Mechanisms and Regulation of the Gene-Expression Response to Sepsis

This is the second in a series of 3 state-of-the-art review articles focused on sepsis.

AUTHORS: Timothy T. Cornell, MD,a James Wynn, MD,b Thomas P. Shanley, MD,a Derek S. Wheeler, MD,c and Hector R. Wong, MDc
aDivision of Critical Care Medicine, C. S. Mott Childrens Hospital, University of Michigan, Ann Arbor, Michigan; bDivision of Neonatology, Duke University Childrens Hospital, Durham, North Carolina; and cDivision of Critical Care Medicine, Cincinnati Childrens Hospital Medical Center, Cincinnati, Ohio

abstract
Sepsis is dened as the systemic inammatory response of the human host that is triggered by an invading pathogen. Despite tremendous advances in both our knowledge of and treatment strategies for this syndrome, sepsis remains among the major causes of morbidity and mortality in children worldwide. Thus, we hypothesize that an improved mechanistic understanding obtained via basic and translational science will continue to identify novel therapeutic targets and approaches. As a result, given the central importance of the alterations in gene expression in regulating the human hosts physiologic response to a pathogen, we review the complex factors genetics, transcriptional expression, and epigeneticsthat regulate unique geneexpression patterns in pediatric sepsis and septic shock. We anticipate that emerging data from genetic, genomic, and other translation studies in pediatric sepsis will advance our biological understanding of this response and undoubtedly identify targets for newer therapies. Pediatrics 2010;125:12481258

KEY WORDS genetics, immunity, sepsis, septic shock, epigenetics ABBREVIATIONS mRNAmessenger RNA SNPsingle-nucleotide polymorphism TLRToll-like receptor TNFtumor necrosis factor ILinterleukin SIRSsystemic inammatory response syndrome PTMposttranslational modication ThT helper www.pediatrics.org/cgi/doi/10.1542/peds.2009-3274 doi:10.1542/peds.2009-3274 Accepted for publication Mar 2, 2010 Address correspondence to Thomas P. Shanley, MD, C. S. Mott Childrens Hospital, F-6892, 1500 E Medical Center Dr, Ann Arbor, MI 48109. E-mail: tshanley@med.umich.edu PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright 2010 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: The authors have indicated they have no nancial relationships relevant to this article to disclose. Funded by the National Institutes of Health (NIH).

1248

CORNELL et al

STATE-OF-THE-ART REVIEW ARTICLES

FIGURE 1
After exposure to an acute infectious insult, the pediatric patient mounts an immune response to defend against microbial invasion. Pathogen-recognition receptors on local and peripheral leukocytes are activated by the presence of microbial products. Leukocyte activation results in transcription of genomic DNA to mRNA, which leads to protein production (eg, ILs, acute phase reactants). Isolation of nucleic acids permits high-throughput sequencing of DNA and expression proling of mRNA via array-based analyses. The presence of altered DNA sequences (eg, polymorphisms) and patterns of mRNA production can then be correlated with a clinical outcome such as mortality to gain insight into genetic inuences on and involved pathways in this complex pathobiology.

Sepsis reects an extraordinarily complex response of the host to an invading pathogen. Historically, preclinical and clinical studies in sepsis have generally examined only 1 to a few genes and also typically focused analyses only on increased expression levels in affected hosts compared with healthy cohorts. The development of highthroughput sequencing capability has afforded the ability to determine genetic inuences on disease phenotypes. In addition, the evolution of microarray technology provided an unprecedented opportunity to examine thousands of gene products simultaneously by measuring genome-wide messenger RNA (mRNA) expression (both increased and, equally important, decreased levels) in clinical samples from affected individuals to be compared with exPEDIATRICS Volume 125, Number 6, June 2010

pression levels in unaffected individuals. As a result, investigators have leveraged this approach to more comprehensively understand the pathobiology of pediatric septic shock as a means of both discovery and novel hypothesis generation. The overall approach to conducting array-based studies is outlined in Fig 1 with RNA-based arrays as an example. Although the technical details of this approach are beyond the scope of this article, in general, microarray technology uses a solid support (often glass or silicone) chip on which large numbers (up to thousands) of nucleic acid sequences are immobilized at known locations.1 These sequences can represent any number of targets of inquiry, including specic gene sequences, so-called expressed sequence tags that repre-

sent genes of unknown function, or sequences that represent gene polymorphisms (reviewed below). Robotic technology can apply the sequences to the solid support by a process termed spotting, or an array of shorter, oligonucleotide probes (usually 20 80 nucleotide base pairs long) can be synthesized directly on the solid support to create a chip that contains a highdensity array of sequences that can be used for large-scale analyses of the targeted genome.2 Together these technologies are affording pediatric investigators the opportunity to examine genetic inuences on outcomes from sepsis, as well as geneexpression proles that may provide novel mechanistic insight into functional pathways involved in the pathobiology of sepsis.
1249

GENETIC FACTORS THAT INFLUENCE THE HOST RESPONSE

Susceptibility to sepsis and the clinical course of patients with sepsis are both highly heterogeneous, which raises the strong possibility that the host response to infection is, at least in part, inuenced by heritable factors (ie, genetics).3 A landmark study by Srensen et al,4 published more than 20 years ago, provides strong evidence that links genetics and susceptibility to infection. A longitudinal cohort of more than 900 adopted children born between 1924 and 1926 and both their biological and adoptive parents were followed through 1982. If a biological parent died of infection before the age of 50 years, the relative risk of death from infectious causes in the child was 5.8 (95% condence interval: 2.513.7), which was higher than for all other causes studied. In contrast, the death of an adoptive parent from infectious causes did not confer a greater relative risk of death of the adopted child, which led these investigators to hypothesize that there exists a strong link between genetics and susceptibility to infection. As a result, investigators recently attempted to identify a link between genetics and predisposition to sepsis by largely focusing on genome-wide association studies and gene polymorphisms. A gene polymorphism is dened as the regular occurrence (1%), in a population, of 2 or more alleles at a particular chromosome location. The most frequent type of polymorphism is called a single-nucleotide polymorphism (SNP): a substitution, deletion, or insertion of a single nucleotide that occurs in 1 every 1000 base pairs of human DNA. SNPs can result in an absolute deciency in protein, an altered protein, a change in the level of normal protein expression, or no discernible change in protein function or expression. As these studies
1250 CORNELL et al

have proceeded, a number of SNPs in several genes responsible for regulating the inammatory, coagulation, and other key immune responses to sepsis have been identied to be associated with variable outcomes in the setting of infections (reviewed in refs 59). Here, we highlight some key examples of this genetic regulation of the host responses in sepsis. The signaling mechanisms involved in pathogen recognition, the immune response, and inammation have been reviewed extensively,1015 and SNPs have been identied in many of the genes involved in these signaling mechanisms. For example, mutations in Toll-like receptor 4 (TLR4), a key pathogen-recognition receptor, have been shown to increase susceptibility to infections secondary to Gramnegative organisms.16 Although several SNPs in the TLR4 receptor gene have been described, few have been found to be associated with an increased risk of septic shock or septic shockrelated mortality in children. For example, an adenine-for-guanine substitution 896 base pairs downstream of the transcription start site for TLR4 (896) results in replacement of aspartic acid with glycine at amino acid 299 (Asp299Gly). The Asp299Gly polymorphism has been associated with reduced expression and function of the TLR4 receptor in vitro.16,17 Furthermore, adults who carry the Asp299Gly polymorphism seem to be at increased risk for septic shock and poor outcome in several cohort studies.9,18,19 Although children who carry the Asp299Gly polymorphism seem to be at increased risk of urinary tract infection, this SNP does not seem to inuence either the susceptibility or severity of meningococcal septic shock in children.20,21 These results were further corroborated in a cohort study that involved more than 500 Gambian children.22

SNPs related to other members of this lipopolysaccharidereceptor complex (eg, CD14, MD-2, and MyD88) have been studied in adult populations, but no such studies have been performed in children.19,2326 SNPs in other classes of TLRs have also been studied. For example, gene polymorphisms of TLR2, the primary pattern-recognition receptor for Gram-positive bacteria, have been associated with increased risk of infection in both children and adults.2729 Several SNPs that affect cytokine expression have been described, although results of gene-association studies of critically ill adults with septic shock have been conicting.7,8,30 For example, 2 allelic variants of the tumor necrosis factor (TNF-) gene have been described: the wild-type allele TNF1 (guanine at -308A) and TNF2 (adenosine at -308A). The TNF2 allele has been associated with higher expression of TNF- and increased susceptibility to septic shock and mortality in at least 1 study that involved critically ill adults.31 Nadel et al32 found an increased risk of death in critically ill children with meningococcal septic shock who carried this TNF2 allelic variant. Several additional SNPs in TNF-, interleukin 1 (IL-1), IL-6, IL-8, and IL-10 have also been shown to inuence susceptibility and severity of septic shock in children.3339 Thus, there are genetically based inuences on the expression of many cytokines established to play key roles in the response to sepsis that may profoundly inuence clinical outcomes in this setting. Dysregulation of the coagulation cascade also plays an important role in the pathophysiology of septic shock, and several studies have examined polymorphisms of key genes involved in coagulation. The 4G allele of a deletion/insertion (4G/5G) SNP in the promoter region of the plasminogenactivator inhibitor type 1 (PAI-1) gene has been associated with higher

STATE-OF-THE-ART REVIEW ARTICLES

plasma concentrations of PAI-1. The 4G allele increases susceptibility to and severity of septic shock and increases the risk of mortality in children with meningococcal septic shock.4043 In addition, an SNP in the protein C promoter has been associated with susceptibility to meningococcemia and illness severity in children.44 SNPs in genes involved in phagocytosis and the complement cascade have also been studied in the context of septic shock. SNPs that affect function have been described in virtually all family members of the Fc receptor, and several of these SNPs have been associated with susceptibility to meningococcal sepsis, severity of meningococcemia, and poor outcome from meningococcal septic shock.4551 Finally, an SNP of the bactericidal permeability increasing protein gene has also been associated with increased mortality rates from septic shock in children.52 Together, these specic examples demonstrate how variations in an individuals genetic sequences can profoundly inuence the expression of the encoded gene. This genetic inuence on expression may have causal effects on the outcomes of individuals infected by a given pathogen. As more studies aim to prove such links between SNPs, gene expression, and susceptibility and/or outcome of pediatric septic shock, the studies need to be carefully considered and evaluated. With respect to evaluating the validity of these studies, the ideal genome-wide association study requires several important qualities that have been emphasized in the genetics literature.53,54 Important factors to evaluate such studies include matching by ethnic-geographic origin as well as other potentially confounding variables, validating by replicate studies, limiting analysis to the primary hypothesis and avoiding
PEDIATRICS Volume 125, Number 6, June 2010

posthoc analyses and multiple comparisons, and reporting associations as odds ratios with condence limits, all complemented by a commitment to publish both positive and sound, negative genetic-association study results to avoid publication bias. By following these and other established principles, teams of investigators who execute replicate studies should succeed in identifying important genetic inuences on gene expression that ultimately alters the outcome (favorably and unfavorably) in the setting of sepsis.

GENOME-WIDE EXPRESSION PROFILING IN CHILDREN WITH SEPTIC SHOCK

The rst publication that involved genome-wide expression patterns in pediatric septic shock consisted of 42 children with septic shock who were compared with 15 normal controls.55 RNA samples were derived from whole blood samples obtained within 24 hours of admission to the PICU for septic shock. As would be predicted, this comparison between 2 biological extremes (ie, healthy children versus those with septic shock) demonstrated a large number of genes (2000) that were differentially expressed/repressed in patients with septic shock relative to normal controls. Also, as would be expected, the upregulated genes in the patients with septic shock corresponded signicantly with functional annotations related to immunity and inammation. The most novel nding of this initial genome-wide expression-proling study surrounded the genes that were signicantly repressed in the patients with septic shock relative to controls (1000 genes). It is surprising that a large number of these repressed genes corresponded to functional annotations related to zinc biology. Thus, this initial approximation of the

genomic response of pediatric septic shock was characterized by repression of a large number of genes that either directly participate in zinc homeostasis or directly depend on zinc homeostasis for normal functioning. Consistent with this observation, the patients with septic shock who did not survive had signicantly lower serum zinc levels compared with the patients who survived. These data have generated the hypothesis that altered zinc homeostasis may play a role in the pathobiology of septic shock and is consistent with the known links between zinc and immune function.56 Recently, Knoell et al57 independently corroborated this hypothesis by demonstrating that zinc depletion led to increased death in a murine model of sepsis and that zinc supplementation partially reversed this phenotype. The second microarray-based study of pediatric septic shock focused on longitudinal expression proles58 from RNA samples from 30 children with septic shock obtained within 24 hours of their admission to the PICU (day 1) and 48 hours later (day 3) compared with those from 15 normal controls. The upregulated genes in the patients with septic shock again corresponded to multiple signaling pathways and gene networks associated with inammation and immunity. One of the most notable signaling pathways represented by the upregulated gene lists was the canonical anti-inammatory cytokine, IL-10, with a large number of genes being persistently upregulated in children with septic shock on both days 1 and 3. The downregulated gene-expression patterns in this longitudinal study again demonstrated large-scale repression of genes that correspond to zinc-related biology. This pattern of repression persisted, to a similar degree, from days 1 to 3. Another notable
1251

gene-repression pattern involved T-cell receptor signaling and the antigen-presentation pathway. Again, this pattern of gene repression was evident on day 1 and persisted well into day 3. In summary, the results of this study demonstrated persistent repression of genes that correspond to the adaptive immune system. This observation suggests that the pathobiology of pediatric septic shock may represent a failure of the adaptive immune system and is well in line with recent experimental and clinical data.5967 The translational approaches reviewed above continue to provide unique opportunities for advancing the diagnostics, prognostics, and therapeutics in pediatric and neonatal septic shock. To that end, the third microarray-based study in pediatric septic shock focused on validation of previous observations by way of formal class-prediction modeling and by the application of alternative ltering and statistical approaches.68 This studys authors made use of the original data as the training data set and a new cohort of patients with septic shock enrolled through ongoing accrual as the test or validation data set. Using 2 distinct class-prediction algorithms, the gene list derived from the training data set was able to identify the separate cohort of patients in the validation data set with 100% accuracy. In addition, the application of alternative gene-ltering and statistical approaches to the validation cohort of patients yielded similar observations to that derived from the original data. Namely, the validation data cohort of children with septic shock was also characterized by large-scale repression of genes that correspond to zinc-related biology and adaptive immunity. These data strengthen the validity of previous observations regarding the genomelevel response of pediatric septic shock.
1252 CORNELL et al

Results of genome-wide expression studies of adults with sepsis and septic shock have also demonstrated widespread and early repression of adaptive immunity-related genes, but the observations regarding repression of zinc biologyrelated gene programs have not been reported from these adult-centered studies.6972 Although it is tempting to conclude that repression of zinc biologyrelated gene programs is a unique feature of pediatric septic shock, this assertion needs to be tested directly by conducting direct comparisons of adult and pediatric expression data based on similar microarray platforms, RNA sources, and bioinformatic approaches. Results of the most recent microarraybased studies of pediatric septic shock have begun to rene our understanding of 2 distinct questions: (1) Are the reported gene-expression proles discussed above specic to patients with septic shock, or are they more generic manifestations of critical illness? and (2) Can gene-expression proling be leveraged to identify subsets or subclassications of children with septic shock? The question of specicity was addressed by comparing the geneexpression proles between children with the systemic inammatory response syndrome (SIRS) versus sepsis versus septic shock.73 These comparisons identied patterns of conserved gene expression across the pediatric SIRS, sepsis, and septic-shock spectrum, particularly with regard to upregulation of genes that correspond to innate immunity. There were also several notable gene-expression patterns that were unique and persistent in the patients with septic shock relative to the patients with SIRS or sepsis. Specically, patients with septic shock had the most prominent and persistent upregulation of genes within the IL-10 signaling pathway. In addition, repres-

sion of zinc biology and adaptive immunity-related genes was most prominent and persistent in the patients with septic shock. These data indicate that the previous observations surrounding zinc biology and adaptive immunity are relatively specic to pediatric septic shock rather than being generic manifestations of critical illness. The more challenging question of subclassication and prognostication was addressed by using microarray data from 98 children with septic shock.74 Through a discovery-oriented ltering approach and unsupervised hierarchical gene clustering, 3 putative subclasses of children with septic shock were identied solely on the basis of their gene-expression proles. One subgroup was characterized by significantly greater repression of genes that corresponded to zinc biology and adaptive immunity. It should be noted that patients in this subgroup were signicantly younger, more severely ill, and had a signicantly higher mortality rate compared with the other 2 subgroups. These data suggest that clinically relevant, molecular subclassication may be possible through gene-expression proling.

CONTRIBUTION OF EPIGENETICS TO THE REGULATION OF THE HOST RESPONSE IN SEPSIS

As reviewed in the previous section, the successful application of a genomic approach in pediatric sepsis has provided novel insight to the geneexpression patterns that characterize the complex biological human response to a pathogen. In the process of interrogating the entire genome through microarray and complex bioinformatic approaches, we have shown that pediatric septic shock is characterized by early repression of a large number of genes (notably those related to T-cell receptor signaling and

STATE-OF-THE-ART REVIEW ARTICLES

antigen presentation55,58) and signicant upregulation of other sets of genes (notably components of the IL-10 signaling pathway). As a result, it is now incumbent on investigators to further understand the molecular mechanisms that either drive or repress the expression of these genes. It is clear that receptor activation, signal transduction pathways, and transcriptional activators, all affected by genetic polymorphisms, will inuence this expression. It is important to note that results of more recent investigations (notably in developmental and cancer biology) have elucidated an additional mechanism that is capable of regulating the activation or inactivation of gene expression, termed epigenetics. Scientists have grappled with a specic definition for epigenetics, because it incorporates a number of mechanisms. Perhaps the most accepted current operational denition of epigenetics is a stably inherited phenotype resulting from changes in a chromosome without alterations in the DNA sequence, which was proposed by Berger et al.75 At the most basic level, epigenetics refers to mechanisms by which expression of a gene encoded by a specic DNA sequence can be altered by changes on the chromatin induced by a variety of signals such as environmental exposures including infections. This epigenetic regulation of gene expression depends on modications in the manner by which specic genes are packaged in chromatin and how the chromatin may subsequently be remodeled by posttranslational modications (PTMs) of the amino acids that make up histone proteins.76 This PTM of the chromatin structure can give rise to a heritable transcription state known as an epigenetic alteration, which in some cases seems capable of conferring long-term stabilization of a particular gene-expression pattern.77 These epigenetic regulatory
PEDIATRICS Volume 125, Number 6, June 2010

FIGURE 2
Cartoon depiction of the relationship between DNA (chromatin) and interacting nucleosomes that contain histone proteins. Histones can undergo PTMs (eg, methylation patterns) that alter the ability of transcriptional activation factors to access promoter regions. Depending on the pattern, these methylation signatures can result in either increased (eg, me2H3K27) or repressed (eg, me3H3K4) gene expression.

mechanisms can be related to PTMs of histones with subsequent effects on transcriptional silencing/repression or activation.78 Although not specically covered in this review, another epigenetic mechanism that inuences gene expression is direct methylation of DNA, which can alter the binding of transcriptional activating factors. Finally, many investigators include the expression of micro-RNA, small segments of RNA that can bind to DNA to alter transcriptional activation of genes, as an additional epigenetic mechanism. At the foundation of the maintenance of a particular gene-expression phenotype that is potentially maintained over a long period of time is our understanding that DNA is incompletely stripped of its nucleosomes after replication. This mechanism allows PTM of histones to serve as replicate tem-

plates to initiate identical modications of new histones and results in the transfer of modied histones to daughter cells during replication to transfer the new histone code. This process is shown in Fig 2, in which nucleosomes are composed of DNA associated with 8 histones that can be targeted for a variety of PTMs. PTMs of histones include acetylation, methylation, and phosphorylation, which are regulated by complex enzymatic machinery. Examples of this epigenetic regulation relevant to gene expression in sepsis include the nding that dimethylation or trimethylation of lysine 27 (K27) of histone 3 (H3) keeps the chromatin in a conformation such that the promoter for a specic gene is not available to transcription factors; thus, the gene is silenced (also known as gene-off).79 In contrast, methylation of lysine 4 (K4) on histone 3 (H3)
1253

is known to open up the conformation of the chromatin and allow transcription factors access to the promoter and initiation of gene expression (also known as gene-on).80 These epigenetic alterations are important, because they can stamp history into the gene-expression repertoire of cells in a heritable manner. As reviewed above, IL-10 is a canonical signaling molecule for adaptive immunity that represses T-helper 1 cytokine production, antigen presentation, and costimulatory molecules. Recent data have shown that IL-10 is epigenetically regulated in the context of polarized Th1 versus Th2 cells.81 Specically, association of the IL-10 promoter with gene-on histone modications (H3K4me3 and AcH4) correlates with increased IL-10 production in Th2 cells, whereas association of the IL-10 promoter with the gene-off histone modication correlates with decreased IL-10 production in Th1 cells. IL-12 is also a canonical signaling molecule for the adaptive immune system that induces Th1 cytokine production and differentiation. Kunkel and coworkers82 recently demonstrated that dendritic cells of mice that survive a model of experimental sepsis possess a blunted adaptive immune response and acquire a phenotype that consists of decreased IL-12 production and increased IL-10 production. They further showed that the epigenetic signature on the promoters for the p35 and p40 subunits of IL-12 are consistent with an increased gene-off signal and a concomitant decreased gene-on signal.82,83 As-yet-unpublished preliminary data suggest that similar epigenetic regulation of IL-12 repression and IL-10 expression occurs in the circulating monocytes of patients with septic shock. Thus, it is very likely that substantial changes in epigenetic signatures occur across a broad spectrum of genes. These resulting signatures
1254 CORNELL et al

are likely to be pathogen dependent, tissue specic, and, importantly, inuenced by age and have long-term effects on subsequent gene-expression patterns. Therefore, it will be imperative that neonatal and pediatric clinician-scientists examine these changes in their cohort of patients with sepsis to fully understand how this regulation of gene expression inuences both long-term and shortterm clinical outcomes.

their study, mice subjected to cecal ligation and puncture were stratied to likelihood of dying on the basis of serum IL-6 levels 6 hours after the procedure, and those with the highest predicted mortality rate were provided immune-modulating therapy, which had a signicant impact on survival.85 These preclinical studies support the concept of trying to stratify children with septic shock on the basis of early identication of septic-shock subclasses by using genome-wide expression patterns. As described above, subclasses of children with septic shock have been identied exclusively on the basis of gene-expression proling conducted within 24 hours of admission, and these subclasses have highly relevant differences in illness severity and mortality.74 A similar strategy was recently demonstrated in adult patients suffering from trauma.86 As high-throughput technologies evolve and validation studies are rigorously performed, the ability to conduct real-time expression-based subclassication and stratication could very well become a clinical reality. Another potential strategy for stratifying children with septic shock (and perhaps more technologically feasible in real time than at present) is to base such an approach on the ultimate proteins expressed as a result of the complex genetic and transcriptional machinery reviewed above. These proteins, usually referred to as biomarkers, can be readily measured in the blood compartment, thus providing a clinically feasible strategy for early stratication of patients. For example, IL-8 can be readily and rapidly measured in small-volume blood samples. Recently, IL-8 was found to be a robust outcome biomarker in children with septic shock.87 Specically, an IL-8 level measured within 24 hours of admission to the PICU was found to have a

USING EXPRESSION DATA TO DRIVE STRATIFICATION OF PEDIATRIC SEPTIC SHOCK

Although a certain gene-expression pattern seems to predict a subclass of patients with worse outcomes from septic shock, it is also clear that substantial heterogeneity in gene expression occurs across a broad spectrum of clinical presentations that fulll septic-shock criteria. We hypothesize that the failure of the vast majority of interventional clinical trials in septic shock is not related to a fundamental aw of the biological or physiologic principle being tested but, rather, lies in the inability to effectively address the substantial heterogeneity that characterizes this syndrome. Septic shock is a heterogeneous syndrome with the potential to negatively and directly affect all organ systems. This heterogeneity consistently challenges investigators who are attempting to evaluate the efcacy of various experimental interventions. As astutely stated by Marshall,84 a key challenge in the eld is to reduce and manage this heterogeneity by more effectively stratifying patients for the purposes of more rational and effective clinical research and clinical management. The concept of preintervention stratication in sepsis, and its positive impact on the efcacy of an experimental therapy, was recently corroborated by Remick and co-workers85 by using a murine model of polymicrobial sepsis. In

STATE-OF-THE-ART REVIEW ARTICLES

95% negative predictive value for mortality in a derivation cohort of patients. Thus, this particular IL-8 level was able to predict survival in pediatric septic shock, with standard care, with 95% probability. The reliability of this assertion was supported by prospective, formal validation in an independent cohort, which demonstrated an identically robust negative predictive value.87 A similar observation (98% negative predictive value for mortality) was found by measuring chemokine ligand 4 serum levels on admission to the PICU for septic shock.88 It has been proposed that these types of sepsis-outcome biomarkers (ie, biomarkers that have high negative predictive values for mortality) could be used to stratify patients who are eligible for interventional septic-shock trials.87,88 Patients who have a high likelihood of survival but otherwise meet entry criteria for a given interventional trial could be excluded from the trial on the basis of these biomarkers. Such a stratication strategy would serve to derive a study population with a more optimal risk-to-benet ratio, thus improving the ability to demonstrate efcacy for a given test agent. This type of strategy would be particularly useful for a test agent that carries morethan-minimal risk. Although single biomarker-based patient stratication is clinically appealing, it may not be sufciently robust to meet all clinical and research needs. Indeed, the aforementioned studies that involved IL-8 and chemokine ligand 4 had clinically unacceptable specicities, sensitivities, and positive predictive values relative to the high negative predictive values. To serve a wide range of clinical and research needs, the ideal biomarker-based stratication tool would simultaneously have high specicity, high sensitivity, high positive predictive
PEDIATRICS Volume 125, Number 6, June 2010

TABLE 1 Proposed Gene List of Biomarkers for Sepsis/Septic Shock Derived From Array-Based
Determination of Gene-Expression Patterns
GenBank No. NM002983 NM005564 NM002424 NM020415 NM003246 J03189 NM005346 NM000607 NM002343 NM001972 NM000575 AL133001 NM006682 NM000584 NM002984 Ofcial Gene Symbol CCL3 LCN2 MMP8 RETN THBS1 GZMB HSPA1B ORM1 LTF ELA2 IL1A SULF2 FGL2 IL8 CCL4 Description/Gene Name Chemokine (C-C motif) ligand 3, MIP1 Lipocalin 2 (oncogene 24p3), neutrophil gelatinaseassociated lipocalin Matrix metallopeptidase 8 (neutrophil collagenase) Resistin Thrombospondin 1 Granzyme B (cytotoxic T-lymphocyte-associated serine esterase 1) Heat-shock 70-kDa protein 1B Orosomucoid 1 Lactotransferrin Elastase 2, neutrophil IL-1 Sulfatase 2 Fibrinogen-like 2 IL-8 Chemokine (C-C motif) ligand 4, MIP1

value, and high negative predictive value. Given the biological complexity of pediatric septic shock, a stratication strategy based on a panel of multiple biomarkers has more potential to meet the needs of an ideal biomarkerbased stratication tool. We recently launched a multi-institutional study to derive and validate a multibiomarkerbased stratication tool for pediatric septic shock. The foundation of this study is a panel of 15 biomarkers (Table 1) derived from a genome-wide expression database that identied differences in outcomes among nearly 100 children with septic shock (unpublished data). The database focused on the gene-expression patterns representing the rst 24 hours of admission to the PICU, which is an ideal time frame for clinically useful stratication. An initial working list of candidate biomarkers was systematically and objectively derived by using 2 complementary statistical tools to determine genes differentially regulated between survivors and nonsurvivors, and then classprediction modeling targeted at identication of survivor and nonsurvivor classes. The nal panel of 15 biomarkers was rened from the

initial working list on the basis of biological plausibility in the context of sepsis and the ability to readily measure the biomarkers (proteins) in blood samples. The resulting list will be used to derive the pediatric sepsis biomarker risk model (PERSEVERE [Pediatric Sepsis Biomarker Risk Model]) that is intended to predict outcome and illness severity for patients with septic shock. This recently launched, multi-institutional study provides the unprecedented opportunity to test the hypothesis that biomarker determination, informed by previous gene-expression studies, can provide a real-time decision and stratication tool that will affect the care of children with septic shock.

CONCLUSIONS
Clinicians and investigators have enabled improved observational and interventional studies by dening consensus denitions that help identify the pediatric cohort affected by sepsis. Through these observations, it is clear that the neonates and childs host response to an invading pathogen is an extraordinarily complex biological process. This process will be inuenced by both genetic
1255

and epigenetic mechanisms that ultimately impact the gene-expression patterns observed among affected hosts. It is hoped that understanding how these patterns vary according to outcome inform clinical investigators in identifying complementary stratication strategies (eg, realtime biomarker determination) as REFERENCES
1. Wong HR, Shanley TP. Genetic Basis of Acute Lung Injury. New York, NY: Saunders; 2007 2. Heller MJ. DNA microarray technology: devices, systems, and applications. Annu Rev Biomed Eng. 2002;4:129 153 3. Cooke GS, Hill AV. Genetics of susceptibility to human infectious disease. Nat Rev Genet. 2001;2(12):967977 4. Srensen TI, Nielsen GG, Andersen PK, Teasdale TW. Genetic and environmental inuences on premature death in adult adoptees. N Engl J Med. 1988;318(12):727732 5. Arcaroli J, Fessler MB, Abraham E. Genetic polymorphisms and sepsis. Shock. 2005; 24(4):300 312 6. Waterer GW, Wunderink RG. Science review: genetic variability in the systemic inammatory response. Crit Care. 2003;7(4): 308 314 7. Holmes CL, Russell JA, Walley KR. Genetic polymorphisms in sepsis and septic shock: role in prognosis and potential for therapy. Chest. 2003;124(3):11031115 8. Stu ber F, Klaschik S, Lehmann LE, Schewe JC, Weber S, Book M. Cytokine promoter polymorphisms in severe sepsis. Clin Infect Dis. 2005;41(suppl 7):S416 S420 9. Barber RC, Chang LY, Arnoldo BD, et al. Innate immunity SNPs are associated with risk for severe sepsis after burn injury. Clin Med Res. 2006;4(4):250 255 10. Bone RC. Sepsis and coagulation: an important link. Chest. 1992;101(3):594 596 11. Deitch EA. Animal models of sepsis and shock: a review and lessons learned. Shock. 1998;9(1):111 12. Flierl MA, Schreiber H, Huber-Lang MS. The role of complement, C5a and its receptors in sepsis and multiorgan dysfunction syndrome. J Invest Surg. 2006;19(4):255265 13. Funk DJ, Parrillo JE, Kumar A. Sepsis and septic shock: a history. Crit Care Clin. 2009; 25(1):83101, viii 14. Oberholzer A, Oberholzer C, Moldawer LL. Sepsis syndromes: understanding the role of innate and acquired immunity. Shock. 2001;16(2):8396

well as novel pathways for therapeutic targeting. Only through successful collaboration among multiple centers will mechanistic, diagnostic, and therapeutic studies be executed, thereby affording our ability to substantially impact the outcomes of children affected by this complex clinical entity.

ACKNOWLEDGMENTS This work was supported by National Institutes of Health grants K12HD047349 (to Dr Cornell), R01GM064619 and 1RC1HL100474-01 (to Dr Wong), and RO1HL097361, RO1GM066839, and UL1RR024986 (to Dr Shanley) and K08GM077432 and R03HD058246 (to Dr Wheeler).

15. Weighardt H, Holzmann B. Role of Toll-like receptor responses for sepsis pathogenesis. Immunobiology. 2007;212(9 10): 715722 16. Arbour NC, Lorenz E, Schutte BC, et al. TLR4 mutations are associated with endotoxin hyporesponsiveness in humans. Nat Genet. 2000;25(2):187191 17. Schwartz DA. The role of TLR4 in endotoxin responsiveness in humans. J Endotoxin Res. 2001;7(5):389 393 18. Lorenz E, Mira JP, Frees KL, Schwartz DA. Relevance of mutations in the TLR4 receptor in patients with gram-negative septic shock. Arch Intern Med. 2002;162(9): 1028 1032 19. Agnese DM, Calvano JE, Hahm SJ, et al. Human Toll-like receptor 4 mutations but not CD14 polymorphisms are associated with an increased risk of gram-negative infections. J Infect Dis. 2002;186(10):15221525 20. Karoly E, Fekete A, Banki NF, et al. Heat shock protein 72 (HSPA1B) gene polymorphism and Toll-like receptor (TLR) 4 mutation are associated with increased risk of urinary tract infection in children. Pediatr Res. 2007;61(3):371374 21. Read RC, Pullin J, Gregory S, et al. A functional polymorphism of Toll-like receptor 4 is not associated with likelihood or severity of meningococcal disease. J Infect Dis. 2001; 184(5):640 642 22. Allen A, Obaro S, Bojang K, et al. Variation in Toll-like receptor 4 and susceptibility to group A meningococcal meningitis in Gambian children. Pediatr Infect Dis J. 2003; 22(11):1018 1019 23. Hubacek JA, Stu ber F, Fro hlich D, et al. The common functional C(-159)T polymorphism within the promoter region of the lipopolysaccharide receptor CD14 is not associated with sepsis development or mortality. Genes Immun. 2000;1(6):405 407 24. Gibot S, Cariou A, Drouet L, Rossignol M, Ripoll L. Association between a genomic polymorphism within the CD14 locus and septic

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

shock susceptibility and mortality rate. Crit Care Med. 2002;30(5):969 973 Heesen M, Bloemeke B, Schade U, Obertacke U, Majetschak M. The -260 C3 T promoter polymorphism of the lipopolysaccharide receptor CD14 and severe sepsis in trauma patients. Intensive Care Med. 2002;28(8): 11611163 Gu W, Shan YA, Zhou J, et al. Functional signicance of gene polymorphisms in the promoter of myeloid differentiation-2. Ann Surg. 2007;246(1):151158 Kutukculer N, Yeniay BS, Aksu G, Berdeli A. Arg753Gln polymorphism of the human Tolllike receptor-2 gene in children with recurrent febrile infections. Biochem Genet. 2007; 45(7 8):507514 Tabel Y, Berdeli A, Mir S. Association of TLR2 gene Arg753Gln polymorphism with urinary tract infection in children. Int J Immunogenet. 2007;34(6):399 405 Lorenz E, Mira JP, Cornish KL, Arbour NC, Schwartz DA. A novel polymorphism in the Toll-like receptor 2 gene and its potential association with staphylococcal infection. Infect Immun. 2000;68(11):6398 6401 van Deventer SJ. Cytokine and cytokine receptor polymorphisms in infectious disease. Intensive Care Med. 2000;26(suppl 1): S98 S102 Mira JP, Cariou A, Grall F, et al. Association of TNF2, a TNF-alpha promoter polymorphism, with septic shock susceptibility and mortality: a multicenter study. JAMA. 1999; 282(6):561568 Nadel S, Newport MJ, Booy R, Levin M. Variation in the tumor necrosis factor-alpha gene promoter region may be associated with death from meningococcal disease. J Infect Dis. 1996;174(4):878 880 McArthur JA, Zhang Q, Quasney MW. Association between the A/A genotype at the lymphotoxin-alpha250 site and increased mortality in children with positive blood cultures. Pediatr Crit Care Med. 2002;3(4): 341344 Read RC, Cannings C, Naylor SC, et al. Varia-

1256

CORNELL et al

STATE-OF-THE-ART REVIEW ARTICLES

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

tion within genes encoding interleukin-1 and the interleukin-1 receptor antagonist inuence the severity of meningococcal disease. Ann Intern Med. 2003;138(7):534 541 Endler G, Marculescu R, Starkl P, et al. Polymorphisms in the interleukin-1 gene cluster in children and young adults with systemic meningococcemia. Clin Chem. 2006;52(3): 511514 Lehrnbecher T, Bernig T, Hanisch M, et al. Common genetic variants in the interleukin-6 and chitotriosidase genes are associated with the risk for serious infection in children undergoing therapy for acute myeloid leukemia. Leukemia. 2005; 19(10):17451750 Michalek J, Svetlikova P, Fedora M, et al. Interleukin-6 gene variants and the risk of sepsis development in children. Hum Immunol. 2007;68(9):756 760 Balding J, Healy CM, Livingstone WJ, et al. Genomic polymorphic proles in an Irish population with meningococcaemia: is it possible to predict severity and outcome of disease? Genes Immun. 2003;4(8):533540 Artifoni L, Negrisolo S, Montini G, et al. Interleukin-8 and CXCR1 receptor functional polymorphisms and susceptibility to acute pyelonephritis. J Urol. 2007;177(3): 11021106 Westendorp RG, Hottenga JJ, Slagboom PE. Variation in plasminogen-activatorinhibitor-1 gene and risk of meningococcal septic shock. Lancet. 1999;354(9178): 561563 Haralambous E, Hibberd ML, Hermans PW, Ninis N, Nadel S, Levin M. Role of functional plasminogen-activator-inhibitor-1 4G/5G promoter polymorphism in susceptibility, severity, and outcome of meningococcal disease in Caucasian children. Crit Care Med. 2003;31(12):2788 2793 Geishofer G, Binder A, Muller M, et al. 4G/5G promoter polymorphism in the plasminogen-activator-inhibitor-1 gene in children with systemic meningococcaemia. Eur J Pediatr. 2005;164(8):486 490 Hermans PW, Hibberd ML, Booy R, et al. 4G/5G promoter polymorphism in the plasminogen-activator-inhibitor-1 gene and outcome of meningococcal disease. Meningococcal Research Group. Lancet. 1999; 354(9178):556 560 Binder A, Endler G, Rieger S, et al. Protein C promoter polymorphisms associate with sepsis in children with systemic meningococcemia. Hum Genet. 2007;122(2):183190 van Sorge NM, van der Pol WL, van de Winkel JG. FcgammaR polymorphisms: implications for function, disease susceptibility

46.

47.

48.

49.

50.

51.

52.

53. 54.

55.

56. 57.

58.

and immunotherapy. Tissue Antigens. 2003; 61(3):189 202 Bredius RG, Derkx BH, Fijen CA, et al. Fc gamma receptor IIa (CD32) polymorphism in fulminant meningococcal septic shock in children. J Infect Dis. 1994; 170(4):848 853 Domingo P, Muniz-Diaz E, Baraldes MA, et al. Associations between Fc gamma receptor IIA polymorphisms and the risk and prognosis of meningococcal disease. Am J Med. 2002;112(1):19 25 Platonov AE, Kuijper EJ, Vershinina IV, et al. Meningococcal disease and polymorphism of FcgammaRIIa (CD32) in late complement component-decient individuals. Clin Exp Immunol. 1998;111(1):97101 Drogari-Apiranthitou M, Fijen CA, Thiel S, et al. The effect of mannan-binding lectin on opsonophagocytosis of Neisseria meningitidis. Immunopharmacology. 1997;38(12): 9399 van der Pol WL, Huizinga TW, Vidarsson G, et al. Relevance of Fcgamma receptor and interleukin-10 polymorphisms for meningococcal disease. J Infect Dis. 2001;184(12): 1548 1555 Fijen CA, Bredius RG, Kuijper EJ. Polymorphism of IgG Fc receptors in meningococcal disease. Ann Intern Med. 1993;119(7 pt 1): 636 Michalek J, Svetlikova P, Fedora M, et al. Bactericidal permeability increasing protein gene variants in children with sepsis. Intensive Care Med. 2007;33(12):2158 2164 Stalets E, Wong HR. Critically associating. Crit Care Med. 2009;37(4):14921493 Cooper DN, Nussbaum RL, Krawczak M. Proposed guidelines for papers describing DNA polymorphism-disease associations. Hum Genet. 2002;110(3):207208 Wong HR, Shanley TP, Sakthivel B, et al; Genomics of Pediatric SIRS/Septic Shock Investigators. Genome-level expression proles in pediatric septic shock indicate a role for altered zinc homeostasis in poor outcome. Physiol Genomics. 2007;30(2): 146 155 Rink L, Haase H. Zinc homeostasis and immunity. Trends Immunol. 2007;28(1):1 4 Knoell DL, Julian MW, Bao S, et al. Zinc deciency increases organ damage and mortality in a murine model of polymicrobial sepsis. Crit Care Med. 2009;37(4): 1380 1388 Shanley TP, Cvijanovich N, Lin R, et al. Genome-level longitudinal expression of signaling pathways and gene networks in pediatric septic shock. Mol Med. 2007; 13(9 10):495508

59. Ayala A, Chaudry IH. Immune dysfunction in murine polymicrobial sepsis: mediators, macrophages, lymphocytes and apoptosis. Shock. 1996;6(suppl 1):S27S38 60. Felmet KA, Hall MW, Clark RS, Jaffe R, Carcillo JA. Prolonged lymphopenia, lymphoid depletion, and hypoprolactinemia in children with nosocomial sepsis and multiple organ failure. J Immunol. 2005;174(6): 37653772 61. Hotchkiss RS, Swanson PE, Knudson CM, et al. Overexpression of Bcl-2 in transgenic mice decreases apoptosis and improves survival in sepsis. J Immunol. 1999;162(7): 4148 4156 62. Hotchkiss RS, Swanson PE, Freeman BD, et al. Apoptotic cell death in patients with sepsis, shock, and multiple organ dysfunction. Crit Care Med. 1999;27(7):1230 1251 63. Hotchkiss RS, Tinsley KW, Karl IE. Role of apoptotic cell death in sepsis. Scand J Infect Dis. 2003;35(9):585592 64. Hotchkiss RS, Karl IE. The pathophysiology and treatment of sepsis. N Engl J Med. 2003; 348(2):138 150 65. Hotchkiss RS, Coopersmith CM, Karl IE. Prevention of lymphocyte apoptosis: a potential treatment of sepsis? Clin Infect Dis. 2005; 41(suppl 7):S465S469 66. Hotchkiss RS, Osmon SB, Chang KC, Wagner TH, Coopersmith CM, Karl IE. Accelerated lymphocyte death in sepsis occurs by both the death receptor and mitochondrial pathways. J Immunol. 2005;174(8): 5110 5118 67. Hotchkiss RS, McConnell KW, Bullok K, et al. TAT-BH4 and TAT-Bcl-xL peptides protect against sepsis-induced lymphocyte apoptosis in vivo. J Immunol. 2006;176(9): 54715477 68. Cvijanovich N, Shanley TP, Lin R, et al. Validating the genomic signature of pediatric septic shock. Physiol Genomics. 2008;34(1): 127134 69. Pachot A, Cazalis MA, Venet F, et al. Decreased expression of the fractalkine receptor CX3CR1 on circulating monocytes as new feature of sepsis-induced immunosuppression. J Immunol. 2008;180(9): 6421 6429 70. Pachot A, Lepape A, Vey S, Bienvenu J, Mougin B, Monneret G. Systemic transcriptional analysis in survivor and non-survivor septic shock patients: a preliminary study. Immunol Lett. 2006;106(1):6371 71. Tang BM, McLean AS, Dawes IW, Huang SJ, Lin RC. The use of gene-expression proling to identify candidate genes in human sepsis. Am J Respir Crit Care Med. 2007;176(7): 676 684

PEDIATRICS Volume 125, Number 6, June 2010

1257

72. Tang BM, McLean AS, Dawes IW, Huang SJ, Lin RC. Gene-expression proling of peripheral blood mononuclear cells in sepsis. Crit Care Med. 2009;37(3):882 888 73. Wong HR, Cvijanovich N, Allen GL, et al. Genomic expression proling across the pediatric systemic inammatory response syndrome, sepsis, and septic shock spectrum. Crit Care Med. 2009;37(5):1558 1566 74. Wong HR, Cvijanovich N, Lin R, et al. Identication of pediatric septic shock subclasses based on genome-wide expression proling. BMC Med. 2009;7:34 75. Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A. An operational denition of epigenetics. Genes Dev. 2009;23(7):781783 76. Jenuwein T, Allis CD. Translating the histone code. Science. 2001;293(5532):1074 1080 77. Reik W. Stability and exibility of epigenetic gene regulation in mammalian development. Nature. 2007;447(7143):425 432 78. Wysocka J, Swigut T, Milne TA, et al. WDR5 associates with histone H3 methylated at K4 and is essential for H3 K4 methylation and

vertebrate development. Cell. 2005;121(6): 859 872 79. Cao S, Fernandez-Zapico ME, Jin D, et al. KLF11-mediated repression antagonizes Sp1/sterol-responsive element-binding protein-induced transcriptional activation of caveolin-1 in response to cholesterol signaling. J Biol Chem. 2005;280(3):19011910 80. Dou Y, Milne TA, Ruthenburg AJ, et al. Regulation of MLL1 H3K4 methyltransferase activity by its core components. Nat Struct Mol Biol. 2006;13(8):713719 81. Su RC, Becker AB, Kozyrskyj AL, Hayglass KT. Epigenetic regulation of established human type 1 versus type 2 cytokine responses. J Allergy Clin Immunol. 2008;121(1): 57.e3 63.e3 82. Wen H, Dou Y, Hogaboam CM, Kunkel SL. Epigenetic regulation of dendritic cell-derived interleukin-12 facilitates immunosuppression after a severe innate immune response. Blood. 2008;111(4):17971804 83. Ishii M, Wen H, Corsa CA, et al. Epigenetic regulation of the alternatively activated 84.

85.

86.

87.

88.

macrophage phenotype. Blood. 2009; 114(15):3244 3254 Marshall JC. Sepsis: rethinking the approach to clinical research. J Leukoc Biol. 2008;83(3):471 482 Osuchowski MF, Connett J, Welch K, Granger J, Remick DG. Stratication is the key: inammatory biomarkers accurately direct immunomodulatory therapy in experimental sepsis. Crit Care Med. 2009;37(5): 15671573 Warren HS, Elson CM, Hayden DL, et al. A genomic score prognostic of outcome in trauma patients. Mol Med. 2009;15(7 8): 220 227 Wong HR, Cvijanovich N, Wheeler DS, et al. Interleukin-8 as a stratication tool for interventional trials involving pediatric septic shock. Am J Respir Crit Care Med. 2008; 178(3):276 282 Nowak JE, Wheeler DS, Harmon KK, Wong HR. Admission chemokine (C-C motif) ligand 4 levels predict survival in pediatric septic shock. Pediatr Crit Care Med. 2010;11(2): 213216

A Unique Approach to Medication Compliance: Only about half of patients prescribed medication for a chronic condition are still taking it a year later, according to an article in The Wall Street Journal (Mathews AW, February 28, 2010). As a result a company in Saint Louis is about to test an electronic top for a standard pill bottle called a GlowCap. This cap is equipped with a wireless transmitter such that when its time for a dose of medicine, the GlowCap is programmed to give off a pulsing orange light for an hour, and after an hour, it also starts beeping every few minutes in arpeggios that become more complicated and insistent. If that doesnt work, the high-tech cap then sends an automated phone or text message to patients and can also let the doctor know how often the medication is or is not taken. While the cost of the device is not provided, this may make taking medicine an easier pill to remember to swallow.
Noted by JFL, MD

1258

CORNELL et al