Mucosal healing and a fall in mucosal pro-inflammatory cytokine mRNA induced by a specific oral polymeric diet in paediatric Crohn’s disease


Fell Dr Academic Department of Child Health, Imperial College School of Medicine, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, UK. E-mail:



Although enteral nutrition is a recognized form of treatment for intestinal Crohn’s disease, there are persisting problems with feed palatability and only limited data as to its mode of action.


To assess the effects of a specific oral polymeric diet (CT3211; Nestle, Vevey, Switzerland), which is rich in transforming growth factor β2, on the mucosal inflammatory process.


Twenty-nine consecutive children with active intestinal Crohn’s disease were treated with CT3211 as the sole source of nutrition for 8 weeks. Patients were assessed clinically, and endoscopically, whilst cytokine mRNA was measured in mucosal biopsies before and after treatment by quantitative reverse transcriptase polymerase chain reaction.


After 8 weeks 79% of children were in complete clinical remission. Macroscopic and histological healing in the terminal ileum and colon was associated with a decline in ileal and colonic interleukin-1β mRNA (pre-treatment to post-treatment ratio 0.008 and 0.06: P < 0.001, P = 0.006). In the ileum there was also a fall in interferon γ mRNA (ratio 0.15, P < 0.001) with a rise in transforming growth factor β1 mRNA (ratio 10, P = 0.04), whilst in the colon interleukin-8 mRNA fell with treatment (ratio 0.06, P < 0.05).


The clinical response to oral polymeric diet CT3211 is associated with mucosal healing and a down regulation of mucosal pro-inflammatory cytokine mRNA in both the terminal ileum and colon. In the ileum there was also an increase in transforming growth factor β1 mRNA.


The relative merits of corticosteroids and enteral nutrition in the treatment of Crohn’s disease remains an area of controversy. Although meta-analysis of the clinical trials comparing enteral nutrition therapy with corticosteroids suggests that enteral nutrition may not be as effective in inducing clinical remission, enteral therapy has the significant advantage over corticosteroids in having fewer side-effects.1[2]–3 In children in particular, the relative sparing of linear growth achieved with nutritional therapy, confirmed in several trials is a major consideration in deciding which treatment to use.4[5]–6

A significant problem with enteral nutritional therapy, however, is compliance; feeds in the past have been unpalatable, requiring a naso–gastric tube for administration.7 Recently more palatable polymeric diets, which use whole protein as the nitrogen source have become available, and have been shown to be as effective as elemental feeds.1 We have previously demonstrated that a polymeric diet used in infant nutrition (AL110; Nestle, Vevey, Switzerland) is effective in inducing remission and mucosal healing, in some children with active small bowel Crohn’s disease.8 Based on these preliminary results, a new specific oral polymeric formula, CT3211, was developed. It is casein-based and rich in transforming growth factor β2 which may be contributing to its therapeutic action.

The possible mode of action of enteral nutrition has been the source of some debate. There is however, growing evidence that it can reduce mucosal inflammation. Beattie et al. reported histological evidence of mucosal healing with enteral nutrition, whilst Breese et al. demonstrated that enteral nutrition reduced the number of mucosal cells producing interleukin-2, interferon γ and tumour necrosis factor α.8[9]–10 Further evidence of down-regulation of mucosal pro-inflammatory cytokines has been reported by Ferguson et al.11 By measuring cytokine levels in whole gut lavage fluid it was demonstrated that enteral nutrition reduced the loss of IgG and interleukin-1β into the intestinal lumen. These results emphasize the importance of focusing on the mucosa when assessing the response to treatment in Crohn’s disease.

This study documents the effect of an 8 week course of treatment with CT3211 as the sole source of nutrition in children with active Crohn’s disease. Children were assessed both clinically and endoscopically before and after treatment. The effect of enteral nutritional therapy on the mucosal inflammatory process was evaluated histologically and by more detailed studies of mucosal cytokine mRNA. These latter observations confirmed the anti-inflammatory properties of enteral nutritional therapy, and demonstrated that this effect involves both the colonic and ileal mucosa.



Twenty-nine consecutive children with active Crohn’s disease, presenting at St Bartholomew’s Hospital and the Royal Free Hospital, London, were treated with a polymeric diet CT3211 (Nestle, Vevey Switzerland). Seventeen were newly diagnosed, previously untreated cases, and 12 were children with relapsed disease. Three were being treated with prednisolone (doses ranged from 20 mg/day to 15 mg alternate days) and four were being treated with mesalazine (400 mg t.d.s.). Four had had previous enteral nutrition therapy. Patients where drug therapy had been changed within the last month were excluded, as were those who were receiving prednisolone at a dose of greater than 0.5 mg/kg, other immunosuppressive therapy such as azathioprine or cyclosporin, or had radiological evidence of stricturing.

The children aged from 8.1 to 17.1 years (median 13.6 years, 21 male). Twenty-three were pre-pubertal or in early puberty (Tanner stage 1–3). All had histological evidence of Crohn’s disease in intestinal mucosal biopsies, apart from one child who had radiological evidence of the disease in the proximal small bowel. Seven had disease affecting the small bowel alone, 17 had small and large bowel disease, and five had large bowel disease as defined by barium meal and follow through and macroscopic appearances at colonoscopy. Further clinical details of the cases studied are available from the authors on request.

Control mucosal biopsy specimens

Control mucosal biopsies for cytokine mRNA measurement (seven terminal ileal, nine transverse colonic) were obtained from 16 children aged 2.2–15.5 years (median 10.5 years). They were undergoing colonoscopy for a variety of indications such as chronic abdominal pain and rectal bleeding. After their assessment, which included histological examination of mucosal biopsies, they were found not to be suffering from intestinal pathology apart from a Meckel’s diverticulum in two, and distal proctitis associated with constipation in four. None of these children were receiving immunosuppressive therapy or 5-amino salicylic acid derivatives.


CT3211 is a polymeric diet with casein as its protein source, which is rich in transforming growth factor β2 (> 24 p.p.m.). The protein content is 14%, the carbohydrate content 44% and the fat content 42%. It is lactose-free with glucose polymer and sucrose as the carbohydrate source. Its lipid content is made up of milk fat (55.6%), corn oil (13.9%) and medium chain triglycerides (26.1%). The calorie density of the feed is 1 kcal/mL with an osmolarity of 312 mosm/L. It has been formulated to contain adequate amounts of vitamins, minerals and trace elements. A summary of the composition of the diet is given in Table 1.

Table 1.  . Constituents of CT3211 (osmolarity: 312 mosm/L) Thumbnail image of

The study patients were treated with CT3211 as the sole source of nutrition for eight weeks. The volume was prescribed to match the patient’s daily nutritional requirements by a paediatric dietitian, who also monitored the intake of feed during the treatment period. The prescribed volume was taken within the first week in all cases, and this intake was sustained for the 8-week period in the 27 out of 29 children who completed the treatment phase of the study. Only one child required a nasogastric tube for just 2 weeks, the others all managed to take their nutritional therapy of 1950–3000 mL (median 2500) orally. Following the 8-week treatment period, foods were re-introduced at a rate of one new food every 2 days, according to a protocol described by Sanderson et al.4 Over this period CT3211 therapy was slowly withdrawn.

Some minor symptoms attributed to the treatment were reported, namely nausea in two cases and constipation in three cases. These symptoms responded readily to medical treatment in the form of metoclopramide and lactulose, respectively. Drug treatment was otherwise left unchanged during the period on enteral nutrition, apart from the two children on long-term maintenance prednisolone medication, who discontinued this therapy over the first 4 weeks.


Prior to treatment children were assessed clinically to document disease activity using the Paediatric Crohn’s Disease Activity Index (PCDAI).12 This index scores clinical and biochemical parameters, and has been specifically designed to monitor paediatric patients by including measurements of both weight and height in the assessment process. Anthropometric data were analysed with reference to age and sex as standard deviation scores.13, 14 Patients were also assessed serologically for evidence of inflammation, with measurements of serum C-reactive protein and tumour necrosis factor α. The disease site was evaluated by radiology, in the form of a barium meal and follow through, and colonoscopy, at which mucosal biopsies for histological and cytokine assessments were obtained. At 2 weeks the paediatric Crohn’s disease activity index and C-reactive protein were re-evaluated, and at 8 weeks, when the period of full enteral nutrition treatment was completed, all the initial assessments were repeated with the exception of radiology. There was then a further 10-month follow up period during which time any relapses were documented.

Colonoscopy was performed under the supervision of J.F. Endoscopic appearance of the bowel was scored 0–3 to produce a ‘colonoscopy score’ as described previously.15 The terminal ileum and colon were scored separately, the colon score being derived from the appearance of the most severely affected area. A score of 0 represented no inflammation, whilst a maximum score of 3 represented severe inflammation, with extensive deep ulceration.

Multiple mucosal biopsies were taken from the terminal ileum and large bowel (caecum, ascending colon, transverse colon, descending colon, sigmoid colon, and rectum) before and after treatment and included the macroscopically most severely affected areas in all cases. These were reviewed ‘blindly’ by one of the authors (P.K.) who was unaware of the relationship of the samples to the study protocol. They were scored 0–3 using parameters previously described to produce a ‘histology score’: 0 = no inflammation; 1 = chronic inflammation only ± slight active inflammation; 2 =  more severe active inflammation with crypt distortion or crypt abscess formation; 3 = severe active inflammation with ulceration.9 The presence or absence of granulomas did not necessarily influence the score. As with the endoscopic assessment, the terminal ileum and colon were scored separately, the colonic score being derived from the appearance of the microscopically most severely affected biopsy.

Measurement of serum tumour necrosis factor α

Following venesection, clotted blood was centrifuged at 2000 r.p.m. for 10 min. Serum supernatant was separated and stored at −20 °C until analysis. Fifty microlitre aliquots of serum were analysed for tumour necrosis factor α by the enzyme linked immunoabsorbant assay technique (ELISA) using a commercial assay (R&D Systems Europe, Abingdon, UK). According to the manufacturer’s specifications the lower limit of detection for this assay is 1 pg/mL.

Quantification of mucosal cytokine mRNA

Mucosal biopsies obtained at colonoscopy from terminal ileum and transverse colon were snap-frozen in liquid nitrogen, and stored at −70 °C. Subsequently biopsies were thawed and the RNA extracted using TRIZOL reagent (Life Technologies, Paisley, UK) as described by McDonald et al.16

Total RNA (1–3 μg) was reverse transcribed to cDNA in a competitive reaction with plasmid pHcQ1 kindly provided by Dr Martin Kagnoff (University of San Diego).17 The plasmid was added in four different plasmid dilutions: 100, 10, 1 and 0.1 fg per reaction. Polymerase chain reaction amplification was performed with primers specific for interleukin-1β, interleukin-8, interleukin-10, transforming growth factor β1, and interferon γ together with T cell receptor α as a positive control (5′ TATCCAGAACCCTGACCCTGCC 3′, 5′ GATTCGGAACCCAATCACTGACA 3′).17 PCR products were visualized on a 1% agarose gel (Sigma Chemical Co., St Louis, MO) stained with ethidium bromide (Sigma). The intensity of brightness of bands under ultraviolet light was measured using an image analyser (Seescan). With this method the ratios of the band intensities of the PCR products from the standard RNA and the target RNA, allows quantification of the number of cytokine transcripts in a tissue sample down to as few as 1000 and up to 108 transcripts per microgram of total RNA.

Statistical analysis

The responses to treatment in paediatric Crohn’s disease activity index, C-reactive protein, endoscopic and histological scores were analysed by Wilcoxon signed rank test. Weight standard deviation score, body mass index standard deviation score, serum tumour necrosis factor α and mucosal cytokine RNA responses were analysed by one sample t-test, the RNA data having previously undergone log transformation. Skewness and kurtosis for these parameters was not significantly different from normal (P < 0.05).

Ethical consent

Ethical approval for this study was obtained from the ethics committees of St Bartholomew’s and Royal Free Hospitals. Written informed consent was obtained from the parent/guardian and the subject in all cases.


Clinical response

At initial assessment 12 children had moderate or severe disease (PCDAI > 30) and 17 had mild disease (PCDAI 11–30). After 8 weeks of treatment 23 (79%) were in clinical remission (PCDAI ≤ 10). There were two treatment failures, one child with colonic disease who subsequently responded to corticosteroid therapy, and one with an appendiceal abscess which required surgical drainage. The median paediatric Crohn’s disease activity index, initially 30 (range 12.5–72.5) declined with treatment by a median of 15 at 2 weeks and 25 at 8 weeks (P < 0.00001, P < 0.00001). Similarly the median C-reactive protein, initially 19 mg/L (range 1–93; normal range < 5) declined with treatment by a median of 14 mg/L at 2 weeks and 13.5 mg/L at 8 weeks (P < 0.0001, P < 0.001).

Children gained weight in response to treatment (mean weight gain at 8 weeks 3.2 kg, P < 0.001), their weight standard deviation score increasing by a mean of 0.42 from –1.49 (s.d. 1.36, P < 0.001), and their body mass index standard deviation score increasing by a mean of 0.73 from –1.48 (s.d. 1.60, P < 0.001).

Endoscopic response

All children underwent an initial colonoscopy. This was repeated in 26 out of 29 at 8 weeks. Two cases had been withdrawn from the protocol as treatment failures, and one case with isolated small bowel disease and a normal initial colonoscopy was not re-endoscoped. In total, paired observations (before and after treatment) were obtained in 26 cases for the colon, and in 22 cases for the terminal ileum. The overall rate of visualization of the terminal ileum was 49 out of 55 colonoscopies (89%). The macroscopic appearances improved in the terminal ileum in 15 out of 22 and were unchanged in seven out of 22, whilst in the colon, appearances improved in 13 out of 26 and were unchanged in 12 out of 26. There was a median improvement in colonoscopy score of one point at both sites (P < 0.0001 and P < 0.001, respectively). Macroscopic healing occurred in eight cases in the terminal ileum, and also in eight cases in the colon.

Histological response

Paired histological observations (before and after treatment) were obtained in 25 cases for the colon, and 20 for the terminal ileum. The histological appearances improved in the terminal ileum in 13 out of 20 and were unchanged in five out of 20 (worse in two), whilst in the colon appearances improved in 12 out of 25 and were unchanged in 13 out of 25. There was a median improvement in histology score of one point at both sites (P < 0.05 and P < 0.001, respectively). Histological healing occurred in eight cases in the terminal ileum, and in two cases in the colon. Details of the endoscopic and histological changes in individual cases are available from the authors on request.

Serum tumour necrosis factor α

Serum samples for tumour necrosis factor α estimation were obtained in 26 out of 29 before treatment and 20 out of 27 after treatment. The mean level prior to treatment was 15.3 pg/mL (s.d. 8.48). This fell by a mean of 4.7 pg/mL with treatment (P < 0.05).

Mucosal cytokines

Cytokine mRNA was successfully estimated in terminal ileal and transverse colonic mucosal biopsies before and after treatment in between 14 and 18 cases depending on the specific cytokine. A further seven ileal and eight colonic biopsies were analysed as controls. T cell receptor α mRNA was measured in all samples and used as a positive control, thus cytokine mRNA was only quantified when greater than 1000 transcripts/μg RNA of T cell receptor α mRNA were recorded. In some instances the full range of cytokines could not be quantified due to insufficient RNA being recovered from the biopsy.

Interleukin-1β mRNA before treatment was elevated in both the ileum and colon when compared with controls (P = 0.04, P < 0.0001). Levels fell at both sites in response to treatment. The ratio of pre- to post-treatment interleukin-1β mRNA was 0.008 for the ileum (P < 0.001), and 0.06 for the colon (P = 0.006).

Interferon γ mRNA before treatment was elevated in the ileum and colon when compared with controls (P = 0.004, P = 0.001). Levels fell in the ileum in response to treatment, with a ratio of pre- to post-treatment interferon γ mRNA of 0.15 (P < 0.001). On the other hand, ileal transforming growth factor β1 mRNA rose with treatment (ratio 10, P = 0.04). In the colon there was no significant change in either interferon γ or transforming growth factor β1 mRNA in response to treatment.

Interleukin-8 mRNA before treatment was elevated in the colon (P = 0.002), but not in the ileum, when compared with controls. Colonic levels fell in response to treatment, with a ratio of pre- to post-treatment interleukin-8 mRNA of 0.06 (P = 0.02).

There was no significant change in interleukin-10 mRNA in response to treatment in either the ileum or colon. The mucosal cytokine mRNA levels from patients before and after treatment, together with levels measured in control specimens are summarized in Table 2.

Table 2.  . Mucosal cytokine mRNA in ileal and colonic biopsies (transcripts/μg total RNA). The response to treatment is expressed as the ratio of pre- to post-treatment mRNA. Statistical analysis was performed on log transformed data Thumbnail image of

Clinical relapse

Over the 10-month follow-up period nine of the 23 cases (39%) who had achieved a clinical remission (PCDAI ≤ 10), relapsed (defined as a rise in PCDAI above 10). One child (4%) had already relapsed within 2 months of the treatment phase of the study.


This prospective cohort study into the therapeutic action of nutritional therapy with a polymeric diet CT3211, has shown that this oral treatment is effective in inducing a clinical remission in children with active Crohn’s disease. The clinical response was associated with a decrease in serum C-reactive protein and tumour necrosis factor α levels, whilst in the intestinal mucosa there was evidence of mucosal healing together with a down regulation of the pro-inflammatory cytokines interleukin-1β, interleukin-8 and interferon γ.

The remission rate of 75% after 8 weeks of CT3211 is an encouraging preliminary indicator as to the efficacy of the treatment. It is in keeping with the results of the large Canadian paediatric comparative trial of nutritional and corticosteroid treatment which despite excluding children with colonic disease, reported similar remission rates of 75% for enteral nutrition and 89% for corticosteroids.18 It also compares favourably with the remission rates of 58% and 79% for nutrition and corticosteroids calculated by meta-analysis of trials in adults.3 To fully evaluate the relative efficacy of CT3211 compared to other therapies will however, require formal randomized trials. These would also help answer the question of whether the transforming growth factor β2 in CT3211 is contributing to its therapeutic action. Such trials should ideally be performed separately in both children and adults since there appears to be a discrepancy between the results from studies performed on paediatric or adult subjects, the former tending to include more newly diagnosed cases. Although a separate meta-analysis has yet to be performed specifically on paediatric studies, the fact that all paediatric studies have found no difference in remission rates for steroid or nutrition treated groups contrasts with the more disappointing results from adult studies.3[4]–5, 18, 19

Endoscopic and histological assessment of mucosal biopsies demonstrated significant improvement, and in some cases healing, of the mucosa in both the terminal ileum and colon. This colonic response, particularly in the light of the mucosal cytokine changes, demonstrates the therapeutic efficacy of enteral nutrition in colonic disease, a conclusion previously reached by Griffiths et al. following meta-analysis of the clinical trials.1

The improved palatability of CT3211 meant that the children could take the volumes required to meet their full daily nutritional requirements orally. This is a significant advance, particularly over elemental and semi-elemental feeds, since a naso–gastric tube is now no longer required for the 2 to 3 months of therapy. As well as this relatively short-term usage of enteral nutrition to achieve remission, there is now growing evidence for the beneficial effects of long-term nutritional therapy both in promoting linear growth and possibly also in maintaining remission.20[21]–22 In fact with CT3211, the relapse rate within 10 months of completion of CT3211 treatment was relatively low at 39%. This compares favourably with relapse rates by 1 year of 65% and 67% for enteral nutrition and corticosteroids, respectively, derived by pooling the results of the two adult randomized trials with follow-up data.7, 23 Whether the CT3211 relapse rate could be further improved by long-term administration is, however, worth exploring. Traditionally such continuing therapy has required naso-gastric administration, which represents a significant imposition on the child. Long-term oral supplementation would obviously be preferable, and in view of this a prospective trial into the effect on relapse of chronic administration of CT3211 is currently being undertaken.

The clinical response to CT3211 was associated with a fall in inflammatory mediators documented both in the circulation and in the mucosa. Elevated levels of C-reactive protein and ESR in Crohn’s disease are believed to be caused by the release of intestinal inflammatory mediators into the circulation and their subsequent action on the liver.24 This was also demonstrated in this study, with a fall in both C-reactive protein and ESR in response to treatment being associated with a fall in circulating levels of one such inflammatory mediator, tumour necrosis factor α. The changes in tumour necrosis factor α levels are however, much less striking than the changes in C-reactive protein or ESR, once again emphasizing the need wherever possible to monitor cytokines in the mucosa in Crohn’s disease rather than peripheral blood.

The fall in mucosal interleukin-1β mRNA in response to treatment observed in both the colon and ileum is consistent with the concept that enteral nutrition influences the underlying mucosal inflammatory process of Crohn’s disease. Although mucosal mRNA estimation is by definition not a direct measure of cytokine release, this finding is consistent with the observations of Ferguson et al. who reported a fall in interleukin-1β in the intestinal lumen following nutritional therapy.11 It is also in line with our earlier observations which demonstrated that mucosal release of another macrophage-derived cytokine, tumour necrosis factor α, goes down in response to treatment.9

Whereas the changes in interleukin-1β were similar in the ileum and colon the results for interferon γ, transforming growth factor β1 and interleukin-8 differed in the two sites. The fall in interferon γ mRNA in the ileum was consistent with our previous report of falls in the number of mucosal interferon γ and interleukin-2 secreting cells in response to enteral nutrition, and is compatible with the model of Crohn’s disease as a chronic inflammatory condition driven by lymphocytes of a predominantly T helper-1 (Th-1) phenotype.25 The rise with treatment in transforming growth factor β1 mRNA in the ileum has not previously been reported in Crohn’s disease, although the suppressive effect of transforming growth factor β on the Th-1 inflammatory response is well recognized in animal models.26 Whether this increase in ileal transforming growth factor β occurs with other treatments, and other diets, or is an effect of this specific diet which is itself rich in transforming growth factor β, is a matter for speculation.

In the case of interleukin-8, the fall in mRNA levels were only significant in the colon. This site however, had higher pre-treatment levels, which may go some way to explaining the predominance of infiltrating neutrophils, for which interleukin-8 is a chemo-attractant, affecting the colon in Crohn’s disease.27 An underlying mechanism to explain the observed differences between colon and ileum is however, not obvious. The presence in the terminal ileum of organized lymphoid tissue in the form of Peyer’s patches does, however, mean that biopsies from the ileum are more likely to include these specialized structures than those taken from the colon.

The mechanism by which enteral nutrition exerts the therapeutic and immunological effects in Crohn’s disease remains unknown although several hypotheses have been proposed. The traditional hypothesis that enteral nutrition works by the exclusion of dietary antigens seems unlikely since specific disease-associated foods have only rarely been identified, and in our study, despite the systematic re-introduction of foods following therapy, in no case was a specific food identified which caused symptoms.28 Another possible explanation is that enteral nutrition has an effect on the bowel flora, although specific alterations in bowel flora have not yet been identified. An alternative mechanism for enteral nutrition to exert its therapeutic action, could be by the repletion of nutritional deficiencies, thus promoting mucosal healing. Unraveling the relative importance of these, and other putative mechanisms will however, require further investigation.


J.M.E.F. is supported by the Crohn’s in Childhood Research Association. T.T.M. is supported by the Welcome Trust. We are grateful for the assistance of Nestle Clinical Nutrition.

Statistical assistance was provided by R. Mansourian (Nestec, Switzerland).