Systematic review: nutritional support in acute pancreatitis

Authors


Dr M. S. Petrov, Department of Surgery, Nizhny Novgorod State Medical Academy, PO Box 568, Nizhny Novgorod, 603000 Russia.
E-mail: max.petrov@gmail.com

Summary

Background  There has been controversy concerning the merits of enteral and parenteral nutrition compared with no supplementary nutrition in the management of patients with acute pancreatitis.

Aim  To perform a systematic review of the data from randomized controlled trials (RCTs) in acute pancreatitis that compares enteral nutrition with no supplementary nutrition, parenteral nutrition with no supplementary nutrition and enteral nutrition with parenteral nutrition.

Methods  A search was undertaken in the MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials.

Results  Fifteen RCTs were included. Enteral nutrition, when compared with no supplementary nutrition, was associated with no significant change in infectious complications: ratio of relative risks (RR) 0.56, 95% confidence interval (CI) 0.07–4.32, = 0.58, but a significant reduction in mortality: ratio of RR 0.22, 95% CI 0.07–0.70, = 0.01. Parenteral nutrition, when compared with no supplementary nutrition, was associated with no significant change in infectious complications: RR 1.36, 95% CI 0.18–10.40; = 0.77, but a significant reduction in mortality: RR 0.36, 95% CI 0.13–0.97, = 0.04. Enteral nutrition, when compared with parenteral nutrition, was associated with a significant reduction in infectious complications: RR 0.41, 95% CI 0.30–0.57, < 0.001, but no significant change in mortality: RR 0.60, 95% CI 0.32–1.14, = 0.12.

Conclusions  The use of either enteral or parenteral nutrition, in comparison with no supplementary nutrition, is associated with a lower risk of death in acute pancreatitis. Enteral nutrition is associated with a lower risk of infectious complications compared with parenteral nutrition.

Introduction

Since acute pancreatitis was established as a disease entity in 1889 by Reginald Fitz, there has been a little progress towards the development of specific remedies for its treatment.1 Given a lack of pathogenetic therapies currently proven to be effective in acute pancreatitis, supportive care has become a cornerstone of the early management of these patients. In particular, the quest for the optimal nutritional support strategy in patients with acute pancreatitis has been going on for decades and essentially includes three approaches: enteral nutrition, parenteral nutrition and no supplementary nutrition. Ideally, the best nutrition strategy should fit the current basic thinking on the pathogenesis of acute pancreatitis and also be in accordance with the key postulates of evidence-based medicine.

Historically, the oldest living doctrine related to nutrition in acute pancreatitis is a concept of ‘pancreatic rest’, which implies avoidance of alimentary stimulation of the pancreatic exocrine secretion in the attempt to attenuate/resolve the inflammatory process in the pancreas.2, 3 Another contemporary doctrine is the importance of maintaining intestine integrity (‘gut motor’ concept). This concept advocates luminal nutrition to prevent the rupture of intestinal barrier, bacterial translocation and subsequent development of systemic inflammatory response syndrome.4, 5 In addition, to meet the necessities of the times, the efficacy of each nutrition strategy should be judged against a control group without nutrition in a randomized controlled trial (RCT). At the same time, the current recommendations on nutrition in acute pancreatitis are mainly based on the findings of previous RCTs and meta-analyses on enteral nutrition vs. parenteral nutrition,6–8 whereas the systematically assessed evidence from RCTs on either enteral nutrition or parenteral nutrition vs. no supplementary nutrition is still scarce.

To fill the fundamental gap in the available evidence on absolute value of both enteral nutrition and parenteral nutrition in the setting of acute pancreatitis and to update the data on relative value of enteral nutrition vs. parenteral nutrition, the effect of nutrition on main clinical outcomes in three randomized comparisons (parenteral nutrition vs. no supplementary nutrition, enteral nutrition vs. no supplementary nutrition, and enteral nutrition vs. parenteral nutrition) was systematically reviewed.

Methods

Study selection

We searched online for studies published until 1 January 2008, using MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials. The search was limited to studies on adult men and women. The following search terms were used in MEDLINE: ‘Pancreatitis’ [Medical Subject Heading (MeSH)] and ‘Enteral Nutrition’ (MeSH) or ‘Parenteral Nutrition’ (MeSH) or ‘Parenteral Nutrition, Total’ (MeSH) and ‘Randomized Controlled Trials’ (Publication Type). The terms used for the search in EMBASE and Cochrane Central Register of Controlled Trials were: ‘acute pancreatitis’ and ‘enteral nutrition’ or ‘parenteral nutrition’ and ‘randomized trial’. No language restrictions were applied. Additionally, we scanned the reference lists of the primary and review articles to identify publications not retrieved by electronic searches.

Two independent investigators (MSP, RDP) reviewed studies for relevance. RCTs comparing enteral nutrition with no supplementary nutrition, or parenteral nutrition with no supplementary nutrition, or enteral nutrition with parenteral nutrition in acute pancreatitis were eligible for inclusion in the present systematic review if they reported on the following outcomes: total infectious complications and/or in-hospital mortality. Both enteral nutrition and parenteral nutrition were defined as a delivery of standard nutrition formula not supplemented with any immune enhancing ingredients (glutamine, arginine, omega-3 fatty acids, prebiotics, probiotics, etc.).

Data extraction

Two investigators (MSP, RDP) independently extracted data using standardized data extraction forms. The following characteristics were extracted from each included study: bibliographic data, study population, information pertinent to the trial quality and study outcomes data. Each patient population was used only once in this review. In case of repeated publications of the same studies, one article that provided the most complete data was selected.

The quality of the included trials was defined according to the Jadad scale, which assesses the following items: double blinding, treatment assignment and description of withdrawals.9

Statistical analysis

Chosen a priori, if the same intervention and comparator were assessed in two or more individual RCTs, a direct (conventional) meta-analysis was performed. Der Simonian and Laird (random effects) model for direct meta-analysis was chosen on the assumption that the true effect differs between the included studies.10 Treatment effect in each comparison was expressed as relative risk (RR) together with its 95% confidence intervals (CI). Heterogeneity among trials for every outcome was investigated using the I2 measure, with values of <25% regarded as low.11 Possible publication bias was explored with funnel plot. Analyses were performed using a meta-analysis software [Review Manager (RevMan) (Computer program). Version 5.0.: The Nordic Cochrane Centre, The Cochrane Collaboration 2008, Copenhagen, Denmark].

Chosen a priori, if an intervention and a comparator were assessed in less than two individual RCTs, an indirect adjusted meta-analysis was performed. In general terms, this methodology implies that indirect comparison of intervention A and B is adjusted by the results of their direct comparisons with a common intervention C.12 In the present systematic review, an indirect comparison of enteral nutrition with no supplementary nutrition was adjusted by the results of their direct comparisons with a common reference treatment (parenteral nutrition). The pooled estimates of enteral nutrition vs. no supplementary nutrition were calculated as a ratio of RRs for enteral nutrition vs. parenteral nutrition (stratum 1) and no supplementary nutrition vs. parenteral nutrition (stratum 2). We compared the results between the two strata with regard to each outcome as described by Glenny et al.12 Ratio of RRs of <1.0 represented a benefit of enteral nutrition in comparison with no supplementary nutrition. P-values <0.05 were considered statistically significant.

Results

Overall study data

The titles and abstracts of all identified papers were screened and irrelevant reports were discarded. Figure 1 details study selection process. Altogether, 15 RCTs met the inclusion criteria.13–27 Funnel plot showed no evidence of publication bias (data not shown). Thirteen studies were available as full-text articles13–16, 18–21, 24–27 and two – as abstracts.17, 22 Nine trials included only patients with predicted severe acute pancreatitis.14–16, 18, 19, 22–24, 27 In six trials, the study population consisted of both patients with predicted mild and predicted severe acute pancreatitis.13, 17, 20, 21, 25, 26Table 1 presents the study characteristics and quality assessment of trials included in the systematic review. A total of 266 patients were randomly allocated to the enteral nutrition group, 280 patients to the parenteral nutrition group and 71 patients to the no supplementary nutrition group.

Figure 1.

 Flow diagram of study selection.

Table 1.   Study characteristics and quality assessment of the included trials
StudyYear of publicationSettingStudy designPopulationDouble blindingTreatment assignment Withdrawals*
  1. * Intervention/Comparator (see column ‘Study design’).

  2. EN, enteral nutrition; PN, parenteral nutrition; NN, no supplementary nutrition.

Abou-Assi et al.132002USEN vs. PNMild and severe APNoNot stated0/0
Casas et al.142007SpainEN vs. PNSevere APNoComputer-generated0/0
Eckerwall et al.152006SwedenEN vs. PNSevere APNoNumbered envelopes1/1
Gupta et al.162003UKEN vs. PNSevere APNoNumbered envelopes3/1
Hyde et al.171984USPN vs. NNMild and severe APNoNot statedNot stated
Kalfarentzos et al.181997GreeceEN vs. PNSevere APNoNumbered envelopes2/0
Louie et al.192005CanadaEN vs. PNSevere APNoComputer-generated0/0
McClave et al.201997USEN vs. PNMild and severe APNoNot stated0/0
Olah et al.212002HungaryEN vs. PNMild and severe APNoPatient’s date of birth0/0
Paraskeva et al.222001GreeceEN vs. PNSevere APNoNot statedNot stated
Petrov et al.232006RussiaEN vs. PNSevere APNoComputer-generated0/1
Powell et al.242000UKEN vs. NNSevere APNoNot stated1/0
Sax et al.251987USPN vs. NNMild and severe APNoNot stated0/0
Windsor et al.261998UKEN vs. PNMild and severe APNoHospital number0/0
Xian-li et al.272004ChinaPN vs. NNSevere APNoNot stated2/1

Enteral nutrition vs. parenteral nutrition

A total of 11 RCTs in 453 patients with acute pancreatitis compared enteral nutrition with parenteral nutrition. The use of enteral nutrition resulted in statistically significant 59% reduction in the risk of infectious complications (Figure 2) and statistically nonsignificant 40% reduction in the risk of death (Figure 3). Heterogeneity between study results in these analyses was mainly explicable by random variation (I2 = 0% and 10%, correspondingly).

Figure 2.

 Meta-analysis (random effects model) comparing total infectious complications in patients with acute pancreatitis who received enteral nutrition vs. those who received parenteral nutrition.

Figure 3.

 Meta-analysis (random effects model) comparing mortality in patients with acute pancreatitis who received enteral nutrition vs. those who received parenteral nutrition.

Parenteral nutrition vs. no supplementary nutrition

Three RCTs comprised 113 patients with acute pancreatitis, compared parenteral nutrition with no supplementary nutrition. The use of parenteral nutrition resulted in statistically nonsignificant increase of 36% in the risk of infectious complications (Figure 4) and statistically significant reduction of 64% in the risk of death (Figure 5). There was a moderate heterogeneity between study results in the former analysis (I2 = 52%) and no heterogeneity in the latter analysis (I2 = 0%).

Figure 4.

 Meta-analysis (random effects model) comparing total infectious complications in patients with acute pancreatitis who received parenteral nutrition vs. those who received no supplementary nutrition.

Figure 5.

 Meta-analysis (random effects model) comparing mortality in patients with acute pancreatitis who received parenteral nutrition vs. those who received no supplementary nutrition.

Enteral nutrition vs. no supplementary nutrition

One RCT in 27 patients with acute pancreatitis compared enteral nutrition with no supplementary nutrition. As there were no enough data for direct meta-analysis, the methodology of indirect adjusted meta-analysis has been applied. Using this statistical technique, enteral nutrition, when compared with no supplementary nutrition, resulted in 78% reduction in the risk of death (Table 2). This difference was statistically significant (= 0.01). Using the same methodology, the risk of infectious complications was reduced by 44% with the use of enteral nutrition over no supplementary nutrition (Table 2). This difference was not statistically significant (= 0.58).

Table 2.   The results of indirect adjusted meta-analysis in patients with acute pancreatitis who received enteral nutrition vs. those who received no supplementary nutrition
ParameterTotal infectious complicationsMortality
EN vs. PNNN vs. PNEN vs. PNNN vs. PN
  1. RR, relative risk; CI, confidence interval; EN, enteral nutrition; PN, parenteral nutrition; NN, no supplementary nutrition.

RR (95% CI)0.41 (0.30 to 0.57)0.73 (0.10 to 5.61)0.60 (0.32 to 1.14)2.77 (1.03 to 7.48)
Difference between log RRs (95% CI)−0.58 (−2.62 to 1.46)−1.53 (−2.71 to −0.35)
Test of interaction−0.55−2.54
Ratio of RRs (95% CI)0.56 (0.07 to 4.32)0.22 (0.07 to 0.70)
P-value0.580.01

Discussion

The present systematic review yielded a number of key findings. First, the use of parenteral nutrition, when compared with no supplementary nutrition, significantly reduced the risk of death, whereas it had no influence on the risk of infectious complications (possibly due to the higher incidence of catheter-associated sepsis in the parenteral group). Second, the use of enteral nutrition, when compared with no supplementary nutrition, resulted in a statistically significant reduction of mortality. The risk of infectious complications was reduced by 44% with the use of enteral nutrition over no nutrition; however, this difference was not significant probably because of the fairly small sample size. Third, the use of enteral nutrition, when compared with parenteral nutrition, was associated with a significant reduction in the risk of infectious complications. The risk of death was reduced by 40% with the use of enteral nutrition, although this difference just missed a conventional level of significance.

For the first time, the findings mentioned above provide the highest level of evidence (i.e. evidence derived from a systematic review of RCTs) that nutrition (either enteral or parenteral) is an essential component in the management of patients with acute pancreatitis. Given that this disease elicits a hypercatabolic state promoting nutritional deterioration, this inference is not much surprising per se. However, as all previous recommendations on the use of nutrition in acute pancreatitis were based solely on the results of observational studies or expert opinions, some authors argued that in the era of evidence-based medicine, any kind of nutrition should be considered as a therapeutic intervention and, thus, the recommendations on its use should be based only on the level 1 evidence.28–30 Therefore, we believe that our findings will consolidate the research community and provide a compelling argument for the use of nutrition in patients with acute pancreatitis ubiquitously.

Furthermore, the present systematic review also incorporates an updated meta-analysis on enteral nutrition vs. parenteral nutrition in acute pancreatitis, which, in addition to the trials included in the previous meta-analyses, comprised three most recent RCTs14, 15, 23 and one previously missed RCT.22 Consistent with the previous systematic reviews,7, 8 the present one demonstrated the clinical benefits of enteral nutrition in comparison with parenteral nutrition. The mounting data evidence that, most likely, this is not because of better nutritional effect of enteral nutrition over parenteral nutrition, but rather because of the marked differences between the groups with regard to intestine function.31–33 In its turn, it suggests that (in)adequate intestine function may be either a confounder or an effect of nutrition. In other words, the former point of view implies that the ability of patients to tolerate enteral nutrition may impact a clinical outcome because successful enteral nutrition confirms adequate intestine function and this leads to reduced morbidity and mortality in comparison with patients who are fed parenterally because of the impaired intestine function.33 The latter point of view implies that artificial nutrition (either enteral or parenteral) is essentially a therapeutic intervention the application of which leads to (in)adequate intestine function.4 While the first opinion may be valid in a study with observational design, it is likely to be invalid in the setting of a RCT because randomization assumes the random allocation of all possible (known and unknown) confounders. Consequently, the probability of having a patient with (in)adequate intestine function is equal in both enteral group and parenteral group at baseline. As the present systematic review comprised only RCTs, we believe that the benefits of enteral nutrition over parenteral nutrition in our meta-analysis cannot be ascribed to the differences in intestine function between the groups at baseline, but rather may be attributable to the prolonged lack of luminal nutrition in the parenteral group with subsequent intestine failure leading to increased morbidity and mortality.

Apart from this, it is known that hyperglycaemia is associated with increased risk of infectious complications and mortality.34, 35 Indeed, having found a nearly twofold risk reduction of hyperglycaemia and insulin requirement with the use of enteral vs. parenteral nutrition, a recent meta-analysis of RCTs demonstrated a higher hyperglycaemic potential of parenteral nutrition in previously nondiabetic patients with acute pancreatitis.36 The observed risk reduction may be explained in light of the earlier studies, which showed that the route of artificial nutrition affects anabolic effect of insulin. In particular, it was shown that insulin and total parenteral nutrition produced a positive nitrogen balance and decreased urine 3-methylhistidine excretion, whereas these results were not observed when insulin was administered in conjunction with total enteral nutrition.37, 38

While the usefulness of enteral nutrition over parenteral nutrition is fully consistent with the ‘gut motor’ concept, the validity of another paradigm in the early management of patients with acute pancreatitis – putting the pancreas to rest – is yet to be proven. A recent observational study in healthy volunteers demonstrated that all kinds of enteral nutrition, with the exception of mid-jejunal feeding (40–80 cm past the ligament of Treitz), increase pancreatic enzyme secretory response.39 At the same time, two randomized trials on nasogastric vs. nasojejunal feeding in patients with acute pancreatitis showed no difference between the groups with regard to the tolerance of nutrition and mortality.40, 41 However, it was argued that a true jejunal placement of the feeding tube would have been difficult with the types of feeding tubes and placement techniques used in the UK trial.42 Furthermore, a nasojejunal feeding tube was, in fact, placed in the third part of the duodenum in the RCT from India. Thus, pancreatic secretion in both groups might have been equally stimulated. To characterize definitively the effect of enteral feeding with and without pancreatic rest on clinically meaningful outcomes in patients with acute pancreatitis, a National Institutes of Health funded multicentre trial on gastric vs. mid-jejunal feeding has been initiated.

This systematic review has several limitations. First, there was a moderate heterogeneity between the study results in some comparisons. To address this issue, Der Simonian and Laird model for meta-analysis was used, as it was reported to be the most conservative.10 Second, one may question the inclusion of unpublished studies into the systematic review. In contrast, others believe that identification and inclusion of the grey trials (i.e. that has not been formally published) improve the quality of systematic review as they help overcome some of the problems of publication bias, which can arise because of selective availability of data.43 Moreover, the exclusion of data from two RCTs published in the abstract form only did not principally change the results of this systematic review (data not shown). Third, the findings of indirect adjusted meta-analysis may not completely correspond to the results of meta-analysis of direct head-to-head randomized comparisons. However, the former statistical technique was successfully validated by different authors and applied in a number of clinical settings.44–46

In conclusion, this systematic review demonstrates the benefits of artificial nutrition (either enteral or parenteral) over no nutrition management in patients with acute pancreatitis. Furthermore, the use of enteral nutrition, when compared with parenteral nutrition, is associated with better clinical outcomes. The quest for optimal supplement to nutrition may further advance its use in patients with acute pancreatitis.

Acknowledgement

Declaration of personal and funding interests: None

Ancillary