Preoperative Enteral Nutrition and Surgical Outcomes in Adults with Crohn’s Disease: A Systematic Review

Up to 50% of Crohn’s disease (CD) patients have been classically described to require surgery within 10 years after diagnosis [1]. More recently, according to a population-based cohort study from Denmark, the 9-year cumulative risk of surgery has been reduced from 50% (1979–1986) to 23.3% (2003–2011), along with an increased use of thiopurines and TNFα inhibitors [2]. The same trend was found in a similar study from Sweden – the risk of surgery at 5 years fell from 65.8% (1963–1975) to 34.6% (1991–2005) [3].

Rates of postoperative complications and stoma creation after abdominal surgery are influenced by a number of factors, both clinical and therapeutic, and vary widely across published series [4, 5]. Intra-abdominal septic complications (IASCs) may reach 50%, the risk factors for which the literature mostly reports therapy with corticosteroids (CSTs), malnutrition, intra-abdominal abscesses, low serum albumin levels, and emergent surgery [6-11].

Among medications used in the treatment of CD, CSTs are the one most consensual risk factor for anastomotic complications and, in some series, for global morbidity [4, 6, 9, 10]. As to purine analogues, there is no evidence they might increase the risk of postoperative complications [9, 11, 12]. Differently, TNFα inhibitors increase infectious and global surgical morbidity [11, 13-15].

In what concerns preoperative malnutrition, undernourished CD patients should receive perioperative nutritional support, as in other pathologies of the digestive tract, even if this requires postponing surgery. The enteric route is preferred, whereas parenteral nutrition (PN) is only indicated when the digestive tract is insufficient or unavailable [16-19]. Whenever the digestive tract does not provide enough calories – in patients who tolerate enteral nutrition (EN) but in insufficient quantities – the use of total or supplementary PN becomes inevitable [16]; operating undernourished patients, with consequent increases in postoperative complications and stomas, is to be avoided. Interestingly, there are some early pioneer studies that have found potential advantages in the application of EN to malnourished patients, in whom they seem to produce better surgical results [20, 21].

Concerning therapy with CSTs, exclusive EN (EEN) has largely substituted prednisolone as first-line remission-inducing therapy for moderate to severe disease flares in children [22, 23]. In striking opposition, the former remain the mainstay of therapy for these situations in adults, in whom they seem more effective over EEN [24-26]. In this age group, the therapeutic value of EN in CD is controversial outside the bounds of malnutrition. It may therefore be advised as remission-inducing therapy in case of refusal or intolerance to CSTs, in mild extensive small bowel disease (here as primary or as adjuvant therapy), and as adjuvant therapy in the undernourished which maintain inflammatory stenosis, despite maximized adequate pharmacological therapy [24, 27-29]. The grade of evidence concerning the superiority of CSTs over EEN in adult patients is, however, of very low quality [24].

The mechanisms by which EEN induces remission of CD are not well established [29]. A number of studies suggest that EN promotes the balancing of mucosal cytokines – by decreasing those which are pro-inflammatory [30, 31], restores mucosal integrity and its barrier function [24, 30, 32], decreases adjacent mesenteric fat inflammation [33, 34], and alters the composition of the intestinal microbiome [35, 36]. It is possible that bowel rest and interruption of exposure to alimentary and some bacterial allergens may be contributing mechanisms [29].

To determine whether preoperative EN impacts postoperative complications of adult CD, either by means of nutritional or therapeutic effects.

A systematic review of research articles including the use of preoperative EN in adults with CD submitted to abdominal surgery, from January 1990 to November 2017, was conducted.

An initial search was carried out through PubMed-MEDLINE, ScienceDirect, and Google Scholar. Medical subject headings (MeSH) applied were the following:

• for PubMed-MEDLINE, “Crohn’s disease” AND “surgery” AND “postoperative” AND “complications” AND “adults” AND “preoperative” AND “enteral nutrition” AND “not review”; filters applied were “full text,” “English,” and “humans”;

• for ScienceDirect, “Crohn’s disease” and “enteral nutrition” (Title, abstract, keywords);

• for Google Scholar, “Crohn’s disease and preoperative enteral nutrition and adults.”

Two further searches were conducted:

• upon references of all articles selected as eligible, as result of the initial search;

• of “similar articles” of the two most recent studies of those already selected as eligible.

Methodology respected Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) recommendations [37].

Screening and testing for eligibility were performed upon English written abstracts. A grid was constructed to assess articles for eligibility and select them for inclusion, according to specific inclusion and exclusion criteria.

Inclusion criteria:

• Original clinical research articles wherein preoperative EN was administered to adult CD patients operated on for intestinal complications, as long as EN impact on postoperative complications is discriminated.

• English full-text articles published in peer-reviewed journals, including those published online ahead of print, within the time frame considered.

• Retrospective and prospective studies, whether cohort, case series, case control, or randomized controlled trials, regardless of aims and number of patients.

• Studies concerning adults – patients considered adults by the authors of the study were similarly assumed.

• All formulas, means of administration, and regimens of EN – exclusive or supplemental, as complement to PN or a conventional diet.

• Data referring to each type of nutritional regimen is accordingly discriminated and treated separately from that respecting patients and impact on postoperative morbidity.

• Paucity or even absence of data regarding repercussions of EN on patients’ preoperative state will not be criterion for exclusion.

Exclusion criteria:

• Review articles (both systematic and non-systematic), case reports, comments, letters, and book chapters.

• Abstracts of conferences, oral presentations, and posters presented in meetings.

• Papers concerning children or adolescents or whose authors belonged to pediatric institutions.

• Patient overlap among articles – the article containing the most patients was selected.

• Articles or results referring to the analysis of postoperative recurrence of CD.

• Absence of separate data concerning CD patients as opposed to those with ulcerative, undetermined, and unclassified colitis.

The quality of the studies was rated by awarding stars in each domain following the guidelines of the Newcastle-Ottawa Scale (NOS) [38].

Two authors (A.R. and I.B.) undertook the bibliographic search, independently screened the articles for eligibility, and assessed them thoroughly for inclusion. Discussion took place until consensus was reached whenever accordance was not met initially. Three authors (A.R., I.B., and P.L.) reviewed the included articles.

One hundred and ninety-six references were identified in PubMed-MEDLINE, 61 in ScienceDirect, 4,880 in Google Scholar, and 157 through secondary search (Fig. 1). After the irrelevant titles and duplicates were removed, 257 records were screened, 22 of which were eligible. Of these, one had no English abstract [39] and three had no English full text [40-42]. The main features of the 14 full-text excluded articles are summarized in Table 1 [43-56] – eight of them had no separate results on different nutritional regimens or on treated patients [44, 47-50, 52-54] and four had no data on EN regimen [43, 45, 55, 56]; two other articles [46, 51] met the inclusion criteria but overlapped patients with a third study which contained a larger number of patients and was therefore detrimentally included. All studies in which supplemental EN was used with conventional diet or with PN did not meet the inclusion criteria.

The main characteristics of the four articles selected for review [57-60] are summarized in Tables 2-5. Table 2 refers to the general characteristics, aims, methods, indications for and information concerning preoperative EEN regimen, and main conclusions of the articles under review. Table 3 discriminates the characteristics of operated patient groups, the variables studied in each, studies’ limitations, and the NOS quality assessment – criteria for which, as applied to cohort studies, being presented as online supplementary material (for all online suppl. material, see www.karger.com/doi/10.1159/000494674) [38]. In Tables 4 and 5, the impact of preoperative EEN on clinical and surgical outcomes is detailed. No mortality was reported. Relevant data from these trials not expressed in the tables will be detailed below.

Li et al. [57] analyzed risk factors for postoperative complications after classifying 708 surgeries into four groups according to patients’ preoperative management (Table 3). EEN was associated with a longer immunosuppressant-free interval and fewer emergent surgeries, primary stomata, anastomotic leaks, intra-abdominal abscesses, wound infections, ileus, and reoperations on univariate analysis (Tables 4, 5). The authors identified EEN as an independent protective factor against postoperative complications (both infectious and non-infectious). Preoperative immunosuppressive therapy, perforation or abscess, emergent surgery, previous surgery for CD, and older age were independent risk factors for infectious complications (Table 5). The first two risk factors and longer operative time were the independent risk factors identified for non-infectious complications. The authors argued that the prospective nature of the database and the cohort nature of this study may have mitigated some of the limitations and biases inherent to retrospective studies.

Heerasing et al. [58] retrospectively revised 51 patients treated with preoperative EEN after nutritional assessment. Thirteen patients (25%) ended up not requiring surgery. For each of the 38 operated EEN patients, two matched non-EEN controls were sought (Table 2). The latter were found to have higher basal body mass index (BMI), lower basal C-reactive protein (CRP) values and lower prevalence of treatment with antibiotics (Table 3). EEN was found to be an independent factor against anastomotic leaks and/or abscesses and led to significantly lower global postoperative morbidity and shorter operative times (Tables 4, 5). Patients with IASCs had significantly longer lengths of hospital stay and a higher readmission rate. The authors point out some limitations to this study such as missing data on medical records and incomplete matching of disease severity. Also, they did not clarify the reasons why control-matched patients did not receive nutritional support – it may even have been due to the baseline differences in BMI and/or serum CRP levels, but this was not mentioned.

Beaupel et al. [59] compared two groups of patients, 21 “low-risk” versus 35 “high-risk” for postoperative complications, distinct regarding CD behavior, severity and complications, nutritional state, and CST treatment (Tables 2, 3). As the rate of postoperative complications, for which no protective or risk factors were identified, was not statistically different between the two groups (Table 5), the authors admitted this indicated that TGF-β2-enriched EEN was beneficial for high-risk CD patients (Table 2).

In the study published by Zhu et al. [60], 54 of 83 CD patients were treated with EEN for abdominal abscesses not amenable to percutaneous drainage, either until abscess resolution or until surgery (from 7 to 1,000 days). The remaining 29 patients did not tolerate or comply with EEN and had a significantly higher cumulative 1-year surgical rate than EEN patients. EEN led to a higher rate of abscess resolution and was an independent protective factor against surgery. Previous abdominal surgery and abscess diameter > 3 cm were independent risk factors for surgery. Patients’ nutritional status was not mentioned. Forty-six EEN-treated and 28 EEN-untreated patients required surgery, the former having significantly fewer postoperative IASCs (Tables 2, 5). The authors noted limitations to the study beyond its retrospective nature and reduced number of patients: some of the latter suspended EEN and/or restarted immunosuppressants after abscess resolution, for unclarified periods of time.

In all studies, preoperative EEN was well tolerated by adults with complicated CD and a patent intestinal route – 338 out of 1,024 operated patients received preoperative EEN. Compared to their respective controls, patients under preoperative EEN achieved significantly lower levels of preoperative inflammatory markers in three of the reviewed studies [58-60], in two of which with concomitant increases in serum albumin values [59, 60]. The two studies with the greatest number of patients found preoperative EEN to be an independent factor against postoperative infectious and non-infectious complications [57] and against anastomotic leaks or abscesses [58].

However, quality grading for risk of bias according to NOS criteria [38] (Table 3) qualifies only one of the studies as “fair quality” [58] – the only one to match compared patient groups. The remaining studies were considered “poor quality,” essentially for not scoring in the comparability domain (see online suppl. material).

All reviewed articles were institutional retrospective cohort studies with heterogeneous methods and inclusion criteria. Patients received preoperative EEN with different indications and regimens and for different periods of time. Those belonging to control groups were also heterogeneous between studies and even within the same study, namely regarding baseline nutritional state and CD activity, behavior, and medication. The type of postoperative complications analyzed also varied between studies. Meta-analysis was not done due to the small number of studies with varying designs, posing issues over the estimation of the between-study variance [61].

A non-intentional absence of studies referring to supplemental EN among those reviewed resulted from those which were eligible and referring to supplemental EN meeting exclusion criteria [44, 45, 47-50, 52-56].

From the reviewed articles, two were from China [57, 60], one of them being the study with the largest number of patients [57]. This latter study contains more patients than the sum of all three other reviewed papers, possibly raising transferability issues to other institutions and countries, namely, the Western ones. Although EN is not indicated as treatment for CD in adults by Chinese guidelines [62], eight of the excluded studies originated from China. The article by Li et al. [57] was produced by the same institution as that of seven [46-48, 50-52, 54] of the previous 8 studies, the latter being well known for its publications regarding this topic [31, 34, 63]. One of the excluded articles is Japanese [45]. The Japanese consensus indicates EN for remission induction and maintenance in adult CD [64] but, for the most part, studies from Japan focus on the possible impact of this nutrition on the reduction of postoperative relapse and remission maintenance – both as monotherapy and as an adjunct to pharmacotherapy [65-67].

Both reviewed Western studies made use of orally administered polymeric EEN [58, 59]. In both cases, an enriched formula with transforming growth factor beta-2 (TGF-β2) was used, the advantages of which have been studied for this pathology but not yet proved [17, 24, 68]. In the West, the administration of nutritional support by means of a nasogastric tube is poorly accepted by adults [24]. However, tolerance and acceptance towards the oral route has grown with improvements in the quality and variety of flavors of the polymeric formulations [24, 27].

In all studies, preoperative EEN was well tolerated by adults with complicated CD and a patent intestinal route, even in the presence of active inflammation and abdominal abscesses, leading to the possibility of postponing and even preventing surgery [58, 60]. A preoperative lag allowed for amelioration of patient conditions – reducing the inflammatory environment [58-60], increasing serum albumin [59, 60], weaning or suspending CST therapy [57, 59], and treating abscesses [59, 60] – as opposed to patients operated on immediately [58, 59]. These properties have also been attributed to nutritional support, both enteral and parenteral, by other authors [44, 47, 48, 50, 52-54, 69-72]. Also, in one of the reviewed studies [59], going straight to surgery (non-EEN patients) was an independent risk factor for anastomotic leaks and/or intra-abdominal abscesses.

The authors of one of the reviewed studies [60] refer that EEN led to a significant improvement in nutritional status based solely on a significant rise in serum albumin. Three articles specify BMI values [57-59], one of which also referring to weight loss above 10% [59] and another referring to basal and preoperative weights [58]. Nevertheless, in truth, none of the papers explains the methodology used for nutritional assessment. Standing alone, these anthropometric and laboratory parameters are not indicators of nutritional status in the presence of an active inflammatory syndrome, seeing they are highly influenced by variation in and retention of fluids [16, 49]. Preoperative serum albumin levels may reflect variations in a patient’s inflammatory status and/or nutritional state and, therefore, for one or both reasons, values below 3 g/dL are a risk factor for postoperative complications in abdominal surgery [4, 5, 7, 11, 16]. The inflammatory environment elicited by the disease and surgery actively contributes to a descent in serum albumin and perioperative undernutrition, the control of the former being determinant in the improvement of the latter two [63, 73, 74]. In so being, nutritional support may be an essential weapon against the systemic inflammatory response syndrome and in promoting the restoration of the intestinal mucosal barrier, of immunological balance and tissue scarring, thereby leading to fewer postoperative complications and a faster recovery of the patient [16, 73, 74].

Two studies found preoperative EEN to be an independent factor against infectious and non-infectious complications [57] and against anastomotic leaks or abscesses [58]. However, only the study involving 780 surgeries [57] found EEN to lead to a significantly lower occurrence of postoperative emergent surgery, primary stomata, anastomotic leaks, and reoperations, as opposed to groups under immunosuppression alone. Although anastomotic dehiscence rates among the EEN and immunosuppressant-free groups were low (4.1 and 3%, respectively), stoma rates were notably higher (17.8 and 22%, respectively) [57]. In the other three studies [58-60], stoma rates were similar between EEN and control groups but, in one of them, they were in fact very high (43.5 and 46.4%, respectively) [60].

It is noteworthy that no study reported on ostomies resulting from anastomotic dehiscence, on temporary ostomies that became permanent, or on ostomy closure morbidity. Yet, one of the main concerns of CD patients is that surgery may result in the need for an ostomy bag [75], which can lead to fear or even refusal of operations. Therefore, the possibility that preoperative EEN may lead to a lower stoma rate – whether primary or resulting from postoperative complications – is one of the most important aspects to consider in the surgery of these patients.

Despite the fact that all reviewed studies are retrospective and, consequently, of limited relevance, there is no doubt EEN was well tolerated and the preferred nutritional support, in the absence of intestinal obstruction, in the presence of complicated disease. In addition to its nutritional value, however, it is necessary to ascertain whether in fact this nutritional support conduces to mitigation of postoperative complications, reductions in stoma rates, decreased disease activity, removal of harmful drugs, and avoidance of emergent surgery.

The studies do collectively, nonetheless, call the attention of the scientific community to the potential benefits of preoperative EEN on postoperative outcomes in adult CD. They also carry enough impact to call for multi-institutional studies and randomized controlled trials. The design of this kind of trials, however, is a painstaking endeavor, since a great number of variables – relating to patients, disease, medication, surgical indications, and technique – must be considered. Also, ethical issues concerning randomization of malnourished CD patients towards receiving or not receiving nutritional support before surgery may limit the feasibility of such studies.

The determination of the properties of preoperative EEN is of importance since, as of now, this strategy has been very little or not at all used in most referral centers worldwide, outside the bounds of nutritional support. Only then will the contribution of preoperative EN on postoperative morbidity and stoma rates of adult CD be defined.

In summary, preoperative EEN in adult CD surgery can be a valid option in specific situations – to improve nutritional status in the absence of intestinal obstruction, as an adjuvant to the treatment of existing abscesses and/or fistulas, and to attempt CST suspension.

However, none of the reviewed studies clarifies the possible existence of a cause-effect relationship between preoperative EEN and surgical morbidity in adult CD, their conclusions being far from definite. The few existing studies either carry obvious biases, varying risks of systematic errors, or questionable transferability of conclusions to other institutions. Prospective studies are required to establish whether or not preoperative EEN can in fact improve surgical outcomes in these patients.

The authors thank Isabel Fonseca for her counsel on statistical methodology.

The authors declare no conflicts of interest.

A. Rocha conceived and designed the study. A. Rocha and I. Bessa undertook the bibliographic research and, independently, screened the articles. A. Rocha, I. Bessa, and P. Lago selected for eligibility, determined inclusion, and reviewed the articles included. A. Rocha and I. Bessa drafted the manuscript. P. Lago, M.D. Santos, J. Leite, and F. Castro-Poças revised the draft and approved the final manuscript.