Analysis of 61 exclusive enteral nutrition formulas used in the management of active Crohn’s disease—new insights into dietary disease triggers

Summary Background Exclusive enteral nutrition (EEN) is an effective treatment for Crohn's disease. Aims To investigate the hypothesis that ingredients of EEN formulas are unlikely to initiate a disease flare and that their dietary elimination is not essential for disease amelioration. Methods We performed compositional analysis of EEN formulas with evidence of efficacy in management of active Crohn's disease. Macronutrient content was compared against the dietary reference values (DRV), the UK National Diet and Nutrition Survey (NDNS) and intake of Crohn's disease children. Food additives were cross‐referenced against the FAO/WHO database. Results Sixty‐one formulas were identified with variable composition (carbohydrates [22.8%‐89.3%], protein [7.8%‐30.1%], fat [0%‐52.5%]). Maltodextrin, milk protein and vegetable/plant oils were the commonest macronutrient sources. Their n‐6:n‐3 fatty acid ratio varied from 0.25 to 46.5. 56 food additives were identified (median per formula: 11). All formulas were lactose‐free, gluten‐free, and 82% lacked fibre. The commonest food additives were emulsifiers, stabilisers, antioxidants, acidity regulators and thickeners. Food additives, implicated in Crohn's disease aetiology, were present in formulas (modified starches [100%], carrageenan [22%], carboxymethyl cellulose [13%] and polysorbate 80 [5%]). Remission rates did not differ between EEN formulas with and without those food additives. Analysis including only formulas from randomised controlled trials (RCTs) retained in the latest Cochrane meta‐analysis produced similar findings. EEN formulas contained less energy from saturated fat than NDNS intake. Conclusion We have identified food ingredients which are present in EEN formulas that are effective in Crohn's disease and challenge perceptions that these ingredients might be harmful.


| INTRODUC TI ON
Inflammatory bowel disease (IBD) is more prevalent in Western countries and its incidence is increasing, particularly in low-and medium-income countries which are in financial transition, and in those who adopt a Western lifestyle. 1 This points to an environmental disease trigger, as the rapid increase in incidence outpaces changes in transcriptional human genetics.
Epidemiological evidence from observational studies has implicated certain dietary macronutrients, such as polyunsaturated fatty acids (PUFA) and fibre, in IBD onset, 2 but dietary intervention studies, based on this evidence, have been unable to reverse the disease course and induce or maintain disease remission in patients with IBD. 3,4 In preclinical studies in animal models, a diet high in total fat and sugars aggravated inflammatory response, induced intestinal dysbiosis, promoted overgrowth of pro-inflammatory Proteobacteria with a parallel decrease in protective members of the gut microbiome, reduced production of short chain fatty acids and suppressed the mucosal expression of their G-protein coupled receptor 43. 5 Our habitual diet has evolved enormously in recent decades, not only in terms of nutrient composition but also, importantly, in terms of nonnutrient food ingredients used in food preservation and processing. Although food industrialisation has safeguarded humans from infectious diseases and increased food durability, availability and global accessibility, the effect this may have on host health and the onset of noncommunicable diseases is now becoming clearer. Recent studies in animal models have indicated that food additives might affect colonic and cardiovascular health, mediated by their effects on the gut microbiome, the mucus layer, the gut barrier function and the gut associated immune system. 6,7 Dietary emulsifiers, such as polysorbates and carboxymethyl cellulose have been shown to increase intestinal permeability, alter microbiome composition, promote Escherichia coli translocation across the intestinal epithelium and cause gut inflammation in M cells in vitro and in animal experiments. [6][7][8][9][10] Such epidemiological and preclinical observations have quickly been translated to dietary recommendations and presumptive therapies for the management of IBD. 7 However, it is important to accept that there are currently no well-controlled intervention studies in humans to prove that exposure to food additives increases risk of IBD onset, or indeed aggravates gut inflammation in those with the illness.
Likewise, there are no intervention studies to prove that exclusion of food additives, sugars or milk fat mitigates colonic inflammation in IBD patients or that their introduction to the diet of patients with disease in remission initiates disease flare. On the contrary, data from a dietary intervention showed that among patients with Crohn's disease in remission, the levels of red and processed meat consumption were not associated with time to symptomatic relapse. 11 Exclusive enteral nutrition (EEN), a liquid-only diet using proprietary formula, is the only effective and established dietary treatment which induces remission in active Crohn's disease. It is the first-line treatment for children with active Crohn's disease throughout Europe, Oceania and parts of North America. 12 Treatment with EEN for 8 weeks induces clinical remission and disease improvement in approximately 80% of children with active Crohn's disease and suppresses colonic inflammation. [13][14][15] The EEN formulas used for the management of Crohn's disease are extensively industrialised products which vary considerably in terms of their composition of nutrients and nonnutrient food ingredients, including food additives. It is therefore possible to use the compositional profile of EEN formulas, shown to induce remission in Crohn's disease, to elicit clues about the role of various food components in the disease course. We hypothesise that nutrients and food ingredients which are included in one or more EEN formulas with clinical efficacy data in the management of active Crohn's disease are unlikely to play a major role in the disease course and are therefore unlikely to initiate or sustain a disease flare in Crohn's disease.
Conversely, nutrients and food additives, or other nonnutrient food ingredients, which are both specifically absent from the composition of EEN formulas (eg lactose, gluten), and whose hypothetical inflammatory mechanism is supported by preclinical data, warrant further exploration of their role in Crohn's disease.
We performed an extensive nutrient, nonnutrient food ingredi-

Conclusion:
We have identified food ingredients which are present in EEN formulas that are effective in Crohn's disease and challenge perceptions that these ingredients might be harmful. OR "inflammatory bowel disease") AND ("enteral nutrition" OR "nutritional support"), Figure 1.  16 The GSFA is a standardised database of food additives, permitted to be used in foods, grouped into 27 different functional classes. Each food additive identified within the EEN formulas was assigned to its functional class. Food additives implicated in IBD pathogenesis were identified from major recent reviews in the topic 6,17 and any additional evidence from published primary research. 18 From the studies which reported remission rates (eg percentage of patients entering remission), median differences in remission rates between EEN formulas which contain these implicated food additives and those which do not were compared.
Subgroup analysis was performed, in which we examined only the EEN formulas which were used in at least one of the RCTs retained in the most recent Cochrane meta-analysis of efficacy of EEN in active Crohn's disease. 19 We followed this approach to select only EEN formulas trialled within an RCT.

| Nutritional composition analysis of EEN formulas
Sixty-one formulas used for induction of remission in active Crohn's disease were included in the final analysis. Of these formulas, 39/61 (64%) were polymeric, 16/61 (26%) were semi-elemental and the re-

| Food additives in EEN formulas
Of the 61 identified EEN formulas, information pertinent to food additives was available from 55 (90%) of these. In total, 56 unique

| Food additives and other ingredients implicated in IBD in preclinical research
In recently published research, maltodextrin, carboxymethyl cellulose, polysorbate 80, carrageenan, inorganic phosphates, sucralose and microparticles such as aluminium silicate and titanium dioxide were reported as causative agents in gut inflammation in animal models and in vitro experiments, and by extension have been implicated in IBD onset and disease management. 6,17 Modified starch, including maltodextrin, was present in all EEN formulas, carrageenan was present in 12/55 (22%), carboxymethyl cellulose was present in 7/55 (13%) and sucralose and polysorbate 80 were present in 3/55 (5%) EEN formulas, Table 1. No EEN formula contained titanium dioxide or propionic acid salts; the latter has been shown to serve as a potential mediator for pro-inflammatory action of adherent-invasive E coli, 18 Table 1. Inorganic phosphates, compounds such as phosphoric acid or phosphate salts (sodium, magnesium, calcium and potassium phosphates) were found in 49/54 (91%) of the EEN formulas, Table 1. Soy lecithin has been implicated in IBD pathogenesis, due to its pro-inflammatory potential in animals fed high fat diets, 23   food additive, present in 38/55 EEN formulas (69%). Gluten, which has been associated with disease pathogenesis in preclinical studies, 6 was not present in any of the EEN formulas surveyed here,

| Comparison of EEN macronutrient composition with the NDNS intake of children
Compared with the NDNS data of UK children on a Western diet, which describe the dietary intake of a representative population of UK children, there was no significant difference in the median proportion of energy from carbohydrates, protein, total fat and sugars, between EEN formulas and the intake from the NDNS. Conversely, the median percentage of energy derived from saturated fat in EEN formulas was lower by 7.7% (P = 0.001), and the median fibre intake by 25 g (P < 0.001), compared to the NDNS intake, Figure 5.

| Comparison of EEN macronutrient composition with the UK DRV for children
The macronutrient composition of the EEN formulas was also compared against the UK DRV, the national dietary recommendations, Figure 5.

| Subgroup analysis including only EEN formulas from RCTS retained in cochrane meta-analysis
Twenty unique EEN formulas had been used in at least one of the RCTs retained in the latest Cochrane meta-analysis. 19 This subset of EEN formulas had similar sources of macronutrients as with the full set of 61 EEN formulas presented above. Maltodextrin and milk protein were the commonest sources of carbohydrates and protein, respectively, while coconut oil was the most common source of fat used, Figure 2. One single formula contained fibre (7%), in the form of fructo-oligosaccharides.
There was a substantial variation in the macronutrient content of these EEN formulas, as was similarly identified when considering  (Table S3). Information pertinent to food additives content was available for 15/20 (75%) of the Cochrane meta-analysis subset, with a total of 43 food additives present, Figure 3.
Apart from carrageenan and sucralose, all other food additives currently implicated in Crohn's disease pathogenesis were present in all these EEN formulas, Table 1. Within this group of 20 formulas, we found no difference in median remission rates between those EEN formulas which contained food additives implicated in Crohn's disease and others which did not, Figure 4.

| D ISCUSS I ON
In this study, we report the macronutrient, nonnutrient food ingredi- Patients with Crohn's disease are often at high risk of undernutrition, 13,34 particularly those with active disease, and are reported to have poor food-related quality of life and often introduce food aversions. 35 The results of this study would help to remove a significant degree of food-related anxiety. This study provides a list of macronutrients, along with their sources of origin, nonnutrient food ingredients and food additives, that health professionals can use as a guide to advise their patients on permitted food ingredients; at least within the amounts these are contained in EEN formulas and consumed during an EEN course.
A secondary aim of this study was to compare the macronutrient content of EEN formulas with the national DRV, the intake of heathy children following a Western diet from the UK NDNS and a group of children with Crohn's disease, whose dietary intake had been described in a previous study. 21 Differences were observed when comparing EEN formula macronutrient composition with the median intake of both the healthy children and children with Crohn's disease. One may assume that EEN works by limiting the amount of sugars and total fat, and in particular saturated fat. However, the fact that EEN formulas have a wide variation in the content of these macronutrients and still demonstrate clinical effectiveness contradicts this assumption. Once again, subgroup analysis with inclusion of EEN formulas from RCTs retained in Cochrane meta-analysis mirrored the findings of the initial set of all formulas.
Here, we provide a list of food ingredients which are unlikely to be harmful, within the amounts these are consumed during an EEN course. However, we are unable to comment on food additives and other nonnutrient food ingredients which have been implicated in gut inflammation in Crohn's disease but are not present in at least one of the EEN formulas. 18 For example, calcium propionate has been implicated in Crohn's disease in preclinical research 18 but it is not present in EEN formulas. It is possible that such food additives are important to Crohn's disease pathogenesis and future studies should explore whether their selective exclusion is associated with a therapeutic signal in Crohn's disease patients. Similarly, to the best of our knowledge, all EEN formulas were gluten and lactose-free. It is therefore possible that the mode of action of EEN is mediated by exclusion of gluten, lactose or other ingredients in food which coincide with the presence of gluten. In support of this claim, novel dietary treatments of Crohn's disease with early signals of clinical efficacy, including reduction of inflammatory markers, eliminate gluten and lactose intake. 7,36 Similarly, a cross-sectional study in 1,647 patients with IBD showed that adherence to a gluten-free diet was associated with improvement in gastrointestinal symptoms. 37 Fibre has long been advocated as a beneficial nutrient for Crohn's disease patients despite well-designed RCTs failing to prove any profound benefit. 3 However, despite their clinical effectiveness and amelioration of gut inflammation, the large majority of the EEN formulas analysed here (80%) lack any fibre. This observation does not mean that fibre is essentially harmful for patients with Crohn's disease but, certainly, its elimination does not exacerbate symptoms in patients with active Crohn's disease. However, it is worth reporting that of the few EEN formulas which contained fibre, the predominant ones were inulin, fructo-oligosaccharides, pectin and gum arabic which do not encompass the variable range of fibres included in our habitual diet.
There is extensive interest in the role of PUFA in the aetiopathogenesis of Crohn's disease and inflammatory bowel disease. 38 High intake of n-6 pro-inflammatory PUFA has been associated with an increased risk of IBD onset. 2 However, the wide range of fat content (ie 0% to 50%) in EEN formulas, the majority of which originates from n-6-containing vegetable oils, suggests that their role in disease activity is likely negligible. This assumption is further supported from a recent Cochrane review. 19 There are limitations to this study. The exact concentration of food additives in some EEN formulas were unavailable and would not be disclosed by the manufacturers of EEN formulas, despite authors' requests. The fact, though, that food products for medical use are closely regulated by health and food standard authorities meant that we were able to obtain a large amount of nutritional information which is not normally available from the manufacturers of "ordinary" food. Although we provided a list of nutrients, food ingredients and food additives contained within EEN regimes, we are unable to comment on other food additives implicated in the pathogenesis of IBD which were not included in the composition of the EEN formulas described here. Furthermore, there may have been some changes in the composition of EEN formulas over the years. EEN formulas used in older studies may have had different composition to those used in this analysis which is based on their current ingredients. It was also not possible to explore associations between the concentration of food additives within EEN formulas and the magnitude of clinical response to EEN. As mentioned above, this information was unavailable in most of the cases and the methods used to assess specific disease activity and biomarkers varied considerably between studies preventing more complex data synthesis on efficacy (Table S2). However, differential analysis between EEN formulas containing these food additives and others which do not, showed no difference in remission rates between these two groups, thus further challenging their role in disease management.
Although we performed an extensive literature search, there is also a possibility that we may have missed EEN formulas which were used for treatment of active Crohn's disease in routine clinical practice but have never been described in a peer-reviewed publication.
Furthermore, in studies which used more than one EEN formulas, and which reported cumulative efficacy signals only, it was not possible to assign remission rates for each type of formula separately.