EEN and gut microbiota composition in patients with CD

In Europe and elsewhere, EEN is the first-line treatment of active CD, in children, achieving clinical remission rates up to 85 % and improving mucosal healing in a substantial proportion of patients ^{(Reference van Rheenen, Aloi and Assa2)}. There are currently two main doctrines of how EEN works. The first, by the exclusion of components from the diet of people with CD causing disease, and a second mechanism involving dietary modulation of inflammatory microorganisms of the human gut^{(Reference MacLellan, Moore-Connors and Grant3–Reference Shah and Kellermayer5)}. We identified a total of 24 original articles which investigated the effect of treatment with EEN on the composition or function of the gut microbiota of patients with CD (Table 1). The current literature is comprised predominantly of studies in paediatric patients (N = 20)^{(Reference Leach, Mitchell and Eng6–Reference Pigneur, Lepage and Mondot25)} with four studies reporting the impact of EEN on the gut microbiota of adults with CD^{(Reference Jia, Whitehead and Griffiths26–Reference He, Gao and Jie29)}. The duration of the EEN course varied from two^{(Reference Jia, Whitehead and Griffiths26,Reference Walton, Montoya and Fowler28,Reference He, Gao and Jie29)} to 12^{(Reference Dunn, Moore-Connors and MacIntyre17,Reference Dunn, Moore-Connors and MacIntyre18,Reference Jones, Connors and Dunn24)} weeks; most of the studies (N = 19) used polymeric feeds^{(Reference Leach, Mitchell and Eng6–Reference Guinet-Charpentier, Lepage and Morali14,Reference Ashton, Colquhoun and Cleary16–Reference Diederen, Li and Donachie23,Reference Pigneur, Lepage and Mondot25,Reference He, Gao and Jie29)} . All studies recruited patients with active CD with sample sizes varying from one^{(Reference D'Argenio, Precone and Casaburi8)} to 55^{(Reference de Meij, de Groot and Peeters21)} patients.

Table 1. Human studies investigating the effect of exclusive enteral nutrition on the gut microbiota of patients with CD

CD, Crohn's disease; EEN, Exclusive enteral nutrition; TTGE, Temporal temperature gradient gel electrophoresis; DGGE, Denaturing gradient gel electrophoresis; PCDAI, Paediatric Crohn's disease activity index; qPCR, quantitative PCR; T-RFLP, Terminal restriction fragment length polymorphism; SCFA, Short-chain fatty acids; BCFA, Branched-chain fatty acids; RNA, RNA; OTU, Operational taxonomic unit; FCP, Faecal calprotectin; KEGG, Kyoto Encyclopedia of Genes and Genomes; HBI, Harvey Bradshaw index; OPLS-DA, Orthogonal Projections to Latent Structures Discriminant Analysis; CS, Corticosteroids; LPS, Lipopolysaccharide; ASV, Amplicon sequencing variant; AUC, Area under the curve; RF, Random forest; MGS, Metagenomic sequencing.

Changes in gut microbiota were investigated using an array of molecular methods profiling its composition at the various levels of microbial hierarchy. The most common methods used to characterise the gut microbiota were 16S rRNA sequencing (N = 10)^{(Reference D'Argenio, Precone and Casaburi8,Reference Kaakoush, Day and Leach10,Reference Quince, Ijaz and Loman12,Reference Guinet-Charpentier, Lepage and Morali14–Reference Ashton, Colquhoun and Cleary16,Reference Dunn, Moore-Connors and MacIntyre18,Reference Levine, Wine and Assa22–Reference Jones, Connors and Dunn24)} , shotgun metagenomics (N = 6)^{(Reference Kaakoush, Day and Leach10–Reference Quince, Ijaz and Loman12,Reference Dunn, Moore-Connors and MacIntyre17,Reference Jones, Connors and Dunn24,Reference He, Gao and Jie29)} , quantitative real-time PCR (qPCR) (N = 3)^{(Reference Gerasimidis, Bertz and Hanske9,Reference Jia, Whitehead and Griffiths26,Reference Shiga, Kajiura and Shinozaki27)} , temperature gradient gel electrophoresis (TGGE) or denaturing gradient gel electrophoresis (DGGE) (N = 4)^{(Reference Leach, Mitchell and Eng6,Reference Lionetti, Callegari and Ferrari7,Reference Gerasimidis, Bertz and Hanske9,Reference Pigneur, Lepage and Mondot25)} and terminal restriction fragment-length polymorphism (N = 1)^{(Reference Shiga, Kajiura and Shinozaki27)}. Fewer studies assessed changes of the gut microbiota function by measuring targeted metabolites, such as short-chain fatty acids (SCFA), lactate, sulphide and ammonia (N = 3)^{(Reference Gerasimidis, Bertz and Hanske9,Reference Tjellstrom, Hogberg and Stenhammar13,Reference Ashton, Colquhoun and Cleary16)} or by performing untargeted metabolomics analysis using liquid chromatography (N = 2)^{(Reference van der Hooft, Wandy and Young19,Reference Alghamdi, Gerasimidis and Blackburn20)} , GC (N = 1)^{(Reference Walton, Montoya and Fowler28)} or NMR (N = 1)^{(Reference Diederen, Li and Donachie23)}. The tissue studied was faeces for the vast majority of the studies with the exception for a single study in which ileal biopsies were used^{(Reference D'Argenio, Precone and Casaburi8)} to assess the mucosal-associated microbiota and changes during EEN. In these 24 studies, remission rates varied between 45 % and 100 %, but different groups defined remission using different clinical disease indices and disease biomarkers outcomes (Table 1).

Even though, the majority of the studies reported significant effects on microbiota composition, the directions of these effects varied. Few studies observed a decrease in microbial diversity following treatment with EEN proposing that its mode of action may include suppression of either the global gut microbiota or selective bacterial subpopulations causing inflammation^{(Reference Leach, Mitchell and Eng6,Reference Gerasimidis, Bertz and Hanske9,Reference Quince, Ijaz and Loman12,Reference Diederen, Li and Donachie23)} (Fig. 1). Leach et al. showed that EEN significantly decreased the diversity of global microbiota, and other groups within Bacteroides-Prevotella and Clostridium coccoides ^{(Reference Leach, Mitchell and Eng6)}. In accordance to these findings, Gerasimidis et al., also found a decrease in global bacterial diversity, and absolute concentration of Bacteroides-Prevotella and Faecalibacterium prausnitzii ^{(Reference Gerasimidis, Bertz and Hanske9)}; with the decrease of F. prausnitzii being replicated by others^{(Reference Pigneur, Lepage and Mondot25,Reference Jia, Whitehead and Griffiths26)} . This latter observation is counterintuitive and challenges perceptions that F. prausnitzii is a critical micro-organism in the pathogenesis of CD and that its diminished abundance increases the risk of post-surgical relapse^{(Reference Sokol, Pigneur and Watterlot30)}. A follow-up study of the aforementioned study by Gerasimidis et al. using 16S rRNA amplicon and shotgun metagenomics sequencing, also observed a significant decrease of the faecal microbiota Shannon α-diversity; for every 10 days on EEN, 0⋅6 genus diversity equivalents were lost. This decrease was accompanied by a shift of the microbiota composition to a direction opposite to that of healthy controls^{(Reference Quince, Ijaz and Loman12)}. Similarly, Lewis et al., described that EEN shifted the gut microbial community structure of CD children away to that of the healthy reference^{(Reference Lewis, Chen and Baldassano11)}. In contrast, five other studies reported no significant changes in microbiota diversity during EEN, hence suggesting that the underlying mode of EEN action is less likely to be mediated by such effects and it might be associated with changes in the function or abundance of selective members of the microbial community^{(Reference Schwerd, Frivolt and Clavel15,Reference de Meij, de Groot and Peeters21,Reference Levine, Wine and Assa22,Reference Jones, Connors and Dunn24,Reference Shiga, Kajiura and Shinozaki27)} . In previous research, such microorganisms included a decrease of Bacteroides fragilis ^{(Reference Shiga, Kajiura and Shinozaki27)}, Bacteroidetes, Bacteroidaceae and Porphyromonadaceae ^{(Reference Schwerd, Frivolt and Clavel15)}, Haemophilus, Veillonella, Bifidobacterium, Prevotella, Proteobacteria, Anaerostipes, and Lachnospira ^{(Reference Levine, Wine and Assa22)}. Only a study by D'Argenio et al., reported that EEN increased the ileal mucosal bacterial diversity and Bacteroidetes abundance in a single paediatric CD patient^{(Reference D'Argenio, Precone and Casaburi8)}, and Ashton et al., reported increased faecal microbiota richness, diversity, and abundance of Bacteroides and Bifidobacterium in three children with active CD during EEN^{(Reference Ashton, Colquhoun and Cleary16)}.

Fig. 1. A proposed microbial mediated mechanism of action of EEN and disease recurrence following habitual diet reintroduction.

EEN and gut microbiota function in patients with CD

The effect of EEN on gut microbiota function was also investigated but in a smaller number of studies than those which explored compositional shifts. Organic compounds in faeces, including SCFA were consistently found to decrease following treatment with EEN, suggesting a suppressive effect of the therapy on bacterial metabolism, and fermentation capacity in particular^{(Reference Gerasimidis, Bertz and Hanske9,Reference Tjellstrom, Hogberg and Stenhammar13,Reference van der Hooft, Wandy and Young19,Reference Walton, Montoya and Fowler28)} . Gerasimidis et al., also reported a substantial increase in faecal hydrogen sulphide levels after 8-week treatment with EEN^{(Reference Gerasimidis, Bertz and Hanske9)}. Interestingly, the magnitude of changes in faecal bacterial metabolites was larger when in subset analysis patients who did not achieve clinical remission were excluded. In 1H NMR metabolomics analysis, EEN affected the global metabolome with differences noted prior to initiation between treatment responders v. non-responders. Following EEN, several metabolites (i.e. leucine, propionate, valine, lactate, alanine, cadaverine, trimethylamine, tyrosine, phenylalanine, isovalerate, urocanate, succinate) were normalised in responders but not in non-responders^{(Reference Diederen, Li and Donachie23)}. These observations contradict the compositional shifts observed during EEN treatment (Table 1) and may suggest that changes in microbiota function might be more relevant in disease improvement.

Microbiota changes during EEN and association with disease activity markers

Several of the studies above also explored relationships between disease activity response to EEN and changes in gut microbiota characteristics. Improvement in disease activity indices, inflammatory and immunological markers correlated with changes in the abundance of bacterial taxa, KEGG modules or pathways^{(Reference Leach, Mitchell and Eng6,Reference Gerasimidis, Bertz and Hanske9,Reference Kaakoush, Day and Leach10,Reference Guinet-Charpentier, Lepage and Morali14,Reference Schwerd, Frivolt and Clavel15,Reference Dunn, Moore-Connors and MacIntyre17,Reference Dunn, Moore-Connors and MacIntyre18,Reference Jones, Connors and Dunn24)} . Quince et al., found that 35 different OTUs (operational taxonomic unit) correlated with faecal calprotectin changes, 14 of which explained 78 % of the variance in calprotectin levels^{(Reference Quince, Ijaz and Loman12)}. In accordance, Kaakoush et al. reported that a decreased OTU richness was associated with disease remission, and reversely, disease recurrence was associated with increased richness^{(Reference Kaakoush, Day and Leach10)}. Interestingly, clinical response was associated with the distance of the CD microbiota from the centroid of the healthy microbiota; however, paradoxically, as children who clinically responded to EEN moved further away from the healthy reference status, hence their community became even more dysbiotic than those children who did not show improvement^{(Reference Lewis, Chen and Baldassano11)}.

The exact opposite pattern was reported by two other studies that investigated the faecal microbial composition, at EEN completion, as a predictor of remission duration. The authors found that patients who sustained remission for 12 weeks post-treatment with EEN had a more similar gut microbiota to that of healthy controls, than patients who experienced disease relapse sooner^{(Reference Dunn, Moore-Connors and MacIntyre17,Reference Dunn, Moore-Connors and MacIntyre18)} .

Microbiota changes following treatment with EEN and during food re-introduction

There is limited literature about microbiota changes following EEN completion and during reintroduction of habitual diet. Such data are important since they offer insight into the role of diet triggering disease relapse and the mediating role gut microbiota may have in this process. Studies in healthy participants showed that one week of EEN was long enough to significantly decrease total faecal bacterial load and faecal SCFA and influence drastically microbiota composition. In reverse, food reintroduction rapidly reverted these changes to the pre-treatment level^{(Reference Whelan, Judd and Preedy31,Reference Svolos, Hansen and Nichols32)} . Consistent with this observation, recurrence of CD following the end of EEN is associated with a remarkable reversal of microbial shifts to pre-treatment levels, when participants return to the habitual diet^{(Reference Leach, Mitchell and Eng6,Reference Gerasimidis, Bertz and Hanske9,Reference Kaakoush, Day and Leach10,Reference Quince, Ijaz and Loman12,Reference Levine, Wine and Assa22)} . It has been proposed that exclusion of dietary components from the habitual diet, rather than enteral nutrition consumption itself, is critical to successful treatment with EEN; underlined by the limited efficacy of partial enteral nutrition in which only a part of the habitual diet has been replaced by enteral nutrition^{(Reference Johnson, Macdonald and Hill33)}. This is potentially due to the less drastic microbial effects of enteral nutrition when consumed along with a free diet^{(Reference Leach, Mitchell and Eng6,Reference Lionetti, Callegari and Ferrari7,Reference Lewis, Chen and Baldassano11,Reference Guinet-Charpentier, Lepage and Morali14)} . Monitoring microbiota changes upon food reintroduction following successful treatment with EEN treatment may allow the development of evidence-based advice for maintenance of EEN-induced remission.

Proposed mechanism of EEN mediated by the gut microbiota

Several of the microbial signals observed in previous research may be implicated in the therapeutic action of EEN and may open opportunities to develop novel microbial therapeutics for CD management, such as the CD-TREAT (Crohn's Disease TReatment-with-EATing) diet^{(Reference Svolos, Hansen and Nichols32)}. The most consistent effects observed in the current literature include a decrease in bacterial diversity^{(Reference Leach, Mitchell and Eng6,Reference Gerasimidis, Bertz and Hanske9,Reference Kaakoush, Day and Leach10,Reference Quince, Ijaz and Loman12,Reference Dunn, Moore-Connors and MacIntyre18)} and the development of a microbial community structure even more dissimilar to that associated with health^{(Reference Lewis, Chen and Baldassano11,Reference Quince, Ijaz and Loman12)} . Paradoxically, the development of such suppressed and dysbiotic microenvironment coincides with mucosal healing, reduction in colonic inflammatory markers, and disease activity improvement^{(Reference Ruemmele, Veres and Kolho34)}. The limited efficacy rates reported for therapeutic strategies aiming to promote or restore a healthy gut microenvironment in patients with CD, including faecal material transplantation^{(Reference Paramsothy, Paramsothy and Rubin35)}, probiotics, prebiotics and fibre supplementation^{(Reference Benjamin, Hedin and Koutsoumpas36,Reference Chermesh, Tamir and Reshef37)} further support the doctrine that suppression of the global microbiota or selective inflammatory members of the community might be needed to promote disease remission and induce mucosal healing^{(Reference Paramsothy, Paramsothy and Rubin35,Reference Ghouri, Richards and Rahimi38)} (Fig. 1). Opposite to that, microbial suppression with antibiotics has benefitted gut inflammation in CD^{(Reference Levine, Kori and Kierkus39)}. The same is also the case for total parenteral nutrition which has been found to ameliorate CD gut inflammation but also to suppress the gut microbial community through substrate depletion for bacterial growth and gut rest^{(Reference Shiga, Kajiura and Shinozaki27)}. From a nutrition perspective, most of the above microbial effects are to be expected since the composition of feeds used in EEN are of low residue, including fibre, and comprise rapidly absorbed carbohydrates. These factors could modulate the gut microbial dynamics, suppress bacterial growth and activity hence reducing antigenic stimulation, repressing aberrant activation of the gut-associated immune system and consequently promoting mucosal healing^{(Reference Cuiv, Begun and Keely40)} (Fig. 1).

Future studies should aim to develop and test strategies to sustain EEN-induced microbial effects by using dietary and pharmacological therapies. Should such strategies be successful, they will confirm the hypothesis proposed here that modulation of microbiota is a plausible mechanism of action of EEN. The CD-TREAT diet is a novel dietary therapy for an active CD which has a similar nutritional and food component profile to EEN and effects on gut microbiota characteristics^{(Reference Svolos, Hansen and Nichols32)}. In animal models of gut inflammation, CD-TREAT improved ileal inflammation, both histologically and in terms of inflammatory biomarkers, and in a pilot trial of 5 children with active CD receiving CD-TREAT, 80 % showed a clinical response, three (60 %) entered remission, with significant 50 % decreases in faecal calprotectin levels^{(Reference Svolos, Hansen and Nichols32)}. These early signals await confirmation by the results of a major multicentre trial (ClinicalTrials.gov identifier NCT03171246).

Food-based dietary therapies and the gut microbiota in IBD

There is a high interest from patients with IBD and healthcare professionals to develop dietary therapies for the management of CD and UC. A recent review identified a total of 24 food-based dietary therapies for the management of CD and UC with variable proposed modes of action^{(Reference Gerasimidis, Godny and Sigall-Boneh1)}. Nine clinical trials met the inclusion criteria of the current review (Table 2). Of these, six studies recruited only patients with CD^{(Reference Levine, Wine and Assa22,Reference Zhang, Taylor and Shommu41–Reference Lewis, Sandler and Brotherton45)} , one study recruited only patients with UC^{(Reference Fritsch, Garces and Quintero46)} and two studies^{(Reference Cox, Lindsay and Fromentin47,Reference Suskind, Cohen and Brittnacher48)} had a mixed IBD population. Most studies recruited adult patients (n = 6)^{(Reference Zhang, Taylor and Shommu41–Reference Walters, Quiros and Rolston43,Reference Lewis, Sandler and Brotherton45–Reference Cox, Lindsay and Fromentin47)} , while children and young adults were recruited in the other three studies^{(Reference Levine, Wine and Assa22,Reference Suskind, Lee and Kim44,Reference Suskind, Cohen and Brittnacher48)} .

Table 2. Human studies investigating the effect of food-based diets on the gut microbiota of patients with CD or UC

CD, Crohn's disease; SCD, Specific carbohydrate diet; LRD, Low residue diet; RNA, RNA; FODMAP, fermentable oligosaccharides, disaccharides, monosaccharides and polyols; qPCR, quantitative PCR; SCFA, Short-chain fatty acids; FCP, Faecal calprotectin; UC, ulcerative colitis; CDED, Crohn's disease exclusion diet; PEN, Partial enteral nutrition KEGG, Kyoto Encyclopedia of Genes and Genomes; LFHF, Low fat high fibre; iSAD, improved Standard American diet; CRP, C-reactive protein.

Like in EEN, the composition of the gut microbiota was examined in most cases using high-throughput sequencing (n = 8)^{(Reference Levine, Wine and Assa22,Reference Zhang, Taylor and Shommu41,Reference Walters, Quiros and Rolston43–Reference Suskind, Cohen and Brittnacher48)} , with one study using qPCR^{(Reference Halmos, Christophersen and Bird42)}. Microbial metabolic activity was assessed in five studies using gas (n = 4)^{(Reference Zhang, Taylor and Shommu41,Reference Halmos, Christophersen and Bird42,Reference Cox, Lindsay and Fromentin47)} and liquid (n = 1)^{(Reference Fritsch, Garces and Quintero46)} chromatography, while meta-proteomic analysis was performed in one study using liquid chromatography and MS^{(Reference Suskind, Lee and Kim44)}. All studies used faecal samples for profiling the gut microbiota.

CD exclusion diet and partial enteral nutrition

The CD exclusion diet (CDED) in combination with 50 % partial enteral nutrition (CDED + PEN) is a new dietary regime which aims to alleviate gut inflammation through modification of inflammatory gut microbiota^{(Reference Levine, Wine and Assa22)}. The premise behind the CDED + PEN dietary regime is the exclusion of food constituents (i.e. animal fat, food additives) thought to aggravate gut inflammation and cause dysbiosis, solely based on evidence from preclinical and epidemiological research. This hypothesis has recently been challenged in the literature^{(Reference Logan, Gkikas and Svolos49)} since the inclusion of 50 % PEN, an integral component in the CDED + PEN dietary regime, inevitably increases the consumption of food components (e.g. emulsifiers and maltodextrin) and nutrients (milk fat) that the authors aim to avoid with CDED. The authors also recommend the consumption of foods, such as fruits and vegetables, thought to provoke beneficial effects on the gut microbiota including production of SCFA. Improvements in disease activity and quality of life indices were observed after a 6-week, multicentre intervention in children with active CD. Some microbial compositional shifts were similar between patients on the CDED + PEN and the EEN group. CDED + PEN decreased Haemophilus, Veillonella, Bifidobacterium, Prevotella, and Anaerostipes, and increased Oscillibacter and Roseburia whereas during EEN more bacteria were influenced including a decrease in Lachnospira and increases in Subdoligranulum, Blautia, Ruminococcus, and Erysipelotrichaceae. Both diets decreased Proteobacteria, with this effect lost when EEN patients returned to their habitual diet but sustained in patients on CDED + 25 %PEN. Irrespective of dietary intervention, non-responders had a lower overall change in microbiota composition but more Gammaproteobacteria. Interestingly, although the rationale behind CDED + PEN is to increase consumption of dietary fibre through fruits and vegetables, Bifidobacterium, whose levels are positively influenced by fermentable fibre, were decreased following both dietary interventions, most likely indicating that not only was it not possible for patients to increase fibre intake, but on average they had a lower consumption compared to prior treatment initiation^{(Reference Holscher50)}.

Low FODMAP diet

A diet that limits the intake of low fermentable, oligo-, di-, monosaccharides and polyols (FODMAP) has been tested as an option for alleviating functional gastrointestinal symptoms in patients with IBD. Two studies assessed the impact of low FODMAP diets on the gut microbiota of patients with CD. In the first pilot crossover study in eight patients with CD in remission, Halmos et al. showed that a 3-week low FODMAP diet reduced the concentration of Clostridium XIVa cluster, Akkermansia muciniphila and increased levels of Ruminococcus torques, compared to a FODMAP containing diet^{(Reference Halmos, Christophersen and Bird42)}. No significant differences were observed in total bacteria or faecal SCFA between the two groups.

In the second RCT, patients with CD and UC in remission following a 4-week low FODMAP diet, experienced a decrease in the relative abundance of B. adolescentis, B. longum and F. prausnitzii, compared to a sham diet^{(Reference Cox, Lindsay and Fromentin47)}. No differences were observed neither in the microbiota α- and β-diversity between the two groups, nor in the functional metagenomic capacity using KEGG orthologues, or SCFA levels. The abundance of F. prausnitzii, A. muciniphila were negatively impacted by the low FODMAP diet in both studies, species which are considered to exert favourable effects on host immunity^{(Reference Sokol, Pigneur and Watterlot30,Reference Earley, Lennon and Balfe51)} . Importantly, the effects observed in the IBD microbiota, were similar to the effects of a low FODMAP diet on the microbiota of healthy people or of patients with irritable bowel syndrome, hence stressing that the underlying mechanism might be disease independent^{(Reference Halmos, Christophersen and Bird42)}. In both studies, low FODMAP diets improved certain functional symptoms and these effects were not associated with recurrence of intestinal inflammation, as measured with faecal calprotectin levels.

Specific carbohydrate diet

The specific carbohydrate diet (SCD) is a diet which excludes complex carbohydrates which in principle could escape absorption and lead to bacterial fermentation, bacterial overgrowth and intestinal inflammation. Although popular among the IBD community, until recently, there was a lack of robust scientific evidence to recommend the SCD for the induction or maintenance of remission in IBD^{(Reference Sasson, Ananthakrishnan and Raman52)}. Four clinical trials on SCD and its effects on the gut microbiota were identified and included in this review. In the first pilot, cross-over study, five patients with CD in remission followed the SCD or a low-residue diet for 30 days^{(Reference Walters, Quiros and Rolston43)}. SCD appeared to increase bacterial diversity and the abundance of over 100 species, of which, more than 20 species were Clostridia. These microbial signals compared to a more stable bacterial composition following a low-residue diet. Nonetheless, alterations in the gut microbiota were not linked to changes in clinical activity suggesting that any effect of SCD might be independent of microbial modification. Similar results were also observed in another open-label trial in nine children and young adults with active IBD where considerable inter-individual variation in microbial changes was observed, potentially reflecting the diversity of the dietary regime or normal biological variation^{(Reference Suskind, Cohen and Brittnacher48)}.

Suskind et al. investigated the effect of the SCD and a modified version of SCD (with added oats and rice), on the gut microbiota of children with mild to moderate CD and in comparison, to a food-based exclusion diet for 12 weeks^{(Reference Suskind, Lee and Kim44)}. Comprehensive ‘omics analysis, examining the effect of the different diets on the faecal metagenome, metabolome and metaproteome was performed; albeit in only 5/10 patients (modified SCD, n = 3, food-based exclusion diet, n = 2). Although species richness did not change throughout the 12-week intervention, inter-individual shifts in the microbiota community structure were observed. The relative abundance of various taxa (i.e. F. prausnitzii and R. hominis) increased in 4/5 patients, whereas E. coli decreased in 3/5 patients although it is unclear how these signals related to disease activity. A decrease in 1,2-propandiol, and sterol metabolites was observed, while certain fatty acids and metabolites involved in amino acid biosynthesis decreased after the initial 2 weeks of SCD intervention. Meta-proteomics analysis revealed a reduced enzymatic activity linked to starch catabolism and sugar metabolism, after 2 weeks of SCD, and highlighted the decreased enzymatic functionality related to amino acid biosynthesis, complementing the metabolomics findings. All patients achieved clinical remission, while normalisation of CRP levels was achieved in patients following the SCD and modified SCD. Faecal calprotectin levels decreased with the modified SCD and the food-based exclusion diet, while a non-significant increase was observed after the SCD.

In a large recent study of 191 patients, Lewis et al. compared the efficacy of the SCD against a Mediterranean diet on clinical outcomes and their effects on the faecal microbiota, in adults with mild-to-moderate active CD for 12 weeks^{(Reference Lewis, Sandler and Brotherton45)}. There were no significant differences in clinical remission or faecal calprotectin normalisation rates between the two groups, although very few patients had raised faecal calprotectin levels at study enrolment. None of the two diets had a significant effect on microbiota α-diversity. Bacterial richness and Shannon diversity were comparable between the groups and remained stable throughout the study period. Beta diversity changed slightly over the course of the study. This was not related to the diet or symptomatic remission but was weakly associated with FC concentration; an effect which was no longer significant following adjustment for multiple comparisons.

Other dietary interventions and their effect on the gut microbiota of patients with IBD

Fritsch et al. assessed in a randomised, crossover trial, the impact of a catered, low-fat, high-fibre diet (LF/HF) compared to a typical American diet enriched with fruit, vegetables and fibre on the gut microbiota of patients with quiescent UC^{(Reference Fritsch, Garces and Quintero46)}. Microbial α-diversity was not impacted by either of the two diets, while a trend for increased β-diversity was observed only after the LF/HF diet. Adherence to the LF/HF was associated with a higher abundance of Bacteroidetes, Prevotella and fewer Actinobacteria, compared to baseline. The relative abundance of F. prausnitzii was also higher after the LF/HF compared to the improved standard American diet. The increase in the relative abundance of Bacteroidetes and Prevotella are in accordance with the effects of plant-based diets on the gut microbiota composition^{(Reference Wu, Chen and Hoffmann53,Reference De Filippo, Cavalieri and Di Paola54)} . Faecal metabolome profiles showed a clear separation between the two dietary interventions and a significant increase in levels of acetate and tryptophan was noted, along with a reduction in lauric acid after the LF/HF. Using regression models, dietary changes had more pronounced effects on the faecal microbiota compared to the metabolome. Significant improvements in quality of life scores and serum amyloid A were observed only following the LF/HF compared to baseline, while faecal calprotectin levels remained low following both diets suggesting that in patients in remission with conventional medication a LF/HF does not exacerbate gut inflammation.

Zhang et al. assessed the impact of a Mediterranean-style diet, which is believed to protect from the development of CD, on the faecal microbiota of patients with quiescent CD^{(Reference Zhang, Taylor and Shommu41)}. Patients who followed a diet enriched in red and processed meat and low in fibre, fruit and vegetables were allocated to the intervention arm, while every other patient followed their standard of care medication and unrestricted diet. Although the dietary intervention did not influence α-diversity, the baseline differences observed in β-diversity between patients on the intervention arm and those on the control group were no longer significant after 12 weeks. In the dietary intervention group, an increase in F. prausnitzii levels was observed after 12 weeks, along with a significant reduction in Escherichia/Shigella and overall Proteobacteria, compared to baseline. Although these changes represented a shift towards a less dysbiotic microbiota in CD, there was no effect of the diet on faecal SCFA levels. Faecal calprotectin levels remained in normal ranges over the course of the intervention in both groups.