The exact mechanisms of action of EEN remain unknown and involve several synergistic factors. These include the exclusion of many “offending foods” or other potentially damaging dietary components (such as food preservatives and emulsifiers), direct effects on host immune response, microbial modulation, and enhanced intestinal barrier function with decreased inflammation; these mechanisms have been reviewed extensively in other papers [28,39,43]. Direct benefits from the availability of a rich and balanced diet (as enterocytes obtain most of their nutrients from the luminal content) have also been suggested to contribute to the effectiveness of EEN, with specific effects on enterocyte differentiation [44].

The principles of dietary exclusion are common to other diets reviewed below (i.e., Mediterranean diet, specific carbohydrate diet, and CD-TREAT diet) and are best supported by the lower efficacy of partial enteral nutrition (PEN), where only 15% of children achieved remission with 50% formula and 50% regular diet [45,46,47]. Some dietary components have already received attention as potentially “‘offending foods”; however, most studies supporting these effects are conducted in animal models or cells in a culture, given the difficulty of conducting dietary research in humans with IBD [48].

One example is emulsifiers, which have been shown to have negative impacts on the gut environment, including impaired mucosal immunity, intestinal permeability, and the microbiome [49]. A 2015 study by Chassaing et al. in mice showed the ability of two commonly used emulsifiers, carboxymethylcellulose and polysorbate-80, to induce inflammation and metabolic changes, mediated by gut microbes [32]. Multiple studies have investigated the negative effects of a “WD,” including animal-based proteins and fats, and have shown that replacing these with plant-based food sources is beneficial in IBD animal models. Again, these have been hypothesized to be mediated by effects on the microbiota [37].

The theory that processed food is a primary driver of inflammation in IBD has been somewhat challenged by work published by Logan et al. in 2020, who demonstrated that some components of successful EEN formulae include preservatives and processed compounds [50]. Furthermore, the role of fibre (lack of fibre) in IBD is supported by multiple observations, such as damage to the intestinal mucosal layer when gut microbes are starved of fibres [29]. Following these observations, one would expect that a liquid formula diet (as used in EEN) that is low in fibre would not confer benefit; however, dietary fibres are diverse and have various effects on bowel homeostasis and IBD pathogenesis [51]. These examples highlight the complexity of nutritional science in IBD and the need for better-controlled studies.

Data supporting the effectiveness of EEN have been reviewed in several meta-analyses and indicate at least equivalent clinical response to steroids (18 studies, N = 1329, OR = 1.35, 95% CI: 0.90–2.10, favouring EEN; p = 0.14) with most studies showing a remarkable 80–85% remission rate [52,53]. EEN was superior to steroids for inducing mucosal healing in a subset of the studies (OR = 4.5, 95% CI: 1.64–12.32) [52,53]. Emerging data from studies across multiple centres, including adult patients with CD, have shown similar results [54,55,56]. Positive effects of EEN (and other dietary therapies) can be seen as early as two weeks after initiation, including improvement in clinical disease activity scores, fecal calprotectin (FC), C-reactive protein (CRP), and extraintestinal manifestations of CD such as bone health [57,58,59]. EEN is recommended as first-line therapy to induce remission in paediatric CD [42]. While these recommendations differ in adult CD management [2,60], a prospective, open-label trial from 2018 in Japan assessing the role of combination infliximab dose escalation with elemental diet versus infliximab dose escalation alone had to be stopped early due to clear benefits in the combination therapy over monotherapy treatment group in the midterm analysis [61].

Early experience suggested that EEN was more likely to be effective in patients with small bowel disease than colonic CD, based on higher stooling frequency during initial treatment in colonic CD, likely owing to osmotic effects of formula feeds in the affected colon. The effectiveness of EEN has since been demonstrated regardless of the site [62]. Nevertheless, the experience of EEN for the treatment of UC is limited. A recent open-label RCT from 2021 by Sahu et al. assessed the role of EEN in acute severe ulcerative colitis (ASUC) [63]. A total of 62 adult patients were randomized to semielemental formula for seven days along with intravenous (IV) corticosteroids or IV corticosteroids alone. Patients receiving EEN + IV corticosteroids compared to IV corticosteroids alone had lower rates of treatment failure (19% versus 43%; p = 0.04), shorter length of stay in hospital (median 10 versus 13 days; p = 0.04), and a lower composite outcome of colectomy/hospitalization at six months (16% versus 39%; p = 0.045).

Many of these studies help focus attention on the role of gut microbes in the pathogenesis of IBD, especially the presence of pathobionts, and how EEN might dampen such adverse effects and reduce pathobionts. Observed changes in microbiota with EEN are diverse and depend on study design and technical consideration. Nonetheless, some taxonomic alterations are reproducible [38,75]. These changes can be rapid, occurring even within one week of initiating EEN therapy [76]. Indeed, dramatic effects are observed, some of which are counterintuitive to our understanding of the impact of microbiota on CD pathogenesis. For example, EEN appears to reduce microbial diversity, lower SCFA concentrations (including butyrate), and reduce Faecalibacterium prausnitzii, which is usually considered beneficial in IBD [77]. In another study, responders to EEN showed lower bacterial richness than nonresponders [78]. Quince et al. showed a decrease in Shannon diversity with EEN, but this returned to pretreatment levels two months after EEN was stopped, as did decreases in Bifidobacterium, Ruminococcus, and Faecalibacterium [79]. It is hypothesized that some of these decreases in specific taxa and diversity are simply due to the lack of fibre in EEN [80,81]. Furthermore, a baseline reduction in dysbiosis-associated taxa could potentially be a marker of better response to EEN, where patients who respond to therapy may subsequently have paradoxical increases in pathobiont taxa.

It remains unclear whether microbial shifts observed with EEN are responsible for the effectiveness of therapy or are a consequence of successful therapy and resolution of inflammation. One recent study showed an improvement in microbial diversity with both EEN and steroids, associated with response to therapy, suggesting that at least some of the microbial changes may simply be due to reduced inflammation [86]. Collectively, these studies do suggest that EEN does act through compositional and functional microbial changes, but much remains to be understood in this growing field of research.

EEN has been shown in numerous high-quality studies to be an effective treatment for the induction of remission in CD. Large systematic reviews have indicated that EEN and CS are equally effective in inducing remission in CD, and EEN is recommended as first-line therapy for induction for paediatric CD. A recent open-label RCT suggests a short course of EEN and CS in combination may be more effective than CS alone for ASUC. While there are small studies to indicate EEN is superior to CS for mucosal healing, further work is needed before this can be concluded.

The mechanism of action of PEN remains unknown, but similar mechanisms underlying the effectiveness of EEN mediate the effects of PEN. Preclinical studies assessing modulation of colonocyte inflammation using polymeric formula demonstrate improved cell viability and survival with exposure to polymeric formula, compared with cells stimulated by tumour necrosis factor-alpha (TNF-α) [88]. Additionally, the same group demonstrated both enriched nuclear factor kappa B subunit 1 (NFκB1)-associated pathways via microarray analysis after applying the polymeric formula in TNF-α stimulated colonocytes and significant downregulation of TNF signalling. In a small study of 34 patients, Marques et al. described metabolomic differences between paediatric patients with minimally active or quiescent disease receiving PEN for maintenance of remission as compared to those receiving a regular diet [89]. Baseline differences in metabolomic profiles between treatment groups may have affected results. Further research is needed in this area, including a head-to-head comparison of microbial composition and function in patients receiving PEN versus a regular diet.

There is considerable heterogeneity among studies assessing the clinical effectiveness and efficacy of PEN. This may stem from the lack of standardization with the amount administered (percentage of calories per day), type of food allowed, and duration of treatment across centers [68]. Without a concomitant exclusion diet, there is a lack of convincing data supporting the use of PEN to induce clinical remission in CD.

Despite this, PEN has been used successfully to maintain clinical remission in CD [45,66,67,90,91,92,93,94,95,96]. Additionally, patients on PEN show improvements in nutritional status, defined by improved weight gain, linear growth and levels of micronutrients such as vitamins A, B12, D and E and minerals [92,93,94,96,97]. A systematic review (n = 429) of prospective studies assessing the effectiveness of PEN for maintenance of remission in CD demonstrated that PEN was superior at preventing clinical relapse (RR 0.67, 95% CI: 0.54–0.82, p < 0.01) and achieving clinical remission (RR 1.32, 95% CI: 1.07–1.64; p = 0.01) than those not receiving nutritional therapy [68].

A recent abstract presented at the 2021 European Crohn’s and Colitis Organization’s annual congress reported 100 treatment-naïve paediatric CD patients randomized to cyclic EEN (N = 49), or daily supplementary nutrition (PEN) combined with normal access to food (N = 51) for 12 consecutive months [98]. Fifty-one percent (25/49) of patients in the cyclic EEN group versus 24% (12/51) in the PEN group (p = 0.004) remained in remission at 12 months, suggesting that paediatric patients with CD can be maintained in remission with cyclic EEN without immunosuppression or biologics. Patients in the PEN group showed a relapse rate of 76%, suggesting PEN was not as effective as cycling EEN, nor was an effective long-term maintenance strategy. More extensive multicentre randomized–controlled trials (RCT) are needed to compare PEN protocols and define minimal percent of total calories required from enteral formula for clinical benefit in paediatric patients.

Current data do not demonstrate the efficacy of PEN for the induction of remission in CD. However, PEN may have a role in the maintenance of remission for paediatric CD based on a recent systematic review. Existing studies have varied in methodology and inclusion criteria, particularly in defining percent calories that constitute “partial” enteral nutrition. Further research applying a common, standardized methodology will be required to further describe the role of EEN as maintenance therapy [98].

The first results of CDED were published in 2014 with 33 children and 14 adults [70]. Participants followed a 12-week protocol, as described above. Remission was defined as Paediatric Crohn’s Disease Activity Index (PCDAI) <7.5 or Harvey Bradshaw index (HBI) ≤3. Taken together, remission was seen in 70% of children and 69% of adults at week six, 84% of patients maintained remission at week 12, and 70% normalized their CRP. Disease severity was the only predictor of lack of response. Only 33% of patients with severe disease at baseline (PCDAI > 40, HBI > 9) reached remission, compared with 71–75% of patients with mild-to-moderate patients (PCDAI 7.5–27.5 or HBI 4–6, mild disease; PCDAI 30–37.5 or HBI 7–8, moderate disease).

A second study was published in 2017 and assessed the use of CDED in patients who lost response to infliximab or adalimumab [99]. The study included 11 adults and 10 children, and all participants received 12 weeks of CDED with 50% PEN (rather than the two-step protocol described above). There was a significant decrease in HBI and CRP at weeks six and twelve and an improvement in albumin at week six. Clinical response occurred in 90% of patients at week six, and clinical remission in 62%. The authors also describe 18 patients who received CDED alone (without PEN), with 77% reaching clinical remission (14 patients).

In 2019, the first multicentre RCT was conducted exclusively in children comparing CDED + PEN to EEN for remission induction [69]. A total of 74 patients with mild to moderate CD were included and either received the 2-phase 12-week CDED protocol (as described above) or six weeks of EEN followed by 75% unrestricted diet +25% PEN, gradually returning to normal diet at week 12. The primary outcome was the tolerability of the intervention. Tolerability was significantly higher in the CDED group (97% CDED versus 74% EEN). At week six, CDED and EEN were found to be equally effective at inducing remission, while at week 12, there was a significantly higher proportion of patients in corticosteroid-free sustained remission in the CDED group (76% CDED versus 45% EEN groups); 87.5% of patients in remission at week six, maintained remission at week 12. This was associated with sustained improvement in markers of inflammation (CRP and FC) in the CDED group. Data from this group were also analyzed for response at week three (rapid responders) and week six (sustained response). At week three, 82% of the CDED group and 85% of the EEN group showed a response (PCDAI decrease > 12.5) or remission (PCDAI < 10). Remission at week three significantly increased the odds of remission at week six (OR 6.37, 95% CI: 1.6–25, p < 0.05), and week 12 (OR 4.5, 95% CI: 1.2–16.5, p < 0.05) [59,69].

Two further CDED studies were published in 2021. Real-world data was collected by Niseteo et al. to compare CDED to EEN [100]. The majority of patients received one to two weeks of EEN prior to starting CDED. There was comparable efficacy between CDED and EEN for induction of remission (66% EEN versus 75% CDED groups); however, patients who received CDED showed significantly higher weight gain. Yanai et al. performed the first adult CDED + PEN versus CDED study and showed that CDED (even without additional PEN) is a viable treatment option in mild-to-moderate (biologic naïve) CD [101]. Patients on CDED showed significant improvement in calprotectin and CRP, with sustained remission in 80% at 24 weeks and endoscopic remission in 35%.

Clinical response in CDED was associated with changes in microbial composition and function, based on 16 s ribosomal ribonucleic acid (rRNA) and metagenomic sequencing analyses. In the Levine et al. 2019 RCT, remission in the CDED + PEN and EEN groups was associated with changes in microbial diversity, a decrease in Proteobacteria and an increase in Firmicutes [69]. However, changes at the genus level for Firmicutes phyla were not consistent across diets. Achieving remission at week six with CDED + PEN was associated with a significant increase in Firmicutes, particularly Clostridiales, driven by Roseburia, Oscillibacter, Anaerotruncus and Ruminococcus. Remission also led to a significant decrease in Proteobacteria, particularly Gammaproteobacteria. After week six (as patients entered Phase 2 of CDED + PEN), changes continued to be seen in the microbiome. A further decrease occurred in Proteobacteria up to week 12. Patients who did not achieve remission with CDED + PEN at week six showed fewer microbial changes; however, this was not statistically significant given the small number of children who failed to achieve remission [69,102].

When comparing children achieving remission, both CDED + PEN and EEN diets showed a decrease in Proteobacteria and an increase in Firmicutes. Although Proteobacteria abundance decreased, relative abundance was still high compared to healthy controls [76], particularly in Escherichia coli [102]. The increase in Firmicutes, although significant, was not consistent across genera between the two diets. Children receiving EEN showed increases in Firmicutes driven by Clostridiales, Erysipelotrichaceae and Veillonellaceae [102]. As both diets show a decline in Proteobacteria, this suggests that Proteobacteria might be important taxa in dietary interventions that affect intestinal inflammation in CD. The increase in the relative abundance of Firmicutes may therefore be due to secondary niche expansion following the decrease in Proteobacteria. This is supported by the observation that Firmicute expansion occurs with EEN, despite EEN not containing fibre, which is thought to be an important substrate that ordinarily supports Firmicute expansion [103].

Emerging data suggest that CDED may be superior to EEN for the induction of clinical remission, based on improved tolerability profiles. Additional studies are needed to further describe the role of CDED as maintenance therapy and an adjunctive treatment alongside PEN and medications.

There is a lack of high-quality studies assessing the effect of SCD. Suskind et al. reported results of an exploratory multiomics pilot study comparing the SCD, modified Specific Carbohydrate Diet (mSCD) and Whole Food diet (WF) in 18 paediatric patients with CD, of which only 10 completed the study [68]. All participants spent two weeks on the SCD prior to being randomized into one of three dietary treatment groups: SCD, mSCD, or WF. The mSCD allowed rice and oats while the WF eliminated wheat, corn, sugar, milk, and food additives. All meals were prepared by a study chef, and all patients were counselled by a dietician who specialized in SCD throughout the study. Clinical remission was seen more often in children completing the SCD after 12-weeks, but not with mSCD or WF diet. After 12 weeks, SCD was associated with decreases in PCDAI (21.9 ± 5.5 to 1.9 ± 3.8), whereas most children participating already had normal CRP at baseline (1.3 ± 0.7 mg/dL to 0.9 ± 0.5 mg/dL) and normal or mildly elevated erythrocyte sedimentation rate (ESR) (15.6 ± 13.7 mm/h to 13 ± 14.2 mm/h).

Patients in the mSCD group showed the greatest decrease in FC compared with the other groups (697 ± 520 mg/kg to 157 ± 156 mg/kg). Although children who followed the WF diet also showed a decrease in inflammatory markers, it was associated with weight loss, lethargy, hunger, and a higher rate of allergies. Participants from all three dietary interventions showed some improvement; however, the statistical significance of the results is limited due to the small sample size (n = 16), with even fewer patients completing the study (n = 10). Therefore, further research is required to determine the clinical response of SCD to Crohn’s disease. [68].

When compared to the Mediterranean Diet (MD) (see: Mediterranean Diet), Lewis et al. concluded that SCD is not a superior therapeutic option in the dietary treatment of IBD [47]. This 2021 RCT evaluated symptomatic remission, clinical remission and inflammatory biomarkers in adult patients with CD experiencing mild to moderate symptoms (n = 197). Participants were randomized to the SCD group (n = 101) or MD group (n = 96) and followed over 12 weeks. The primary outcome was symptomatic remission, and the secondary outcomes were FC response and CRP response. All participants were provided with prepared meals and snacks until week 6 of their assigned diet, after which they were asked to continue their diet independently. At week 12, symptomatic remission was achieved in 42.4% of the SCD group and 40.2% of the MD group (p = 0.87), while clinical remission was seen in 40.4% of the SCD group and 46.7% of the MD group (p = 0.28). Further, 26.1% and 7.7% achieved an FC response in the SCD and MD groups, respectively (p = 0.20). A CRP response was seen in only 10.8% of the SCD group and 7.1% of the MD group (p = 0.55).

There was no significant difference in the primary outcome between both groups, and combined symptomatic remission with the concurrent resolution of inflammation was rare. Although both diets were well tolerated, the SCD provides no additional benefit than the MD, and its restrictive nature may limit long-term adherence to the diet [47].

Microbiome changes were variable among participants from the 2020 study by Suskind et al. [68]. Some studies have identified increased microbiome diversity in CD patients on the SCD, while others have used murine models to show that carbohydrate monotony provides a selective pressure favouring certain intestinal bacterial species while decreasing the total diversity [106,108]. Microbial composition in all three groups (SCD, mSCD, WF) showed a trend towards increased microbial diversity (Inverse Simpson Index), from 4.67–14.09 at enrollment to 10.93–17.76 at the end of the two-week study period [68]. Conversely, two patients had a modest decrease in microbial diversity (13% and 19%), and two patients had low diversity prior to initiating the study.

Across all cohorts, Blautia, Lachnospiraceae, Faecalibacterium, Roseburia, Anaerobutyricum and Eubacterium were enriched; Escherichia coli and Faecalibacterium prausnitzii had decreased abundance. Although less pronounced, stool metagenomics analyses showed increased metabolic activity in the mSCD cohort associated with starch and sucrose metabolism and increases in amino acid metabolism. The WF diet cohort showed increased amino acid metabolism and variable changes in glutamate synthase, methionine synthase, and 2-dehydro-3-deoxy-D-pentonate aldolase enzyme activity. These alterations in microbiome activity are indicative of adaptions to dietary changes and impact the immune system by abating intestinal inflammation. However, the mechanism by which these changes occur remains unclear [68].

There is a lack of high-quality studies supporting SCD. Existing data are limited by small sample sizes, and in the largest study (DINE-CD), no significant differences were found between the SCD and MD [47]. Current evidence does not suggest a role for the SCD in the induction or maintenance of remission for CD or UC.

The CD-TREAT diet is still in the very early stages of assessment. Therefore, no conclusions about its efficacy for the induction or maintenance of remission in CD or UC should be made yet.

The DINE-CD trial (see: SCD) compared the efficacy of the SCD to the MD in adults with CD. This study was designed to test the hypothesis that the SCD is superior to the MD and hence was powered as a superiority trial. Patients with mild to moderate CD, defined as a short Crohn’s disease Activity Index (sCDAI) score >175 and <400, were randomized to receive either the SCD (n = 101) or the MD (n = 93). Patients received six weeks of delivered meals and snacks, followed by six weeks of independent meal preparation based on instructions from the study team with recommended ingredients. SCD was not found to be superior to the MD. DINE-CD showed no significant difference in clinical remission rates or change in FC. However, only a minority of patients had elevated FC at baseline, suggesting that patients who entered the trial had very mild disease. Given the comparative effectiveness design of this trial and the lack of a control arm, this study does not help assess the overall efficacy of either diet or show any benefit or inferiority compared to an individual’s usual diet [47].

The association of low adherence to the MD and TMAO may be significant. Decreased TMAO levels are seen in IBD compared to a non-IBD population [116]. TMAO is produced by gastrointestinal anaerobes through the digestion of dietary phosphatidylcholine and carnitine. This compound may serve as a biomarker for IBD [116]. TMAO has been found to increase following fecal microbiota transplant (FMT) in patients with Crohn’s disease prior to their second dose of FMT [117]. In a study of eight adults with CD on six weeks of a “Mediterranean-inspired diet,” significant changes in gene expression were observed. While patients self-reported adherence, results were compelling. While no single gene stood out, metagenomics analysis showed a change in expression of >3500 genes whose cumulative effects have been associated with “normalizing” intestinal microbiota to a composition more similar to healthy controls than patients with IBD, with an increased abundance of Bacteroidetes and a decrease in Proteobacteria and Bacillaceae [83].

With the exception of CDED, high-quality data to guide real food dietary recommendations are lacking [118]. If the CDED is not being considered to treat an individual with CD, then the MD, given its relative ease and its additional reported health benefits, may be a sensible dietary recommendation for patients seeking general advice on nutrition choices in IBD [112,114,115].

There are insufficient data to support the role of the MD for the induction or maintenance of remission in CD or UC. The primary study assessing MD was a head-to-head study comparing MD to SCD, which also does not demonstrate efficacy. Without further prospective data or randomized controlled trials, the MD cannot be recommended in the management of IBD at this time.

In a recent randomized controlled trial at two large gastroenterology clinics in the United Kingdom, the LFD resulted in significant symptomatic improvement of IBS in adult patients with quiescent IBD [71]. Overall, 52% of participants in the LFD cohort experienced relief of symptoms after four weeks, compared with only 16% in the sham (placebo) diet control. Furthermore, the effects of GI symptoms on health-related quality of life, as determined by the Bowel II domain score on the IBD questionnaire, was significantly greater with the LFD.

There was no change seen in inflammatory markers. Specifically, there was no statistically significant difference in FC or CRP at the end of the trial with either the LFD or control. Interestingly, the IBD-control score demonstrated that participants with ulcerative colitis experienced greater patient-perceived control of their disease with the LFD (88.3, SEM 4.3) compared with the sham diet (74.3, SEM 4.5). Short-chain fatty acids were also lower in the UC group on LFD. These findings were not seen in the CD population [71].

A second RCT assessed the LFD (n = 26) versus a standard diet (n = 29) on IBD patients over a six-week study period [72]. There was a significant difference in median FC levels, which decreased by 34.4% in the LFD group and 4.4% in the standard diet group, but there were no differences in CRP levels. Although the LFD may reduce the symptomatic burden of IBS in IBD, more extensive series and longer study periods are needed to understand its effect, if any, on inflammatory pathways.

Functional metagenomics highlighted 34 Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthology (KO) groups that varied between LFD and sham diet controls. Out of 616 identified species in the intestinal microbiome, 29 were significantly affected by the LFD (p ≤ 0.05). Following targeted microbiome analysis, the total population of Bifidobacteria remained unaltered; however, the composition varied within the LFD group. Strains such as Bifidobacterium longum and Bifidobacterium adolescentis were significantly lower, while Bifidobacterium dentium was higher. Additionally, the total population of Faecalibacterium prausnitzii decreased without any change in specific strains [71].

Existing data do not support the efficacy of the LFD for the induction or maintenance of remission in CD or UC; however, LFD may have a role in symptomatic improvement. In two RCTs that have been conducted, there was no change in inflammatory markers, but the symptomatic improvement was shown. Neither study described the impacts on nutritional micronutrient status in study participants. Given the restrictive nature of the LFD, nutritional sufficiency would be an important outcome to assess in future studies.