Review article: the gut microbiome in inflammatory bowel disease avenues for microbial management

Transcription

1 Received: 5 May 2017 First decision: 29 May 2017 Accepted: 25 September 2017 DOI: /apt Review article: the gut microbiome in inflammatory bowel disease avenues for microbial management J. McIlroy 1 G. Ianiro 2 I. Mukhopadhya 1 R. Hansen 3 G. L. Hold 1 1 School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Aberdeen, UK 2 Internal Medicine, Gastroenterology and Liver Unit, Gastroenterology Area, Fondazione Policlinico Universitario Gemelli, Universita Cattolica del Sacro Cuore, Rome, Italy 3 Royal Hospital for Children, Glasgow, UK Correspondence Dr G Hold, Gastroenterology Research Group, Institute of Medical Sciences, University of Aberdeen, Aberdeen, UK. g.l.hold@abdn.ac.uk Funding information None. Summary Background: The concept of an altered collective gut microbiota rather than identification of a single culprit is possibly the most significant development in inflammatory bowel disease research. We have entered the omics era, which now allows us to undertake large-scale/high-throughput microbiota analysis which may well define how we approach diagnosis and treatment of inflammatory bowel disease (IBD) in the future, with a strong steer towards personalised therapeutics. Aim: To assess current epidemiological, experimental and clinical evidence of the current status of knowledge relating to the gut microbiome, and its role in IBD, with emphasis on reviewing the evidence relating to microbial therapeutics and future microbiome modulating therapeutics. Methods: A Medline search including items intestinal microbiota/microbiome, inflammatory bowel disease, ulcerative colitis, Crohn s disease, faecal microbial transplantation, dietary manipulation was performed. Results: Disease remission and relapse are associated with microbial changes in both mucosal and luminal samples. In particular, a loss of species richness in Crohn s disease has been widely observed. Existing therapeutic approaches broadly fall into 3 categories, namely: accession, reduction or indirect modulation of the microbiome. In terms of microbial therapeutics, faecal microbial transplantation appears to hold the most promise; however, differences in study design/methodology mean it is currently challenging to elegantly translate results into clinical practice. Conclusions: Existing approaches to modulate the gut microbiome are relatively unrefined. Looking forward, the future of microbiome-modulating therapeutics looks bright with several novel strategies/technologies on the horizon. Taken collectively, it is clear that ignoring the microbiome in IBD is not an option. The Handling Editor for this article was Professor Jonathan Rhodes, and this commissioned review was accepted for publication after full peer-review John Wiley & Sons Ltd wileyonlinelibrary.com/journal/apt Aliment Pharmacol Ther. 2018;47:26 42.

2 MCILROY ET AL INTRODUCTION The gut microbiota comprises a large collection of microbes and the largest within the body, reaching cells/g of luminal contents in the colon. 1 In healthy individuals, it is dominated by 2 major bacterial phyla: Bacteroidetes and Firmicutes, with smaller representation from Proteobacteria and Actinobacteria. 2,3 We are relatively ignorant of the functional and structural contributions of microorganisms outside of gut bacteria, with only a handful of publications looking at the gut virome (viruses and bacteriophages), mycome (fungi) and other micro-eukaryotes including protozoa. 4-6 The gut microbiota performs a number of crucial functions for the host including priming the immune system, breakdown of dietary substrates inaccessible to host enzymes, and detoxification of xenobiotics. 7 The healthy gut microbiome exhibits considerable functional diversity and possesses far greater genomic potential compared to its host. Perhaps of most interest, it is inherently modifiable. The implication of this interplay between humans and microbes is that pharmacological therapies, nutrient modifications and associated interventions that are targeted at the host will also significantly impact on the gut microbiota. Genome-wide association studies (GWAS) have now identified 235 inflammatory bowel disease (IBD)-associated susceptibility loci, substantially expanding our understanding of the biology underlying these diseases. 8,9 Early genetic studies focused on searching protein coding sequences, although it is now recognised that coding variation explains only ~20% of genetic variation associated with IBD GWAS loci. 10 These studies highlighted the pivotal role of host:microbial interactions in IBD pathogenesis, specifically identifying T-cell activation, IL-23/T helper 17 pathway, autophagy and microbial recognition The most recent studies have undertaken low-coverage whole genome sequencing to interrogate low-frequency variants and define how much these variants contribute to IBD susceptibility. This approach has identified a missense variant in ADCY7, with mechanistic interpretation being that loss of function reduces camp production leading to an excessive inflammatory response that predisposes to ulcerative colitis but not Crohn s disease (CD). 8 2 DEFINING THE GUT MICROBIOTA IN INFLAMMATORY BOWEL DISEASE A change in the normal gut microbiota with a shift away from host:microbial mutualism has been reported in many IBD studies; however, the vast majority of studies to date have focussed on bacterial changes and only recently has consideration of fungal and viral constituents been forthcoming. Some bacterial changes appear to be clearly linked to either CD or ulcerative colitis, while others appear to be attributed more generally to IBD The most consistent changes are reduction in biodiversity (lower number of species), with lower proportions of Firmicutes, consistently reported alongside increases in Proteobacteria 16,21-24 and Bacteroidetes phylum members 19 although reductions have also been reported. 15 Spatial re-organisation of Bacteroides species has also been documented with higher proportions of Bacteroides fragilis being seen in IBD patients. 20 In CD, changes in Firmicutes have particularly documented changes in Faecalibacterium prausnitzii. 25 Faecalibacterium prausnitzii levels have been consistently lower in ileal CD patients, although increases were noted in paediatric CD, potentially suggesting that further assessment is required Additional interrogation of F. prausnitzii has identified that F. prausnitzii populations are patient-specific and there are differing functional capabilities between strains with the tantalising potential that differing phylotypes may be having opposing effects. 28 Increases in Proteobacteria, especially Escherichia coli, including pathogenic variants, are also reported in ileal CD. 29 Changes in bacterial functional capabilities have been seen in ileal CD with alterations in bacterial carbohydrate metabolism, bacterial:host interactions and host secreted enzymes noted. 30 Perturbations to the structure of the gut microbiota, termed dysbiosis, have also been shown to impart functional changes within the host creating a pro-inflammatory state. 31 More broadly, microbial functional changes associated with inflammatory bowel disease have identified enrichment in host metabolite uptake, oxidative stress tolerance and immune evasion, alongside decreases in microbial metabolism, including short chain fatty acid (SCFA) biosynthesis and amino acid biosynthesis. Put simply, the changes seen in gut microbial structure are associated with major metabolic impairments, which impart huge functional consequences to the host. Therefore, modulation of gut microbiota, through diet, antibiotics, prebiotics, probiotics and faecal microbiota transplantation (FMT), represent, at least theoretically, a promising therapeutic avenue for the management of inflammatory bowel disease, although available evidence suggest that fine tuning of our therapeutic offerings is still needed (Figure 1). 32 Being able to harness this potential is the fundamental basis of emerging therapeutic approaches to inflammatory bowel disease management, with microbial modulation becoming a major consideration in inflammatory bowel disease management over the last decade. Potential drawbacks of most of the current literature, in terms of defining microbial changes in inflammatory bowel disease, are that studies have primarily reported on cohorts with established disease and they often fail to appreciate that mucosal and faecal microbiota are different. The confounding impact of therapeutic regimes (often complex and evolving), disease chronicity and also surgical intervention make it challenging to decipher whether taxonomic changes reflect disease-driven changes or are merely a response to a drastically altered intestinal environment. In order to address this a small number of studies looking at the microbiota of newly diagnosed patients have been published. The first study looked at mucosa-associated changes in newly diagnosed children (13 CD and 12 ulcerative colitis patients), all of whom were assessed at first presentation of active disease. However, importantly for microbiome analysis, the patients had not received systemic antibiotics or steroids in the 3 months prior to investigation or immunosuppression at any time. 33 Microbial diversity was significantly reduced in CD patients, compared to ulcerative colitis and control patients. Higher abundance of F. prausnitzii was also seen in CD patients, in marked contrast to numerous other reports. 26,28,34-37 A larger study comprising of 447

3 28 MCILROY ET AL. Dramatically alters structure and function of gut microbial community Case reports only, needs Randomised control trial evidence Faecal Microbial Transplantation Faecal microbial transplantation Emulsifiers Probiotics Ulcerative colitis Inflammatory bowel disease Crohn s disease Exclusion diets Reduce bacterial substrates Alter bacterial diversity Reduce bacterial diversity Antibiotics Pouchitis Inflammatory bowel disease unclassified Antibiotics Reduce bacterial diversity Microbial therapy approaches currently unclear Microbial pathogenic potential Inflammatory bowel disease Missing or depleted key species Roseburia Faecalibacterium prausnitzii Butyrate producing species Excess of pathogenic species Proteobacteria Fusobacteria Mucin degrading species Inflammatory bowel disease FIGURE 1 Current microbial therapeutic mechanisms of action. Modulation of gut microbiota through diet, antibiotics, prebiotics, probiotics and faecal microbiota transplantation, which reflects the existing microbial therapeutic avenues for inflammatory bowel disease management. The evidence for use of the existing approaches depends on the type of disease present, with ulcerative colitis appearing to benefit most from approaches, such as faecal microbiota transplantation, which dramatically alters gut microbial community structure and function. Dietary modulation through exclusive enteral nutrition demonstrates success in paediatric Crohn s disease and is thought to work by reducing the bacterial diversity within the gut. Overall, however, the balance of evidence suggests that fine tuning of our therapeutic offerings is still needed paediatric CD patients, aged 3-17 years, alongside 221 controls demonstrated that several taxa could discriminate disease phenotype, including Enterobacteriaceae, Bacteroidales, Clostridiales, Haemophilus spp., Veillonellaceae, Neisseriaceae and Fusobacteriaceae. 34 Taken collectively, these studies highlight that the microbiota changes associated with treatment naive, new-onset inflammatory bowel disease are different from established disease and require additional/further consideration. Serial follow-up of such patient cohorts potentially offers a unique opportunity to study the purest profile of inflammatory bowel disease-related microbiota changes. 3 MUCOSAL VS FAECAL MICROBIAL POPULATIONS When considering the issue of mucosal vs faecal populations, there is limited comparison of both sample types within the same patient cohort, highlighting a gap in our understanding. The current evidence would indicate that only limited differences exist between different mucosal sites, or between inflamed and uninflamed mucosa, but significant differences exist between mucosal and luminal microbial populations even within healthy individuals. 34,38,39 Table 1 summarises publications which describe microbial composition between IBD (either CD or ulcerative colitis) and healthy controls, where next-generation sequencing was used as the principal analysis tool (Table 1). In CD, a systematic review, published in 2015, documented the microbial alterations seen in luminal (faecal) vs mucosal samples. 40 In total 73 studies were included with 6 or more CD patients compared to healthy controls. Post-operative CD was not included in the review. In both luminal and mucosal samples, a reduction in microbial richness was apparent and mainly attributed to reduction of diversity within the Firmicutes phylum. Increases in Bacteroidetes and Enterobacteriaceae (Proteobacteria phylum) were seen, with E. coli specifically noted to rise in both tissue and faecal

4 MCILROY ET AL. 29 TABLE 1 Microbial alterations in IBD as determined by high-throughput sequencing. Only studies including 10 or more patients per group have been included 30 Reference Cohort description Sample type Findings Rehman Adult (28 CD, 30 UC, Mucosal colonic biopsies et al 183 control) Reduced Faecalibacterium prausnitzii, Bacteroides, Blautia, Ruminococcus, Roseburia, Coprococcus, Lachnospiraceae, UC microbial signature Limited changes noted Eun et al 184 Adult (35 CD, 15 control) Mucosal colonic biopsies (T) and subgroup of faecal samples (F; n = 25) Increased Enterobacteriaceae, Fusobacteriaceae (T) Increased Enterobacteriaceae, Pseudomonadaceae, Streptococcaceae and Erysipelotrichaceae (F) Reduced Bacteroidaceae, Prevotellaceae, Lachnospiraceae and Ruminococcaceae and Veillonellaceae (T, F) Reduced microbial richness (F) Paediatric (n = 13 CD, 10 Alipour Mucosal ileal biopsies et al 185 UC, 12 control) Limited changes noted UC microbial signature Reduced microbial richness Mar et al 186 Adult (30 UC, 13 control) Faeces UC microbial signature Reduced microbial richness Reduced Bacteroides, Prevotella and a number of unclassified Lachnospiraceae and Ruminococcaceae Enrichment of Streptococcus, Bifidobacterium and Enterococcus genera Shah et al 187 Paediatric (10 UC, 13 control) Quince et al 170 Paediatric (23 CD, 21 control) Haberman et al 188 Gevers et al 34 Tong et al 189 Rajilic- Stojanovic et al 190 Paediatric (n = 243 CD, 73 UC, 43 control) Paediatric (n = 447 CD, 221 control) Adult (n = 16 CD, 16 UC, 32 control) Adult (n = 15 UC, 15 control) Mucosal colonic biopsies Faecal Mucosal ileal biopsies Mucosal biopsies (T) and subgroup of faecal samples (F; n = 223) Colonic lavage Faeces UC microbial signature Reduced Verrucomicrobia, Roseburia, Akkermansia Increased Haemophilus Reduced microbial richness Reduced Lachnospiraceae, Subdoligranum and Faecalibacterium Increased Peptostreptococcus, Atopobium and Enterobacteriaceae Increased Neisseriaceae, Gemellaceae, Fusobacteriaceae, Veillonellaceae, Pasturellaceae, Enterobacteriaceae and Epsilonproteobacteria Reduced Bifidobacteriaceae and Firmicutes including Lachnospiraceae, Clostridiales and Erysipelotrichaceae. UC microbial signature Limited changes noted Increased Neisseriaceae (T), Gemellaceae (T), Fusobacteriaceae (T), Veillonellaceae (T, F), Pasturellaceae (T), Enterobacteriaceae (T) Reduced Bacteroidales (T, F), Clostridiales (T, F), Erysipelotrichaceae (T) and Bifidobacteriaceae Reduced microbial richness Increased Actinobacteria UC microbial signature Reduced microbial richness Increased Actinobacteria UC microbial signature Reduced Ruminococcus bromii, Eubacterium rectale, Roseburia sp. and Akkermansia sp Increased Fusobacterium sp., Peptostreptococcus sp., Helicobacter sp., Campylobacter sp. and Clostridium difficile (Continues)

5 30 MCILROY ET AL. TABLE 1 (Continued) Reference Cohort description Sample type Findings Hansen et al 25 Paediatric (n = 13 CD, 12 UC, 12 control) Kaakoush Paediatric (n = 19 CD, 18 et al 191 control Morgan et al 21 Adult (n = 121 CD, 75 UC, 27 control) Willing et al 192 Adult twin pairs (n = 29 CD, 15 UC, 46 control) Mucosal biopsies Faeces Mucosal biopsies (n = 95) and faecal samples (n = 136) Faeces (F) and subgroup of biopsy samples (T; n = 18) Reduced microbial richness Increased Faecalibacteria prausnitzii UC microbial signature Limited changes noted Reduced Firmicutes including Clostridiales, Roseburia, Coprococcus, Ruminococcaceae Increased Bacteroidetes and Enterobacteriaceae Reduced Roseburia, Phascolarctobacterium, Ruminococcaceae, Faecalibacteria prausnitzii (ileal disease) Increased EnterobacteriaceaeUC microbial signature Reduced Roseburia, Phascolarctobacterium, Leuconostocaceae, Odoribacteriaceae Ileal Reduced F. prausnitzii, Roseburia, Bacteroidetes, Collinsella (F, T) Increased Enterobacteriaceae, Ruminococcus gnavus Colonic Increased Faecalibacteria prausnitzii, Roseburia, Bifidobacteriaceae, Coriobacteriaceae, Ruminoccocaceae, Anaeroplasmataceae (F, T) UC microbial signature Limited changes noted CD, Crohn s disease; UC, ulcerative colitis; F, depicts changes associated with faecal samples, T = depicts changes associated with tissue/mucosal samples. samples, relative to healthy controls. Reductions in the Firmicutes member F. prausnitzii were also noted, particularly in ileal CD. 40 Abnormalities in the intestinal microbiota have also been reported in ulcerative colitis, although to a lesser degree compared to CD. Less diverse microbiota profiles have been demonstrated in ulcerative colitis patient samples, and in particular, the finding of increased Clostridium perfringens in faeces suggests a role in disease exacerbation. 41 A decrease in Fusicatenibacter saccharivorans in patients with active ulcerative colitis has also been reported, in contrast to the increase observed in patients with quiescent disease THE ROLE OF THE VIROME Newer methodological techniques in the metagenomic assessment of the gut virome as a composite entity has opened new avenues in our understanding of its role in the aetiopathogenesis of inflammatory bowel disease. Expansion of annotated viral databases and improved bioinformatic analysis have also increased the ability to correctly identify a larger cohort of these viruses. An initial study utilising epifluorescence microscopy was the first to suggest that there was an increase of bacteriophages in patients with CD as opposed to controls. 43 The predominant gut viruses identified were double-stranded DNA viruses in the Caudovirales order (including Podoviridae, Siphoviridae and Myoviridae). This finding was subsequently confirmed by a study utilising metagenomic assessment techniques on biopsy samples and gut washes in paediatric CD patients, wherein Wagner et al demonstrated that the maximum number of viral hits were seen in patients with CD. 44 Subsequently, a study by Wang et al looked at these changes in colonic biopsy samples from patients with CD and controls and showed alteration of virome abundance and diversity in the former group. 45 The study by Perez-Brocal was the first to document parallel changes between the bacterial and viral components of the gut microbiota in patients with inflammatory bowel disease. They demonstrated that there was a decrease in diversity and abundance of both the bacteria and viruses in patients with CD. 46 The most comprehensive analysis of the disease-specific alteration of the virome in IBD was reported by Norman et al. 47 The study demonstrated an expansion of Caudovirales bacteriophages in patients with both ulcerative colitis and CD but more importantly documented an inverse correlation with the bacterial component in both these diseases. However, the inverse relationship between the bacteria and the virome could not be reliably replicated in the validation USA cohorts in the study. 47 Subtle distinctions were noted between patients with CD and ulcerative colitis, wherein in the former group of subjects, the presence of Caudovirales were positively correlated with Enterobacteriaceae, Pasteurelloacaeae and Prevotellaceae but there was no such relationship noted with UC. Perez-Brocal also demonstrated the positive representation of Synechococcus phage S CBS1 and Retroviridae family of viruses in patients with CD, with this propensity suggesting that they could be potential biomarkers. 46

6 MCILROY ET AL. 31 Both these observations suggest that phenotypic differences in manifestations of ulcerative colitis and CD might be explained by the different changes in the gut virome. However, all these findings need to be consistently validated in other studies, and the bridge from association to causation still needs to be crossed. 5 THE ROLE OF THE MYCOBIOME Similar to virome analysis, fungi have long been suspected to play a role in IBD pathogenesis. Anti-Saccharomyces cerevisiae antibodies are well-known markers for CD. In mice, gut inflammation and antibiotic usage have been shown to promote fungal proliferation. 48 Other studies have shown that fungi can modulate susceptibility to inflammation in a negative (Candida albicans) or positive (Saccharomyces boulardii) manner. 49 Finally, mice lacking major genes involved in fungal sensing, such as Dectin-1 or Card9, have an increased fungal microbiota load and are more susceptible to colitis, with human genetic studies also confirming a role for CARD 9. 50,51 Recently, the pace of investigation has gathered momentum, although most clinical studies to date involve small patient cohorts and therefore lack definitive power. Looking at fungal diversity, the consensus of findings indicates that the 2 major fungal phyla detected in the human gut are Ascomycota and Basidiomycota 50,52-54 and that, numerically, fungal DNA is not a major constituent of the microbiome, with 99.1% of the genetic catalogue from the gut lumen being of bacterial origin, whereas fungal DNA accounts for around 0.02% of the entire mucosa-associated microbiota Mukhopadhya et al were the first to look at the mycobiome in de novo paediatric inflammatory bowel disease patients and noted that there was an overwhelming predominance of the Basidomycota phylum in IBD patients, although sample size was limited, and each patient demonstrated a unique fungal signature. 4 Interestingly, the study was undertaken on mucosal biopsy samples rather than faecal samples. Higher levels of Basidiomycota were recently reported in mucosal biopsies from a treatment na ıve paediatric CD cohort from Saudi Arabia compared to control patients. 59 Interestingly, the mycobiome diversity analysis of faecal samples from the same cohort did not show the same fungal profile, highlighting similar findings to bacterial diversity studies which clearly demonstrate differing microbial consortia between mucosal and luminal samples. A mucosa-based study on 23 CD adult patients also demonstrated similar findings. 60 Sokol et al recently published a larger faecal-based adult inflammatory bowel disease study comprising 235 well-phenotyped inflammatory bowel disease subjects and 38 healthy subjects. 61 They identified a disease-specific fungal dysbiosis with shifts in composition involving the 2 dominant fungal phyla, and several fungal species including S. cerevisiae, Malassezia sympodialis and C. albicans. Bacterial biodiversity was investigated and shown to decrease in both CD and ulcerative colitis; however, fungal biodiversity was decreased only in ulcerative colitis, indicating that a CD -specific gut environment may favour fungi at the expense of bacteria. They explored the equilibrium between bacterial and fungal diversity in the gut and determined that the fungi-to-bacteria diversity ratio was increased in IBD, particularly in ileal CD and flares. Interestingly, an increase in Basidomycota abundance was associated with disease flare, matching the findings of the de novo paediatric studies, which exclusively assessed active disease. Inter-kingdom analysis (bacterial and fungal) was also conducted by Sokol et al, and this demonstrated correlations between bacterial and fungal components with differences seen between inflammatory bowel disease and healthy subjects, suggesting the existence of disease-specific inter-kingdom alterations. 6 MICROBIOTA AS A THERAPEUTIC TARGET IN INFLAMMATORY BOWEL DISEASE 6.1 Antibiotics Generally, antimicrobial drugs are known to influence the natural history of several disorders, including inflammatory bowel disease, by modulating the host microbiota, which in turn drives beneficial or harmful consequences in clinical practice. 62,63 Antibiotics can potentially ameliorate the microbial environment of patients with inflammatory bowel disease, both by decreasing proinflammatory bacteria and by increasing beneficial ones, throughout the intestinal lumen. 64 Several antimicrobial drugs have been investigated in patients with ulcerative colitis for the induction and/or maintenance of disease remission. Long-term ciprofloxacin usage was found to improve clinical outcomes, when given as adjuvant therapy to steroids and salicylates, for the induction and maintenance of disease remission, 65 without any advantage in ulcerative colitis as short-term therapy. 66 Metronidazole did not show any benefit in the induction of remission, 67 despite achieving similar results to salicylates in the remission maintenance of mildly-to-moderately active ulcerative colitis. 68 Alternate results come from the use of oral tobramycin, as this drug was shown to be effective in achieving clinical remission 69 but not in maintaining it. 70 Rifaximin has been shown to be capable of improving symptoms, but not activity of disease, in patients with moderatesevere ulcerative colitis. 71 Finally, a combination of different antibiotics (metronidazole, amoxicillin, doxycycline and vancomycin) achieved promising results in children with moderate-severe refractory ulcerative colitis. 72 According to several systematic reviews and meta-analyses of randomised placebo-controlled trials, antibiotics appear to be effective in the induction of remission in patients with ulcerative colitis However, these meta-analyses pooled together studies that investigated different antibiotics, which is arguably not appropriate for amalgamated analysis, therefore their conclusions should be considered carefully. Based on available evidence, the use of antibiotics for patients with ulcerative colitis has only been approved in clinical practice for a limited number of specific indications, including Clostridioides (formerly Clostridium) difficile superinfection, before surgical intervention, toxic megacolon, and if an

7 32 MCILROY ET AL. infection is suspected. 77 There is no evidence to support a routine role for antibiotics in the induction and/or maintenance of remission. 78 Although antibiotics have a well-defined role in the management of complicated CD, mainly in perianal disease and septic complications, 79 their value in uncomplicated disease has not yet been established. Several antibiotic classes have been investigated in inducing or maintaining disease remission in patients with CD. Rifaximin, metronidazole, ciprofloxacin and anti-mycobacterial drugs have been investigated more extensively than other antibiotics, mainly in RCTs against placebo or active comparators. Rifaximin has been shown to be more effective than placebo in achieving clinical remission of mild-to-moderately active CD. 80,81 Metronidazole was shown to improve CD activity index (CDAI), and to prevent disease recurrence after surgery, 82 if combined with azathioprine, but not to achieve clinical remission. 83 Moreover, data show that when used in combination with azithromycin, metronidazole was effective in inducing remission in paediatric CD. 63 Ciprofloxacin has been shown to reduce CDAI significantly, 84 but not to prevent post-surgical recurrence. 85 There has been long-standing interest in Mycobacterium avium subsp. paratuberculosis as a potential trigger of CD. Consequently, many studies have addressed the role of anti-mycobacterial drugs such as rifampicin, clofazimine, ethambutol, isoniazid, sulphadoxine-pyrimethamine, dapsone alone or in combination, in several clinical settings, including induction of remission (either alone or as adjuvant therapy to steroid) and maintenance of remission. 87 However, in a large randomised control trial (RCT) of patients with active CD, a 2-year combination therapy of clarithromycin, rifabutin and clofazimine did not provide any clinical benefit. 90 This suggests that M. avium subsp. paratuberculosis does not have a role as a chronic infection in CD. In a meta-analysis of RCTs, antibiotics showed an overall significant advantage over placebo in the induction of disease remission, although, at sub-analysis, only rifaximin and anti-mycobacterial drugs achieved significant success in induction of remission (RR = 0.78; 95% CI = ) and in maintenance (RR = 0.62; 95% CI = ) respectively. 74 The inconsistencies of these results mean that antibiotics are still not considered a standard treatment option for CD. To date, antibiotics are not recommended for the management of uncomplicated CD because of possible side effects. Their use is recommended only in particular situations, including sepsis, perianal disease and bacterial overgrowth. 79 As well as being a potentially useful therapeutic intervention, there is evidence that suggests antibiotics play a role in the pathogenesis of inflammatory bowel disease, in particular CD, through development of dysbiosis. In the RISK study, an assessment of treatment-na ıve microbiomes of children with CD showed that the use of antibiotics increased the microbial dysbiosis associated with the background disease. 34 In a recent meta-analysis, antibiotic therapy was identified as a significant risk factor for the development of CD (odd ratio [OR] 1.74, 95% CI ), mainly in childhood (OR 2.75, 95% CI ), without being significant for ulcerative colitis (OR 1.08, 95% CI ) Probiotics and prebiotics Many probiotic strains, administered as single strains or probiotic combinations, have been investigated in inflammatory bowel disease, with the aim of modulating the microbiota and reversing intestinal dysbiosis. Escherichia coli Nissle (ECN) 1917, a nonpathogenic strain, is the most frequently investigated. ECN 1917 achieved comparable efficacy and safety outcomes to salicylates in maintenance of remission in subjects with quiescent ulcerative colitis In the same population setting, another single strain, Lactobacillus rhamnosus GG, achieved a longer time free from relapse than salicylates. 95 In few pilot studies, the probiotic yeast S. boulardii was effective both in inducing and in maintaining remission in subjects with mildto-moderately active ulcerative colitis. 96,97 VSL#3, a probiotic mix of 4 Lactobacilli, 3 Bifidobacteria and a Streptococcus, is the product with the most available evidence to date. In a small sample of patients with ulcerative colitis unsuitable for salicylates, VSL#3 was found to be effective in maintenance of remission. 98 Moreover, VSL#3 has been shown to be effective in inducing remission in subjects with mild-to-moderately active ulcerative colitis, either combined with standard treatment 99 or alone. 100 In a meta-analysis, VSL#3, added as adjuvant to standard treatment, achieved better results than standard treatment alone in the induction of clinical remission (OR = 2.4) and clinical response (OR = 3.03; NNT: 3-4). 101 Other combinations of probiotic strains have not achieved results as consistent as those of VSL# Several systematic reviews and meta-analyses tried to address the role of probiotics in the management of ulcerative colitis, without conclusive results. A Cochrane review showed probiotics were not more effective than placebo or active comparators in inducing the remission of active ulcerative colitis. 105 However, when only RCTs were pooled, probiotics were shown to be effective in the induction of remission (RR 1.80), although only VSL#3 was confirmed to achieve a significant advantage over placebo (RR 1.74) after analysis of strain subgroups. 106 A further Cochrane review found that probiotics were ineffective in the maintenance of remission in ulcerative colitis, 107 although a meta-analysis of RCTs showed their efficacy in the prevention of pouchitis. 108 Overall, these meta-analyses suffer from several biases, as the authors pooled together studies that included different species and strains, therapeutic dosages and periods of treatment. In the light of this, probiotics are not currently recommended for the induction of remission in adults patients with ulcerative colitis, although the value of VSL#3 has been highlighted. Moreover, ECN 1917 has been shown to be as effective as salicylates in maintenance of remission. 77 ECN 1917 and VSL#3 have been suggested, with caution, to be useful alone for the induction of remission of mild ulcerative colitis in children not tolerating salicylates, or as an adjunct to standard treatment in patients not completely achieving remission. 78 Finally, probiotics have also been shown to cause sepsis in patients with ulcerative colitis. 109 This finding suggests that probiotics are not always free from side effects, especially in high-risk patients (such as immunosuppressed subjects with severely active inflammatory bowel

8 MCILROY ET AL. 33 disease); therefore, their administration should be approached with caution, particularly in severe disease. Only limited data are available for probiotic application in CD. A number of probiotic strains have been used; however, none of them have shown efficacy in different outcomes of CD (induction of remission, maintenance of remission, prevention of relapse after surgery) either in single studies or meta-analysis of RCTs Moreover, Lactobacilli did not show any effect on the maintenance of remission as an adjunct to standard treatment. 80,114 Other probiotic strains have not achieved better results. ECN 1917 may have an advantage over placebo in achieving a faster induction of remission, but without any effect on remission rates. 115 Saccharomyces boulardi has shown some efficacy in decreasing recurrence rates after clinical remission is achieved; 116 however, it did not improve remission rates in subjects with active disease. 117 Finally, 2 Cochrane reviews found that probiotics conferred no advantage over placebo in maintaining or inducing remission of CD. 118,119 Therefore, probiotics are not suggested in the routine maintenance of remission in patients with CD. 79 The available evidence on the efficacy of prebiotics in CD is still poor. As suggested in a systematic review of RCTs, 120 further high-quality studies are needed to address the role of probiotics and prebiotics in the management of CD. The disappointing results from probiotics to date are perhaps not surprising given that they were generally studied during a period with limited comprehension of, or ability to monitor, the microbiota in inflammatory bowel disease. Moreover, currently available probiotics were not rationally designed to correct the dysbiosis which can underlie the disease; therefore, any therapeutic effect from a single strain is a chance finding. Further study of targeted, specific probiotics informed by the modern microbial pathogenesis paradigm of inflammatory bowel disease is badly needed. The application of metagenomics may help to identify specific strains with biologically plausible efficacy in inflammatory bowel disease. 6.3 Faecal microbial transplantation Faecal microbiota transplantation is a medical treatment that involves the administration of faecal microbiota into the intestinal tract of a recipient. Various methods have been utilised for delivery of FMT, such as nasogastric tube, nasoduodenal tube, rectal enema, the biopsy channel of a colonoscope and more recently via entericcoated capsules. 121,122 The optimal method of delivery remains unclear. The most researched and widely practiced form of FMT is allogenic and involves the transfer of faecal microbiota from a healthy donor into a patient. However, it should be noted that FMT can also be autologous in nature, 123 where faecal material is banked by a patient and reinstated a later date. In modern medicine, FMT was first described in the literature in 1958 as a treatment for fulminant pseudomembranous colitis, where Eiseman et al reported the successful treatment of 4 patients using FMT enemas. 124 Over the subsequent decades, there were several scattered case reports and case series of FMT for C. difficile infection (CDI), the majority of which were successful; however, there was still a lack of controlled clinical trial data for clinicians to reference. 125 In a landmark paper, van Nood et al published the first randomised control trial of FMT in C. difficile infection. 126 The effects were dramatic, with 81% of patients cured after a single FMT, given through nasoduodenal tube, compared with a cure rate of <31% in control groups. Since then, a large body of controlled and noncontrolled evidence has accumulated that reports a primary cure rate of 85%-90% in recurrent C. difficile infection where antibiotic treatment has failed. 127,128 Although the mechanism of action of FMT in the treatment of C. difficile infection has yet to be fully defined, several theories have been put forward. These include direct activity against C. difficile by the bacteria in the donated sample and restoration of secondary bile acid metabolism. 129 Interestingly, the notion that the bacteria in the donated sample play an indispensable role in the efficacy of FMT recently came into question in the light of a clinical case series that reported that sterile faecal filtrate was also effective in treating patients with C. difficile infection. 130 The success of FMT as a treatment for recurrent C. difficile infection has prompted a surge of interest in evaluating its efficacy as a treatment intervention in other diseases and disorders where perturbations to the microbiota are thought to play a role in pathogenesis and severity, including inflammatory bowel disease, where the first case report was published in Since then, some authors have reported positive results, others have reported variable results or no observed clinical improvement The strongest evidence for FMT in inflammatory bowel disease comes from 4 double-blind randomised control trials, 136,137,142,146 all of which were performed on patients with ulcerative colitis. A recent systematic review and meta-analysis published by Costello et al, in this journal, found that in cohort studies, 24% of patients achieved clinical remission. 147 In the 4 published randomised control trials, clinical remission was achieved in 28% of patients who received FMT from healthy donors and 9% of patients in the placebo groups. These results indicate that FMT appears to be a moderately effective treatment for patients with active ulcerative colitis. There are, however, several methodological differences between the studies that make the results challenging to translate into clinical practice. These differences, along with the reported microbiome changes associated with FMT in inflammatory bowel disease, are explored below. Rossen et al randomised 50 patients suffering from ulcerative colitis to undergo 2 FMTs via nasoduodenal tube, containing faeces from either a healthy donor or patients own stool. 142 The authors reported that there was no statistically significant difference in clinical or endoscopic remission between the 2 arms of the study. Moayyedi et al allocated 75 patients to 6 FMT s, using stool prepared from a healthy donor or drinking water (placebo) via retention enema. 136 Patients who received FMT met the robust primary endpoint (clinical and endoscopic remission Mayo <3 with endoscopic Mayo 0) in a higher percentage of patients than placebo (24% vs 5%; P =.03). An interesting observation is that stool from 1 out of the 6 donors induced remission in 39% of patients, which was remarkably higher than the other donors (10%), suggesting that there

9 34 MCILROY ET AL. may be a donor-patient compatibility effect for FMT in inflammatory bowel disease. Moayyedi et al reported that disease <1 year was also more associated with remission. However, the systematic review and meta-analysis performed by Costello et al found no association between disease duration and remission. Paramsothy et al randomly allocated 85 patients with active ulcerative colitis to receive either FMT or placebo (isotonic saline with added brown food colourant and odorant) administered once by colonoscopy, followed by 39 self-administered enemas over 8 weeks. 137 The faecal microbiota administered to patients was prepared from a mix of between 3 and 7 donors. The authors noted that this was a deliberate attempt to increase microbial diversity in each infusion and indeed increased diversity was confirmed using 16S amplicon sequencing. However, clearly this approach makes identifying microbial signals of efficacy more challenging and concurrently runs the risk of inadvertent friendly fire between the combined donor microbial ecosystems in the pooled FMT. Furthermore, the risk of infection transmission is theoretically higher in multi-donor FMT. The primary outcome (steroid-free clinical remission and endoscopic improvement of Mayo 2) was achieved in 27% of patients allocated to FMT with only 8% of the 40 patients allocated the placebo achieving the primary outcome. Costello et al allocated 73 patients with active ulcerative colitis to receive multi-donor faecal microbiota or autologous FMT (placebo) via colonoscopy on day 0 followed by 2 enemas on day The authors reported that in the intentionto-treat analysis, 32% of patients that received FMT achieved the primary endpoint (steroid-free remission, Mayo 2 and endoscopic subscore 1), as compared to 9% who received autologous FMT. In contrast to all prior FMT in inflammatory bowel disease studies, Costello et al processed the faecal microbiota in anaerobic conditions, which could theoretically influence the efficacy of the FMT. Some bacteria such as F. prausnitzii, which have anti-inflammatory effects, are known to only survive in strictly anaerobic conditions and therefore may be lost over the course of aerobic processing. 148 There are several methodological differences between each trial that make it challenging to recommend an optimal protocol. However, taken collectively, the data from the randomised control trials that have been published to date suggest that distal administration, frequent dosing and use of diverse faecal microbiota could all be factors that could influence a positive response in ulcerative colitis. Interestingly, a recent study of FMT in CD reported that frequent dosing was associated with increased efficacy. 149 Future research should investigate the efficacy and safety of capsule-delivered FMT in inflammatory bowel disease as this would be more suitable for frequent dosing and would simplify the design of placebo-controlled trials, though dose limitation may become a factor. 150 Across all studies FMT appear to be safe in the short term, with the majority of reported adverse events being mild, self-limiting and gastrointestinal in nature. 151 However, serious adverse events such as bacteraemia, perforations and death have been reported. Furthermore, there have been reported instances of flares, significant escalation of therapy and development of perianal disease in patients with concomitant inflammatory bowel disease and C. difficile infection treated with FMT Overall, the rate of serious adverse events appears to be higher in recipients of FMT through the upper gastrointestinal tract as a result of procedure-induced aspiration pneumonia. 151 The causality between nonprocedural adverse events and FMT has yet to be fully delineated and therefore further research from controlled trials is necessary to establish the factors involved. 151,155 The long-term effects of FMT are yet to be established. However, as FMT involves the infusion of a largely uncharacterised active microbial suspension, there is a theoretical possibility that diseases linked to gut bacteria could be transferred. Faecal microbiota transplantation is known to elicit significant changes to the structure and function of the microbiome of the recipient. Paramsothy et al reported that the presence of several taxa in the donor were associated significantly with the primary endpoint. In particular, Barnesiella spp., Parabacteroides spp., Clostridium cluster IV, Ruminococcus spp., Blautia spp., Dorea spp. and Clostridium cluster XVIII. These results are broadly consistent with Moayyedi et al, who reported that donor B was enriched with Ruminococcus and Lachanospiraceae (Clostridium cluster IV and XIVa). Interestingly, Rossen et al reported that remission was associated with donor enrichment in IV, XIVa and XVIII, which contain known butyrate-producing species. Taken collectively, it appears that Clostridium clusters IV, XIVa and XVIII are consistently associated with remission. Remarkably, these observations are supported by animal models showing that cocktails of Clostridial species belonging to clusters XIVa, IV and XVIII promote an anti-inflammatory immune responses by activating regulatory T (T reg ) cells 156 as well as data that show that inoculation of clostridia during early life reared mice resistant to colitis. 157 Cluster XIVa contains highly oxygen-sensitive organisms of clinical importance in UC pathogenesis, particularly Roseburia hominis. Interestingly, this renewed focus on cluster XIVa also reinvigorates interest in the role of butyrate in ulcerative colitis pathogenesis. FMT is a relatively crude approach to treating inflammatory bowel disease, which is likely a complex continuum of disorders. However, these results suggest that when combined with robust microbial analysis of both the donor and the recipient, FMT can be an efficacious treatment option and a powerful research tool which may facilitate patient stratification and personalised microbial therapeutics. 6.4 Nutritional therapy in inflammatory bowel disease The use of supportive nutrition in inflammatory bowel disease is outside the scope of this review, which will instead focus on the use of nutrition as therapy in inflammatory bowel disease and the microbial signals that emerge from its use. The role of particular foods in driving inflammation and triggering flare in inflammatory bowel disease is also of huge interest; however, while research in this area may well point to microbial mechanisms of action, insufficient targeted microbial research has been undertaken in this area for us to develop specific microbial arguments further. Such food/ microbiome research, particularly coupled with disease activity/flare

10 MCILROY ET AL. 35 data, represents an important unmet need in inflammatory bowel disease. The potential utility of nutrition as therapy in inflammatory bowel disease was first demonstrated in an observational study by Voitk et al who recognised that use of an elemental diet prevented the need for surgery in some patients where operative management was planned. 158 While this observation prompted an initial interest in elemental and semi-elemental protein sources in early trials of exclusive enteral nutrition, including the seminal control trials from O Morain et al and from Sanderson et al. subsequent meta-analysis has highlighted that whole protein approaches are equally effective Similarly, although the initial observation was not disease phenotype-specific, very limited work has looked at the role of enteral nutrition in ulcerative colitis, although anecdotally it is thought to be ineffective. The focus has therefore fallen on enteral nutrition in CD, with recognition that exclusive liquid diets for periods of 6-8 weeks are highly effective in inducing remission in ~80% of paediatric patients with active luminal CD, leading to exclusive enteral nutrition being promoted as first-line therapy in paediatric CD by the European and North American Societies of Paediatric Gastroenterology, Hepatology and Nutrition and the European Crohn s and Colitis Organisation. 162,163 The ethics of repeat colonoscopy limits studies of mucosal healing (and mucosal microbial change) in paediatric CD; however, small series support mucosal healing rates comparable to described rates of remission (75% vs 81% at week 8 in Rubio et al 164 ). The lack of before/after mucosal microbiota studies impacts our understanding of the fundamental biological mechanism of action of exclusive enteral nutrition, with most work to date relying on serial faecal samples instead. Exclusive enteral nutrition also helps with nutritional rehabilitation and lean body mass, hence is an obvious choice in paediatric inflammatory bowel disease practice where weight loss is a common feature at presentation. 165,166 Use of exclusive enteral nutrition within adult practice remains limited and probably relates to the cumbersome social aspects of a liquidonly diet. A description of the efficacy of exclusive enteral nutrition preceded any meaningful understanding of its mechanism of action, which has only recently begun to be understood with the advent of high-throughput, culture-free, microbial ecology tools. One early microbial study of exclusive enteral nutrition suggested either the low residue nature or potential prebiotic properties of the diet as potential modulators of the microbiome. 167 Recent data, however, demonstrate that the mechanism of action is paradoxical when compared to our understanding of CD pathogenesis. Exclusive enteral nutrition appears to reduce bacterial diversity further and reduces the proportion of key species, including F. prausnitzii, suggesting that its main mechanism of action is a reduction in the availability of bacterial substrate in the gut lumen (Figure 1) This offers the tantalising prospect of potentially identifiable key pathogenic organisms, which might be amenable to long-term remission by directed microbial therapy. Limited data support the use of partial enteral nutrition as maintenance therapy for CD, with only 1 randomised control trial published to date describing a 35% relapse rate on half enteral nutrition vs 64% on free diet, over nearly 12-month follow-up, after induction of remission with exclusive enteral nutrition in 92%. 172 Unsurprisingly, the message of larger volumes of enteral formula being associated with higher rates of remission emerges from published work. 173 Although partial enteral nutrition is not thought to be effective as an induction agent in CD, 174 interest has turned recently to solid food alternatives to the socially challenging liquid-only exclusive enteral nutrition currently in use. Exploration of such strategies has the support of patients and their families. 175 Exciting work from Sigall-Boneh et al in Israel demonstrates proofof-principle of this approach, where a structured specific food elimination diet was used in children and adults with active CD, supplemented with 50% polymeric formula. 176 Remission rates were encouragingly high at ~70%. The exact constituents of the diet were not reported, but it involved reduced exposure to animal fat, dairy, gluten and emulsifiers while allowing fruit- and vegetablederived fibre sources. It is difficult to comment on the potential mechanism of action without further dietary data, specifically regarding the substrate availability for colonic bacteria, particularly with regard to potential fibre and SCFA sources. Interestingly, 6/7 patients given the elimination diet without supplementary formula also entered remission, suggesting a food-only approach might achieve high efficacy rates. Other approaches and hypotheses for food-based CD therapy are discussed in detail in Lee et al. 177 Dietary emulsifiers are one area of significant interest that sit outside of the microbial substrate hypothesis, potentially opening avenues for increased pathogenic potential within the host microbiota. 178 The exact mechanisms of CD dietary pathogenesis may of course be multifactorial and complex. The potential efficacy of solid food elimination diets is exciting in proposing a coeliac-type approach to CD management in the future, potentially more acceptable to adults than traditional exclusive enteral nutrition. Hopefully further study will help stratify the underpinning mechanisms to help refinement of dietary approaches. While it is unlikely that a single fundamental agent as simple as gluten is the key to CD dietary response, further dietary study and control trials of interventions might unlock a greater understanding of the disease pathogenesis and open other avenues for targeted microbial therapy as a byproduct. Further study of diet and CD should be an area of high research priority, underpinned with detailed microbial interrogation of the mechanisms at play. 7 CONCLUSIONS/FUTURE PERSPECTIVES It is now clear that inflammatory bowel disease is intimately linked to changes in the composition of the gut microbiota. The existing therapeutic approaches presented in this review broadly fall into 3 categories, namely: accession (faecal microbial transplantation and probiotics), reduction (antibiotics) or indirect modulation (exclusive enteral nutrition and prebiotics; Figure 2). 179 In paediatric patients,

11 36 MCILROY ET AL. Current microbial therapeutic options Indirect modulation Dietary modulation Future perspectives Personalised dietary manipulation Genetically Manipulated strains Delivery of targeted microbiome modulators eg SCFAs, cytokines, phage EEN Prebiotics/ Assession Prebiotics Small molecules eg oligosaccharides Uncharacterised community transplant eg FMT Reduction Antibiotics Next generation FMT Standardised protocols, optimal methods of processing and administration, stratification of patients Live biotherapeutics Single strain or consortia of microbes rationally selected based on evidence FIGURE 2 The future of microbiome-modulating therapeutics. Existing therapeutic approaches to modulate the gut microbiome are relatively unrefined and primitive, broadly falling into 3 categories, namely: accession (faecal microbiota transplantation and probiotics), reduction (antibiotics) or indirect modulation (exclusive enteral nutrition and prebiotics). Future therapeutic approaches will be more targeted and personalised to treat the underlying disease pathophysiology. Strategies that are currently being investigated include live biotherapeutics containing single strains or consortia of bacteria that have been rationally selected and bacteria that have been engineered to produce therapeutic proteins. Alongside this, developing personalised dietary manipulation strategies and small molecule delivery will also likely feature exclusive enteral nutrition is an established treatment in CD and the focus should now be on maximising the evidence supporting this strategy and bringing this into clinical practice. The available evidence for other strategies is sparse and should only be implemented within the clinical trial setting. In adults, faecal microbial transplantation appears to hold the most promise, with several randomised control trials reporting clinical efficacy. However, the differences in study design and methodology mean that it is currently not possible to elegantly translate these results into clinical practice. Faecal microbial transplantation dramatically alters the structure and function of the whole microbial community, which means that faecal microbial transplantation could have therapeutic effects on numerous host targets. There are, however, many unanswered questions and, in the light of the fact that our understanding of what constitutes a healthy microbiota is still in its infancy, a robust donor selection and screening program is essential. Future research should focus on standardising faecal microbiota processing protocols and placebo selection. Furthermore, it is now incumbent on researchers to focus on the stratification of responders and detailed interrogation of the associated microbial changes in these patients. This will be fundamental in unravelling the pathogenesis of inflammatory bowel disease and the mechanism of action of faecal microbial transplantation. Multifaceted longitudinal observation of patients before, during and after treatment as well as robust randomised controlled trials will need to be performed to refine and improve on the evidence that is currently available. The existing therapeutic approaches to modulate the gut microbiome are relatively unrefined and primitive (Figure 2). However, looking forward, the future of microbiome modulating therapeutics looks bright. There are several novel strategies and technologies hurtling towards the clinic. In an effort to develop a more specific form of faecal microbial transplantation, many groups are pursuing single strains of live organisms or characterised communities of microbes (Figure 2). 180,181 Other groups are taking an even more reductionist approach by mining bacteria for small molecules. 182 Finally, advances in the field of synthetic biology may mean that one day each of us will be able to colonise our gut with genetically modified bacteria that knock out bad bugs with bacteriophages or secrete antiinflammatory small molecules. Taken collectively, it is clear that ignoring the microbiome in inflammatory bowel disease is not an