Table 3 The microbiome of the critically ill: future research directions

Literature gaps, data sharing and reporting consensus

Systematic review and meta-analysis

– Chronic critical illness

– Loss of biogeography

– Utility of serum or faecal metabolites

– Microbiome reconstitution

– Post-intensive care syndrome and long-term follow-up

– Vulnerable diagnostic groups such as solid-organ transplant

– The sepsis spectrum and role of anti-microbials

– Extremes of age

Comparability of studies

– Standardisation of microbiome terms

– Prepublication of statistical analysis plan

– Uniformity of reported outcomes (see STORMS)

– Harmonisation of post-processing with testing of robustness and/or validated adjustments based on methodological differences

Open-access biorepositories

– Publishing to existing repositories or digital platforms (such as American Gut Project)

– Mandatory sharing of relevant code (such as github)

– Expansion of creative commons licencing

Benchwork models

In vitro

– Organoid models such as SHIME seeks to duplicate ultra-low diversity conditions which could facilitate ‘matching’ of candidate interventions

– Cell culture approaches

Animal models

– Use of C. elegans 2-member community model with organisms derived from ICU patients (Zaborin et al.) to probe destructive versus protective signals for phenotype switching

Systems biology

– Complementary approaches stress-tested in other fields such as soil and forensic science may elucidate relationships which prevent or correct dysbiosis, versus maintenance of a maladaptive steady state

Exploratory

Abstract influences, surrogate markers, extrapolation from healthy populations

– Adjusting for time of day of sampling, patient mobility, and bed location

– Identification of bedside urinary or faecal surrogate markers for dysbiosis as seen in Kuo et al. (2021)

Testing of robust relationships seen in healthy individuals may unravel risk factors for mortality in critical illness, e.g. Christensenellaceae shows strongest association to healthy body mass index and secondary bile acid formation (Waters and Ley, 2019) yet conferred poorer prognosis in three ICU studies—raising the question of whether this organism may play a role in the ICU obesity paradox (Sakr et al. 2015)

Clinically occult subgroups

Enterotypes and endotypes

– Clustering across genotype to phenotype levels has been attempted with utility in both prognostication and mechanism assembly

– Enterotypes are based upon patterns in gut microbiome structure have been found to improve performance of traditional scores in predicting diagnosis (Gu et al.), mortality [33] and surgical complications (Schmitt et al. 2019)

– Endotypes represent a meaningful, synergistic synthesis of biochemical and physiologic biomarkers and represent an emerging science in this population

Novel clinical trials

Embedded, adaptive, multifaceted designs

– Large multicentre embedded observational studies generalisable to the ICU population are crucial first step

– Adequately powered adaptive trials which integrate existing therapies such as artificial feeding with safe candidate therapies such as pre- pro- or syn-biotics

– Synchronous metagenomic profiling or enterotype prediction may underlie these precision medicine trials in the short to medium term and incorporate lessons learned from FMT and other vulnerable cohorts such as those with active Crohn’s disease

– Later trial phases dedicated to synthetic formulation SER-109, antimicrobial peptides, competitive inhibition approaches such as non-toxigenic Clostridium difficile strains and bacteriophage formulations