Introduction
Acute pancreatitis (AP) is a common disease that affects about 270,000 people annually in the United States, and its incidence has been increasing by 5% every year in the United States and Europe [
1,
2]. Severe acute pancreatitis (SAP) accounts for 10% to 20% of AP cases and has unacceptably high morbidity and mortality rates [
3,
4]. Necrotic tissue infection is one of the principal causes of complications and death in SAP patients. It is believed that intestinal barrier dysfunction and subsequent bacterial translocation from the intestinal tract to the bloodstream and necrotic tissues play a critical role in the infection of necrotic tissues [
5‐
7]. Evidence derived from animal studies suggests that probiotics could stabilize the intestinal barrier and thus minimize bacterial translocation and prevent infection in AP [
8‐
11]. Moreover, clinical trials have documented the benefits of probiotics in some critically ill patients [
12‐
18]. The results of several studies indicate that probiotics can enhance intestinal barrier function, stimulate host cell production of antimicrobial peptides and produce antimicrobial factors, so probiotics could potentially be effective in the treatment of SAP [
19‐
22]. Clinical trials have been conducted on the use of probiotics in AP patients at risk of developing SAP. However, disparate results were obtained from the trials; therefore, there is still no consensus about the use of probiotics in SAP [
23‐
28], and their use is rarely recommended in clinical practice guidelines [
2,
29‐
38].
The results of the PROPATRIA trial (probiotic prophylaxis in patients with predicted severe acute pancreatitis), published in 2008 (a multicenter randomized controlled trial (RCT) dominated by the Dutch Acute Pancreatitis Study Group), showed that probiotics had harmful effects [
23], which deterred the initiation of other trials on probiotics. In recent years, however, two other RCTs have been completed, with no negative consequences were in patients treated with probiotics [
24,
28]. We therefore consider that the results of the PROPATRIA trial are questionable and that further meta-analyses of the more recent RCTs is required. To this end, we performed this meta-analysis on six select RCTs in order to determine the effects of probiotics on the rate of pancreatic and total infection, operation rate, length of hospital stay and mortality. In addition, we tried to determine the reason for the heterogeneous results across the different trials.
Discussion
In this meta-analysis of RCTs that compared probiotics with placebo in patients with SAP, we found no evidence for probiotics’ being either harmful or beneficial with regard to all important outcomes. However, significant heterogeneity was found between different trials, so the results need to be validated further.
Following the harmful effects of probiotics in SAP reported in the PROPATRIA trial [
23], Sun
et al. reported a meta-analysis of four RCTs with significant heterogeneity [
41]. In recent years, two other RCTs have been conducted on the same subject; therefore, we conducted our present meta-analysis of all six trials with a focus on the source of the heterogeneity. The results indicate that the type and duration of treatment may contribute greatly to the heterogeneity of the results. The results of the subgroup analysis would be underpowered, with only six RCTs included in the meta-analysis. To compensate for this, we also analyzed the anti-infective effect of probiotics on the treatment of critical illness to serve as a reference.
The performance of strains differs, as different bacteria have different adherence sites and divergent immunological effects [
42,
43]. Therefore, the types of probiotics used may have led to differences in the clinical outcomes.
L. plantarum 299, Synbiotic 2000 Forte and golden bifid, which were used in RCTs for the treatment of predicted SAP [
25‐
28] have also been used in RCTs for the treatment of critical illness [
13,
14,
17,
44‐
50]. The probiotic mixture used in the Cui
et al. trial contained probiotic strains similar to golden bifid [
24]. These probiotics resulted in significant improvement of infection or showed a trend in that direction. However, we found no RCTs on critical illness in PubMed in which Ecologic 641 (Winclove Bio Industries, Amsterdam, the Netherlands) was used for treatment (Besselink
et al. [
23], PROPATRIA trial), and none of the six strains of probiotic bacteria in Ecologic 641 were used in the other five trials studied in the present meta-analysis. Further, probiotic mixtures that contained probiotic strains similar to Ecologic 641 were studied in only two of the thirteen RCTs on critical illness (Jain
et al. [
51] and Barraud
et al. [
52]), and both of these RCTs showed that probiotics had a detrimental effect on infection, although the effect was not significant (see Table
2, Figure
2 and Additional file
4). All these findings suggest that the type of probiotic used plays an important role in the heterogeneity between trials.
The second aspect to be considered is the dose of probiotics, as the same probiotics can have opposite effects at different doses [
53]. The dose of probiotics varied greatly among the six RCTs in this meta-analysis. Even when the same probiotics were used in different trials, the doses were different. Thus, the dose may have also contributed to the heterogeneity.
The final factor to be considered is the duration of treatment, although it is still not clear whether the effects of probiotics are influenced by this variable [
54]. An interesting finding was that the subgroup in which the treatment duration was within 15 days or less showed significant improvement with regard to almost all outcomes in the probiotics groups (see Additional file
3). Similarly, trials for the treatment of critical illness with treatment duration of no more than 15 days also showed higher efficacy with regard to decreased infection and reduced heterogeneity (see Additional file
5). We tentatively put forth the notion that prolonged treatment duration may lead to an overload of probiotics, which might be harmful to patients with SAP and critical illness who have intestinal barrier dysfunction. However, there is a need for evidence to confirm this hypothesis, as the results could be artefactually positive.
The PROPATRIA trial, which was the best in terms of methodological quality (Jadad score of 5), also had the highest number of participants, and the investigators used Ecologic 641 for the treatment of patients with predicted SAP, but with no beneficial effects. Some ongoing studies were abandoned after the dismal results of the PROPATRIA trial [
55], and even the executors of the PROPATRIA trial thought that new randomized trials of probiotics in patients with predicted SAP were not warranted [
56]. However, the dose, type and treatment duration, which have been documented to be critical factors, vary in clinical trials. Ecologic 641, which has rarely been used in other clinical trials, induced high concentrations of interleukin 10 (IL-10) and low concentrations of IL-2 compared with the other probiotics. This effect is considered to have contributed to bowel ischemia in the probiotic-treated patients in the PROPATRIA trial [
57]. Therefore, the results of the PROPATRIA trial are not sufficient to draw the conclusion that probiotics are associated with adverse effects in the treatment of patients with predicted SAP.
The history of probiotic treatment of SAP is similar to that of immune formula treatment of critical illnesses. Both treatments were documented to be beneficial in animal experiments and primary clinical trials, but subsequent multicenter, prospective, randomized clinical trials showed adverse effects, which made the efficacy of the treatments questionable [
58]. However, subsequent studies of immune formulations indicated that certain populations of patients did benefit from the treatment [
59,
60]. Similarly, because the trials were heterogeneous with regard to their treatment strategies and outcomes, it is still possible that patients with SAP can benefit from particular probiotics if administered at the appropriate dose and for the correct duration. Moreover, given that prophylactic antibiotics have been tried with limited or no success, the effects of probiotics with regard to their potential in decreasing necrotic tissue infection warrant further study.
In the interpretation of the results of this meta-analysis, the limited number of available trials, the methodological quality of the included trials and the heterogeneity of the included trials should be noted. Only six RCTs with relatively low methodological quality were included (that is, Jadad score <3 for four of the six included trials), and considerable heterogeneity was observed among these trials with regard to the effect of probiotics. The limited number of RCTs hampered the subgroup analysis conducted to understand the cause of the heterogeneity. Upon reviewing 11 major clinical practice guidelines for AP [
2,
29‐
38], we found that the use of probiotics was mentioned only in the Society of Critical Care Medicine [
32] and International Association of Pancreatology and American Pancreatic Association guidelines [
29]. Therefore, because of insufficient clinical data, there is clearly a need for further investigations.
Acknowledgements
The authors thank the library staff of HUST, China and Mayo Clinic, Rochester, MN, USA, for their support during the revision process. The authors thank Dr Tao Yin and Pengfei Shi at the PDI, Union Hospital of HUST, China, for their help with the study design and data acquisition. The authors also thank Dr Wenchun Qu at the Department of Physical Medicine Rehabilitation, and Dr Zhen Wang at the Center for the Science of Healthcare Delivery, at the Mayo Clinic, Rochester, MN, USA, for their statistical help and advice on study design.
Competing interests
The authors declare that they have no competing interests.
Authors’ contribution
SG, ZY, TL, HW and CW were involved with study conception and design. SG, ZY and TL were involved in data acquisition. SG, ZY, HW and CW interpreted the data and results of the analyses. SG and ZY drafted the manuscript, which was critically revised for intellectual content by TL, HW and CW. All authors read and approved the final manuscript.