Chapter 88: Critical Care Medicine in the Era of Omics

Within the human body there exists an incredible complexity, making the practice of medicine extremely challenging. No two individuals manifest illness identically. In critical care medicine, where illness strains human physiology to the brink of collapse, this becomes even more apparent. With critical illness accounting for nearly 39% of total hospital costs, now more than ever there is a demand to understand the multifaceted pathology that contributes to individual critical illness.¹

For centuries, illness has been studied on a macrolevel with gross anatomy being the epicenter of medicine. However, with the completion of the Human Genome Project (2003), a new paradigm of medicine that focuses on molecular interactions is evolving. Researchers and physicians are now examining human health and disease in the context of the human genome, epigenome, transcriptome, proteome, and microbiome.^2,3,4,5,6 Furthermore, these “omes” are being integrated to create new network models of physiology and pathology (Figure 88–1).

Since the inception of the omic era, multiple disease states have been examined including Alzheimer’s disease, cancer, and obesity. However, the application of omics and network medicine to critical illness remains vastly unexplored. The intensive care unit is an uncharted frontier. As omics continues to rapidly expand, the molecular derangements of critical illness will become increasingly apparent. Such insight will evolve the practice of critical care medicine—allowing intensivists to augment their understanding of disease, and to more efficiently diagnose and treat the individual with critical illness. (See the “Glossary and Abbreviations” section for key terms.)

For ages mankind has been fascinated by human variability (Table 88–1). Scientists have extensively hypothesized about the origins and variability of man. One of the earliest recognized publications addressing this query came from Dr. William Harvey, an English physician.⁷ In 1651 he published, De Generatione Animaliumi, which examined the egg and early embryo in different species. In this doctrine, he pronounced, “ex ovo aminum—all things from the egg.” His publication was the first to denounce the concept of spontaneous generation and propose the theory of epigenesis.

Nearly 200 years later, Austrian monk Gregor Mendel expounded further upon the origins of variability.⁸ In 1865, Mendel published his well-popularized findings examining flower color and texture of the seed of the garden pea. Through his cross breeding of plants he defined the theory of inheritance, demonstrating that each parent contributes an allele to the offspring for a certain characteristic forming a pair. These pairs further segregate independently of one another.

In the mid-1800s researchers further localized hereditary information to a cellular level. In 1869, Swiss physician and researcher, Friedrich Miescher, isolated nuclein (DNA) from leukocytes collected from soiled bandages from a nearby hospital.⁹ In 1882, Walter Flemming, a German physician and professor of anatomy, published the first images of chromosomes in tumor cells (Figure 88–2). Many decades later in 1944, researchers Oswald Avery, Colin MacLeod, and Maclyn McCarty demonstrated that the elemental blocks of inheritance were contained within DNA through experiments with Streptococcus pneumoniae.

Following the Avery, MacLeod, and McCarty findings, interest rapidly grew in the scientific community to characterize DNA further. Utilizing x-ray crystallography, English chemist, Rosalind Franklin, and English physicist, Maurice Wilkins, first described the structure of DNA in 1953.⁷ Together they revealed a repeating helical structure. In the same year, researchers James Watson and Walter Crick published on the structure of DNA (Figure 88–3).¹⁰ They first described DNA as a right-handed double helix, with base pairing of cytosine with guanine and thymine with adenine.

In the next 30 years, interest in deciphering the DNA sequences of various organisms, including Homo sapiens, quickly expanded. In 1989, international efforts collaborated to form the Human Genome Organization (HUGO), which sought to map the entire human genome. At that time, it was estimated that such a project would cost 200 million dollars per year and take 15 years to complete. In 1989, the National Institutes of Health (NIH) also established the National Human Genome Research Institute to participate in the international Human Genome Project (1989). James Watson led the efforts, which were funded by both the NIH and the Department of Energy.

Simultaneously, simpler organisms were being sequenced. In 1996, yeast, Saccharomyces cerevisiae, was the first eukaryote deciphered. The roundworm, Caenorhabditis elegans, followed in 1989 and was found to contain 19,000 genes.¹¹ By 2000, the genome of the fruit fly, Drosophila melanogaster, was sequenced and noted to have 14,000 genes.¹² The mustard plant, Arabidopsis thaliana, was sequenced in the same year and contained 26,000 genes.¹³ Comparing the genomes of these different organisms began to reveal remarkable insight into the functionality and complexity of DNA.

After much anticipation, the first draft of the Human Genome Project (HGP) was announced ahead of schedule in 2000, and made publically available in 2001.¹⁴ Surprisingly, the number of protein coding genes identified was much less than anticipated. Due to the complexity and variation displayed in humans, it was predicted that 100,000 protein-coding genes would be identified. However, only approximately 30,000 protein-coding genes were derived in the first publication of the human genome. A second draft completed in 2003 further delineated the number of protein coding genes. Finally, in 2004 the International Human Genome Sequencing Consortium performed an additional revision of the human genome, further refining the number of protein coding genes in the range of 20,000 to 25,000. After this revision, the Human Genome Project was deemed complete.¹⁵

Through the Human Genome Project, a newfound wealth of knowledge was revealed, a blue print for human variability. However, researchers quickly realized that the human genome was a very basic structure, and the protein coding genes could not completely explain the vast differences seen among humans. Researchers consequently began to search for other parameters that contributed to human variability. In addition to examining the protein-coding regions of DNA, researchers began to analyze the structure, function, and products of the entire human genome. Multiple disciplines quickly developed and flourished, ushering in a new phase of discovery—the era of omics.

Following the completion of the Human Genome Project interest reignited in the already established discipline of epigenetics. Originally termed by British researcher, Conrad Waddington in the early-1940s, epigenetics describes the mechanisms through which undifferentiated cells develop into differentiated cell types such as myocytes, neurons, adipocytes, etc.^16,17 It is now well established that DNA undergoes reversible, non-encoded modifications, which ultimately influence phenotype.¹⁸ Multiple factors, including developmental stage, age, and environment have been linked to epigenetic changes. Research has also revealed that these modifications can be maintained and propagated to daughter cells.

Within the past two decades, efforts have focused on examining the epigenomes (epigenetic changes) of the nearly 200 cell types contained within the human body. Researchers have identified several epigenetic modifications affecting transcription, including DNA methylation, histone modification, nucleosome/chromatin packaging, and RNA transcripts (Figure 88–4).¹⁹ DNA methylation describes the addition of a methyl group to cytosine, creating 5-methylcytosine (5 MeC).²⁰ Less commonly, cytosine can also undergo hydroxymethylation. Numerous studies have demonstrated that hyper-methylation represses transcription at promoter regions.²¹ Histones, which are the scaffolding proteins that support DNA packaging, similarly undergo methylation, as well as acetylation, phosphorylation, and ubiquitination. These changes alter the structure of the histone tail, which restricts or facilitates the transcription machinery’s access to nucleotides. Nucleosome positioning within the chromatin has also been demonstrated to influence transcription. Akin to histones, the packaging of the nucleosomes exposes certain areas of DNA, which influences binding of transcription factors such as enhancers, silencers, and insulators.

With renewed interest in epigenetics, The National Human Genome Research Institute established two large-scale research projects to examine the human epigenome. In 2003, the ENCyclopedia of DNA Elements (ENCODE) Project was created, with the objective of identifying all functional elements: RNA-transcribed regions, protein-coding regions, transcription-factor-binding sites, chromatin structure, and DNA methylation.^22,23 At its inception, 35 international research groups examined 30 million bases of human DNA, equivalent to about 1% of the genome. In the second phase of the project, researchers analyzed 1640 genome wide data sets, from 147 cell types.

Through the ENCODE project an intricate regulatory system was revealed.^24,25 Among the notable findings was pervasive transcription, meaning that the majority of the genome was transcribed. Areas outside of the protein coding regions that were previously thought to be silent were found to undergo transcription. Researchers further revealed that the majority of the human genome (80.4%) participated in a biochemical function. Much of which was thought to be inert DNA, was found to contain regulators of expression, including RNA elements and transcription factor binding sites. Analysis further demonstrated regulatory elements acting both locally (cis) and distally (trans).

In 2008, the National Human Genome Research Institute initiated the Roadmap Epigenomics Program (REP) to more thoroughly characterize the human epigenome (Figure 88–5).²⁶ Specifically, the project sought to examine the temporal changes of the epigenome from stem cells to mature cells in human tissues. The project further aimed to examine the epigenetic changes associated with diseased states. In 2015, the REP published its preliminary integrative analysis of 127 reference human epigenomes.²⁷

REP analysis revealed several epigenetic modifications including histone marks, DNA methylation, DNA accessibility, and RNA expression. As confirmed by previous studies, REP demonstrated that low DNA methylation states were associated with high accessibility for transcription, whereas hypermethylated states were associated with low accessibility. Notably, on average 68% of the reference epigenome was found to be quiescent. Further analysis revealed that embryonic-stem-cell-derived cells and pluripotent cells often exhibited methylation near regulatory elements, whereas differentiated cells exhibited methylation loss. Finally, the REP demonstrated specific epigenetic modifications that are associated with diseased states, including enrichment of enhancers, promoters, and open chromatin.

Following the Human Genome Project, international efforts further collaborated to catalog human genomes of various populations, forming the International HapMap Project (2002–2007). This project sought to identify and catalog human genomes to assist with linking genetic variants to disease states.^28,29 To accomplish this objective the project mapped single nucleotide polymor—phisms (SNPs) of 1184 individuals across 11 populations. Similar to fingerprints, representative SNPs (tag SNPs) were recognized as unique identifiers, allowing for identification of areas linked to individual genes (Figure 88–6).

The database created by the HapMap project has subsequently been utilized to conduct genome wide association studies (GWAS) between healthy and diseased populations. GWAS have allowed researchers to grossly sift through entire genomes and identify allele variants associated with diseases. Thus far, GWAS have facilitated the isolation of genes associated with cancer, diabetes, obesity, and dyslipidemia.^30,31,32 Additional GWAS have examined autoimmune diseases such as Crohn disease, ulcerative colitis, and psoriasis. Despite their success, GWAS have demonstrated limitations. Tag SNPs do not precisely localize involved genes; they only serve as general landmarks. Furthermore, disease processes are multifaceted, involving numerous interactions between genes and gene products rather than a single locus.

To better understand human variability and the complexity of disease states, researchers have also examined the human transcriptome. In 2000, the National Human Genome Research Institute initiated the Mammalian Gene Collection (MGC). This project was tasked with creating a database containing at least one complimentary DNA (cDNA) per gene for both human and mouse genomes.³³ The MGC later broadened its initial objective, further constructing rat and cow cDNA databases (completed 2009). The MGC and similar international consortiums anticipated that these databases would facilitate functional and comparative genomics. Specifically, transcript analysis would provide insight into the process of gene transcription, protein expression, and the networks of communication occurring at the molecular level.

Within the human transcriptome, approximately 80,000 transcript products have been identified originating from the 20,000 to 25,000 protein coding genes.³⁴ This disproportionate relationship of genes to transcription products, spurred considerable research focusing on transcription initiation and termination.^35,36,37 Researchers have since discovered that an individual gene may contain several transcription start sites (TSS), which can then produce various transcript products. Selection of a specific start site has been found to depend on multiple factors, including a cell’s developmental phase, cell-cell signaling, and tissue type. Variation in transcription has also been linked to the terminal processing of transcripts through the addition of adenosine monophosphate moieties. This modification, referred to as polyadenylation, influences processing and nuclear transport of transcripts and has been linked to various developmental phases of the cell. After transcription occurs, transcription products undergo additional processing, where certain areas are spliced out to create various alternative spliced products.

Numerous studies have compared the human transcriptomes during healthy and diseased states.³⁸ Transcriptomes of neurodegenerative diseases (eg, Alzheimer’s disease), malignancies (eg, breast, prostate cancer), and respiratory diseases (eg, asthma, COPD) are a few of the many disease states that have been analyzed.³⁹ Transcriptome analysis of brain tissue from patients with Alzheimer’s disease has revealed alternative promoter regions and transcription start sites.⁴⁰ In various malignancies, transcript studies have also identified abnormal fusion transcripts and alternative splicing (Table 88–2). It is theorized that these alternate transcript products result in cellular dysfunction and disease. Although incredibly informative, transcriptomics has limitations. Specifically, transcript products may be extremely fragile and of such small concentrations that current technologies cannot characterize them. Furthermore, transcriptomics primarily identifies genomic metabolites and has limited capabilities to identify the functional properties of transcript products.

After the completion of the Human Genome Project, researchers also focused efforts towards examining the protein products of the human genome. In 2009, the Human Proteome Organization (HUPO) announced an international effort to expand upon previous studies examining the human proteome in healthy and diseased states.⁴¹ In September 2010, HUPO officially initiated the Human Proteomic Project (HPP) to construct a comprehensive library of human proteins. The HPP was also tasked with examining protein expression, splice variants, post-translational modifications, and localization of proteins in cells, tissues, and organs. Furthermore, the project planned to analyze proteins during all developmental stages of adult life and under various physiologic and pathologic conditions. Nearly five years after the induction of the HPP, the first draft of the human proteome was published in 2014.⁴² Ongoing efforts continue internationally examining each chromosome to complete a more comprehensive human proteome database.

In the last two decades, research has extended beyond the human chromosome to examine the microorganisms that inhabit the human body, establishing the discipline of microbiomics. One of the first projects examining the human microbiome was the France Human Intestinal Metagenome Initiative (HIMI) established in 2005. The National Institutes of Health (NIH) soon thereafter followed, initiating the Human Microbiome Project (HMP) in 2007.⁴³ At its inception, the project’s initial objective was to examine the microbiomes of the mouth, gastrointestinal tract (stool), skin, and vagina. These databases could then be utilized to identify dysfunction, develop treatments, and possibly prevent illnesses linked to dysbiosis. The HMP project also included an initiative to examine ethical, legal, and social implications associated with genomics. Approximately, one-year after the establishment of the HMP, the International Human Microbiome Consortium formed in 2008 to further foster collaboration worldwide.

Microbiomes of multiple disease processes have since been analyzed^44,45 Most research to date has focused on the microbiomes of the gut and skin.⁴⁶ Numerous studies have repeatedly shown a strong association between diseased states and alterations in the gut microbiome.⁴⁷ For example, in obesity and the Crohn disease, there are associated changes in the diversity and composition of gut flora. Environmental stressors have also been shown to induce virulent behavior of bacteria in the gut. Microbiomic research suggests that there are numerous symbiotic and dysbiotic relationships between the human body and microorganisms which appear to directly influence human health and disease.

As omic research develops, it becomes increasingly apparent that the human genome exists within a highly integrated functional complex. Researchers are only beginning to understand the intricacies of the human genome, epigenome, transcriptome, proteome, and microbiome. Omic research is in its formidable years, and growing exponentially. Each discovery raises new questions and hypotheses, giving rise to an ever-expanding discipline.

Throughout the era of omics, the human genome has revealed itself as a multifaceted structure that exists in a constant state of flux. Genomic studies have repeatedly illustrated that the human genome orchestrates thousands of functional elements and metabolites that interact on multiple levels. In order to comprehend what appears chaotic, a new type of medicine has emerged, network medicine. Specifically, this type of medicine focuses on integrating various omic disciplines to create networks that explain healthy and diseased states (Figure 88–7).^19,48,49

Underlying network medicine are several principles utilized to construct network models.⁵⁰ These models assume that biological networks are not random, there is a certain order, and interactions are connected through links or edges. Network medicine designates entities with frequent interactions as nodes. Highly linked nodes are referred to as hubs. Numerous studies have identified hubs as essential functional elements in an organism. Within the nodes there may be subnetworks, or groups of interactions that are referred to as motifs. Networks ultimately create a theoretical model with inherent plasticity amenable to integration of new information.

Various network models have been constructed to explain the functionality of the human genome. Included among them are human diseasome networks (HDN), where various diseases are linked to the same gene. For example, multiple cancers, including breast, head, neck and bladder have been linked to tumor protein 53 gene dysfunction. In another model, referred to as a metabolic disease network (MDN), diseases are linked by a common enzymatic pathway that results in illness. Models have also been constructed based upon phenotype or co-morbidities. Phenotypic disease models (PDM) consider environmental exposures as integral components of variability. These models are relevant to population studies, allowing for analysis in a retrospective manner, from phenotype to genotype.

Newer models have begun to integrate genomic information from multiple platforms to examine common complex diseases utilizing various network algorithms, such as Bayesian models and co-expression gene models.^49,51 Coined network enabled wisdom (NEW) or systems biology, these models utilize supercomputers and distributed processing to systematically integrate vast amounts of data generated by omic technology. Through the integration of omic data, network models illustrate the relationships between the human genome and its metabolites. Once fabricated, these networks are validated by comparison to a new human sample population or by knockout mice models. Several network models for common complex diseases are in development including atherosclerosis, diabetes, Alzheimer disease, allergic rhinitis, asthma, obesity, and coronary artery disease (Figure 88–8).^52,53

Through the construction and manipulation of network models, the underlying pathology of common complex diseases can be understood. With such information, disease diagnosis and treatment may become infinitely more accurate and personalized. Diseases may also be identified at earlier stages. Network biology can also be utilized to guide development of new pharmaceuticals. By identifying consensual genes or pathways, treatments could be better targeted to address pathologic processes as well as avoid pathways that could result in deleterious side effects. Although in an early phase of development, network medicine appears to have an incredible potential to expand our understanding of the underlying molecular pathology of illness.

The application of omics to critical illness remains vastly unexplored. Most research to date has examined critical illness from a phenotype perspective. Syndromes with multi-organ failure, such as sepsis, trauma, and acute respiratory distress have received the most attention.^54,55,56,57 Genome wide association studies and transcription studies have been the most widely applied to critical illness. The majority of studies have been performed in mice models and in vitro. Few studies have been conducted in the intensive care unit, and of those completed the population size has predominately been small. In this expanding stage, the application of omics to critical care medicine is just starting to reveal the molecular pathology and complex relationships of critical illnesses.

Sepsis has been the most studied phenotype, and justifiably as it constitutes approximately 37% of ICU admissions and carries the highest mortality in general medical and surgical intensive care units.⁵⁸ Efforts have predominately focused on elucidating the molecular pathway of the inflammatory cascade associated with sepsis, with leukocytes being the primary substrate for analysis.^59,60 In transcription studies, several genes involved in pathogen recognition have displayed up-regulation. Included among them are genes encoding toll-like receptors (TLR) and genes involved in cluster of differentiation 14 (CD14) pathways. Upregulation has also been identified in genes involved in signal transduction pathways that induce immune response elements, including nuclear factor kappa-B (NF-KB), mitochondrial activated protein kinase, and Janus kinase (JAK).

Several studies have examined transcription during sepsis in attempt to substantiate the hypothesis of a two-phased model, which describes a pro-inflammatory state followed by an anti-inflammatory state. Transcription studies have proposed another potential mechanism involved in early sepsis, adaptive immunity dysfunction. In a small study examining expression profiles of patients with community-acquired pneumonia resulting in severe sepsis or septic shock, genes associated with wounding, inflammatory response, defense response, chemotaxis, and regulation of interleukin-6, exhibited higher expression compared to healthy patients.⁶¹ Notably, this study also found marked variation between gene expression profiles of survivors and non-survivors. In non-survivors, several genes involved in immune response were markedly down-regulated after one week compared to initial response (Table 88–3). The deranged gene expression of the immune response system in non-survivors may be reflective of adaptive immunity dysfunction.

In the setting of sepsis, acute kidney injury (AKI) has been analyzed utilizing GWAS.⁶² In one large study examining 887 patients with sepsis, GWAS identified several SNPs that appeared to be linked to AKI. Three of the polymorphisms rs8094315 (BCL2 gene), rs12457893 (BCL2 gene), and rs2093266 (SERPINA4 gene) were associated with decreased incidence of AKI. Two additional SNPs, rs625145 (SIK3 gene) and rs1955656 (SERPINA5), were also identified; the former being associated with an increased risk of AKI and the latter noted to be in complete disequilibrium with SERPINA4 gene. In previous research, the BCL2 gene has been shown to express BCL-2, an anti-apoptosis protein. Previous studies have also demonstrated that the SERPINA4 gene encodes kallistatin, which contains vasodilatory, anti-inflammatory, antioxidant, and antiapoptotic properties. The functional role of the SIK3 gene has yet to be delineated in the context of AKI. These findings correspond with previous research, which has identified apoptosis as a key mechanism contributing to the development of AKI in sepsis.⁶³

Omic research has also been utilized to investigate the multi-organ dysfunction that occurs in severe injury. In a multicenter study, examining transcription in severe blunt trauma and burn injuries, lymphocytes displayed remarkably similar patterns of expression as seen in sepsis.⁶⁴ After severe blunt trauma, genes involved in innate immunity, pathogen recognition, and inflammatory response had the greatest increase in expression (Figure 88–9). Conversely, genes involved in T-cell receptor function/proliferation, antigen presentation, apoptosis, and natural killer (NK) cell function were downregulated. Examination of the leukocyte transcriptome in burns (> 20% of total surface area), revealed a nearly identical pattern of transcription to severe blunt trauma. These findings suggest similar mediators (ie, damaged cellular component) or receptors (ie, toll-like receptors) that are instrumental in initiating the inflammatory cascade.

Clinical outcomes after severe blunt trauma and burns have been further examined utilizing transcriptomics. Similar to sepsis models, there also appears to be an exaggerated response of gene expression in complicated recoveries. Furthermore, in the above-described multicenter study, uncomplicated recoveries had a return to baseline expression of upregulated and downregulated pathways within 7 to 14 days after insult. In complicated recoveries, gene expression at 28 days continued to be deranged. These results may be interpreted to suggest that resolution of gene expression derangement is associated with recovery. Alternatively, these findings may suggest that adverse outcomes are associated with exaggerated responses of innate immunity and prolonged derangement of adaptive immunity. Further studies are needed to delineate the mechanisms that contribute to outcomes in both sepsis and trauma.

Critical respiratory illnesses, such as acute respiratory distress syndrome (ARDS) and chronic obstructive pulmonary disease (COPD), have also been examined utilizing omic technology. Several proteomic studies have been performed, examining bronchoalveolar lavage fluids (BALF) in humans to elucidate the molecular pathology of ARDS. In one small study examining the expression of proteins in BALFs of ARDS patients (n = 8), 22 proteins exhibited increased expression compared to controls (n = 9) over seven days. Utilizing gene ontology analysis, these proteins were linked to inflammation, immunity, response to microbials, response to stress/injury, and enzyme inhibitor activity (Table 88–4).⁶⁵

Applying network medicine to these findings has revealed connectivity between multiple proteins, including TNF-α, IL1β, LPS-binding protein, p38 MAPK, β-estradiol, retinoic acid, and the S100 proteins (Figure 88–10). Analysis has also demonstrated temporal changes in protein expression during the progression of ARDS. On day one, complement proteins, antiproteases, annexin A3, S100 proteins, actin, and extracellular matrix proteins increase, whereas surfactant protein-A, annexin A1, fibrinogen, and fatty acid-binding proteins decrease. On day seven annexin A3 and actin decrease, whereas surfactant protein-A increases; these findings may represent regeneration of damaged lung tissue. Network model analysis of ARDs has also revealed redundancy in protein connectivity, with several proteins involved in multiple pathways. These proteins may also assume various functional roles depending upon the time course of ARDS. Specifically, TNF-α has been found to function as both a pro-inflammatory mediator as well as a mediator of lung repair. Thus far, ARDS has been the only critical illness to be examined utilizing network medicine.

Among critical illnesses, chronic obstructive pulmonary disease (COPD) is the first phenotype to be examined for epigenetic modifications that influence disease progression. Until recently, omic research of COPD had predominately focused on identifying risk loci utilizing GWAS.⁶⁶ However, more and more omic research has begun to focus on epigenetic changes. Comparison of lung samples from healthy controls to those with COPD has demonstrated increased methylation of multiple genes in COPD (EP300, EPAS1, FOXF1, FOXA2, KDR, LAMA5, SHH, NKX2-1, VEGFA, FZD1, NUMB, and PKDCC).⁶⁷ Hypermethylation has been shown to be associated with down-regulation of cellular communication, multicellular development, and tissue morphogenesis. It has also been associated with up-regulation of co-translational protein targeting to membranes, protein targeting to endoplasmic reticulum, translational initiation, translation termination, and cellular protein complex disassembly. These changes in transcription induced by methylation appear to be associated with the pathogenesis and progression of COPD.

Microbiomics in critical illness has rapidly gained interest, resulting in numerous studies examining the gut microbiome during this time of extreme physiologic stress. Within the gut microbiome, there are five major phyla of bacteria including actinobacteria, bacterioidetes, firmicutes, fusobacteria, and proteobacteria.⁴⁷ During critical illness it has been repeatedly demonstrated that antibiotics, route of nutrition (enteral vs parenteral), vasoactive agents, acid-reducing agents, and opioids induce stress on the gut microbiome altering its phyla composition. Subsequently, during prolonged critical illness, ultralow diversity communities emerge as well as multidrug-resistant bacteria.⁶⁸ It is hypothesized that these multidrug-resistant bacteria are responsible for the development of late sepsis during critical illness.

Additional studies have demonstrated that bacteria and fungi, specifically Pseudomonas aeroginosa and Candida albicans, can convert from commensal to virulent during physiologic stress. Administration of opioids, hormones, and steroids has been shown to induce virulence through the bacterial quorum-sensing signaling systems. Analysis of critical illness in animal and in vitro models has demonstrated that virulence can be mitigated with phosphate supplementation. These findings suggest that the gut microbiome plays an influential role in critical illness and preserving its integrity can improve the outcomes of patients.

During this rapidly developing era of omics, criticalomic medicine is just starting to establish its presence in the intensive care unit as a new conceptualized paradigm of medicine. More and more basic research needs to be done, followed by integration of genomic, epigenomic, transcriptomic, proteomic, microbiomic, and other omic data to create more network models of critical illness. Once established, these models can then be manipulated and applied to the bedside through translational medicine. The ethics of omic research and its applications have also yet to be defined, raising many questions about health privacy and determinism. Omic research holds incredible promise to revolutionize the delivery of health care by individualizing prevention, diagnosis, and treatment of illness. However, it must be done thoughtfully and meticulously to augment the full potential of the secrets contained within our nucleotides.