Impact of Intensive Care Unit Nutrition on the Microbiome and Patient Outcomes

INTRODUCTION

The intestinal ecosystem represents a complex interface where our own intestinal and immune cells work in concert with trillions of microorganisms to provide key homeostatic and regulatory functions. The composition of these commensals and the genes they harbor (collectively referred to as the microbiome) influences not only our intestinal barrier and the absorption of necessary vitamins and nutrients but also local and systemic immune responses as well as virtually every other organ system.¹ In critical illness and other acute inflammatory states, the intestinal ecosystem is disrupted, resulting in a loss of barrier defenses and a cascade of derangements that are thought to further contribute to the progression of sepsis and multiple organ dysfunction.² Adequate nutrition has long been viewed as integral to the support of both intestinal defense mechanisms and the increasing metabolic demands of critical illness.³ However, only recently are we beginning to understand and value the role of the microbiome as an integral modulator of key processes relevant to perioperative and intensive care. Are the microbes that are present—or absent—responsible for development of infection after major intra-abdominal surgery? Are they in part responsible for the dysregulated immune responses characteristic of sepsis? By adjusting the content of enteral supplementation, can we influence our patient’s microbiomes in such a way that alters clinical trajectory? The role of intensive care unit (ICU) nutrition on the microbiome is largely unexplored; however, emerging literature suggests that the microbiome may have a more significant role in critical illness than previously realized.

In this review, we first address the more familiar topic of how ICU nutritional strategies affect outcomes, summarizing the recent available evidence in this field. Next, we transition to exploring the rapidly growing field of microbiome medicine. As this area of study is new to many clinicians, we first aim to provide an introduction on composition and function of the microbiome in the context of perioperative and critical care, then move to a summary of recent clinical data implicating our microbiota as potential modulators of outcomes. Finally, we address the intersection of nutrition and the microbiome, and the growing data exploring the use of supplemental pre-, pro-, and synbiotics to influence microbial composition as a means of enhancing host responses and improving outcomes in critically ill and post-surgical patients. Our aim is to highlight the importance of medical nutritional therapy and alterations in the microbiome in critical and perioperative care (Fig. 1), bringing these emerging fields toward the forefront of the minds of intensivists and anesthesiologists.

Fig. 1.

Schematic of nutrition- and microbiome-associated factors that have been linked to clinical outcomes (ICU length-of-stay, duration of mechanical ventilation, or mortality) across multiple independent investigations. Factors in green boxes (up) associated with significant improvement; factors in red (below) associated with increased morbidity or mortality. Supplementation with pre-, pro-, and synbiotics, and fecal microbial transplant (FMT) lie at the intersection of nutrition and the microbiome, and represent potential emerging therapies for critically ill patients, though further studies are needed to delineate which patients most stand to benefit.

Effects of Intensive Care Unit Nutritional Therapy on Patient Outcome

Malnutrition, or inadequate nutrient intake relative to an individual’s metabolic demand, is associated with worse patient outcomes. This holds true for preoperative malnutrition, with adverse outcomes found following virtually all major surgeries, or that occurring in the ICU.^3,4 Preclinical studies have demonstrated impaired intestinal barrier defenses in the setting of enteral nutrient deprivation, such as reduced mucin (MUC2) secretion from goblet cells, villous atrophy, and altered cytokine production by local immunocytes.^5,6 The effects on barrier permeability seem augmented when protein malnutrition accompanies enteral nutrient deprivation.⁷ However, although clinicians have long known that prevention of malnutrition is essential, recent research suggests that the provision of medical nutrition is far more nuanced than simply a question of presence or absence of therapy. Rather, content, delivery mechanism, and timing are all important factors that should be considered. There is increasing recognition that both underfeeding and overfeeding as well as providing insufficient protein may prolong ICU stay and duration of mechanical ventilation, whereas survival may be improved if therapy targets 70% of caloric needs estimated by indirect calorimetry as well as sufficient protein.⁸ Fig. 2 delineates a proposed timeline and step-wise approach for incorporating these recommendations into the ICU treatment plan.

Fig. 2.

Proposed step-wise approach for initiation of medical nutritional therapy for critically ill patients based on current evidence.

Administration route remains important, though the superiority of enteral nutrition (EN) over parenteral nutrition (PN) has become far less definitive in recent years. Despite the negative consequences of enteral nutrient deprivation on the gut barrier, PN prescribed to meet (but not exceed) target caloric and protein needs in the setting of modern infectious control practices and improved lipid formulations may result in similar outcomes.⁴ Indeed, findings from CALORIES, a pragmatic, multicenter, randomized controlled trial (RCT) of 2,400 patients admitted to mixed medical–surgical ICUs demonstrated no difference in mortality or infectious complications with initiation of early PN (EPN)versus early EN (EEN) providing equivalent caloric needs (25 kcal/kg/day to be met by 48–72 hour).⁹ However, timing does matter. EEN versus delayed EN is associated with fewer infectious complications in hospitalized patients,¹⁰ lower mortality in patients on Veno-Arterial Extracorporeal Membrane Oxygenation (VA-ECMO),¹¹ and reduced risk of nosocomial pneumonia in traumatic brain injury (TBI).¹²

To address the growing evidence in this field, the European Society for Clinical Nutrition and Metabolism conducted a meta-analysis and issued an updated guideline for management of medical nutritional therapy in critically ill patients.¹⁰ The guideline advocates for a much more comprehensive approach to initiating nutrition in the ICU, at once considering the decision of route (oral vs EN vs PN) and timing of initiation, alongside a well-defined plan for achievement of target caloric and protein needs. Of note, irrespective of the route, a progressive ramp-up of therapy is recommended, such that less than 70% of measured energy expenditure (or estimated needs) is to be provided in the first 48 hour after admission, with escalation to 80% to 100% after day 3, all in conjunction with a targeted protein intake of 1.2–2 g/kg.¹⁰ A recently published prospective multinational cohort study of 1172 medical ICU patients lends further support for this approach. Irrespective of timing, compared with low-caloric intake, providing 10 to 20 kcal/kg per day was associated with shorter duration of mechanical ventilation and longer survival, with the greatest benefit conveyed after day 5 of admission. Compared with low-protein intake, intake of 1.2 g/kg in the first 15 days was associated with shorter duration of mechanical ventilation.¹³

Regardless of route and timing, if oral intake is not adequate, and barring any contraindications, EN remains preferred over PN, ideally within 48 hour of admission.¹⁰ Although the consensus cedes that the inherent heterogeneity in the critically ill population, coupled with variability in study design, execution, and evaluated outcomes challenges the ability to make generalizable statements, the guideline continues to indicate a superiority of EN over PN in severe acute pancreatitis (SAP), following gastrointestinal (GI) surgery, and in unselected critically ill patients, conveying lower infectious complications and reduced ICU stays.¹⁰ Circulatory shock remains a contraindication to EN due to concern for bowel ischemia in the setting of increased enterocyte oxygen consumption from absorption of enteral nutrients. Nonetheless this contraindication may be debatable. The NUTRIREA-2, a pragmatic RCT comparing outcomes between EEN and EPN in 2410 mechanically ventilated patients on vasopressors admitted to primarily medical ICUs demonstrated an increased risk of nonocclusive mesenteric ischemia in the EEN group (2% vs <1% in EPN, P = .007).¹⁴ However, the mean vasopressor dose in NUTRIREA-2 was 0.56 mg/kg/min norepinephrine and the mean caloric intake was 20 kcal/kg/day, and it remains possible that these findings may not hold true at reduced doses of vasopressor and lower early caloric intake.¹⁵ To this point, data from a retrospective analysis of over 50,000 mechanically ventilated patients with shock on varying doses of vasopressor support indicate an improved survival if EEN is initiated over delayed EN in patients on lower doses of norepinephrine (<0.3 mg/kg/min), but not in patients requiring more pressor support.¹⁶ Similarly, a recent retrospective study evaluating differences in hospital outcomes between EEN and late EN in 1700 patients receiving mechanical ventilation and vasopressor support (dose unknown) revealed no difference in mortality, and a reduction in ICU length of stay, electrolyte disturbances, and renal replacement therapy in the EEN group.¹⁷ Taken together, these studies fall in support of EEN, even in the setting of shock.

What then, if EN fails to meet nutritional requirements? In this instance, the available literature supports supplementation with PN. Although data are limited for what defines “inadequate EN,” when to initiate supplemental PN, and when to achieve target caloric needs, it seems that the best evidence exists for initiating supplemental PN when EN provides less than 60% caloric needs after 3 days following an ICU admission, with a goal of supplemental PN to achieve 100% of target caloric and protein needs (1.2 g/kg/day) between days 4 and 7. This approach showed reductions in nosocomial infections, antibiotic duration, and ICU mortality with the addition of supplemental PN versus EN alone.^4,18,19

The Gut Microbiome in Critical and Perioperative Care

How does the effect of nutritional therapy on the microbiome contribute to the difference in outcomes observed in the clinical trials above? The short answer is simply that we do not yet know. However, before exploring what is known and unknown about the effects of perioperative nutrition on the gut microbiome, and how these interactions may shape a patient’s clinical trajectory, context for this field of inquiry is necessary. Below we review the structure and function of the microbiome as it relates to critical care, why and how the microbiome of critical illness differs from health and data implicating a role for gut microbial alterations in outcomes of critically ill and postsurgical patients.

What is the gut microbiome?

The gut microbiota represents the complex group of microorganisms comprising primarily bacteria but also fungi and viruses that reside along our GI tract.²⁰ Advanced sequencing techniques based on recognition and amplification of the 16S ribosomal RNA gene (rRNA)—expressed by nearly all bacteria—has allowed for exponential growth in our understanding of the role of these organisms in health and disease. The addition of metagenomic sequencing has allowed for identification of the millions of genes expressed by these microbes and the functions that are encoded.²¹ Within the 16s gene, there are 9 hypervariable regions (v1–v9) that differentiate between bacterial species. Primers amplify the hypervariable regions and discovered sequences can then be matched to known identifiers within a reference library, providing information on “who is there” at different taxonomic ranks (phylum, class, order, family, genus, and species), though specificity is often only possible to the genus level.²² Researchers can then understand how much of the bacterial community in a sample is represented by each taxon (or plural taxa), measured as relative abundance, and apply analytical techniques to assess how the numbers of taxa present (diversity) differs across samples. No two individual’s microbiomes are identical, and in fact, even within our GI tract, as intestinal morphology, nutrient absorption, pH, and oxygen supply varies, so too, does the microbial community.^23,24 Owing to such profound interindividual and intraindividual variability, it remains difficult to have a universal definition of a healthy human intestinal microbiome. We do, however, have some generalizable guidelines. Although hundreds of different species exist, in health, greater than 90% of these belong to either Bacteroides or Firmicutes, with less representation from Proteobacteria, Fusobacteria, and Actinobacteria.²⁰ At the genus-level, there seem to exist a set of 15 to 20 core genera that serve as the primary regulators of homeostatic functions affecting nearly all organ systems.^25–27 Importantly, the tens to hundreds of species within each genus exhibit different, though at times, overlapping, traits, and the representation of these species varies from person to person.

We now know that the intestinal microbiome has a role in many aspects of health that are essential in the perioperative period. The microbiome mediates host immune responses and protects against enteric pathogens via defenses such as geographic occupation, competition for nutrients, and stimulation of MUC2 by goblet cells.²⁸ In addition to barrier defenses, the microbiome influences immune responses both in homeostatic conditions and when acute threats are encountered, by signaling to local and distant immune cells including monocytes, macrophages, and intraepithelial lymphocytes, promoting differentiation of T-cell subsets (T_regs, Th₁₇), and stimulating Immunoglobulin A (IgA).²⁹ Importantly, different species can have widely different effects on cytokine production and immune responses.^29,30 These functions include secretion of antimicrobial peptides (Bacteroides species [spp]), and fermentation of dietary fibers into short-chain fatty acids (SCFAs) like butyrate, propionate, and acetate (Clostridium clusters IV and XIVa; Bifidobacterium; Lachnospiraceae).^23,28,31 In turn, SCFAs serve as fuel for colonocytes, and also as immunoregulatory (usually “anti-inflammatory”) signaling molecules for local and systemic immune cells.^31,32 In addition, microbiota produce antioxidants, regulate insulin sensitivity, modulate neurologic processes through a bidirectional communication with the brain (“gut-brain-axis”), and transform drugs into the bioactive form.^33,34 When a disturbance in the intestinal ecosystem reduces the diversity of the microbiota and/or significantly alters the abundance of certain taxa, these downstream homeostatic processes are impaired, resulting in a state known as “microbial dysbiosis.”^1,25

There is increasing evidence that microbial dysbiosis plays an integral role in the pathobiology of sepsis,^35,36 acute respiratory distress syndrome,³⁷ and stress responses to surgery.³⁸ Preclinical models demonstrate that mass reduction in commensals directly results in increased susceptibility to sepsis from enteric pathogens,^29,39 and invasions at distal sites such as pneumonia by Pseudomonas aeruginosa.⁴⁰ Moreover, the isolated loss of certain specific taxa, namely SCFA-producing Clostridiales species (spp), has been shown to result in systemic dissemination of enteric pathogens like Listeria monocytogenes⁴¹ and Salmonella.⁴² The importance of community composition holds true for models of polymicrobial intra-abdominal sepsis, as well: Fay and colleagues⁴³ demonstrated that in a cecal ligation and puncture model, the variability in survival and immunophenotypes observed between mice from different vendors was attributable to differences in their fecal microbial composition. Although the precise mechanisms by which microbial dysbiosis conveys increased susceptibility to infectious insults are not certain in every model, they likely include reduced MUC2 and increased barrier permeability; unbalanced or impaired immune responses and increases in pro-inflammatory transcription factors; and increased virulence or emergence of a dominant species which easily overtakes local defense mechanisms and reaches distant sites.^29,35,44 In this context, it is evident how microbial dysbiosis could result in pathologic mechanisms that overlap with, and further worsen, the immunopathology observed in critical illness, or explain bacteremia or infection in locations unrelated to the site of surgery.

What causes microbial dysbiosis in critical illness and the perioperative period?

Multiple interventions enacted during the perioperative period alter the composition and/or function of the microbiota. Antimicrobials represent perhaps the most important and pervasive culprit, particularly when considering that most patients will receive antibiotics at some point during their hospitalization. In the United States, greater than 90% of patients undergoing major surgery receive perioperative prophylactic antimicrobials, and 50% remain on therapy beyond 24 hour.⁴⁵ Moreover, estimates show that on any given day, ~ 70% of ICU patients worldwide are receiving antimicrobials.⁴⁶ Extensive investigations have demonstrated that antimicrobials reduce overall gut microbial diversity, indiscriminately ablate nearly all commensals, alter metabolic activity, and lead to an increased susceptibility to invading pathogens.^47,48 Studying the effects of a 4-day course of meropenem, gentamicin, and vancomycin on the gut microbiome of healthy men, Palleja and colleagues observed a marked reduction in beneficial taxa and an overgrowth of enteric pathobionts, including Enterococcus faecalis and Klebsiella pneumoniae, with positive selection for species harboring β-lactam resistance genes. Even more concerning, for some taxa, reconstitution had not yet occurred at 180 days after administration.⁴⁹ Murine studies demonstrate that a single dose of antibiotics can cause a profound (30%–90%) loss of indigenous taxa,^50,51 as well as marked susceptibility to Clostridium difficile colitis.⁵¹ Antibiotic-induced depletion of host microbiota is now such a well-established finding that it has become a mainstay of experimental design in preclinical models exploring the role of the microbiome in host response to secondary infection.⁵²

However, antimicrobials are far from the sole perpetrators; in fact, many therapies routinely administered by clinicians in the operating room and ICU have been found to induce microbial dysbiosis. Our own investigations in a mouse model revealed that a 4-hour exposure to either 100% O₂, or isoflurane and 100% O₂, reduces the abundance of multiple beneficial commensals, including Clostridiales spp,⁵³ Ashley and colleagues⁵⁴ further demonstrated that 24 to 72 hour exposures to 100% O₂ result in alterations in gut and lung microbiota that contribute to hyperoxia-induced lung inflammation; interestingly, varying antimicrobial regimens administered before exposure enacted different effects on parameters of lung injury. In addition to reducing abundance of beneficial commensals, certain therapies additionally promote the expansion of pathogenic species. Proton-pump inhibitors taken for 4 weeks resulted in increases in taxa belonging to the Enterococcaceae and Streptococcaceae families in otherwise healthy patients and may predispose to infectious complications.⁵⁵ In preclinical models, exposure to opioids for 24 hour enhanced growth of E faecalis and virulence of P aeruginosa, effects that have been associated with an increased development of sepsis and pneumonia.^56,57

How are these findings translated to the patients we encounter as intensivists and anesthesiologists?

Longitudinal 16S sequencing of fecal and rectal samples from critically ill patients demonstrate profound reduction in community diversity and compositional changes, even within 48 hour of ICU admission.^58,59 Moreover, certain taxonomic alterations have been replicated across multiple groups. Specifically, patients exhibit a profound loss of health-promoting commensals recognized for their roles in SCFA production and maintenance of barrier integrity, including Faecalibacterium,^58,60,61 Ruminococcus, Roseburia, and Blautia.^60,62 Functionally, the loss of SCFA producers is supported by independent reports documenting reduced fecal butyrate, propionate, and acetate in ICU patients with sepsis.^63,64 In addition to the loss of beneficial commensals, critically ill patients exhibit a significant expansion of opportunists and pathogens, which at times represent more than half of the total number of species present.^61,65 The most frequently reported findings include expansion in the phylum Proteobacteria, and dominance of the Enterococcus genus, but increases in Staphylococcus, and Enterobacteriaceae like Escherichia coli and Shigella are also well documented.^58,65–67 Changes in the microbiome are not unique to critically ill patients, and reduced diversity and similar alterations in community structure have also been observed after acute trauma and burns and numerous surgeries.^68–71

These observational human studies have provided “proof” that microbial dysbiosis is rampant in the perioperative period, however the extent that iatrogenic factors versus changes in host physiology each contribute to these observations remains unknown. Although the isolated effects of antimicrobials, oxygen, and therapeutics have been demonstrated independently, we also know that the reduced intestinal perfusion in shock and sepsis increases mucosal inflammation and impairs barrier integrity, and likely also alters the microbial landscape.³⁷ Most findings indicating microbial dysbiosis in patients with critical illness or acute injury are based on data from samples collected at 48 to 72 hour after ICU admission, when the provision of supportive care (oxygen, lack of enteral nutrition, antimicrobials) was well underway, and both timing and site of collection can significantly influence findings.⁷² This by no means undermines the relevance of these discoveries. Irrespective of the cause of dysbiosis, observational studies have additionally been able to detect conserved patterns in taxonomic alterations even within subgroups of patients. For example, expansion of Enterococcus in critically ill patients was independently associated with severe dysbiosis and may serve as a potential surrogate biomarker⁷³; Agudelo-Ochoa and colleagues. identified 24 taxa that were differentially expressed in patients with sepsis compared with non-sepsis, including the distinctive expansion of Bilophilia, Fusobacterium, and Parabacteroides in sepsis, which have all been linked to pro-inflammatory pathologies in other disease states.⁶⁷

Does microbial dysbiosis influence clinical outcomes in critically ill and postoperative patients?

Although alterations to the gut microbiome are prevalent in the postoperative period, it is understandably challenging to derive cause and effect from clinical microbiome studies. Nonetheless, several recent investigations provide compelling evidence implicating the gut microbiome in the development of infectious complications and sepsis. In a large-scale study by Prescott and colleagues, investigators examined records of nearly 11,000 patients, exploring whether three hospitalization types—non-infection-related hospitalization, infection-related hospitalization, and hospitalization with C. difficile infection (CDI)—were associated with increasing risk of sepsis after hospital discharge. Indeed, they found that dysbiosis was an independent predictor for the development of severe sepsis, with adjusted probabilities of 4% in the non-infectious hospitalization group, 7% for infection-related hospitalization, and almost 11% for CDI.⁷⁴ Although purely an associative study, it supports the idea that disruption of the microbial ecosystem impairs host defense mechanisms and heightens susceptibility to infection. As for the source of infection, while not explored by Prescott and colleagues, studies using high-throughput sequencing techniques indicate that enteric bacterial translocation is likely a major contributor. A landmark study of patients undergoing hematopoietic stem cell transplantation (HCT) revealed that perioperative intestinal domination (≥ 30% community membership) by Enterococcus was associated with a 9-fold increased risk of developing vancomycin-resistant Enterococcus (VRE), and domination by Proteobacteria was associated with a 5-fold increased risk of gram-negative rod bacteremia.⁷⁵ Through the development and application of a novel platform (StrainSifter) that provides higher bacterial strain resolution than 16S rRNA sequencing, Tamburini and colleagues⁷⁶ demonstrated that in HCT recipients with bloodstream infections (n = 30), half of the pathogens (with strain level resolution) were present in the patient’s stool in the weeks preceding bacteremia. Moreover, in a prospective cohort study of w300 medical ICU patients, Freedberg and colleagues found that the presence of specific pathogens in rectal swabs (E coli, Pseudomonas spp, Klebsiella spp, C difficile, and VRE) was associated with subsequent infection by the same organism. Importantly, irrespective of VRE detection on admission, Enterococcus domination (≥30% 16S reads) was independently associated with the increased risk of death or all-cause infection.⁵⁹ Another study separately demonstrated that in patients with sepsis, expansion of Enterococcus spp over the course of ICU stay was an independent predictor of death.⁶⁷

Changes in the microbiome may also affect postoperative outcomes like anastomotic insufficiency and peritonitis after intra-abdominal surgery. In a small study of patients undergoing intestinal resection, those with post-op anastomotic disruptions or enteric infections were found to have much lower diversity in bowel specimens taken intraoperatively compared with patients that did not develop post-op complications.⁷⁷ In a longitudinal sampling of 32 patients undergoing pancreatic surgery, the incidence of post-op complications (primarily pancreatic fistula formation) was associated to the presence of distinct taxonomic alterations in fecal samples (increased Enterobacteriaceae, Akkermansia, and Anaeroplasma and less representation from Lachnospiraceae and Prevotella and Bacteroides).⁷⁸ Conversely, van Praagh and colleagues⁷⁹ observed an association between anastomotic disruption and expansion of Lachnospiraceae and Bacteroidaceae families as well as reduced community diversity in tissue samples from patients undergoing colorectal surgery. Although results from the latter two studies seem to be at odds with each other, this information may not be contradictory, but rather a result of combinatorial factors including variable spatial resolution of taxa within and along the intestinal tract, differing mechanisms leading to disruption, and varied sampling and sequencing techniques, but underscore the difficulty in the interpretation and clinical application of microbiome studies. Nonetheless, studies like these represent the first of many exciting findings bringing us one step closer to an understanding of how the perioperative microbial landscape may influence responses to surgery and inflammation.

Intersection of Nutrition and the Microbiome

Research into the effects of nutrition on the gut microbiome in critical illness is very much in its infancy. There are only a few clinical studies exploring the effects of different ICU nutritional therapy practices on the composition of the microbiome, let alone the clinical outcomes that are associated with nutrition-induced alterations in the microbiota. However, preclinical studies have provided insight into how dietary content, as well as enteral nutrient deprivation, affect the gut microbiota. In addition to reviewing these findings, below we also summarize the recent literature from human interventional trials exploring nutrition supplementation with prebiotics, probiotics, and synbiotics to modify the gut microbiome and improve outcomes in critically ill and postsurgical patients.

Diet and enteral nutrient deprivation

Diet is one of the greatest environmental determinants of microbial composition and gene expression. Nutrients (namely indigestible carbohydrates) interact directly with gut microbes to enhance or inhibit their growth.³³ The presence of absence of certain vitamins can modify the intestinal microenvironment which, in turn, influences community structure.⁸⁰ For example, vitamin D enhances expression of epithelial tight junction proteins and decreases the abundance of Bifidobacterium and increases Prevotella.⁸¹ In mice, diets rich in fat or deficient in fiber compromise the integrity of the gut epithelium and contribute to the production of potentially harmful metabolites.⁸²

Not surprisingly, the lack of EN too affects the microbiota. In addition to increased barrier permeability and altering local cytokines (reduced Interleukin-10 (IL-10), increased tumor necrosis factor (TNF) and Interferonγ⁵), EN deprivation alongside PN fosters expansion of proteobacteria and reductions in firmicutes phyla.^83,84 Of note, provision of hypocaloric EN (20% of estimated needs) both reversed some of these changes in microbiota composition and preserved MUC2 and barrier integrity.⁸³ In a separate study, addition of IV butyrate to parenteral formulations also ameliorated some of the changes observed in enteral nutrient deprivation, altering both intestinal microbial composition and increasing expression of ileal mucosal tight junction proteins and antimicrobial peptides.⁸⁵

Soluble fibers and prebiotics.

There is emerging evidence that soluble fiber—either contained within enteral formulations or added as a supplement—may have a potential therapeutic role in the ICU.⁸⁶ Soluble fibers are among the most widely recognized prebiotics, a term that describes any nonviable substrate that promotes growth of beneficial commensals. Prebiotics vary in structure, which dictates the taxa and microbial-mediated metabolites they enrich.²³ When fiber (usually plant polysaccharides), is used in this context, the aim is often to enhance growth of SCFA-producing bacteria.⁸⁷ Recently, Fu and colleagues⁸⁸ demonstrated that the amount of fiber in the enteral formula of 129 critically ill patients administered from time of admission to 72 hour, influenced both fecal SCFA levels and gut microbial composition. Patients who received greater quantities of fiber had higher SCFA levels and increased abundance of SCFA-producers, as well as significantly less Enterococcus than other groups, without changes in GI symptoms. Although other clinical outcomes were not evaluated, the protective effect of fiber is implied by the significant reduction in Enterococcus spp, expansion of which has been independently associated with mortality, as detailed previously.^59,67 Nonetheless, data from RCTs evaluating the effects of fiber supplementation on infectious complications and lengths of stay in critically ill patients⁸⁹ and patients undergoing GI surgery⁹⁰ have failed to show consistent improvement in outcomes.

Probiotics and synbiotics

In contrast to prebiotics, “probiotics” are live microorganisms that confer health benefits when ingested; “synbiotics” represent a mixture of probiotics with prebiotics.⁹¹ Importantly, the microbes within these formulations can have varying effects on immunocytes: certain Lactobacillus spp (L lactis, L acidophilus, L casei, and L rhamnosus) and Bifidobacteria (B infantis, B longum, and B brevis) favor regulatory T-cell (T_reg) differentiation and promote anti-inflammatory responses; L plantarum also increases Th₁₇, whereas L acidophilus, L rhamnosus, B brevis, and E faecalis inhibit Th₁₇ differentiation.⁹² Despite inherent variability in formulation, animal and interventional human studies have demonstrated benefits with probiotic administration, particularly in the treatment of GI diseases.⁹³ For example, supplementation with clostridium butyricum-induced IL-10 production from intestinal macrophages via toll-like receptor 2 (TLR2), preventing development of acute experimental colitis.^94,95 However, a separate study found that in the setting of enteral nutrient deprivation, reduced intestinal macrophage IL-10 is independent of the gut microbiota, but rather, due to the lack of direct stimulation by dietary amino acids.⁹⁶ Together these findings suggest that for IL-10 production and perhaps other microbial-mediated responses, augmentation through pre-/synbiotics may only be possible in the presence of EN, and the deleterious effects of complete enteral deprivation cannot be rescued by microbiota alone.

In the largest meta-analyses evaluating administration of pre-/synbiotics in nearly 3000 critically ill patients, a clear benefit was found in infectious complications, reduction in rates of ventilator-associated pneumonia (VAP), and in duration of antibiotics usage.⁹⁷ Interestingly, the recent multicenter PROSPECT trial evaluating the effects of L rhamnosus GG versus placebo in 2560 patients found no difference in rates of VAP (primary outcome), nor secondary outcomes, though an increased presence of L rhamnosus was observed in cultured sites in the therapy group (1.1% vs 0.1%, P < .001).⁹⁸ However, it is unclear to what extent certain methodologic considerations contributed to these findings, namely that the primary indication for admission was pneumonia in 60% of patients, and the quantity of L rhamnosus administered was much greater than that used by other investigations. Nonetheless, pro-/synbiotics represent a promising therapeutic avenue for critically ill patients,⁹⁹ though explorations targeted to specific subgroups of patients is necessary to better inform clinical practice.

In the postsurgical population, there is increasing evidence to suggest benefit from pro-/synbiotics, although generalizability is challenged by small sample size and variable therapy formulations and treatment lengths. Nonetheless, a meta-analysis evaluating the effects of perioperative pro-/synbiotics on infectious complications after elective intra-abdominal surgery demonstrated a significant reduction in those receiving therapy.¹⁰⁰ Other notable prospective placebo-controlled RCTs exploring pro-/synbiotic use in surgical populations also merit individual mention. In patients with colorectal cancer, consumption of a synbiotic preparation (L acidophilus, L casei, L rhamnosus, B lactis + 6g fructooligosaccharide; n = 36 vs n = 37 placebo) for 7 days before surgery resulted in reductions in IL-6 and CRP (post-therapy/pre-operatively) and fewer postoperative infectious complications (2.8% vs 18.9%, P = 0.02).¹⁰¹ Similarly, in patients undergoing resection of periampullary neoplasms, administration of a synbiotic formulation (L acidophilus, L casei, L rhamnosus, B bifidum + 0.1 g fructooligosaccharide; n = 23 vs n = 23 placebo) 4 days before surgery, until post-op day 10, resulted in reduced infectious complications in the therapy group (26.1% vs. 69.6%, P = .001), as well as reduced incidence of delayed gastric emptying and length of stay.¹⁰² Of note, the reduction in infectious complications included reduced incidence of surgical site infections, pneumonia, and urinary tract infections in the synbiotic group, whereas peripancreatic fistula formation was equivalent.

Nonetheless, although available evidence suggests promising results, there is no universally accepted strategy for provision of perioperative synbiotics, and virtually all studies have varying protocols in terms of timing of initiation (from several days pre- to even weeks post-op), supplement composition, and dose of therapy. Moreover, complications—though rare—can occur. Perhaps the most well-known example remains the PROPATRIA trial which demonstrated increased incidence of bowel ischemia with administration of a synbiotic formulation for 28d in patients with SAP.¹⁰³ Various aspects of this study (from design to execution) have since been questioned, and recent meta-analyses demonstrate either improved outcomes¹⁰⁴ or equivocal findings¹⁰⁵ with the use of pro- and synbiotics in SAP. Nonetheless, reports of bacteremia and sepsis from probiotic administration exist. S boulardii, a probiotic yeast used in the treatment of diarrhea, has been associated with fungemia in critically ill patients and is now avoided.¹⁰⁶ Moreover, in addition to PROSPECT, sepsis from Lactobacillus spp in patients receiving pro-/synbiotics been described.¹⁰⁷ These complications highlight that probiotics likely affect patients in differing ways, perhaps beneficial for some, while fostering selective expansion of some species in others, which may be particularly dangerous in the setting of impaired gut permeability in critical illness. Invasive sampling along the GI tract reveals that in response to probiotic administration, humans demonstrate significant person-, region-, and strain-specific colonization patterns in both microbial composition and microbial-encoded functions, which have little association with correlate analysis done on stool samples.²⁴ To this point, Suez and colleagues¹⁰⁸ demonstrated that following antibiotic administration in 21 healthy humans, probiotic administration resulted in delayed and irregular reconstitution of community structure and transcriptome responses throughout the GI tract, whereas autologous fecal microbial transplant (FMT) induced a rapid and near-complete recovery of both structure and function within days. From this respect, FMT represents perhaps the most promising therapeutic option, though also currently the least well-studied in the critically ill and postsurgical population. FMT is now a primary treatment strategy for refractory and relapsed CDI, and a recent retrospective study of 111 patients with severe CDI revealed a significant survival benefit in those who received early FMT.¹⁰⁹ Clinical benefit has also been demonstrated in inflammatory bowel disease, irritable bowel syndrome, and intestinal cancer¹¹⁰ and may have a potential therapeutic role in ICU-associated dysbiosis.¹¹¹

SUMMARY AND FUTURE DIRECTIONS

Our understanding of the role of the microbiome in critical illness is rapidly evolving, however, there is still much to learn. The impact of ICU nutritional therapy on the microbiome, in particular, represents a major gap in knowledge. Are the outcomes differences observed in EEN over PN and delayed EN in fact partially due to nutritional effects on the gut microbiome? We do not yet know, and exploration of this relationship should be incorporated in future trials on nutritional interventions. Biologic plausibility and recent literature support the idea that the composition of our gut microbiota figures prominently in immune responses and even the clinical course of critically ill patients. Nonetheless, we need a better understanding of the microbial signatures that are beneficial versus harmful in specific subgroups. In the perioperative space, future studies should focus on incorporating microbiome analysis in personalized perioperative risk assessment and care. Nonetheless, the inherent difficulties with execution of high-quality large-scale studies suggest that progress will be gradual: personnel is required for sample collection; rigorous bioanalytical and statistical analysis should only be performed by those with specialized training; and sample collection site, handling, and processing can all influence results. However, a diligent characterization effort could yield a future, in which the microbiome is considered equally alongside other organ systems and routinely integrated in clinical care.

For now, the current literature supports thoughtful practice decisions. A nutritional plan should be in place for every patient, integrating decisions about caloric and protein needs with route and timing. Awareness of the microbiome should be integrated into patient assessment and care: mindful provision of antimicrobials, supplemental oxygen, and other therapeutics that disrupt the microbial ecosystem; EEN when possible; and enhanced awareness that patients recently treated with antimicrobials are at greater risk of infection. In coming years, modulation of the microbiome in targeted critically ill and perioperative patients will likely be a realizable and compelling therapeutic option.

KEY POINTS.

• When possible, nutrition should be initiated early and enterally, but regardless of route, overfeeding and inadequate protein should be avoided.

• The microbiome is increasingly recognized as a contributor to outcomes in critically ill and postsurgical patients.

• Nutritional supplementation with pre-, pro-, and synbiotics has demonstrated therapeutic potential, though better characterization of individual microbial signatures is needed to identify patients that most stand to benefit from these therapies.

CLINICS CARE POINTS.

• The impact of intensive care unit (ICU) nutritional therapy on the microbiome and how this interaction contributes to clinical outcomes represents a major gap in knowledge.

• A plan for nutrition should be made on ICU admission, factoring timing, and route alongside composition. Outcomes may be improved with early (<48 hour) initiation of nutrition (enteral, if possible), avoidance of over- and underfeeding and provision of adequate protein (1.2 g/kg)

• Initiation of nutrition for patients with shock remains debatable, though data suggest this practice may be safe—and preferred over delayed enteral nutrition —when pressor support is low

• Emerging data suggest that alterations in the microbiota—and certain microbial signatures in particular—may increase susceptibility to sepsis from enteric pathogens and may increase the incidence of complications after some intra-abdominal procedures.

• Clinicians should be thoughtful in provision of antimicrobials, supplemental oxygen, opioids, and proton pump inhibitors (PPIs) as these interventions may disrupt the microbial community and impair host defenses.

• Soluble fiber and pro-/synbiotics may improve outcomes in critical illness and after intra-abdominal procedures. However, significant heterogeneity in formulation and therapy duration, and the inherent intraindividual and interindividual variability in patients’ microbiomes challenges our current ability to apply findings to large populations of patients.

• The microbiome has significant potential as a therapeutic target, however, additional rigorously designed and executed studies are needed to identify the patient subgroups that most stand to benefit.