Lonneke Van Vught

CHAPTER 12

Sepsis, defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection, remains an increasing cause of morbidity and mortality worldwide. In this thesis, we have explored the epidemiology, outcome and host response in critically ill patients admitted with sepsis to, or developing sepsis while on, the intensive care unit (ICU). For this, we have used the Molecular Diagnosis and Risk Stratification of Sepsis (MARS) data collection initiative and biobank study as backbone. Two tertiary teaching hospitals in the Netherlands participated in the inclusion of extensive clinical, microbiological, interventional, outcome and follow up data of 8305 consecutively admitted critically ill patients during a study period of 3 years. Additionally, biomarkers reflective of the host response in sepsis were determined in a subset of this cohort, using both a targeted approach (by measuring soluble plasma protein biomarkers reflective of pathways implicated in sepsis pathogenesis) and an unbiased approach (by analyzing the whole genome expression profiles in whole blood leukocytes). The MARS project aimed to generate rapid and accurate tools providing insight in disease severity, host response and causative pathogens in critically ill patients with sepsis, initiating and guiding individualized therapy and predicting course of disease, severity and outcome. In this thesis, consecutively ICU admitted patients with sepsis, or noninfectious disease as a comparator, were analyzed to provide insight into factors contributing to disease severity, host response and outcome in the setting of two Dutch ICUs.

Immune suppression and secondary infections in sepsis
The recent understanding that sepsis is associated with activation of both pro- and anti-inflammatory pathways concomitantly, has resulted in an increased focus on the possibility of a net immune suppressive phase during sepsis. This phase is characterized by exhaustion, apoptosis and hyporesponsiveness of immune cells, increasing susceptibility to the development of nosocomial infections. These alterations are assumed to be sepsis specific, and not or to a more limited extent detectable in patients with noninfectious critical illness. The subject of immune suppression has become of increasing interest now that the majority of sepsis deaths occur more than one week after admission to the ICU. This so called ‘late sepsis mortality’ has been ascribed to the development of nosocomial infections in immunoparalysed patients. Opposite from the use of anti-inflammatory agents in many unsuccessful sepsis trials, immune stimulation has recently been advocated as a novel treatment strategy for sepsis, targeting immune suppression. However, immune stimulatory therapy could have deleterious effects and should ideally only be used in patients who might benefit. Overall, knowledge on the host response of ICU-acquired infections is essential and the consequential attribution to mortality is important when preventive strategies are initiated. Therefore, in Part I of this thesis we focused on immune suppression in critically ill sepsis patients and the subsequent development of ICU-acquired infections. If the assumption holds true that the immunological features of immune suppression reported in sepsis patients increase the susceptibility to secondary infections and are more prominent than in critically ill patients without sepsis, it is expected that ICU-acquired infections occur more often in patients admitted with sepsis than in patients admitted with a noninfectious condition.

Contrary to this hypothesis, ICU-acquired infections complicated sepsis admissions in 13.5%, and noninfectious admissions in 15.1% (Chapter 2). Additionally, after the first week, ICU-acquired infections occurred with a comparable incidence in noninfectious admissions as in admissions for sepsis and occurred even more often in noninfectious admissions in the first week after ICU admittance. These findings suggest that ICU-acquired infections are not merely a feature of sepsis, but are at least equally prevalent in noninfectious admissions. In both sepsis and noninfectious admissions the incidence of ICU-acquired infections was increased in subgroups of patients with organ failure and shock on admission, suggesting the severity of disease and the consequentially increased length of ICU stay with increased exposure to invasive supportive treatments as most influencing factors. In accordance, SOFA scores were higher in patients who went on to develop an ICU-acquired infection, compared to patients not developing ICU-acquired infections. Additionally, a multivariate competing risk analysis identified increased disease severity (higher APACHE-IV scores) as an independent risk factor for the development of such infections. These findings were confirmed in a subgroup analysis including only sepsis patients with a probable or definite infection likelihood (Chapter 3), where disease severity and the proportion of shock were higher in sepsis patients developing ICU-acquired infections.

Importantly, the overall contribution of ICU-acquired infections to mortality was modest, and the attributable mortality was higher in patients with a noninfectious admission diagnosis than in those with a sepsis admission diagnosis, i.e., 21.1% and 10.9% by day 60 respectively. This finding was consistent in subgroup analyses that included patients with more severe disease on admission (Chapter 2). The relatively low population attributable mortality suggests that other factors (e.g., organ failure and shock) contribute more to sepsis mortality. However, since our study was not designed to relate sepsis to noninfectious admissions, we refrained from directly comparing the two groups because of major differences between them. Still, even in the hallmark study identifying immune suppression in sepsis important differences between sepsis and noninfectious patients were present in terms of length of ICU stay, age and preexisting comorbid conditions. Nonetheless, the important conclusion that ICU-acquired infections are not solely a sepsis related problem should be drawn before pleading for immune stimulatory therapy exclusively in critically ill patients with sepsis. Together these data suggest that immunoparalysis is not an exclusive feature of sepsis and that severity of disease and administration of invasive supportive treatments pose critically ill patients at risk for ICU-acquired infections irrespective of their admission diagnosis. Individualized, rapid and guided therapy should support future research and treatment in the heterogeneity of critical illness.

We did find characteristics of ICU-acquired infections in sepsis patients suggestive of immune suppression. In patients with sepsis, more opportunistic pathogens caused ICU-acquired infections, and more patients developed more than one ICU-acquired infection during ICU stay, when compared with patients with a noninfectious admission diagnosis. Also, in sepsis patients whole blood leukocyte gene expression at the time of an ICU-acquired infection showed decreased expression of genes involved in leukocyte glucose metabolism when compared to the leukocyte transcriptome at ICU admission, suggesting immune suppression (Chapter 2). This difference was not identified in patients with sepsis experiencing noninfectious ICU-acquired complications (acute kidney injury, acute respiratory distress syndrome and acute myocardial infarction). This result suggests that during ICU admission impairment of glucose metabolism in immune cells may contribute to the occurrence of nosocomial infections in patients admitted with sepsis.

To obtain insight into pathogenetic mechanisms initiated shortly after ICU admission for sepsis and subsequently contributing to the development of an ICU-acquired infection, we compared the host response in sepsis patients who acquired such an infection with the host response in those who did not (Chapter 2, Chapter 3 and Chapter 4). First, on admission, an exploratory analysis of gene expression profiles in whole blood leukocytes did not reveal differences between those who did and those who did not develop an ICU-acquired infection (Chapter 2). This comparable leukocyte genomic response with up-regulation of multiple pro- and anti-inflammatory pathways, and down-regulation of adaptive immunity pathways in patients with sepsis on ICU admission, suggests the existence of both hyperinflammation and immunoparalysis. Analyses of the responsiveness of whole blood leukocytes obtained the day after ICU admission to a bacterial stimulus (lipopolysaccharide) did not identify a relation with the subsequent development of ICU-acquired infections (Chapter 4). Critically ill patients showed a decreased capacity to mount a proper cytokine response compared to healthy subjects; however, no differences were found between patients who did and those who did not acquire an infectious complication during ICU stay. These findings together argue against an already identifiable difference between these two groups on admission to the ICU.

However, sepsis patients who developed an ICU-acquired infection did demonstrate an increase in plasma biomarkers providing insight into sepsis pathogenesis compared to those who did not (Chapter 3). The more dysregulated proinflammatory, vascular and procoagulant host response on ICU admission remained when corrected for disease severity and source of infection, and these differences increased even further in the days thereafter. Together, this suggests an increased derailment of the immune response in patients going on to develop a nosocomial infection, and that this aberrant reaction is not captured by genomic or functional analyses of blood leukocytes done in this thesis. Our host response analyses in patients with sepsis demonstrate concurrent immune suppression and hyperinflammation, both on ICU admission and upon development of secondary infections. Even though our data do not provide insight in the host response in noninfectious critically ill patients,