These data represent the most comprehensive analysis of longitudinal gene expression in pediatric septic shock to date. As relatively descriptive data, they are subject to interpretation. Below we provide our pathophysiology-based interpretation of the expression data, being cognizant of alternative interpretations.
The genome-level expression profile during the initial three days of pediatric septic shock is complex and heterogeneous. Between four and five percent of the genes encompassing the entire array were differentially expressed (upregulated or downregulated) on either day one or day three, respectively, compared with controls. Among the differentially expressed genes, 11 percent of the day one differentially expressed genes were unique to day one, and 24 percent of the day three differentially expressed genes were unique to day three. Thus, the genome-level expression profile during the initial three days of pediatric septic shock is characterized by a predominant constant component and a relatively smaller temporal component. The genes contributing to the smaller temporal component, however, appear to be biologically significant in that they contribute to differential expression of signaling pathways and gene networks known to be associated with inflammatory disease states.
The day one and day three datasets contain multiple genes contributing to signaling pathways. Many of the pathways identified within these data sets have been linked previously by the experimental literature to septic shock pathobiology (3
). While the demonstration of some of these pathways does not necessarily represent novel concepts, their presence in the datasets greatly strengthens both the validity and relevance of the overall dataset. For example, we were intrigued to find that day three (but not day one) of pediatric septic shock is characterized by upregulation of genes that participate in insulin-like growth factor (IGF)-, insulin receptor-, and PPAR-signaling. These data indicate an important role for insulin-related biology/metabolism in pediatric septic shock, and are made all the more intriguing by the recent advancements, by Van den Berghe and colleagues, regarding the use of intensive insulin therapy in critically ill adult surgical patients (26
The primary novelty of the signaling pathway-related data lies in the temporal dependence of pathway expression. Several of the upregulated signaling pathways were consistently upregulated from day one to day three. More than half of the upregulated pathways, however, were expressed in a time-dependent manner. In fact, some of the signaling pathways were exclusively expressed on day three. These data could have implications for the timing and duration of future therapeutic strategies centered on any one of these pathways.
The importance of timing and duration of therapy in critically ill patients is well illustrated in the recent follow-up study of intensive insulin therapy in critically ill adult medical patients by Van den Berghe and colleagues (27
). This study involving 1,200 critically ill adults demonstrated that intensive insulin therapy conferred a survival benefit only for patients having an intensive care unit length of stay ≥ three days. In contrast, mortality was greater in patients receiving intensive insulin therapy and having an intensive care unit length of stay < three days. These data indirectly support our observations regarding time-dependent expression of signaling pathways related to insulin biology/metabolism.
Using a pathway-based analytical tool (IPA), the day one and day three datasets were found to contain multiple genes corresponding to distinct networks. The biological importance of these networks is broadly supported at two levels. First, all of the networks selected for presentation had exceptionally high significance scores based on the IPA algorithm. Second, many of the functional annotations corresponding to the networks are biologically consistent with the signaling pathways shown in and . Innate immunity and inflammation are the predominant themes of the upregulated gene networks, and these observations are again consistent with some of the current paradigms in the septic shock literature (3
). The network expression patterns were time-dependent, as demonstrated by unique gene networks being expressed on day one and day three, respectively. Thus, the temporal nature of network expression and the implied functionality may also have implications for the timing and duration of future therapeutic strategies.
The most novel feature derived from the network-based analysis is the down-regulation of gene networks related to T cell immunity and the MHC antigen presentation pathway. Importantly, this observation does not appear to be an artifact of relative lymphopenia when analyzed in the context of total lymphocyte counts. The strength of this observation is indicated by the multiple gene networks having these particular annotations (T cell function and the MHC antigen presentation pathway), and the ability to consistently derive these functional annotations using two distinct gene annotation/ontology databases (i.e. D.A.V.I.D., and PANTHER) that are independent from the IPA database. In addition, the analyses focused on discovery of signaling pathways also yielded “T cell receptor signaling” and “antigen presentation” as being predominant themes among the group of genes down-regulated on day one and day three. Collectively, these data indirectly suggest that pediatric septic shock may have a strong component of immune dysfunction secondary to decreased T cell function and antigen presentation.
Our conceptual framework of septic shock pathobiology has evolved over the last decade to include the concepts of immune suppression and “immunoparalysis.” Whereas septic shock has been viewed traditionally as being a reflection of hyperinflammation, it is now thought that septic shock also has a strong, perhaps predominant, anti-inflammatory component that can be manifest as functional immune suppression and the relative inability to effectively clear infection (24
). Our current observation that genes corresponding to interleukin-10 signaling, a major anti-inflammatory pathway (30
), are temporally upregulated in this patient cohort, well supports this anti-inflammatory concept.
Monocyte deactivation related to decreased MHC gene mRNA expression and decreased surface expression of MHC molecules has been demonstrated in patients with septic shock (31
). Regarding lymphocyte dysfunction, Heidecke and colleagues demonstrated that adult patients with intraabdominal infections and septic shock had defective T cell proliferation and T cell-dependent cytokine secretion, consistent with an-ergy/immune suppression (35
). Felmet and colleagues identified prolonged lymphopenia and apoptosis-associated depletion of lymphoid organs as independent risk factors for the development of nosocomial infections and multiple organ failure in critically ill children (36
). Animal studies have documented the requirement of an intact T cell system to adequately combat a septic challenge (29
). More recently, animal-based experiments have demonstrated that experimental septic shock is characterized by widespread T cell apoptosis and that preventing T cell apoptosis positively impacts the outcome of experimental sepsis (38
). Importantly, the concept of T cell apoptosis in human sepsis has been indirectly corroborated by autopsy studies (36
). Thus, while formal studies of T cell function in pediatric septic shock remain to be conducted, our current data support the concept of immune suppression secondary to T cell dysfunction at the transcriptome level and represent a first step toward demonstrating proof of principle on a clinical genomics scale.
Our previous report indicated that day one of pediatric septic shock is characterized by a large scale downregulation of genes that play a direct role in zinc homeostasis or are functionally dependent on zinc homeostasis (6
). We have corroborated this observation and now demonstrate its persistence well into day three of pediatric septic shock. In fact, the current data suggest that the perturbation of genes related to zinc homeostasis is even more profound on day three than on day one. The functional significance of these observations remains to be directly tested and elucidated. However, given the importance of zinc homeostasis for T cell function (47
), coupled with our observations suggesting large scale downregulation of T cell-related genes, the overall data provide the basis for biologically plausible and testable hypotheses surrounding zinc homeostasis and lymphocyte dysfunction in pediatric septic shock.
Many of the limitations inherent to the data in our previous report are applicable to the current data and have been discussed previously (6
), including the possibility that our observations are confounded by differences in the white blood cell populations. Similar to the previous report, we have not found any correlation between our current observations and differential white blood cell counts (at the level of neutrophils, monocytes, and total lymphocytes). We are cognizant, however, that differences in T cell subset numbers (i.e. CD4 cells, CD8 cells, etc.) could account for some of our observations regarding T cell receptor signaling and antigen presentation. Thus, we again maintain that our observations are reasonably representative of genuine genomic responses, rather than sampling artifact related to whole blood-derived RNA.
Two interrelated limitations, more specific to the current data, warrant further discussion. First, the data are descriptive and limited to two time points. The data are, however, unprecedented in scope by addressing the problem of clinical pediatric septic shock using a translational approach at the level of the entire genome. Second, the data require functional validation in the form of directly testing the hypotheses derived from the current observations. We are currently in the process of designing a validation, follow-up study in which we will formally measure lymphocyte function in a large cohort of children with septic shock.
These limitations aside, the current data represent an unprecedented first approximation of longitudinal, genome-level responses of pediatric septic shock. The data demonstrate: 1) the complex nature of this genomic response; 2) a predominance of constant, coordinated gene regulation over the first three days of illness; 3) a relatively small but biologically relevant temporal component of coordinated gene regulation; 4) large scale downregulation of genes corresponding to T cell function and MHC-mediated antigen presentation; and 5) persistent downregulation of a large subset of genes related to zinc homeostasis. Collectively, the data support novel hypotheses that can be tested readily at both the experimental and translational levels.