Secondary Logo

Journal Logo

Clinical Aspects

Identifying Key Regulatory Genes in the Whole Blood of Septic Patients to Monitor Underlying Immune Dysfunctions

Parnell, Grant P.*†; Tang, Benjamin M.*; Nalos, Marek*; Armstrong, Nicola J‡§; Huang, Stephen J.*; Booth, David R.; McLean, Anthony S.*

Author Information
doi: 10.1097/SHK.0b013e31829ee604

INTRODUCTION

There is a recent paradigm shift in our understanding of sepsis. Immunosuppression is increasingly recognized as the overriding host response observed in septic patients (1, 2). Consequently, immunostimulatory therapy has emerged as a promising new treatment modality in drug development for sepsis (3). During the last 5 years, clinical trials aiming at reversal of immunosuppression have produced encouraging results, providing evidence that reversal of immunosuppression could improve clinical outcomes (4, 5). With this new focus on immunotherapy, it has become increasingly important that researchers develop reliable tools to measure immunosuppression at the bedside to guide the use of immunotherapy.

Measurement of immunosuppression in septic patients at the bedside is difficult. This is because sepsis causes multiple dysfunctions at both molecular and cellular levels. For example, molecular level changes include decreased expression of antigen-presenting HLA-DR (6); reduced cytokine production by leukocytes on ex vivo stimulation by LPS (7); and increased production of programmed death 1 (PD-1), cytotoxic T lymphocyte–associated antigen 4, and B- and T-lymphocyte attenuator molecules (1). Changes at a cellular level include lymphocyte depletion (2), an increase in regulatory T cells (8, 9), and a shift in the type 1/2 helper T-cell balance (10). Measurement of these complex immune dysfunctions requires the use of specialized assays that are not readily available in the clinical setting (e.g., assay for monocytic HLA-DR expression), making them impractical for use in routine clinical decision making.

This study tests the hypothesis that a whole-blood gene expression assay can capture complex immune dysfunctions in septic patients. Whole blood is easily accessible in critically ill patients and provides an ideal biological sample for monitoring immune dysfunction. However, whole blood contains a heterogeneous mix of immune cell subsets. The compositions of these subsets undergo highly dynamic changes throughout different phases of sepsis and can also be widely different between individuals under normal physiological conditions. It is therefore unknown whether a whole-blood gene expression assay can reliably capture the full range of immune dysregulation during sepsis. Here we have used microarray analysis on the whole blood of septic patients to identify key genes that could be used as surrogate markers of immunosuppression. In addition, we assessed if changes in the expression levels of these genes reflect clinical outcome (death/survival) in septic patients.

MATERIALS AND METHODS

Study design

This is a prospective cohort study that took place in the combined medical and surgical intensive care unit (ICU) of a teaching hospital in Sydney, Australia, in June 2008 through January 2011. The study was approved by the Sydney West Area Health Service Human Research Ethics Committee, and informed written consent was obtained from all patients or their relatives.

Thirty-five septic patients and 18 healthy controls were recruited into our study (Table 1). Sepsis was defined as documented bacterial infection in addition to the presence of at least two of the following four clinical criteria: (a) fever or hypothermia (temperature >100.4°F [38°C] or <96.8°F [36°C]); (b) tachycardia (>90 beats/min); (c) tachypnea (>20 breaths/min or PaCO2 <4.3 kPa [32 mmHg]) or the need for mechanical ventilation; (d) an altered white blood cell count of more than 12,000 cells/μL, less than 4,000 cells/μL, or the presence of more than 10% band forms (11). The inclusion criteria required microbiological pathology results confirming bacterial infection and consensus by the consulting physician that sepsis was the cause of the patient’s ICU stay. Clinical characteristics, including Acute Physiology and Chronic Health Evaluation II ([APACHE II] an integer score based on disease severity [12]), age, sex, length of ICU stay, and mortality, were collected from all included patients (Table 1).

Table 1
Table 1:
Characteristics of the individuals included in the study

Gene expression profiling

To screen the transcriptome for candidate genes associated with immune dysfunction, we performed microarray analysis on the whole blood of patients with confirmed sepsis. To enable longitudinal gene expression profiling, a whole-blood sample was collected daily for a maximum of 5 days. The first sample from each patient was collected within the initial 24 h of admission to the ICU, henceforth referred to as day 1. As a control cohort, whole-blood samples were collected from healthy individuals at two time points, 5 days apart. Samples were collected between the hours of 8 am and 1 pm to minimize the effect of diurnal variation (13). Whole-blood samples were collected into PAXgene tubes (PreAnalytiX, GmbH, Switzerland) and immediately stored at -20°C. RNA extraction was performed using the standard protocol in batches of 12 to 24 samples at a time (PAXgene Blood RNA kit; Qiagen, Germany). RNA quality was analyzed using the Agilent 2100 Bioanalyser (Agilent Technologies, Santa Clara, Calif), and all samples obtained an RNA integrity number of greater than 6.5, indicating a high sample integrity. Extracted RNA was stored at −80°C until expression profiling using Illumina Sentrix HT-12_v3_BeadChip arrays (Illumina, San Diego, Calif). Sample amplification and labeling were carried out on 200 ng of total RNA using an Illumina TotalPrep Amplification kit (Ambion, Austin, Tex) in batches of 24 samples at a time. Amplified cRNA was assessed using the 2100 Bioanalyser to ensure satisfactory amplification. The samples were then immediately hybridized onto HT-12_v3_BeadChips. Each sample (750 ng) was loaded onto the arrays. The hybridization and washing procedure were identical for each set of arrays processed and, after normalization, no significant batch effects were identified. To minimize experimental artifacts, all of the RNA extraction, sample amplification and labeling, hybridization and washing, and scanning procedures were carried out by the same operator at the same time of day. All microarray data are available on the Gene Expression Omnibus (GEO), in accordance with minimum information about a microarray experiment (MIAME) standards. The microarray data obtained from the healthy controls and the sepsis patients in this study with bacterial pneumonia (n = 16) have been used in a previous publication (14).

Bioinformatic workflow and statistical analyses

Raw data obtained by scanning of the microarray slides were processed using Illumina GenomeStudio V2010.3. Each probe on the array was passed through a filter requiring a detection value of P < 0.0050 in at least one sample to be included in any further analyses. Of the 48,804 probes present on the Illumina HT 12 array, 24,840 probes (henceforth referred to as genes) passed this criterion. Genes that passed the filtering were loaded into BRB ArrayTools (10) where quantile normalization and log transformation of the data were applied. Normalized and log-transformed microarray data were imported into R (v2.12). Genes with a low variance across all samples, defined to be less than the median, were removed from the data set. This left 12,420 genes to be used for statistical analyses.

Validation of the microarray hybridizations was performed by measuring the expression relative to GAPDH (glyceraldehyde 3-phosphate dehydrogenase) for a subset of genes using quantitative reverse transcriptase polymerase chain reaction (qRT-PCR). The R-squared values obtained when comparing (qRT-PCR) and microarray relative fold changes ranged from 0.67 to 0.83, indicating strong concordance between the two platforms.

To compare the whole-blood longitudinal profile of septic patients with that of healthy controls, a linear mixed model was fit to each gene using the R library lme4. Phenotype (septic or healthy), day of sample, and age were all included in the model as independent variables. In addition, Patient ID was included as an independent variable to use the repeated sampling during the 5 days for each individual. To account for the known increase in the proportion of neutrophils in the whole blood of septic patients, neutrophil proportion was also included as an independent variable. Lymphocyte proportion was not able to be fitted to the model as it was confounded by phenotype. This enabled selection of genes that are differentially expressed in sepsis patients compared with healthy controls after accounting for each of the other terms in the linear mixed model (age, day, patient ID). Values of P were adjusted for multiple testing using the Benjamini and Hochberg False Discovery Rate (FDR) method (R library multitest). An FDR of 5% was used as the cutoff for genes deemed to be differentially expressed between the two classes. The differentially expressed gene lists were then uploaded into GeneGo Metacore and tested for overrepresentation in known biological pathways. Significance of all additional comparisons between septic patients and healthy controls was determined using unpaired t tests with Welch correction (assuming unequal variance).

Immune cell deconvolution of whole-blood gene expression data

To elucidate which particular immune cells are likely contributing to the immune dysregulation gene signature in sepsis, we performed an analysis referred to as immune cell deconvolution using two separate data sets. The first data set consisted of 17 characterized human immune cell subsets assayed by microarray and published on the ImmGen Web site (15). The ImmGen database was used to interrogate differentially expressed immune system response genes for cell subset representation. A gene was included in this analysis if it was highly expressed in four or fewer of the 17 cell subsets characterized in ImmGen. This enabled deconvolution of genes into particular cell subsets that were overrepresented or underrepresented. Fisher exact test was used to compare the immune cell subset expression of the upregulated compared with the downregulated immune system response genes.

The second data set used consisted of 20 immune cell subsets isolated from healthy controls and assayed using next-generation RNA sequencing (RNAseq). Use of this data set enabled clustering of the expression levels and visualization by heat map for the immune response genes upregulated and downregulated in sepsis. RNAseq was performed for the following immune cell subsets: peripheral blood mononuclear cell (PBMC) neutrophils, CD4, CD4+CD45RA+, CD4+CD45RO+, and CD8+ T cells purified ex vivo by magnetic beads, Th1, Th2, Th17, Tregs cultured in vitro, B cells, natural killer (NK) cells, plasmacytoid dendritic cells, myeloid dendritic cells, monocytes, monocyte-derived dendritic cells (MDDCs), monocyte-derived macrophages (MDMs), MDDCs treated at day 6 with IFN-γ (1,000 U/mL) to become DC1s (IL12hi), and MDDCs treated at day 6 with IFN-β to become DC2s (IL10hi). RNA was isolated from these cell subsets using Qiagen RNeasy Mini kit, and RNA was prepared for sequencing on Illumina HiSeq 2000 (Illumina, San Diego, Calif) using the Illumina TruSeq RNA sample preparation kit V1 (Illumina, San Diego, Calif). Raw sequence data were aligned to the UCSC human reference genome (hg19) using the Tophat software package (16). Aligned sequencing reads were summarized to counts per gene using the RAEM procedure (17) and RPKM values were calculated in SAMMate (18) (v 2.6.1). RPKM values were transformed by quantile normalization before visualization. Visualization was performed by clustering the dysregulated immune response genes using 1 minus correlation and average linkage.

RESULTS

Study group demographics

The clinical characteristics for the sepsis and healthy control groups are summarized in Table 1. The mean age of patients in the sepsis cohort was significantly higher than that in the healthy control cohort (P = 0.0011). This was accounted for by including age as a covariate in the linear mixed model analyses. All results henceforth have accounted for the difference in age between these two groups.

Immune response genes were significantly affected by sepsis

To identify immune response genes affected by sepsis, we performed biological pathway analysis on 3,677 genes identified by the linear mixed model as differentially expressed between septic patients and healthy controls (1,923 genes upregulated and 1,754 genes downregulated in sepsis). This analysis showed that the immune response was among the most overrepresented ontology in the differentially expressed genes (Fig. 1A). Within the immune response ontology, a total of 13 biological pathways were overrepresented, confirming that numerous immune response pathways were dysregulated by sepsis (Fig. 1B).

Fig. 1
Fig. 1:
Biological pathway analysis reveals dysregulation in immune response pathways. A, Overrepresented pathway ontologies in genes dysregulated in sepsis. B, Overrepresentation of pathways related to immune response. The gene lists used in this analysis consist of genes that are differentially expressed in sepsis patients compared with healthy controls after accounting for each of the other terms in the linear mixed model (age, day, patient ID). Red bars refer to genes upregulated in sepsis compared with healthy controls; blue bars refer to genes downregulated in sepsis compared with healthy controls.

Of these 13 immune response pathways, 12 were downregulated in sepsis and one was upregulated (regulation of granulocyte development pathway was upregulated). Further analysis of the 12 downregulated pathways revealed that these genes reflect reduced T-lymphocyte activation and deficient antigen presentation (Fig. 2). With the exception of the major histocompatibility complex class I molecule, none of the genes present in these downregulated immune response pathways were upregulated in sepsis at 5% FDR.

Fig. 2
Fig. 2:
Downregulation of key genes expressed in antigen-presenting cells and T lymphocytes in the whole blood of sepsis patients. Summary of the antigen presentation and both CD4+ and CD8+ T-lymphocyte signaling pathways overrepresented in sepsis compared with healthy controls. Blue circles indicate that a gene encoding that object is downregulated, with the darkness of the blue color indicating the degree of downregulation. Both upregulated and downregulated genes are present for major histocompatibility complex class I molecule, as indicated by the red and blue colored circle.

Expression levels of immune response genes reflect lymphocyte depletion

In addition to defects in antigen presentation and T-cell activation, another major contributor of immunosuppression in sepsis is the depletion of lymphocytes, as demonstrated extensively in established literature (2, 19–21). To examine if lymphocyte depletion has contributed to the downregulation of the immune pathways previously described, we carried out two independent analyses. The first was immune cell deconvolution, a technique that used the publicly available ImmGen database (15) to analyze immune cell subset composition of our whole-blood gene expression data. This process, previously used to identify genes contributing to T-cell activation in multiple sclerosis (22) and upregulation of T-helper cell–expressed genes in influenza pneumonia (14), enabled us to determine which immune cell subsets are likely contributing to changes in the immune response genes (Fig. 3). This analysis showed overrepresentation of downregulated immune response genes in T lymphocytes (P = 0.003), whereas the upregulated immune response genes were overrepresented in the neutrophil subset (P < 0.0001). Other cell types that were also overrepresented in the downregulated immune response genes were B cells (P = 0.007) and NK cells (P = 0.007). The second analysis used an independent RNAseq data set consisting of 20 immune cell subsets isolated from healthy controls. The RNAseq analysis confirmed that a high proportion of the downregulated immune response genes were expressed in T-lymphocyte cell types (Fig. 4). In contrast, a high proportion of the upregulated immune response genes were more highly expressed in neutrophils and myeloid cell types. To further validate our finding, we analyzed full blood count data for each of the individuals enrolled in the study (Fig. 5 and Figure, Supplementary Digital Content 1, at http://links.lww.com/SHK/A179, which show the neutrophil and lymphocyte cell counts). This analysis confirmed lymphocyte depletion in the sepsis cohort compared with healthy controls and that the difference was consistent throughout 5 days of follow-up.

Fig. 3
Fig. 3:
Representation of cell subsets in the immune response genes upregulated and downregulated in sepsis. Representation of immune cell subsets in the dysregulated immune response genes as inferred from the ImmGen database (15). The P values presented are from Fisher exact test comparing the proportion of each cell type represented in the upregulated as opposed to the downregulated gene lists.
Fig. 4
Fig. 4:
Immune cell subset expression of the immune response genes dysregulated in sepsis. Immune cell subsets isolated from whole blood collected from healthy controls and assayed by RNAseq. Orange color depicts a high expression level, whereas blue depicts a low level of expression. Genes are clustered using 1 minus correlation and average linkage.
Fig. 5
Fig. 5:
Whole-blood composition in sepsis patients compared with healthy controls. For visual comparison across days 1 to 5 in the sepsis cohort, only patients with 5 days of cell count data are included in the illustration.

Changes in immune response genes correlate with clinical outcome

If the identification of dysregulated immune response genes in sepsis is to be clinically useful, they should reflect disease severity and therefore predict clinical outcome. We therefore assessed the association between the expression level of the dysregulated immune response genes and disease progression in septic patients. To achieve this, we summarized the gene expression of the downregulated genes by using a composite score. This composite score, the “Immune Suppression Integer” (ISI) was calculated by adding together the deviation from the mean as a proportion for each downregulated immune response gene (see Table, Supplementary Digital Content 2, http://links.lww.com/SHK/A180, which displays the expression level and fold change for each of the genes used to calculate the ISI). Examination of the relationship between the ISI and disease severity showed that a greater degree of immunosuppression was associated with septic patients who did not survive, whereas a lesser degree of immunosuppression (closer to that of healthy controls) was associated with survival (Fig. 6, and Figure, Supplementary Digital Content 3, http://links.lww.com/SHK/A181, which presents the expression level of the genes contributing to the ISI as a heatmap). Figure 7 shows that a receiver operating characteristic curve separates the sepsis survivors and nonsurvivors with an area under the receiver operating characteristic curve of 0.812. A multiple logistic regression of day 1 samples revealed that the ISI was more strongly associated with survival (P = 0.038; odds ratio, 0.004; and 95% CI, 0.000 – 0.737) than APACHE II (P = 0.138; odds ratio, 0.876; and 95% CI, 0.736 – 1.043) or lymphocyte percentage (P = 0.761; odds ratio, 1.034; and 95% CI, 0.834 – 1.282).

Fig. 6
Fig. 6:
Immune Suppression Integer and clinical outcome across 5 days of ICU stay.
Fig. 7
Fig. 7:
Receiver operating characteristic curve for separation of sepsis survivors and nonsurvivors according to the ISI.

As the downregulated immune response genes used to generate the ISI were shown to be expressed largely by lymphocytes (Figs. 3 and 4), it was expected that lymphocyte count (as a proportion of leukocytes) would also be different between survivors and nonsurvivors. The proportion of lymphocytes was lower in the sepsis nonsurvivor group (see Figure, Supplementary Digital Content 3, http://links.lww.com/SHK/A181, which shows the lymphocyte proportions of whole blood for the 5 days sampled). However, this difference was only significant on days 2 and 3 of the 5 days sampled. The ISI, on the other hand, showed significant differences between survivors and nonsurvivors across days 1 to 4 of the 5 days sampled. This result suggests that the ISI is not only affected by the proportion of lymphocytes present but may also reflect other differences such as activation state of lymphocyte populations (or lack thereof) as well as the increase or decrease of particular lymphocyte subpopulations (e.g., Th1, Th2, Th17, regulatory T).

DISCUSSION

Immunosuppression in sepsis has received increasingly more attention in recent years. This is because of realization that most septic patients present in the late phase of their illness, which is characterized by a state of immunosuppression. This represents a shift in the paradigm, moving away from a traditional focus on perceiving sepsis as a “hyperimmune” state, to a renewed understanding that, in patients with sepsis, inflammation and immunosuppression are proceeding concurrently and for prolonged periods (23, 24). Numerous studies have provided supporting evidence to this new paradigm. Consequently, there has been an increase in the number of experimental antisepsis therapies that aim to restore immune competence in septic patients (3). However, this development is bottlenecked by a lack of reliable methods to accurately quantify immunosuppression at the bedside. Here, we have presented a proof-of-concept study that addresses this important issue. Our study identified a number of genes found in the whole blood of septic patients that corresponds to known immune response pathways, lending support to the contention that a transcriptomic assay is a potential new tool to provide bedside assessment of immune function in sepsis.

Of the 86 downregulated immune response genes identified in our study, many are involved in pathways implicated in sepsis, including lymphocyte depletion, reduced T-lymphocyte activation, and deficient antigen presentation. Several experimental therapies targeting these pathways are currently under development, highlighting the potential usefulness of our approach (3). If validated, these signature genes could be used as a surrogate biomarker to facilitate clinical trials (e.g., when to institute immunotherapy), monitor treatment response (e.g., when does immune function recover), and predict outcome (e.g., death/survival).

The use of immune-specific biomarkers circumvents a long-standing problem in clinical trials of sepsis. For decades, crude physiological markers (blood pressure, body temperature, heart rate) have been used to determine treatment threshold. But these markers do not correlate well with the underlying biological process. Host response markers (e.g., tumor necrosis factor, interleukin 6) have also been used; but they are nonspecific because they are elevated in a variety of inflammatory disorders. Our study represents a practical approach to monitor immune function. It raises the possibility that immune dysfunction could be measured at bedside by assaying a limited number of genes (on a conventional platform such as real-time PCR) using only a small amount of peripheral blood. Such a possibility warrants further validation in future studies.

Several attempts have previously been made at using gene expression data for stratification of sepsis patients into clinically relevant subclasses. For instance, Wong et al. (25) identified a 100-gene signature that separated pediatric septic shock patients into three separate subclasses, with one subclass having a more severe illness and an increased organ failure and mortality rate compared with the other two subclasses. McDunn et al. (26) identified a gene expression pattern from buffy coat samples of ventilated patients that fluctuated with the onset and subsequent recovery from ventilator-associated pneumonia. This was in contrast to a subsequent study in which whole-blood expression profiles were unable to predict or diagnose ventilator-associated pneumonia in a cohort of trauma patients (27). A genomic score summarizing global gene expression has also been shown to be associated with adverse outcome in trauma patients (28). Our present study has taken an alternative approach that was to first identify a set of genes differentially expressed between septic patients and healthy controls across 5 days of ICU stay and subsequently show that a composite score derived from the expression level of these genes was significantly different in sepsis survivors compared with nonsurvivors. These studies show that gene expression profiling for stratification of septic patients shows promise, but each requires further validation in large independent cohorts before use in the clinical setting can be considered.

Our longitudinal analyses focused on genes related to the immune response. This is in light of multiple genome-wide expression studies reporting repression of adaptive immunity genes in both adult and pediatric septic patients (see review by Wong [29]). This decision was also driven by the recent emergence of new antisepsis therapies targeting the immune response. Many of these new therapies aim to augment lymphocyte function (3, 24). Consequently, our analysis was intentionally biased toward immune response genes expressed predominantly in lymphocytes. However, there is equally sound scientific justification for undertaking similar analyses on other ontological groups identified in our data set, such as granulocyte response, inflammatory response, or apoptosis. For example, we identified a list of upregulated genes highly expressed in granulocytes that correlate with disease severity and clinical outcomes (see Figure, Supplementary Digital Content 4, http://links.lww.com/SHK/A182, which shows an immune response integer derived from genes upregulated in sepsis). We have not explored this further because experimental therapies targeting granulocytes are relatively underdeveloped in contrast to immune therapy directed at lymphocyte function.

Because we intended to develop a practical assay that could be used in a standard hospital laboratory, we used whole blood in our study for its easy accessibility in the critically ill patient. This pragmatic approach comes at a cost. First, changes in the circulating immune cells do not necessarily correlate with resident immune cells in vital organs (e.g., spleen, liver, lungs, intestine). Second, the composition of peripheral blood is constantly changing in critically ill patients in terms of the relative proportion and absolute number of each immune cell subset. These changes are determined by stress response, drug therapy, neuroendocrine status, and bone marrow function. The rapidly changing milieu of sepsis further confounds these changes. As a result, assaying immune function by testing peripheral blood faces the challenges of diminishing tissue specificity and poor correlation with underlying biology. Despite these challenges, we found that transcriptome profiling of whole blood did capture several important immune dysfunctions of sepsis (e.g., antigen presentation, T-cell activation, and effects on lymphocyte depletion).

We chose gene expression profiling to assess immune function in this study. In contrast, the conventional way of interrogating immune function is usually protein based (e.g., flow cytometric analysis of cell surface receptors such as CD86) or relying on specific functional assays (e.g., ex vivo LPS challenge of leukocytes). These specialized techniques are important in deciphering the underlying biology of sepsis, but are impractical in a clinical setting. Nevertheless, these traditional immune assays form the gold standard of testing immune competence and provide a benchmark against which to validate our transcriptomics approach. For the purpose of developing a practical bedside assay, however, our transcriptome profiling approach offers several technical advantages over conventional assay. First, it allows a global survey of immune status by interrogating genes involved in multiple cellular pathways, whereas a conventional assay could only interrogate one specific component of the immune system at a time. Second, once identified, marker genes can be quickly assayed by real-time PCR, allowing a more rapid turnaround time than a conventional assay. Third, transcriptional changes can be more dynamic, reflecting moment-to-moment changes at an intracellular level as the host cell responds to stress or infection by rapid upregulation/downregulation of particular RNA species.

The ISI showed significant separation between the sepsis survivor and nonsurvivor groups (Figs. 6 and 7). However, before a composite score such as ISI is considered for use in the clinical situation, it must be validated in an independent cohort of a much larger sample size than present in this study. This will enable the ISI to be more thoroughly compared with other conventional predictors of survival such as the Sequential Organ Failure Assessment score (30), APACHE II (12), and serum lactate level (31). The effect of the presence of comorbidities such as cancer, COPD, recent trauma, and diabetes on the ISI should also be further investigated. It may also be of interest to investigate if a composite score such as the ISI is useful in terms of predicting clinical outcomes other than death, such as secondary infection or organ failure.

Overall, this study provides proof-of-concept evidence that immune dysfunction in sepsis can be identified by gene expression profiling, and this has a potential to be used to monitor immunosuppression in critically ill septic patients and to guide immunomodulatory therapy in clinical trials. Further validation in large independent patient cohorts is needed to test the practical feasibility and clinical utility of transcriptome profiling in assessing immune function in sepsis.

ACKNOWLEDGMENTS

The authors express sincere thanks to the Nepean Hospital ICU research coordination team, led by CNC Leonie Weisbrodt, for patient recruitment and sample collection, and Stephen Schibeci of Westmead Millennium Institute for his laboratory technical support.

ABBREVIATIONS

ICU: intensive care unit

FDR: Benjamini and Hochberg False Discovery Rate

ISI: Immune Suppression Integer

REFERENCES

1. Boomer JS, To K, Chang KC, Takasu O, Osborne DF, Walton AH, Bricker TL, Jarman SD 2nd, Kreisel D, Krupnick AS, et al.: Immunosuppression in patients who die of sepsis and multiple organ failure. JAMA 306: 2594–2605, 2011.
2. Hotchkiss RS, Tinsley KW, Swanson PE, Schmieg RE Jr, Hui JJ, Chang KC, Osborne DF, Freeman BD, Cobb JP, Buchman TG, et al.: Sepsis-induced apoptosis causes progressive profound depletion of B and CD4+ T lymphocytes in humans. J Immunol 166: 6952–6963, 2001.
3. Hotchkiss RS, Opal S: Immunotherapy for sepsis—a new approach against an ancient foe. N Engl J Med 363: 87–89, 2010.
4. Meisel C, Schefold JC, Pschowski R, Baumann T, Hetzger K, Gregor J, Weber-Carstens S, Hasper D, Keh D, Zuckermann H, et al.: Granulocyte-macrophage colony-stimulating factor to reverse sepsis-associated immunosuppression: a double-blind, randomized, placebo-controlled multicenter trial. Am J Respir Crit Care Med 180: 640–648, 2009.
5. Leentjens J, Kox M, Koch RM, Preijers F, Joosten LA, van der Hoeven JG, Netea MG, Pickkers P: Reversal of immunoparalysis in humans in vivo: a double-blind, placebo-controlled, randomized pilot study. Am J Respir Crit Care Med 186: 838–845, 2012.
6. Ditschkowski M, Kreuzfelder E, Rebmann V, Ferencik S, Majetschak M, Schmid EN, Obertacke U, Hirche H, Schade UF, Grosse-Wilde H: HLA-DR expression and soluble HLA-DR levels in septic patients after trauma. Ann Surg 229: 246–254, 1999.
7. Ertel W, Kremer JP, Kenney J, Steckholzer U, Jarrar D, Trentz O, Schildberg FW: Downregulation of proinflammatory cytokine release in whole blood from septic patients. Blood 85: 1341–1347, 1995.
8. Monneret G, Debard AL, Venet F, Bohe J, Hequet O, Bienvenu J, Lepape A: Marked elevation of human circulating CD4+CD25+ regulatory T cells in sepsis-induced immunoparalysis. Crit Care Med 31: 2068–2071, 2003.
9. Brudecki L, Ferguson DA, McCall CE, El Gazzar M: Myeloid-derived suppressor cells evolve during sepsis and can enhance or attenuate the systemic inflammatory response. Infect Immun 80: 2026–2034, 2012.
10. Ferguson NR, Galley HF, Webster NR: T helper cell subset ratios in patients with severe sepsis. Intensive Care Med 25: 106–109, 1999.
11. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, Cohen J, Opal SM, Vincent JL, Ramsay G: 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Intensive Care Med 29: 530–538, 2003.
12. Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: a severity of disease classification system. Crit Care Med 13: 818–829, 1985.
13. Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC, Relman DA, Brown PO: Individuality and variation in gene expression patterns in human blood. Proc Natl Acad Sci U S A 100: 1896–1901, 2003.
14. Parnell G, McLean A, Booth D, Armstrong N, Nalos M, Huang S, Manak J, Tang W, Tam O, Chan S, et al.: A distinct influenza infection signature in the blood transcriptome of patients who presented with severe community acquired pneumonia. Crit Care 16: R157, 2012.
15. Heng TS, Painter MW: The Immunological Genome Project: networks of gene expression in immune cells. Nat Immunol 9: 1091–1094, 2008.
16. Trapnell C, Pachter L, Salzberg SL: TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25: 1105–1111, 2009.
17. Deng N, Puetter A, Zhang K, Johnson K, Zhao Z, Taylor C, Flemington EK, Zhu D: Isoform-level microRNA-155 target prediction using RNA-seq. Nucleic Acids Res 39: e61, 2011.
18. Xu G, Deng N, Zhao Z, Judeh T, Flemington E, Zhu D: SAMMate: a GUI tool for processing short read alignments in SAM/BAM format. Source Code Biol Med 6: 2, 2011.
19. Dahn MS, Whitcomb MP, Lange MP, Jacobs LA: Altered T-lymphocyte subsets in severe sepsis. Am Surg 54: 450–455, 1988.
20. Venet F, Davin F, Guignant C, Larue A, Cazalis MA, Darbon R, Allombert C, Mougin B, Malcus C, Poitevin-Later F, et al.: Early assessment of leukocyte alterations at diagnosis of septic shock. Shock 34: 358–363, 2010.
21. Le Tulzo Y, Pangault C, Gacouin A, Guilloux V, Tribut O, Amiot L, Tattevin P, Thomas R, Fauchet R, Drenou B: Early circulating lymphocyte apoptosis in human septic shock is associated with poor outcome. Shock 18: 487–494, 2002.
22. Gandhi KS, McKay FC, Cox M, Riveros C, Armstrong N, Heard RN, Vucic S, Williams DW, Stankovich J, Brown M, et al.: The multiple sclerosis whole blood mRNA transcriptome and genetic associations indicate dysregulation of specific T cell pathways in pathogenesis. Hum Mol Genet 19: 2134–2143, 2010.
23. Gentile LF, Cuenca AG, Efron PA, Ang D, Bihorac A, McKinley BA, Moldawer LL, Moore FA: Persistent inflammation and immunosuppression: a common syndrome and new horizon for surgical intensive care. J Trauma Acute Care Surg 72: 1491–1501, 2012.
24. Hotchkiss RS, Monneret G, Payen D: Immunosuppression in sepsis: a novel understanding of the disorder and a new therapeutic approach. Lancet Infect Dis 13: 260–268, 2013.
25. Wong HR, Cvijanovich N, Lin R, Allen GL, Thomas NJ, Willson DF, Freishtat RJ, Anas N, Meyer K, Checchia PA, et al.: Identification of pediatric septic shock subclasses based on genome-wide expression profiling. BMC Med 7: 34, 2009.
26. McDunn JE, Husain KD, Polpitiya AD, Burykin A, Ruan J, Li Q, Schierding W, Lin N, Dixon D, Zhang W, et al.: Plasticity of the systemic inflammatory response to acute infection during critical illness: development of the riboleukogram. PLoS ONE 3: e1564, 2008.
27. Textoris J, Loriod B, Benayoun L, Gourraud PA, Puthier D, Albanese J, Mantz J, Martin C, Nguyen C, Leone M: An evaluation of the role of gene expression in the prediction and diagnosis of ventilator-associated pneumonia. Anesthesiology 115: 344–352, 2011.
28. Warren HS, Elson CM, Hayden DL, Schoenfeld DA, Cobb JP, Maier RV, Moldawer LL, Moore EE, Harbrecht BG, Pelak K, et al.: A genomic score prognostic of outcome in trauma patients. Mol Med 15: 220–227, 2009.
29. Wong HR: Clinical review: sepsis and septic shock–the potential of gene arrays. Crit Care 16: 204, 2012.
30. Jones AE, Trzeciak S, Kline JA: The Sequential Organ Failure Assessment score for predicting outcome in patients with severe sepsis and evidence of hypoperfusion at the time of emergency department presentation. Crit Care Med 37: 1649–1654, 2009.
31. Mikkelsen ME, Miltiades AN, Gaieski DF, Goyal M, Fuchs BD, Shah CV, Bellamy SL, Christie JD: Serum lactate is associated with mortality in severe sepsis independent of organ failure and shock. Crit Care Med 37: 1670–1677, 2009.
Keywords:

Sepsis; gene expression; immunosuppression; lymphopenia; immune system

Supplemental Digital Content

© 2013 by the Shock Society