 did not specify treatment, 3 had the wrong patient population, and 1 was found to be a duplicate. Disagreements between reviewers were resolved by consensus with third-party experts (G.H. and S.L.). Twenty-eight studies on faecal microbiota transplantation were out of the scope of this systematic review and, therefore, excluded. Thirty-nine studies met the eligibility criteria and were included in the analysis (Figure 1). microbiota without using high-throughput sequencing, 12 had insufficient microbial data, 11 lacked longitudinal data, 16 had insufficient data on intervention, 3 had the wrong intervention, 2 had the wrong outcomes, 1 included fewer than six patients, 1 did not specify treatment, 3 had the wrong patient population, and 1 was found to be a duplicate. Disagreements between reviewers were resolved by consensus with third-party experts (G.H. and S.L.). Twenty-eight studies on faecal microbiota transplantation were out of the scope of this systematic review and, therefore, excluded. Thirty-nine studies met the eligibility criteria and were included in the analysis (Figure 1).

Description of Included Studies
The characteristics of the studies involving biological therapies, non-biological therapies, and nutritional therapies stratified by disease subtypes (studies with CD patients, UC patients, or both) are summarised in Table 1. All thirty-nine studies were prospective in nature, with thirty-four cohort studies and five RCTs. All studies reported the impact of IBD therapeutics on alpha diversity and/or taxonomic changes by conducting microbial assessment at baseline prior to therapy initiation and at least two weeks post therapy exposure. Some studies also reported longer-term outcomes of more than a year [42,48,49]. Nineteen studies involved only CD patients; five studies involved only UC patients; fifteen studies included both UC and CD patients. Twenty-one studies focused on adult subjects, while eighteen studies evaluated paediatric cohorts. Twenty studies reported on biological therapies: anti-TNF-alpha therapies (infliximab, adalimumab, etanercept, golimumab; n = 18), anti-integrin antibody therapy (vedolizumab; n = 2), and anti-IL-12/23 monoclonal antibody therapy (ustekinumab; n = 1) ( Table 1). One study assessed the impact of 5-ASAs, two corticosteroids, two the impact of antibiotics, and one study the

Description of Included Studies
The characteristics of the studies involving biological therapies, non-biological therapies, and nutritional therapies stratified by disease subtypes (studies with CD patients, UC patients, or both) are summarised in Table 1. All thirty-nine studies were prospective in nature, with thirty-four cohort studies and five RCTs. All studies reported the impact of IBD therapeutics on alpha diversity and/or taxonomic changes by conducting microbial assessment at baseline prior to therapy initiation and at least two weeks post therapy exposure. Some studies also reported longer-term outcomes of more than a year [42,48,49]. Nineteen studies involved only CD patients; five studies involved only UC patients; fifteen studies included both UC and CD patients. Twenty-one studies focused on adult subjects, while eighteen studies evaluated paediatric cohorts. Twenty studies reported on biological therapies: anti-TNF-alpha therapies (infliximab, adalimumab, etanercept, golimumab; n = 18), anti-integrin antibody therapy (vedolizumab; n = 2), and anti-IL-12/23 monoclonal antibody therapy (ustekinumab; n = 1) ( Table 1). One study assessed the impact of 5-ASAs, two corticosteroids, two the impact of antibiotics, and one study the impact of azathioprine (Table 1). There were fourteen studies reporting on nutritional therapies, specifically exclusive enteral nutrition (EEN; n = 8), specific carbohydrate diet (SCD; n = 2), prebiotics (n = 2), adsorptive granulomonocytapheresis (GMA; n = 1), and dietary intervention (n = 1).
Quality assessment using the NOS for observational cohort studies and RCTs was undertaken (Supplementary Tables S1 and S2). Scores of 0-3, 4-6, and 7-9 were considered as low, moderate, and high quality, respectively. Of the cohort studies, fifteen scored the maximum 9 points, twelve scored 8 points, four scored 7 points, and three scored 6 points. All except three studies were deemed to be of high quality. Olbjorn et al. did not include Pathogens 2023, 12, 262 6 of 42 information confirming that the use of the IBD treatment of interest was not present at the start of the study [50]. Valcheva et al. and Suskind et al. did not include information on the selection of the non-exposed cohort [51,52]. Similarly, no study control factors involving the IBD subjects were documented in the three moderate-quality studies. Of the RCTs, two studies attained 9 points, two studies scored 7, and one study scored 6 points, meaning four articles were deemed to be of high quality, while the remaining one was of moderate quality. The study of moderate quality did not have information on the representativeness of the controls and was not blinded, and the patients were recruited in a hospital setting [34].

Biologics
Of the twenty studies on biological therapies, eight studies included CD patients only and twelve studies enrolled both CD and UC patients. Four studies were conducted in China; three studies were from Finland; two studies were from Poland and the United States, respectively. Single studies originated from the United Kingdom, Greece, Italy, Czech Republic, Austria, Germany, South Korea, Spain, and Norway. Eighteen studies evaluated the impact of anti-TNF-alpha agents including infliximab, adalimumab, golimumab, and etanercept; two studies investigated the role of anti-integrin antibody (vedolizumab) and interleukin antibody (ustekinumab), respectively (Table 1).
A meta-analysis was carried out on the prospective cohort studies that examined the effects of infliximab on the gut microbiota in CD patients, measured by the Shannon index [38][39][40]54]. Eighty-two subjects were included in the meta-analysis that demonstrated a statistically significant increase in alpha diversity following commencement of infliximab treatment (SMD −1.16 (−1.50, −0.83), p < 0.00001; Figure 2). The degree of variance resulting from between-study heterogeneity, reported as I 2 , was 65%.

Biologics
Of the twenty studies on biological therapies, eight studies included CD patients only and twelve studies enrolled both CD and UC patients. Four studies were conducted in China; three studies were from Finland; two studies were from Poland and the United States, respectively. Single studies originated from the United Kingdom, Greece, Italy, Czech Republic, Austria, Germany, South Korea, Spain, and Norway. Eighteen studies evaluated the impact of anti-TNF-alpha agents including infliximab, adalimumab, golimumab, and etanercept; two studies investigated the role of anti-integrin antibody (vedolizumab) and interleukin antibody (ustekinumab), respectively (Table 1).
A meta-analysis was carried out on the prospective cohort studies that examined the effects of infliximab on the gut microbiota in CD patients, measured by the Shannon index [38][39][40]54]. Eighty-two subjects were included in the meta-analysis that demonstrated a statistically significant increase in alpha diversity following commencement of infliximab treatment (SMD −1.16 (−1.50, −0.83), p < 0.00001; Figure 2). The degree of variance resulting from between-study heterogeneity, reported as I 2 , was 65%.

Microbial Diversity
Microbial species diversity in terms of alpha diversity was reported in 16 studies (Table 2). Measures of alpha diversity among studies varied with the use of the following: Shannon, Simpson, inverse Simpson, observed operational taxonomic units (OTUs)/species, observed number of phylotypes, Chao estimated richness, phylodiversity, Good's coverage, Faith's phylogenetic diversity, Pielou's measure of species evenness indexes, ACE index, and Jackknife diversity. Of the 14 alpha diversity measures, the Shannon index was used in 10 studies, the inverse Simpson index in 5 studies, Chao1 index in 5 studies, and observed species in 4 studies. Raw alpha diversity data were only reported for eleven studies, and data from the remaining studies were obtained through manual extrapolation of published graphs.
Sixteen studies reported the impacts of biological therapies on alpha diversity with fifteen studies focusing on anti-TNF-alpha agents and one on interleukin inhibitor, Ustekinumab. Eight studies included only CD patients; one study included only UC patients, while seven studies included both UC and CD patients. Two studies demonstrated a statistically significant increase in alpha diversity [40,58]. Six studies involving CD patients and one study assessing an IBD cohort observed non-statistically significant changes in all measures of alpha diversity measured [43,59]. Dovrolis et al., who analysed both CD and UC patients, respectively, found reduced alpha diversity (Pielou index) in CD patients (regardless of treatment outcome) and UC responders on infliximab [59]. Similarly, Kolho and colleagues reported the same trend in non-responders on infliximab/adalimumab when measured with the inverse Simpson index [43]; however, they did not

Microbial Diversity
Microbial species diversity in terms of alpha diversity was reported in 16 studies ( Table 2). Measures of alpha diversity among studies varied with the use of the following: Shannon, Simpson, inverse Simpson, observed operational taxonomic units (OTUs)/species, observed number of phylotypes, Chao estimated richness, phylodiversity, Good's coverage, Faith's phylogenetic diversity, Pielou's measure of species evenness indexes, ACE index, and Jackknife diversity. Of the 14 alpha diversity measures, the Shannon index was used in 10 studies, the inverse Simpson index in 5 studies, Chao1 index in 5 studies, and observed species in 4 studies. Raw alpha diversity data were only reported for eleven studies, and data from the remaining studies were obtained through manual extrapolation of published graphs.
Sixteen studies reported the impacts of biological therapies on alpha diversity with fifteen studies focusing on anti-TNF-alpha agents and one on interleukin inhibitor, Ustekinumab. Eight studies included only CD patients; one study included only UC patients, while seven studies included both UC and CD patients. Two studies demonstrated a statistically significant increase in alpha diversity [40,58]. Six studies involving CD patients and one study assessing an IBD cohort observed non-statistically significant changes in all measures of alpha diversity measured [43,59]. Dovrolis et al., who analysed both CD and UC patients, respectively, found reduced alpha diversity (Pielou index) in CD patients (regardless of treatment outcome) and UC responders on infliximab [59]. Similarly, Kolho and colleagues reported the same trend in non-responders on infliximab/adalimumab when measured with the inverse Simpson index [43]; however, they did not stratify their findings based on IBD subtypes. The statistical significance of alpha diversity changes compared to baseline were not reported in four studies [38,43,54,58]. One study evaluating the impacts of ustekinumab reported an increase in alpha diversity (measured by the inverse Simpson index) in both responder and non-responder groups [41]. A statistically significant increase compared to the baseline was only achieved at Week 22 in the responder group (Table 2).
In summary, increased alpha diversity following treatment was associated with response to infliximab, adalimumab, and ustekinumab [41,43,57]. Based on the included studies, the study period varied between 6 and 30 weeks. Despite the variability in the study periods, both the short-and longer-term data support the notion that an increase in alpha diversity is characteristic of patients commencing biologics.

Microbial Composition Changes
Overall, a consistent reduction in the relative abundance of Proteobacteria including Escherichia coli and Enterococcus species and enrichment of more beneficial genera including short-chain fatty acid (SCFA) producers such as Blautia and Faecalibacterium species were observed following biological therapy in the CD, UC, and IBD cohorts [38,39,[41][42][43]53,58,59]. Most studies investigated microbial changes in faecal samples of CD patients treated with infliximab [38][39][40]53,54,57,59]. Wang and colleagues conducted two similarly designed studies analysing the effects of infliximab in paediatric CD patients. In the earlier study (2018), the authors reported a sustained increase of SCFA-producing genera (Blautia, Faecalibacterium, Odoribacter, and Sutterella) and decreased abundance of Enterobacteriaceae, Enterococcaceae, Planococcaceae, and Streptococcaceae post-treatment [38]. Similar increases in the abundance of Blautia were also reported in their later study; however, Clostridium Cluster IV, Collinsella, Eubacterium, and Ruminococcus species were also increased. The study also observed a decreased abundance of Abiotrophia and Lactococcus species [39]. The findings by Zhuang and colleagues, in an adult CD cohort, corroborated the results reported by Wang et al., with a sustained enrichment of SCFA-producing taxa (Lachnospira, Roseburia and Blautia), as well as a reduction in pathogenic bacteria such as Enterobacteriaceae including Escherichia-Shigella and Fusobacterium [40]. Interestingly, in the only study evaluating mucosal biopsies of adult CD patients following infliximab treatment, Dovrolis and colleagues observed an increased abundance of Proteobacteria and Rubrobacter, which is in disagreement with other studies, thus highlighting the different microbial composition of stool versus mucosal samples [59]. In a small study involving both CD and UC patients, the restoration of normal levels of 14 phylotypes including several SCFA-producing bacteria following etanercept treatment was noted [58].
Doherty et al. evaluated microbiota changes following the use of ustekinumab on CD patients [41]. The study reported an increase in Bacteroidetes species, as well as F. prausnitzii, Blautia, Ruminococcaceae, and Roseburia species, which agrees with the anti-TNF-a studies. Following vedolizumab treatment, a reduction in Bifidobacterium longum, Eggerthella, Ruminococcus gnavus, Roseburia inulinivorans, and Veillonella parvula species' abundance were observed between baseline and Week 14 post-treatment in CD patients. These changes were sustained over the 54-week study period [42].

Microbial Predictors of Treatment Outcomes
The association between microbiota changes following biological therapies and treatment outcomes was investigated in seven studies (Table 3). Five studies exploring associations between faecal microbiota profiles and infliximab therapeutic response identified microbial features that distinguished treatment responders from non-responders. Paediatric CD patients with sustained response had a higher abundance of SCFA-producing taxa, Blautia, Faecalibacterium, Lachnospira, and Roseburia compared to non-sustained responders [38]. In a more recent study by the same authors, an increased abundance of Actinomyces, Atopobium, and Parabacteroides genera were observed in patients with sustained treatment response [39]. Interestingly, the authors also demonstrated an association between baseline microbial composition and sustained treatment response. A higher abundance of Methylobacterium, Sphingomonas, Staphylococcus, and Streptococcus genera at baseline was associated with sustained response, with an increased abundance of Clostridium XI and Clostridium XVIII members, Eggerthella, Lachnospiraceae incertae sedis, Parabacteroides, and Peptococcus genera at baseline associated with a loss of response [39]. Increased proportions of Lachnospiraceae and Blautia genera at Week 6 following infliximab treatment were associated with clinical and endoscopic response in an adult CD cohort, whilst CD patients who had a relatively higher abundance of Clostridiales at baseline responded better to infliximab compared to those with lower abundance [49,57]. Clostridiales levels were also restored to almost healthy levels in those achieving clinical remission [61].
In contrast, in the study that analysed mucosal biopsies, the authors identified enterotypes distinct for responders and non-responders, which was inconsistent with the findings of faecal-based studies [59]. A high abundance of Hungatella, Ruminococcus gnavus, and Parvimonas genera at baseline were associated with clinical and endoscopic response, while a higher abundance of Blautia, Faecalibacterium, Roseburia, and Negativibacillus genera in CD patients at baseline was associated with non-response determined by clinical and endoscopic parameters. Increased prevalence of Chloroflexi, Ruminococcus, Eubacterium hallii, Eubacterium eligens, Escherichia, Shigella, and Butyricicoccus genera and decreased abundance of Fusobacteria were noted in CD patients who responded to treatment. In UC, responders had increased Bacteroidetes populations, Veillonella, Tyzzerella, Ruminococcus torques, Parabacteroides, Erisipela, Clostridium, and Bilophila genera, and loss of Spirochaetes and Plantomycetes. Non-responders had an increased abundance of Actinobacteria, Porphyromonas, Granulicatella, and Corynebacterium genera and a reduction in Anaerostipes abundance after 3 months of infliximab [59].
Higher abundance of Bifidobacterium, Clostridium colinum, Eubacterium rectale, an uncultured Clostridiales, and Vibrio species, as well as decreased abundance of Streptococcus mitis at baseline were associated with response to infliximab or adalimumab treatment compared to non-responders [43]. Similar findings were seen in adult CD patients on ustekinumab, with Bacteroides, Faecalibacterium, Blautia, Ruminococcaceae, and Roseburia species significantly more abundant in subjects in remission compared to those with active disease at 6 weeks post-treatment initiation [41]. In a study involving infliximab refractory CD patients switched onto anti-integrin, vedolizumab, a higher abundance of Roseburia inulinivorans and Burkholderiales genera at baseline was associated with remission at Week 14 compared to non-remitters [42]. Further analysis at 1 year post-treatment revealed that these specific microbial changes observed in patients that achieved remission at Week 14 were sustained and the treatment effect was durable.        • Decreased Actinomycetales abundance. • Functional profiling indicated pathways involved in metabolism (protein digestion and absorption, primary and secondary bile acid biosynthesis) and immune response (adipocytokine signalling pathway) were altered following IFX treatment.  Doherty et al. [41] A-CD(n = 306) Following UST treatment, in subjects in remission 6 weeks after induction compared to those with active disease: • Bacteroides and Faecalibacterium species were significantly more abundant. • Faecalibacterium, as well as Blautia, Clostridium XIVa, Ruminococcaceae, and Roseburia genera were all present at higher median.

•
In non-responders, there was a significant increase in Escherichia/Shigella and decrease in Faecalibacterium and Agathobacter genera compared to healthy controls. • An association was found (p < 0.001) in the F. prausnitzii/E. coli ratio between responders and non-responders.

Biological therapies
Ding et al. [48] Adult CD (n = 76) UC (n = 10) Following anti-TNF-alpha treatment: No differences between responders and non-responders; most likely due to sample size.
Schierova et al. [60] CD (n = 17) UC (n= 10) Following IFX/ADA treatment: • 13 taxa in CD and 10 taxa in UC were differentially abundant between baseline and endpoints. • These differentially abundant taxa were structural zeros, except for the increase of Ruminococcus genera in UC.

•
In CD, NRs had lower abundances of SCFA producers from the class Clostridia compared with responders.

•
Odoribacter and unknown Ruminococcaceae genera were significantly increased in Rs compared to NRs; Granulicatella, Enterobacter, and an unknown genus of Peptostreptococcae were increased in NRs compared to Rs.

•
In CD, orders Bifidobacteriales Micrococcales, Lactobacillales, Burkholderiales, and Pseudomonales were significantly more abundant in NRs, whereas Bacteroidales and Desulfovibrionales were significantly elevated in Rs.

•
In UC, the order Bacteroidales, families Enterococcae, Clostridiaceae, Peptostreptococcaceae, and Ruminococcaceae and seven genera differed significantly between groups NR and R.

Biological therapies
Effenberger et al. [62] CD R (n = 13) CD NR(n = 11) UC R (n = 7) UC NR (n = 5) Following anti-TNF treatment: • Significant decrease of Proteobacteria and associated increase in Bacteroidetes seen in CD responders but not in UC.

•
In CD patients, there were higher Lactobacillus abundances which were associated with non-remission.

•
In CD remission group, the trend of higher Bacteroides abundance at baseline and follow-up time points were noted.

•
In UC cohort, no such associations detected.

•
Increased abundance of Actinobacteria and Ruminococcus similar to that of control group. • Decreased abundance of Enterococcaceae and Enterococcus faecium.

Non-biological therapies
Effenberger et al. [62] CD R (n = 10) CD NR (n = 9) UC R (n = 5) UC NR (n = 5) Following AZA treatment: • Significant decrease of Proteobacteria and associated increase in Bacteroidetes seen in CD responders but not in UC.

•
In CD patients, there were higher Lactobacillus genera abundances, which were associated with non-remission.

•
In CD remission group, trend of higher Bacteroides abundance at baseline and follow-up time points were noted (more pronounced in AZA than anti-TNF group).

•
Klebsiella species was significantly associated with AZA failure in CD patients.

•
In UC cohort, no such associations detected.

•
Recurrence of CD corresponded with an increase in OTUs.
Quince et al. [65] Paediatric CD (n = 23) HC (n = 21) Following EEN: • Reduction in relative abundance in majority of genera, except Lactococcus. • Significant reduction in Bifidobacterium, Ruminococcus, and Faecalibacterium, whose abundance was lower than HC prior to EEN. • OTU analysis identified reduction of Bifidobacterium and increased in R. gnavus in most subjects.

•
Oligotypes of Lachnospiraceae decreased in abundance. Dunn et al. [67] Paediatric CD (n = 10) HC (n = 5) Following EEN: • Characteristic changes differed between sustained remission group and non-sustained remission group over 12 weeks: Decreased proportion of Firmicutes in sustained remission group, but increased proportion in non-sustained remission group. Increased proportion of Bacteroidetes in sustained remission group, but decreased proportion in non-sustained remission group.
• At baseline, non-sustained remission group had a large proportion of Proteobacteria compared to sustained remission group. • Sustained remission group had a larger proportion of Firmicutes and smaller proportion of Bacteroidetes.
Costa-Santos et al. [68] A -CD (n = 10) Following EEN: • Increased relative abundance of Firmicutes (Clostriadiales order). • Decrease relative abundance of Bacteroidetes (Bacteroidales order) and Proteobacteria (Enterobacteriales order). • High inter-individual variation in microbiota composition. • Significant reduction of Blautia and Subdoligranulum genera abundance during 6 weeks of EEN, but the reduction was not significant compared to baseline at the end of 6 weeks of EEN. • Presence of Dorea longicatena, Blautia obeum, Bifidobacterium longum, and Escherichia coli at baseline were associated with NR. • After 9 weeks, higher relative abundances of butyrate-producing genera Roseburia and Faecalibacterium were associated with R. • LEfSe analysis associated responders with increased F. prausnitzii, B. adolescentis, and R. bromii and non-responders with increased E. coli and four Blautia OTUs at 9 weeks.

•
Decreased relative abundance of Escherichia coli and a strain of F. prausnitzii.

•
Halomonas was decreased in both groups.
Jiang et al. [70] CD (n = 7) Following EEN treatment: • Firmicutes had a higher abundance compared to baseline. • Proteus levels decreased in abundance compared to baseline. • Post-treatment microbiota comprised mostly of Proteobacteria, Firmicutes, Bacteroides, Actinobacteria, and Verrucomicrobia. • Increased Bifidobacteriaceae and Lachnospiraceae genera abundance, but not associated with improved disease scores. • Increased colonic butyrate production.

•
Increased Bifidobacterium and Christenellaceae genera in those in remission at baseline.

•
Decreased Dialister genus in those not in remission.

Non-Biological Therapies Microbial Diversity
Seven studies evaluated the impact of non-biologic therapeutics on alpha diversity, and adequate data were available for extraction from four studies ( Table 2). Schierova et al. reported that 5-ASA use in UC patients was associated with an increase in alpha diversity after 14 days of treatment [33]. The extent of alpha diversity increase was greater in treatment responders than non-responders; however, the results did not reach statistical significance. Hart et al. demonstrated that corticosteroid treatment was associated with a statistically significant increase in alpha diversity (Shannon index) in a cohort of CD and UC patients [35]. The increase in alpha diversity measured from baseline continued throughout 8 weeks of therapy and persisted at Week 12. Sprockett and colleagues evaluated the therapeutic impact of three antibiotic regimens: (1) Metronidazole (MET), (2) MET, and Azithromycin (AZ) or (3) initial MET and the combination of MET+AZ over a 12-week period in a CD cohort [41]. Alpha diversity, measured by Faith's PD index, decreased over 8 weeks of treatment, rebounding slightly at Week 12 in all groups. No significant changes in the distance between baseline and Weeks 12 or 30 in either patients in remission or without remission were noted following azathioprine treatment [62].

Microbial Composition
There were six studies on non-biological therapies. Pigneur et al. observed an enrichment of SCFA-producing genera including Roseburia intestinalis, Eubacterium species, Ruminococcus, and Bifidobacterium bifidum, along with reduced numbers of Blautia species at Week 8 following corticosteroid treatment in a small paediatric CD study [34]. A similar increase in uncharacterised Ruminococcaceae was observed in paediatric UC and CD patients, which also reported an increased abundance of Blautia and Sellimonas species and a decreased abundance of Granulicatella, Haemophilus, and Streptococcus genera at Week 12 compared to baseline [35].
Antibiotic treatment is known to induce an ecosystem-wide disturbance and decrease microbial diversity. A Japanese study evaluated the impact of combination antibiotic regimen (amoxicillin, fosfomycin, and metronidazole) in nineteen adult patients with active UC [37]. The treatment resulted in an increased abundance of Proteobacteria, particularly Enterobacteriaceae, and a near complete depletion of Bacteroidetes species. Some recovery in the proportion of Bacteroidetes species was observed at both 4 and 8 weeks post-treatment [37]. A large multinational longitudinal study involving CD paediatric patients investigated the microbial impacts of two specific antibiotic regimens, MET and MET+AZ [36]. The finding of increased abundances of potentially pathogenic Enterococcus in the MET group concurred with the Japanese study's findings, which may indicate the presence of antibiotic resistant microbes and their enrichment predisposing patients to invasive infections. Interestingly, the opposite effect was observed in the MET+ AZ treatment, which might be reflective of the effects of AZ against adherent and invasive E. coli strains [36]. The 5-ASA study failed to observe specific microbial changes over the 12-week study period [33]. CD patients not benefiting from azathioprine treatment had increased abundance of Lactobacillus and Klebsiella genera [62]. In contrast, AZA responders had a significant decrease in Proteobacteria and an associated increase in Bacteroidetes (Table 3).

Nutritional Therapies
Of the fourteen studies on nutritional therapies, five studies were conducted in North America, four studies in China, three studies in Europe, and one study in Australia and Canada, respectively. Nine studies involved solely CD patients; three studies enrolled UC patients; two involved both UC and CD patients. Eight of the studies evaluated the impact of EEN (6 paediatric and 2 adult cohorts); two paediatric studies assessed the impact of SCD; two adult studies assessed the impact of the prebiotic inulin and galactooligosaccharide (GOS) supplement, respectively; one study assessed adsorptive (GMA); another studied dietary intervention on adult cohorts (Table 1).
When the impact of inulin-type fructan supplementation was assessed, Valcheva et al. showed an increase in Bifidobacteriaceae and Lachnospiraceae abundance along with an increase in colonic butyrate levels in UC patients, but this did not correlate with disease activity scores [51]. GOS prebiotic treatment was associated with a reduction in Oscillospira and Dialister abundance and an increase in Alistipes (Table 3).

Discussion
The concept of altering the gut microbiota to improve health is now a well-established concept in medicine. Microbiome-based therapies include dietary interventions, prebiotics, probiotics, antibiotics, phage therapy, FMT, live biotherapeutics, and microbiome mimetics. In the context of IBD, understanding the interaction between gut microbes, IBD therapeutics, as well as environmental factors and disease progression is essential if we want to achieve symptom resolution and/or avoid adverse reactions. Several prospective studies have investigated the role of gut microbiota as a potential biomarker for treatment response; yet, a systematic evaluation of microbiota changes following IBD treatment has not been undertaken. Therefore, the aims of this systematic review and meta-analysis were to define the microbial changes following pharmacological and nutritional IBD treatments and identify microbial predictors of treatment response.
The review demonstrated a consistent increase in alpha diversity indices following biological therapies, 5-ASA, and corticosteroids. Additionally, higher microbial diversity at baseline and/or following treatment was shown to be predictive of treatment response in studies on infliximab, adalimumab, and ustekinumab [41,43,57,59]. Increased microbial richness in the gut has been strongly associated with good health, and a less diverse microbiome is consistently implicated in IBD [19,75]. Therefore, a more diverse gut microbiome at baseline might indicate a greater abundance of microbial taxa with anti-inflammatory properties (directly or metabolite mediated). A subsequent reduction in colonic inflammation and preservation of mucosal barrier function/integrity may lead to stronger treatment response. However, the exact mechanisms by which these therapeutics impact microbial diversity remains to be elucidated. The studies included in this review tracked alpha diversity changes for varying periods ranging from 2-30 weeks. Considering that IBD is a life-long disease that is progressive in nature, it would be of significant interest to evaluate if the increase in microbial diversity can be sustained longitudinally and its association with long-term health outcomes. By generating this insight through extended interrogation of large longitudinal IBD cohorts, which comprise a range of disease presentations, as well as ethnic diversity, will contribute to the improvement of IBD treatment results and also IBD management.
Comparison of the studies looking at biologic therapies demonstrated considerable heterogeneity in microbial changes in response to treatment; however, an enrichment of putative SCFA-producing bacteria was consistently noted. SCFAs such as acetate, butyrate, and propionate exert immunomodulatory and anti-inflammatory effects through mechanisms such as the modulation of intestinal wall permeability and reducing oxidative stress and decreasing inflammation through inhibition of mediators such as NF-κB and IL-8 and are primary sources of energy for colonic epithelial cells [76][77][78]. Therefore, the expansion of SCFA-producing bacteria, following certain treatments, highlights that those alterations in microbial metabolic functions could be more important than identifying individual taxonomic changes in the context of IBD. This concept accords with a recent study by Anathakrishnan et al., which observed that there were significantly greater metagenomic alterations in microbial functions than there were changes in microbial composition in both CD and UC cohorts following treatment with the anti-integrin agent, vedolizumab [42]. This highlights the need for future studies to characterise both compositional and functional alterations in the gut microbiota.
Evidence supporting the role of the gut microbiota as a predictor of treatment response is accumulating, with most studies in the systematic review identifying microbial predictors associated with treatment outcomes. An important consideration is understanding that IBD therapeutics can alter microbial profiles, and conversely, the gut microbiota can influence therapeutic outcomes. Interestingly, the observation in three studies on infliximab and vedolizumab that certain baseline signatures are predictive of treatment response or nonresponse is strongly suggestive that specific gut microbial factors can influence treatment efficacy [39,42,43]. However, the findings were strikingly heterogenous between studies; thus, it was not possible to perform a meta-analysis to evaluate the potential of microbial predictors of treatment response in IBD reported in these studies. Therefore, the microbiota results have to be interpreted with caution, especially since significant variation between studies in terms of subject age, disease subtype and severity, clinical outcome assessments, geography, medications, the length of the study period, and analysis techniques were noted.
The modulation of the gut microbiota by faecal microbiota transplantation (FMT) is now being used to induce remission in UC. Published randomised clinical trials have highlighted that clinical response relies on the enrichment of certain beneficial bacteria including SCFA-producing strains. Interestingly, recent evidence suggests that baseline microbiome profiles are a good predictor of response, with patients achieving a successful outcome having significantly higher microbial richness prior to, during, and after FMT [79]. This further reinforces the potential of baseline or pretreatment microbiome assessment in defining therapeutic outcomes.
There are several limitations in this review. First, the majority of the studies had small subject numbers and, notably, the patients that were included in the detailed longitudinal microbial analysis often represented a significantly smaller subset of the actual cohort. Another significant limitation of the studies included in this systematic review is that most did not account for confounders that are known to impact gut microbiota such as dietary intake and polypharmacy effects. The lack of studies focusing on certain treatments (vedolizumab (n = 1), ustekinumab (n = 1), etanercept (n = 1), 5-ASA (n = 1), antibiotics (n = 2), corticosteroid (n = 2), prebiotic (n = 1), and SCD (n = 2) also limited the comparisons. Additionally, the lack of standardisation in disease phenotyping, microbial analysis methods, defining outcome assessments (clinical/endoscopic response/remission), study periods, and sampling points remain major barriers to understanding the intricate association between the gut microbiota and therapeutics. Overall, the heterogeneity of the gut microbiota findings between different patient populations is challenging to interpret. These limitations must be considered within the design of future studies. This includes the need to focus on well-phenotyped prospective longitudinal patient cohorts, the appreciation of inflammatory and treatment confounders, as well as considering differences due to geography, age, and diet. Finally, harmonisation across studies in terms of establishing a robust standardised scientific methodology is fundamental to validate, reproduce, and ensure the quality of findings in future research.

Conclusions
In conclusion, this systematic review contributes to the growing appreciation of the relationship between gut microbes and IBD therapeutics; however, defining optimal treatment strategies remains a challenge. The identification of microbial predictors of treatment response is fundamental to achieving precision medicine in IBD; however, there is still a long way to go before these findings can be translated into clinical practice. Further validation in large longitudinal prospective cohorts is needed.
Supplementary Materials: The following Supporting Information can be downloaded at: https:// www.mdpi.com/article/10.3390/pathogens12020262/s1, Table S1: Summary of Newcastle-Ottawa Scale of each included cohort study; Table S2: Summary of Newcastle-Ottawa Scale of each included RCT.