Enteral nutrition is commonly used to treat children with newly diagnosed with Crohn's disease in the United Kingdom, but this practice is relatively rare across the Atlantic Ocean (1). The administration of formulated food to treat acute exacerbations of Crohn's disease has a long history of sporadic interest but is predominantly neglected in North America in comparison with drug treatment. In an era when patients and parents increasingly seek alternative therapies, one must ask why a nutritional intervention that has repeatedly achieved at least moderate efficacy is not more often pursued.
The usefulness of enteral nutrition as primary therapy of active intestinal inflammation was discovered fortuitously, when adult patients given an elemental diet before surgery experienced improvement, not only in nutritional status as intended, but also, unexpectedly, in the inflammatory activity of Crohn's disease (2). Efficacy was suggested further by the results of early small, randomized, controlled trials versus corticosteroid therapy conducted in both adults (3) and children (4). Although the North American Collaborative Crohn's Disease Study (NCCDS) Group evaluated only drug therapy, their counterpart, the European Cooperative Crohn's Disease Study (ECCDS) Group, appropriately undertook to examine exclusive enteral nutrition in the setting of a large, multicenter, randomized controlled trial (5,6). Indeed, when the ECCDS III trial encountered overwhelming (32%) intolerance to orally administered semielemental formulated food (5), the European adult gastroenterologists persisted with the ECCDS IV trial involving nasogastric infusion of the liquid diet (6). Although tolerance was improved in this study of 107 adults with active, moderately severe Crohn's disease, the difference between clinical remission rates achieved with enteral nutrition (53%) compared with corticosteroids and sulfasalazine (85%) was clinically and statistically significant (6). This well-designed and well-conducted study provides grade I evidence of a treatment benefit to conventional drug therapy versus semielemental liquid diet therapy in adults with active Crohn's disease of the small and/or large intestine (7).
Proponents of the necessity for specifically elemental formulae to treat intestinal inflammation have argued that the oligopeptide-containing formula used in the ECCDS is inferior to drug therapy because of its composition. Alternatively, the failure of small trials to detect a difference in response rates with elemental formulas versus corticosteroids may be due merely to inadequate sample size (i.e., to type II errors) and not to superior formula composition. The seemingly divergent outcomes between small controlled trials in adult and pediatric patients, each involving roughly 20 patients, and the much larger multicenter ECCDS IV trial justified the use of meta-analytic techniques to examine apparently conflicting data (8-10). The outcomes in randomized trials of enteral nutrition versus corticosteroids, comprising 316 adult and 97 pediatric patients, were demonstrated in these meta-analytic studies to be, in fact, homogeneous; the 95% confidence intervals for the differences in observed remission rates in individual trials overlapped (8-10). In all but one trial of 21 adult patients (3) and another unpublished study of 19 children (11), more patients achieved clinical remission with corticosteroids than with enteral nutrition. Combination of randomized controlled data from adult and pediatric studies suggests that the likelihood of inducing clinical remission of active Crohn's disease is greater with prednisone than with exclusive enteral nutrition (8-10).
Heuschkel et al. have combined in meta-analysis the data accrued in controlled trials conducted exclusively in children and adolescents. They conclude that nutritional treatment and conventional corticosteroids are equally efficacious in pediatric patients, even if not in adults. Although it is useful to summarize the pediatric trials conducted, the methods used to arrive at this conclusion are flawed. Only five randomized trials comprising 147 children have been conducted. This number falls short of the 182 patients calculated by Heuschkel et al. as necessary to detect a 20% difference in treatment effects. Otherwise stated, the results of the meta-analysis of truly randomized controlled pediatric trials are inadequate to exclude a 20% difference.
Inclusion of two nonrandomized trials, although a means of reaching the desired sample size, is inappropriate. Data from well-conducted randomized controlled trials are considered the highest quality of evidence, because in comparison with other study designs, they attempt to minimize bias, which threatens the validity of results (7,12). This is well illustrated by the fact that results in open trials are consistently more favorable than outcomes in subsequent randomized controlled studies. Those with enthusiasm have no controls, and those with controls have no enthusiasm. Conclusions of any meta-analysis are only as good as the data included (13). The results of a single, large, high-quality randomized controlled trial are more convincing than the results of a meta-analysis of multiple small trials, if flaws in design compromise individual trial validity (13). The 100% remission rates achieved in small pediatric nutritional trials are, at best, surprising! Perhaps selection bias or bias in outcome assessment rather than purely therapeutic efficacy of enteral nutrition explain these exceptional clinical remission rates. The Lloyd-Still index used to assess outcome in three of seven pediatric trials has never been validated, nor has a cut score for remission been defined. Even randomized trials may be subject to selection bias, if the method of randomization does not prevent foreknowledge of treatment assignment. Allocation concealment (by opaque envelopes or centralized randomization in multicenter studies) reduces selection bias in results and is particularly important in trial settings, such as those comparing enteral nutrition with corticosteroids, in which physicians and patients cannot be blinded to the type of therapy received (14).
The one multicenter pediatric study involved 78 children and adolescents but has unfortunately hitherto been published only in abstract form (15). The observed difference between the response rates with corticosteroids and with enteral nutrition was greater in this Canadian study than the differences observed in the much smaller, single-center trials. Moreover, in the subset of 21 pediatric patients with disease in relapse, the response rates to enteral nutrition and to corticosteroids were 50% and 85%, respectively, almost identical with those reported in adults in the ECCDS IV trial (6,15,16). This suggests that variations in the nature and severity of Crohn's disease, rather than age per se, affect response rates to enteral nutrition. Such variations are important to note when making treatment selections. Data from the multicenter study, which stratified children before randomization into newly diagnosed versus in relapse, support the comments of Heuschkel et al. that recent-onset, previously untreated inflammation, may be particularly responsive to nutritional therapy. Children with isolated Crohn's colitis were excluded from the multicenter pediatric study and from at least two of the small pediatric trials based on anecdotal experience that, in comparison with patients with small intestinal disease, they respond less well to liquid diet therapy (17). It must be remembered, however, that Crohn's colitis was relatively refractory to corticosteroids in the NCCDS and ECCDS trials of drug therapy, as well (18).
As argued in earlier meta-analyses, enteral nutrition remains an important therapeutic modality, even if corticosteroids induce clinical remission more often (8-10). There may be subgroups of patients, such as children or adults with newly diagnosed terminal ileal or diffuse small intestinal inflammation, in which liquid diet therapy is as likely to achieve clinical remission. Considering the best published evidence from the high-quality adult ECCDS IV trial, the overall 53% remission rate with exclusive enteral nutrition is still in the 50% to 60% range reported in similar multicenter studies for controlled ileal release budesonide in active Crohn's disease of the ileum and/or right colon (19,20). Controlled ileal-release budesonide is an example of a therapy of modest efficacy, which has the appeal of lower risk of adverse systemic effects. Surely enteral nutrition has an even greater advantage, particularly in children, when risks and benefits are considered.
Direct and indirect evidence of a diminution in mucosal inflammation with enteral nutrition has been provided, although the mechanism of action remains conjectural (21-23). The failure of corticosteroids to induce mucosal healing, the unwanted cosmetic side effects with short-term use, and the more serious adverse effects of longer term use are all too well known. Why then has enteral nutrition been largely neglected as a first-line treatment option for active Crohn's disease in North America? The answer presumably relates to the perceived invasiveness and inconveniences of feeding regimens, which usually entail administration through a nasogastric tube and exclusion of regular food. The initial teaching and encouragement required from the medical and nursing team make it a labor-intensive option. However, despite initial reluctance, some young patients, particularly those whose growth is stunted and/or who are underweight, ultimately prefer to treat exacerbations with enteral nutrition rather than with corticosteroids (17). Moreover, although existing randomized controlled trial data are inconclusive about whether formula composition influences efficacy (8-10), improvements in the palatability of oligopeptide-containing formulae and the use of polymeric diets appear to make oral consumption a more realistic option than previously.
The statement of Heuschkel et al., that enteral nutrition should be recommended as first-line therapy in all children with active Crohn's disease would be a hard sell in North America. Nevertheless, we would do well to observe the experience of our colleagues across the Atlantic, where liquid diets seem to be feasible, effective, and accepted by families. We should encourage consideration of enteral nutrition as an alternative to corticosteroids and make recommendations concerning optimal therapy depending on individual disease and patient characteristics.