Gastrointestinal dysfunction in the critically ill: a systematic scoping review and research agenda proposed by the Section of Metabolism, Endocrinology and Nutrition of the European Society of Intensive Care Medicine

GI symptoms occur frequently in the critically ill [1]. No single symptom correlates with mortality, whereas an increasing number of concomitant GI symptoms are associated with increasing mortality [1]. There is no agreed and validated scoring system for the assessment of GI dysfunction [3, 4]. The presence of GI bleeding that has been used as a symptom identifying GI dysfunction in multiple organ failure scores [5, 6] is not necessarily related to gut dysfunction, as there are numerous specific causes and therapeutic modalities [7]. Likewise, delayed gastric emptying leading to increased GRV can occur in the absence of intestinal dysfunction. Moreover, using a feeding strategy based on GRV may lack relevance, as it did not decrease the risk of ventilator-associated pneumonia in ventilated medical patients with full enteral nutrition (EN) [8]. Several methods to assess gastric emptying (e.g. scintigraphy, paracetamol absorption test) are mostly used for the purpose of research (Additional file 4, Table S5).

Diarrhoea has been suggested as a marker of malabsorption [9] and could also be considered as a sign of feeding intolerance, but existing evidence is scarce [10].

Clinical symptoms, including diarrhoea, can signal a non-occlusive mesenteric ischaemia (NOMI) that may occur related to early full EN during acute circulatory failure [11].

Recent studies demonstrated the potential for ultrasound (US) to provide a measure of (1) gastric emptying, (2) bowel peristalsis, (3) bowel diameter, (4) bowel wall thickness and (5) tissue perfusion (US Doppler). The diameter of the gastric antrum measured with US correlates with both GRV and calculations based on CT images [12]. US may also facilitate the placement of feeding tubes and therefore is an imaging technique that could potentially be incorporated into regular abdominal assessment (Additional file 4, Table S5) [13].

Besides blood l-lactate, several novel biomarkers have been proposed [14] (Additional file 4, Table S6). Citrulline levels may represent enterocyte function [15], and citrulline concentrations < 10 μmol/L are associated with increased mortality [16]. Specific aspects and pitfalls for laboratory measurements are summarized (Additional file 4). Despite encouraging preliminary results, several factors may limit the translation of novel biomarkers to clinical practice including (1) the timing of sampling, (2) the extent of surgical damage, (3) the coexistence of other organ dysfunction (e.g. renal), (4) previous gut surgery and length of intact bowel and (5) precision of laboratory technique, threshold values chosen and rapidity of the result [17].

Small cohort studies have demonstrated that absorption of macronutrients is markedly attenuated in the critically ill when compared to health [18‐20]. Nutrient analogues or nutrient labelled with an isotope (e.g. 3-O-methyl-glucose or ¹³C-glucose) can be administered with enteral nutrition and subsequently sampled from the blood and/or other body fluids to quantify nutrient absorption [18‐20]. The results of absorption studies may substantially vary depending on whether markers are administered intragastrically or intraduodenally, especially if gastric emptying is delayed [21]. The duodenal approach will better reflect the actual absorption, whereas the former might be more representative of the actual nutrient (bio)-availability during routine clinical practice.

Utilization of enterally administrated nutrients can be quantified using whole-body balance studies (Additional file 4). However, the precision of this technique requires accurate measurement of intake and output, including output from urine, faeces and drains. Faecal energy loss can be measured as a marker of malabsorption using bomb calorimetry [9], but this method is not widely available and requires the passage of stool, which is infrequent in many critically ill patients [10].

GI barrier dysfunction may be caused by (1) loss of enterocyte integrity, (2) increased transcellular/paracellular permeability, (3) loss of mucus layer integrity and (4) impaired mucosal immunity.

The GI barrier can be visualized using electron microscopy [22], but this invasive approach requires tissue biopsy and only quantifies structure at the place and time tissue is obtained. GI barrier function is the net result of a myriad of interactions between the luminal content, the epithelium and the mucosal immune system [23]. Because any or all of these components may be dysregulated in critical illness, no single biomarker (Additional file 4 Table S6) is likely to capture all of these processes to provide a robust summary score.

Double/triple sugar absorption tests are used to determine paracellular permeability in ambulant populations. However, these tests may be affected by GI dysmotility, renal and/or liver impairment, and administration of antibiotics [24], possibly limiting their usefulness in the critically ill.

Quantification of specific enteral bacteria in the blood is possible. However, confirming translocation from the gut lumen as a direct result of gut barrier dysfunction is challenging due to low rates and contamination. In HIV patients, reverse transcription polymerase chain reaction (RT-PCR) of bacterial 16S rDNA has been reported to correlate with lipopolysaccharide blood concentration [25]. This genetic information, however, does not refer to gut-specific bacteria such as Enterococcus or Bacteroides species. The widely used quantification of endotoxin, corresponding antibodies or binding proteins is neither gut- nor species-specific [26].

Intra-abdominal pressure (IAP) is readily measurable at the bedside, and increased IAP may be both cause and consequence of GI dysfunction. The definition of intra-abdominal hypertension (IAH) and measurement of IAP is described elsewhere [27]. In a study in mechanically ventilated patients, the presence of IAH in the absence of GI symptoms was not associated with mortality [28].

Current options for treating delayed gastric emptying include drugs such as metoclopramide, erythromycin and domperidone [30] (Additional file 4, Table S7). Domperidone is only available for oral administration, limiting its use in ICU patients. The combination of metoclopramide and erythromycin may have synergistic effects and be superior to either drug alone [31]; however, tachyphylaxis and arrhythmias are the limitations.

A recent meta-analysis reported that prokinetic drugs modestly reduce feeding intolerance (absolute risk reduction 17.3% (95% CI 5–26.8%)) and facilitate the placement of post-pyloric feeding tubes, but had no effect on the development of pneumonia, vomiting and diarrhoea; mortality; or length of hospital stay [32]. An even more recent meta-analysis provided similar results, but erythromycin was the only prokinetic drug to reduce feeding intolerance [33]. Due to concerns about adverse effects of erythromycin, there is a considerable interest in the use of non-antibiotic motilin agonists. The pre-emptive administration of such motilin receptor agonist had a negligible effect on nutrition provision in a recent multicentre clinical trial [34].

Neostigmine is shown effective in colonic paralysis [35] and accordingly used as a treatment for acute colonic pseudo-obstruction (Ogilvie’s syndrome) [36]. To prevent GI paralysis, administration of opioid receptor antagonists, osmotic laxatives (e.g. polyethylene glycol) and stool softeners has been proposed (Additional file 4, Table S7), but demonstrated effect of ‘bowel protocols’ is limited [37].

Although the gastric route is the preferred method of providing EN, international guidelines include recommendations for the post-pyloric route option in patients at high risk of aspiration or with gastric feeding intolerance [33, 38]. It is important to note that most of the trials and available meta-analyses [38, 39] have not restricted inclusion to patients with signs of GI dysmotility.

Apart from systemic conditions such as sepsis and shock, several interventions and specific conditions are considered to contribute to GI dysfunction including (1) intravenous fluid, and plasma glucose and electrolyte concentrations; (2) the use of opioids for analgesia; and (3) untreated intra-abdominal hypertension.

Animal models indicate that altered microbiome (see the ‘Microbiome’ section) during critical illness is associated with loss of intestinal barrier [53]. This then allows the translocation of bacterial products across the mucosa to cause further inflammation and, finally, dysfunction of remote organs (Fig. 1) [54]. For instance, acute lung injury can occur following the release of gut-derived products into the lymphatic vessels and/or directly into the lungs, as shown in a murine model and in humans [55]. In animal models, the ligation of the mesenteric lymph duct prevents the development of lung injury [56]. Likewise, intestinal dysbiosis may lead to hepatic impairment [54]. In the clinical setting, inflammation induced by translocation through disrupted gut epithelium will trigger the administration of fluids and vasopressors. Fluid-induced tissue oedema and mesenteric vasoconstriction may amplify the pathophysiological processes in the gut further and possibly lead to NOMI.

The microbiome refers to all of the microbial consortia (both commensal and pathogenic bacteria, viruses and fungi), their genes and gene products (proteins and metabolites), their community structure (distribution, diversity, and evenness) and the particulars of the environment in which they reside. Essential functions of the gut microbiota include the synthesis, modulation and fermentation of gastrointestinal metabolites. Moreover, the microbiome has immunomodulatory properties [57].

Not only antibiotics but also other commonly used drugs in the ICU can interfere with the gut microbiome [58]. Endogeneous bacteria may play a beneficial role in morbidity and mortality of acute illness [59]. However, critical illness leads to disruption of the balance between the intestinal epithelium (increased apoptosis, permeability and mucus alterations all resulting in decreased barrier function) and the microbiome (predominance of pathological bacteria, increased virulence and antibiotic resistance) [57, 60]. This transfer to a critical illness-related ‘disease-promoting microbiome’ or ‘pathobiome’, respectively, may lead to pro-inflammatory downstream events in the intestinal epithelial cells, increased permeability of tight junctions and mucus disintegration, all of which are considered to be associated—both as a cause and consequence—with gastrointestinal injury and multi-organ dysfunction syndrome (MODS) [50, 60]. Emerging, but still preliminary, data in critically ill patients suggest the following: (1) the presence of specific gastrointestinal microbial pathogens at ICU admission is associated with an increased risk for death or all-cause infection, and rectal carriage of common ICU pathogens may predict specific infections [61]; (2) microbiome of critically ill patients undergoes a significant and rapid dysbiosis with loss of diversity, loss of site specificity and a shift toward dominant pathogens as compared to healthy controls [62]; (3) selective decontamination of the digestive tract (SDD)-treated critically ill patients deviate strongly from the gut microbiota of healthy subjects, whereas recolonization of the gut by antibiotic-resistant bacteria may occur upon ICU discharge and cessation of SDD [63]; and (4) lung microbiome is enriched with gut bacteria in acute respiratory distress syndrome [54].

As described (the ‘The role of the gut in multiple organ dysfunction syndrome (MODS)’ section), in animal models, the gut plays a pivotal role to precipitate MODS.

Peterson and Artis have suggested that the intestinal epithelial cells with all the different phenotypes (i.e. enterocytes, goblet cells, Paneth cells, enteroendocrine cells, M cells and intestinal epithelial stem cells) should be recognized as the central regulatory components of barrier function and immune homeostasis [64]. Secretion of epithelial-derived mucins, antimicrobial peptides and IgA create the first line of defence, whilst the tight epithelial lining builds up a physical border. Special cell populations like M cells and dendritic cells act as sensors for pathogens/antigens that activate the local immune response if necessary. Functional or physical loss of this epithelial integrity can lead to further harm [65]. Translocation of (patho)antigens across the epithelial lining may result in the activation of intestinal macrophages and leucocyte recruitment (i.e. intestinal T lymphocytes CD4+ alpha4beta7+ CCR9+) to the intestinal mucosa. The release of cytokines (i.e. tumour necrosis factor, interleukin-1 beta and interleukin-10), reactive oxygen species and nitric oxide may aggravate the intestinal barrier failure and impair gastrointestinal motility by disruption of the tight junctions and smooth muscle contractile elements [66].

Molecules secreted from the GI tract may have local effects to modulate motility, mucosal growth and immune function and/or distal hormonal effects on other systems, particularly metabolism [67, 68]. Plasma concentrations of enterohormones have therefore been evaluated as a technique to monitor GI function, but none of them is currently clinically used for this purpose (Table S6). However, the precise relation between GI hormones and GI dysfunction is insufficiently understood.

Bile acids have been suggested as a mediator for organ dysfunction [69]. Altered bile acid homeostasis in paediatric patients with intestinal failure has been postulated to contribute to liver dysfunction via increased hepatic bile acid synthesis due to a failing feedback mechanism [70]. Intrahepatic cholestasis of the critically ill is a consequence of alterations of bile acid signalling and transportation at the hepatocellular level. Although the clinical association of cholestasis and inflammation are established, recent studies demonstrated that alterations of hepatic transport and metabolism occur early after ICU admission [69, 70].

In case of malabsorption, the reabsorption of bile acids is reduced and the negative feedback for hepatic bile acid synthesis is inhibited [70]. This mechanism gives a rationale to study bile acid signalling molecules as possible markers of malabsorption and the effects of overproduction of bile acids due to malabsorption (gut-liver axis) on organ dysfunction and outcome.

Pathophysiological mechanisms related to GI dysfunction with potentially impaired outcome in ICU patients are bowel oedema and distension. Gut oedema occurs in the setting of inflammation and capillary leak, fluid resuscitation and increased venous pressure, whereas GI dysmotility may cause bowel distension. Both of them may contribute to (further aggravation of) GI dysfunction.