Effect of gut microbiota modulation on feeding tolerance of enterally fed critically ill adult patients: a systematic review

Purpose The objective of this systematic review was to evaluate the effect of pre-, pro-, and synbiotics on feeding tolerance of enterally fed critically ill adult patients. Methods MEDLINE, Science Direct, Web of Knowledge, and the Cochrane Central Register of Controlled Trials were searched up to November 2019. English language randomized controlled trials reporting the effect of pre, pro or synbiotics on the feeding tolerance of enterally fed critically ill adult patients were included. Results Overall, 15 papers were selected for review. Among six studies reporting the energy intake, only two studies showed significantly higher energy intake in the prebiotic-receiving groups. Among four RCTs reporting frequency or time to achieve the target calorie, only one found a significant effect of probiotics to reduce the time to achieve a target dose of calorie. About the prevalence or duration of diarrhea, 7 out of 12 RCTs reported a beneficial effect. All but one study found no beneficial effects for gut microbiota manipulation on clinical endpoints including length of stay (LOS) in hospital and intensive care unit (ICU). Conclusion It should be noticed that the heterogeneity in study designs, product format, and ICU patient populations makes it difficult to draw any general conclusion. Overall, it seems that pre, pro, or synbiotics have no significant beneficial effect on feeding tolerance and clinical endpoints in critically ill adults, but they may reduce the prevalence or duration of diarrhea. Supplementary Information The online version contains supplementary material available at 10.1186/s13643-021-01633-5.

or both of the following conditions: reduced delivery of EN and presence of gastrointestinal (GI) symptoms, including diarrhea, vomiting, regurgitation, abdominal distention, and high gastric residual volume (GRV) [7]. Feeding intolerance often results in failure to achieve the target nutritional dose as well as increased risk of pneumonia and intensive care unit (ICU) stay [8]. Factors associated with feeding intolerance in critically ill patients include stress-induced hyperglycemia; hormonal disturbances (including high levels of cholecystokinin (CCK), and peptide YY (PYY), and low levels of motilin); administration of sedatives, analgesics, and vasopressor agents; and disturbances in gut microbiota. These factors finally result in gastrointestinal dysfunction and manifest as feeding intolerance [9].
Gut microbiota manipulation can affect enteral feeding tolerance and energy homeostasis through several mechanisms. Administration of pre, pro, or synbiotics are different ways of gut microbiota manipulation. Probiotics are live microorganisms that have beneficial health effects if administered in optimum amounts. Prebiotics are non-digestible oligosaccharides that promote growth and/or activity of specific bacteria in the gut. Synbiotics are products with a combination of probiotics and prebiotics [10]. Altering gut muscle contractions, secretion, and absorption [11][12][13]; regulating glucose homeostasis [14,15]; and affecting hormonal and immune responses, host metabolism, and feeding behavior [16] are known mechanisms by which gut microbiota modulation can affect feeding tolerance and energy homeostasis.
Recently, the relationship between gut microbiota and nutrition, especially in critically ill patients has been attracting considerable interest. Many studies have reported the effect of pre, pro, or synbiotics on EN volume, energy intake, or EN-associated complication. Nevertheless, to the best of our knowledge, no systematic review or meta-analysis has been conducted to evaluate the effect of pre-, pro-, and synbiotics on feeding tolerance of enterally fed critically ill adult patients.

Methods
This systematic review was consistent with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (Additional file 1) [17].

Search strategy
A systematic search of randomized controlled trials published until November 10, 2019, was independently conducted by two authors (NS, AJE) on MEDLINE (via PubMed), Science Direct (via Scopus and Embase), Web of Knowledge (via Web of Science), and the Cochrane Central Register of Controlled Trials (via Cochrane Library). The search strategy was designed in accordance with the database orientations using Boolean operators (AND, OR), parenthesis, quotation marks, and asterisks. The following search strategy was used in MEDLINE: ("critical*" OR "critical care" OR "critical illness" OR "critically ill" OR "critically unwell" OR "severely unwell" OR "severely ill" OR "intensive care" OR "ICU" OR "CCU") AND ("tube feeding" OR "enteral*" OR "enteral feeding" OR "enteral nutrition" OR "force-feeding" OR "nasogastric*" OR "nasoduodenal*" OR "nasojejunal*") AND (prebiotic* OR probiotic* OR synbiotic* OR symbiotic) NOT (child OR pediatric OR infant OR preterm OR neonate) OR (("Enteral Nutrition"[Mesh]) AND ("Critical Care"[Mesh] OR ("tolerance" OR "intolerance" OR "tolerant" OR "intolerant") OR ("diarhea" OR "diarhoea" OR "distension" OR "distent*")) AND ("Probiotics" [Mesh] OR "Synbiotics"[Mesh] OR "Prebiotics"[Mesh])). Language restriction was applied to select articles in English. Furthermore, a manual reference check was conducted on the identified articles to find further relevant studies.

Screening and eligibility of records
The Population, Intervention, Comparison, Outcome, and Study design (PICOS) strategy was used to identify inclusion criteria. The inclusion and exclusion criteria are presented in Table 1. In summary, RCTs that were published in English language; included adult critically ill patients undergoing tube feeding; administered pre, pro, or symbiotics in the intervention group and placebo or routine care to the control group; and assessed enteral feed volume, time to reach full enteral nutrition, the prevalence of feed intolerance, and related GI complications were included in the review. Studies that included patients who received partial EN or in vitro studies were excluded. Based on the review protocol, author of studies that seemed to include other outcomes in the study but failed to report the results were contacted and asked for the missing data to avoid reporting bias. Unfortunately, the authors either did not respond to the email in the designated time of the review or the data were not available for further analysis.
The title and abstract of all identified articles were independently screened by two authors (NS, AJE). Randomized controlled trials that assessed the effect of pre, pro, and synbiotics on feeding tolerance in tube-fed critically ill patients were selected. The full text of selected articles was read and assessed regarding compliance with established eligibility criteria. Discrepancies were resolved by discussion with a third researcher (RR).

Data extraction and synthesis
The following variables were considered in data extraction: title, authors, year, country, study aim, population features (sex, age, number of participants), experimental design, intervention (the composition of prebiotic, probiotic and synbiotic, dose, and timing of administration), and main results. The I 2 statistics was used to assess heterogeneity of the studies. The analysis indicated that the heterogeneity of the studies was high (74%). The source of heterogeneity in the included articles in this systematic were study population, inclusion and exclusion criteria, intervention, duration, and reported outcomes. Due to the heterogeneity, further statistical comparison was not possible. Therefore, vote counting was the method used for data synthesis. The vote counting strategy was performed to sum up the findings of the included studies [18]. Vote counting was performed based on the risk of bias, sample size, and the statistical significance of the findings of the included studies. The harvest plot was then used to present the findings of the vote counting with colors representing studies, bar width representing sample size, and bar height representing study quality (JADAD score).

Risk of bias assessment
The 5-point JADAD score was used independently by two authors (NS, AJE) to assess the quality of included studies. Discrepancies were resolved by discussion with a third researcher (MKR). The five domains of the JADAD score included being randomized, appropriately describing of randomization, being double-blind, appropriately describing blinding, and explanation of withdrawal and dropouts.

Clinical outcomes
In this study, feeding intolerance was defined as either or both of the following conditions: reduced delivery of EN and presence of GI symptoms, including diarrhea, vomiting, regurgitation, abdominal distention, and high GRV. So, the clinical outcomes of interest were enteral feed volume; energy intake, which correlates with feed volume; time to reach full enteral nutrition; the prevalence of feed intolerance; and related GI complications (diarrhea, distention, high residual volume). Diarrhea was considered as a sole outcome and was assessed based on the duration of diarrhea or frequency of defecation in patients. Length of stay (LOS) in the ICU or hospital was investigated as a secondary outcome. The treatment effect was assessed based on the presence of a significant difference in the measurement of the outcome variables between the intervention and control groups.

Study identification and selection
A total of 93 relevant articles were identified through a database search and a review of reference lists of related articles. We excluded 78 articles due to the following reasons: 21 were conference abstracts, or the full text was not available, 17 were not RCTs, 13 did not exclusively enroll ICU patients, 6 were not in English, 8 did not report relevant outcomes, 5 had multiple interventions, 2 did not include exclusive EN, 1 did not have a placebo receiving control group, and 5 due to other reasons, e.g., irrelevant intervention or inconsistency in results. Finally, 15 RCTs, with a total of 1139 patients, were included (Fig. 1). As the articles included in this systematic review were heterogeneous in population, inclusion and exclusion criteria, intervention, duration, and reported outcomes, further statistical comparison was not possible. The mean JADAD score of all trials was 4.2. The minimum JADED score was 1 point for randomization. A summary of the findings of the studies is presented in Table 2. The harvest plot was presented for each outcome assessed with the orange color representing non-significant finding of the study and blue color representing significant finding. The length of the bars represents JADAD score for each study (Fig. 2).

Effect on energy intake and feed volume
Six trials examined the effect of pre, pro, or synbiotics administration on energy intake or feed volume in critically ill patients.

Prebiotics
In one trial in 2000, 44 critically ill patients receiving EN and antibiotics were randomized to receive fibercontaining or fiber-free formula and pectin or placebo for 6 days. Mean energy intake ranged from 1200 kcal Results showed that feed volume and mean energy intake significantly increased from baseline to day 21 in both groups. Patients in the intervention group tolerated a significantly higher amount of energy and feed volume on day 1 and 21 [22].

Probiotics
In a trial conducted in 2014, 40 critically ill patients were randomly assigned to receive a multi-strain probiotic or placebo for 7 days. Results showed no significant difference between groups in terms of the mean energy intake (1503.75±231.6 kcal vs. 1617.5±185.51 kcal, p= 0.09). The percentage of patients who met energy requirements in the synbiotic and placebo groups was 84.98±3.6 and 87.24±3.92, respectively (p=0.06) [23].

Synbiotics
Knight et al. investigated the effect of enteral synbiotic on ventilator-associated pneumonia in critically ill patients. They reported an increase in the daily tolerated feed volume from days 1 to 7 in both groups. The feed volume ranged from 488.9± 622.8 ml on day 1 to 1055.6±722.6 ml on day 7 in the synbiotic group and from 360±431.7 ml to 1243.9±810.3 ml in the placebo group. There was no significant difference between the two groups regarding the mean tolerated enteral feed volume [24]. Based on the findings of the studies and the harvest plot, it can be concluded that majority of the studies indicated a non-significant effect for pre, pro, or synbiotics administration on energy intake and feed volume.

Effect on target calorie achievement
Four trials assessed the effect of prebiotics or probiotics on the prevalence of target calorie achievement or time to receive the target calorie.

Prebiotics
In a trial conducted in 2001, severe sepsis or septic shock patients were randomly assigned to receive EN supplemented with partially hydrolyzed guar or fiberfree EN. All patients were on mechanical ventilation, antibiotics, and catecholamine therapy. The time to reach the preconceived protein/calorie goals was 5±3 days in the prebiotic and 6±3 days in the control group. The difference was not statistically significant [25]. In another trial conducted by Tuncay [27]. Based on the findings of the studies and the harvest plot, it can be concluded that majority of the studies indicated a non-significant effect for pre, pro, or synbiotics administration on target calorie achievement but the only one study that reported a significant finding had a large sample size. Therefore, it may outweigh the findings of the non-significant studies. Thus, the findings of the current studies are inconclusive.

Effect on diarrhea Prebiotics
Schultz et al. investigated the effect of pectin on the prevalence of diarrhea in a critical care setting. Diarrhea was more prevalent in the fiber-free/placebo group compared to the fiber-free/pectin group (36% vs. 9%, p= 0.31). Diarrhea was also more prevalent in the fiber/ placebo than fiber/pectin group; however, the differences were not statistically significant (55% vs. 9%, p= 0.06) [19]. Majid et al. also demonstrated that fiber-enriched EN with additional prebiotic had no significant effect on the prevalence of diarrhea. The prevalence of having at least 1 day of diarrhea was 92% in the prebiotic and 90% in the placebo group (p=0.99). The number of days of diarrhea was 3.8±3.5 in the placebo and 3.9±4.1 in the Fig. 2 Harvest plot for the difference between intervention and control groups in terms of achieving target calorie (a), energy intake and feed volume (b), length of stay (c), and diarrhea (d) prebiotic group (p= 0.94) [28]. Tuncay et al. reported that administration of prebiotic-enriched EN was associated with a significant tendency toward lower prevalence (8.7% vs. 56.5%) and faster amelioration of diarrhea (none vs. 52.2% diarrhea prevalence on day 7) [22].
In another trial, the prevalence of having at least 1 day of diarrhea was 46.1% in the prebiotic and 91.7% in the placebo group (p=0.03). Further, the mean frequency of days of having diarrhea was significantly lower in the fiber group. Besides, in the fiber group, diarrhea occurred in 10.8% of feeding days, compared to 31.5% in the controls (p<0.001) [25].
Rushdi et al. also investigated the effect of soluble guar gum on the number of liquid stools during the four days of intervention. The number of liquid stools was significantly lower at day 4 compared to day 1 in the intervention group, while it was significantly higher in the control group. The number of liquid stools on the fourth day was 1.2±0.7 in the intervention group, compared to 2.1±0.8 in the control group (p<0.01) [20].

Probiotics
In the study by Bleichner et al., 128 critically ill patients were randomized to receive Saccharomyces Boulardii or placebo capsules. The prevalence of diarrhea was not significantly different between the two groups. However, treatment with S. boulardii reduced the mean frequency of diarrhea days per feeding days from 18.9 to 14.2% (p= 0.006). The number of days with diarrhea was also significantly lower in the probiotic group (p<0.001) [29]. Barraud et al. also investigated the effect of probiotic administration on the prevalence of diarrhea. They demonstrated no significant effect of probiotic therapy on diarrhea prevalence (55.2% vs. 52.5%, p=0.72) [30]. In another RCT, Morrow et al. demonstrated that probiotic administration had no significant effect on the incidence of ICUassociated diarrhea. However, the number of days with ICU-associated diarrhea was significantly higher in the placebo group compared to the probiotic group (5.9±3.8 vs. 4.1±3.7, p=0.03) [31]. Another trial in 2010 examined the effect of probiotic VSL#3 on diarrhea among 45 critically ill patients. The mean frequency of liquid stool in the probiotic and placebo groups was 0.53±0.54 and 1.05±1.08 episodes per patient per day, respectively (p=0.03) [32]. Ferrie et al. also reported that critically ill patients who received probiotic had more diarrhea episodes compared to the control group, although the difference was not statistically significant. Diarrhea days in the 14-day study period was 7.22±3.63 in the probiotic and 5.72± 2.88 in the synbiotic group (p=0.17) [27].

Synbiotics
Knight et al. reported the overall prevalence of diarrhea to be 5% in the synbiotic group, and 7% in the controls (p= 0.59) [24]. Shimizu et al. also investigated the effect of daily synbiotic therapy on infectious complications, including enteritis in the intensive care unit. Enteritis was defined as acute onset of continuous liquid stool for more than 12 h. The results showed that the incidence of enteritis was significantly lower in the synbiotic group (6.3% vs. 27.0%; p < 0.05) [33].
Based on the findings of the studies and the harvest plot, it can be concluded that majority of the studies indicated a significant effect for pre, pro, or synbiotics on diarrhea in terms of duration or frequency.

Effect on length of stay
The effect of pre, pro, or synbiotics on ICU and hospital LOS was reported in 10 and 5 trials, respectively (Table  3). Malik et al. demonstrated that probiotic administration was associated with significantly lower ICU LOS [26]. Other studies found no significant difference between groups, regarding ICU or hospital LOS [19,21,22,24,27,[30][31][32][33].
Based on the findings of the studies and the harvest plot, it can be concluded that majority of the studies indicated a non-significant effect for pre, pro, or synbiotics administration on length of stay.

Discussion
In this systematic review, 15 randomized controlled trials were reviewed to determine the potential of pre, pro, or synbiotics administration to improve enteral feeding tolerance in tube-fed critically ill patients. Gut microbiota is a key regulator of gut function, host metabolism, and appetite. Microbial metabolites, including SCFAs, bile acids, and various neuroactive agents, interact with the GI tract and peripheral tissue through affecting the enteric nervous system and central appetite pathways or altering bile acid signaling [34]. These effects result in changes in gastric motility and emptying [35,36], which may reduce enteral feeding intolerance. Besides, gut microbiota can influence intestinal barrier function and modulate the immune system, thus indirectly affect metabolism and eating behavior [16].

Effect on energy intake or feed volume
We found six studies that evaluated the effect of pre, pro, or synbiotics on enteral feeding volume or energy intake. Considering the application of probiotics or synbiotics, no significant effect was reported. Only 2 of 4 studies, which used prebiotics (one soluble guar gum for 4 days and the other FOS for 21 days) in the intervention group, found significant beneficial effects [20,22]. It should be noted that in both of these studies, patients in Table 3 Reported feeding tolerance-related outcomes in RCTs evaluating the effect of pre, pro, or synbiotics on feeding tolerance of enterally fed critically ill patients Mean energy intake was reported for the entire intervention duration the intervention group received significantly more volume and energy on the first day. Therefore, it seems that the significant difference between the two groups in terms of received feed volume and energy at the end of the study may not be merely attributed to the effect of prebiotics.

Effect on target calorie achievement
Four trials evaluated the effect of pre or probiotics on frequency or time to achieve the target calorie. All studies but one found no significant effect. In this study, probiotic administration for seven consecutive days was associated with a significantly faster return of the gut function [26]. The included studies were heterogeneous in population features, intervention, duration, eligibility criteria, and EN protocol. Thus, the conflicting results may be attributed to these factors. It is also believed that the beneficial effect of probiotics or synbiotics could be highly strain-specific.

Effect on diarrhea
In the critical care setting, diarrhea is the most common gastrointestinal complication of EN [37], which may result in several unfavorable clinical conditions including enteral nutrition cessation and exacerbation of undernutrition [22]. Factors that contribute to the pathogenesis of diarrhea include altered physiological responses due to EN, antibiotics administration, and altered gut microbiota function [38]. Therefore, gut microbiota manipulation may be an approach for the prevention and management of diarrhea in the critical care setting. For example, gut microbiota manipulation can reverse abnormal colonic water secretion by SCFAs production [39], alter colonic motor activity [40], and interfere with pathogen colonization in the gut, which protects against diarrhea [38]. The effect of prebiotics on diarrhea was evaluated in five clinical trials [19,20,22,25,28]. Four studies investigated the effect of prebiotic on the prevalence of diarrhea. While two studies found a significant [22,25] and one a non-significant decrease [19], the other reported a non-significant increase [28]. The number of days of diarrhea was also investigated in two studies, one of which reported a significant decrease [25], while the other found a non-significant increase [28]. The number of liquid stools was also reported to be lower in the prebiotic group in one trial [21].
It should be noticed that water-soluble fiber like pectin or guar gum exhibits antidiarrheal effect by two mechanisms: (1) production of SCFAs and maintaining gut microbiota homeostasis or (2) reuptake of water and electrolytes [41]. The beneficial effect of water-soluble fibers on SCFAs production is well documented in noncritically ill patients and healthy subjects, but it is not clearly observed in critically ill patients [41]. So, the positive effect of water-soluble fibers in the mentioned studies may be attributed to the increased reuptake of water and electrolytes, not necessarily acting as prebiotics.
Regarding the effect of probiotics on the incidence of diarrhea, two studies reported a trend towards reduced diarrhea incidence in the probiotic group [29,31], and one reported a non-significant increase [30]. The effect of probiotic administration on diarrhea days was demonstrated in three of the included trials. Two of them reported a significant decrease in diarrhea duration [29,31], while one reported a non-significant increase [27]. In the probiotic group, the number of liquid stools per patient per day was reported to be significantly lower in one study [32] but loose stools were non-significantly more in another study [27].
A non-significant decrease in the prevalence of diarrhea [24] and a significant decrease in the incidence of enteritis [33] were reported to be associated with synbiotic administration.

Effect on length of stay
All but one study found no beneficial effects for gut microbiota manipulation on clinical endpoints, including LOS in hospital and ICU. A recent systematic review and meta-analysis by Manzanares et al. also showed that despite the beneficial effects of probiotic and synbiotic administration on overall infections and ventilatorassociated pneumonia, these agents had no significant effect on LOS in hospital or ICU [42].
To the best of our knowledge, this systematic review was the first study to review the effect of pre, pro, and synbiotics on feeding tolerance in enterally fed critically ill patients. As we assessed relevant outcomes in a heterogeneous ICU population, our results could be attributed to a broad spectrum of critically ill patients with sepsis, trauma, or other medical conditions. Although, the inclusion of diverse patient groups in this systematic review may be considered as a limitation for interpretation of the results. There was also great diversity in the type of administered prebiotic or probiotic strains, duration of treatment, and dose. This heterogeneity also made it impossible to quantitatively evaluate the results. Furthermore, most of the included studies reported the energy intake or feeding tolerance as a secondary outcome, not mentioning the EN protocols, while the reported EN protocols were heterogeneous in other studies.

Conclusion
Overall, the heterogeneity in studied product format, ICU patient populations, and study designs make it difficult to draw any general conclusion on the effect of pre, pro, or synbiotics on feeding tolerance of critically ill tube-fed patients. We suggest more new well-designed trials that assess feeding tolerance as a primary endpoint with a unified definition and an invariable enteral nutrition protocol that would make it possible to compare the obtained results. We have recently designed an RCT in which the main purpose is to determine the effect of synbiotics on feeding tolerance and energy homeostasis of critically ill adult patients [43]. New trials should aim to demonstrate the beneficial composition of supplements, dose, and duration to have beneficial effects. There is, moreover, a need to conduct studies that clearly establish the molecular mechanisms by which gut microbiota manipulation is attributed to feeding tolerance in critically ill patients.