Inflammatory response

Nuclear factor kappa-B

The cell surface receptors toll-like receptors (TLRs) and CD14 are crucial for bacterial recognition and induction of innate immune responses to infection, leading to activation of nuclear factor kappa-B (NF-țB) and transcription of inflammatory genes (Medzhitov et al.

1997).

NF-țB is a transcription factor regulating gene expression involved in immune and inflammatory responses. The mammalian NF-țB family consists of five subunits: p50, p52, p65, reL, and reLB, which can homodimerize and heterodimerize in various combinations.

These different combinations have varying activating abilities (Adib-Conquy et al. 2000).

Inhibited NF-țB exists normally in cellular cytoplasm and is activated by various stimuli, such as pro-inflammatory cytokines, leading to its phosphorylation by protein kinases and entry into the nucleus. NF-țB activation increases transcription of chemokines, pro-inflammatory cytokines, adhesion molecules, and antiapoptotic proteins (Chen et al. 2003).

NF-țB activity was increased in mononuclear cells of septic and critically ill patients with SIRS who died in hospital compared with hospital survivors (Arnalich et al. 2000, Paterson et al. 2000). Despite septic nonsurvivors expressing more NF-țB than survivors, the mononuclear cells of septic patients, particularly of nonsurvivors, seemed to have a decreased response to stimuli such as lipopolysaccharide (LPS) (Adib-Conquy et al. 2000). This phenomenon, referred to as endotoxin tolerance, appears to be mediated by the NF-țB subunit p50 (Adib-Conquy et al. 2000). In a mouse model of gut ischemia-reperfusion injury, NF-țB activation led to acute systemic inflammatory response and lung inflammation via tumor necrosis factor-alpha (TNF-Į), but provided protection against enterocyte apoptotic injury (Chen et al. 2003). Strong systemic inflammation response can thus be considered one of the leading mechanisms for MOD, and NF-țB may serve as a potential therapeutic target for reducing overwhelming pro-inflammatory response and organ damage in critical illness.

However, complete blocking may also be detrimental considering the major role of NF-țB in host defense.

Cytokines

Pro-inflammatory cytokines are upregulated in the early stages of inflammation. TNF-Į and IL-1 are secreted rapidly in minutes, whereas IL-6, IL-8, and high mobility group box-1 protein (HMGB-1) contribute later (Sundén-Cullberg et al. 2005) with anti-inflammatory cytokines such as IL-1 receptor antagonist (IL-1ra) and IL-10.

Increased cytokine concentrations can be measured in the early phase of critical illness, and levels can stay high in circulation over several days (Kinasewitz et al. 2004, Kellum et al.

2007, Rivers et al. 2007). However, the ability of monocytes to produce inflammatory cytokines seems to be downregulated in severe sepsis and septic shock (Brunialti et al. 2006).

Septic patients with the most severe global tissue hypoxia had higher levels of inflammatory cytokines IL-1ra, TNF-Į, IL-8, and caspase-3, a marker of apoptosis, at 12-36 hours from hospitalization (Rivers et al. 2007), but the monocyte production of TNF-Į, IL-6, and IL-10 under LPS stimulation in patients with severe sepsis or septic shock was impaired compared with healthy controls or patients with simple sepsis diagnosis only (Brunialti et al. 2006). The severity of disease and stage of infection affects the cytokine profile as well (Oberholzer et al.

2005, Kellum et al. 2007). Interestingly, early temporal decrease in inflammatory cytokine levels can be seen by optimizing hemodynamics (Rivers et al. 2007), highlighting the benefit of early treatment.

27 Several recent studies have analyzed multiple cytokines simultaneously, showing significant, but inconsistent, correlations between cytokine concentrations, such as IL-6, IL-8, and monocyte chemoattractant protein-1, and mortality in patients with severe sepsis (Oberholzer et al. 2005, Bozza et al. 2007). However, clinical applicability is lacking. In 39 patients with sepsis or severe sepsis, IL-10 had the best discriminative power for mortality, with an AUC of 0.90, although there were only five nonsurvivors (Heper et al. 2006). By measuring 17 cytokines simultaneously in 60 patients with severe sepsis, only monocyte chemoattractant protein-1 was an independent predictor for 28-day mortality, with an odds ratio of 1.4, but the AUC was only moderate (0.715) (Bozza et al. 2007). By contrast, baseline IL-6 concentration was an independent predictor for 28-day mortality (p=.019) in a study population of 124 patients with severe sepsis (Oberholzer et al. 2005).

The high mobility group box-1 protein, originally identified as a nuclear DNA-binding protein, can also be secreted into extracellular milieu by endotoxin-stimulated macrophages or under cellular stress, particurlarly in necrosis, acting as a “late” pro-inflammatory cytokine (Scaffidi et al. 2002). Its biological activities and role in critical illness is not fully understood.

In patients with sepsis, severe sepsis, or septic shock, the association of HMGB-1 levels and 28-day survival was dependent on the laboratory methods, even though the levels remained very high for several days (Sundén-Cullberg et al. 2005). In patients with pneumonia, HMGB-1 concentrations were significantly higher in those who developed severe sepsis and died than in survivors in multivariate analysis (Angus et al. 2007). However, in a larger study of 247 patients with severe sepsis or septic shock, HMBG-1 had no predictive power for survival when measured at baseline or 72 hours later (Karlsson et al. 2008).

Discrepancy exists about whether highly upregulated anti-inflammatory cytokine activation is more disadvantageous than strong pro-inflammatory response. Recent studies have searched for an optimal cytokine profile. The highest risk of death among patients with community-acquired pneumonia and sepsis was in the combination of high levels of the pro-inflammatory IL-6 and the anti-inflammatory IL-10 (hazard ratio 20.5, p<.001) (Kellum et al. 2007), whereas in another study with 65 patients with severe sepsis the sustained anti-inflammatory profile, defined as persisting high IL-10 levels, was associated with adverse outcome (Gogos et al. 2000). IL-10 has also been found to be an independent predictor of hospital mortality in sepsis, although the AUC for fatal outcome was only moderate (0.71) (Hynninen et al. 2003).

The inflammatory response itself as well as its regulation and time course may be too

complex for predicting the outcome of critically ill patients by cytokine measurements.

Inadequate sample sizes and study designs are shortcomings and limitations in studies investigating the predictive value of inflammatory markers in critically ill patients. Although a marker may have an independent effect on mortality, the discriminative power may be deficient or a complete ROC analysis lacking. In addition, the magnitude of inflammatory response may be a matter of timing and the dynamic continuum at different stages of disease severity.

2.2.3 Apoptosis

Apoptosis has a major role in the pathophysiological process in sepsis (Hotchkiss et al. 2005).

Apoptotic lymphocytes and gastrointestinal epithelial cells have been found in the spleen and colon of autopsy samples of septic patients (Hotchkiss et al. 1999). Depletion of B-lymphocytes and CD4-positive T-cells by caspase-9-mediated apoptosis, and dendritic cells, but not macrophages, is seen more often in the spleens of patients with sepsis compared with critically ill nonseptic patients or patients with trauma (Hotchkiss et al. 2001, Hotchkiss et al.

2002). Increased lymphocyte apoptosis is also found in the circulation of patients with sepsis (Le Tulzo et al. 2002, Hotchkiss et al. 2005), and this depletion of adaptive and innate immunity cells leads to immunoparalysis in sepsis. Apoptosis can be initiated by two divergent pathways: the death receptor-initiated pathway leading to caspase-8 activation, mediated by for example by Fas, and the mitochondrial pathway leading to caspase-9 activation. Both of these mechanisms are activated in septic patients (Hotchkiss et al. 2005) and lead to activation of caspase-3 in the final common pathway of the apoptotic cell death programme. Increased levels of Fas/Apo-1, a humoral factor involved in the innate immunity response and cell death signal transduction, and other apoptosis-associated molecules have been detected in serum and peripheral blood mononuclear cells of ICU patients with SIRS, sepsis, and MOD (Papathanassoglou et al. 2001, Torre et al. 2003, Freitas et al. 2004). Both apoptotic and necrotic cell death occur in severe sepsis (Hofer et al. 2009).

Cell-free plasma DNA

An increased concentration of cell-free plasma DNA, a possible marker of apoptosis, has been found in various clinical conditions, including trauma, myocardial infarction, stroke, burn, cancer, head injury, and acute abdominal pain (Lo et al. 2000, Jahr et al. 2001, Chang et al.

29 2003, Rainer et al. 2003, Chiu et al. 2006, Yurgel et al. 2007, Rainer et al. 2008). The predictive value of plasma DNA regarding mortality and morbidity has been evaluated in patients with stroke, chest pain, acute abdominal pain, and different trauma or injury (Lo et al.

2000, Rainer et al. 2003, Chiu et al. 2006, Lam et al. 2006, Rainer et al. 2006, Yurgel et al.

2007, Rainer et al. 2008), but several of these studies suffer from shortcomings in adequate statistical methods and patient sample sizes (Table 4).

Cell-free DNA has been detected in the circulation of septic patients (Martins et al. 2000). In addition, patients with severe sepsis and septic shock have been shown to possess increased plasma levels of nucleosomes, in which fragmented DNA is packed during apoptosis (Zeerleder et al. 2003). However, only a few studies have investigated the cell-free plasma DNA concentration in critically ill ICU patients (Martins et al. 2000, Wijeratne et al. 2004, Pachl et al. 2005, Rhodes et al. 2006) (Table 5). Although the results appear promising, inadequate patient sample sizes, no confidence intervals for AUCs, and the preliminary nature of these studies limit their value.

Evidence suggests that DNA is released into the circulation by apoptosis (Jahr et al. 2001, Jiang and Pisetsky 2005, Atamaniuk et al. 2006, Atamaniuk et al. 2008), although the exact mechanism remains obscure. Other possible additive mechanisms, such as necrotic cell death and active secretion, may also exist (Stroun et al. 2001, Jahr et al. 2001). Cell-free plasma DNA derived from patients undergoing hemodialysis and from 6-hour ultra-marathon runners showed typical apoptotic layers in gel electrophoresis (Atamaniuk et al. 2006, Atamaniuk et al. 2008). After the 6-hour run, plasma DNA concentration increased and the mononuclear mRNA expression shifted to pro-apoptotic state compared with the pre-run situation (Atamaniuk et al. 2008). Information from sex-mismatched transplantation patients suggested that most of the plasma DNA is of hematopoietic origin in healthy transplant recipients (Lui et al. 2002, Lui et al. 2003).

The exact characteristics of plasma DNA kinetics and clearance have not yet been specified.

Clearance of fetal DNA in maternal plasma after delivery is rapid, with a mean half-life of 16.3 minutes (Lo et al. 1998). After hemodialysis, increased cell-free plasma DNA concentrations are normalized back to pre-dialysis concentrations 30 minutes after finishing the session (García Moreira et al. 2006). Atamaniuk and co-workers (2008) found that plasma DNA concentration increased after a 6-hour ultra-marathon and returned to normal levels in

24 hours, whereas in ICU patients with severe trauma plasma DNA increased early after injury and concentrations remained high for days, especially if organ failure developed (Lam et al. 2003). The circulating plasma DNA has been proposed to be eliminated in the liver and kidneys. Nucleotides are predominantly metabolized in the liver in mice (Gauthier et al.

1996). In humans, approximately 0.5–2% of the cell-free plasma DNA crosses the kidney barrier and is excreted in urine (Botezatu et al. 2000). The pre-dialysis plasma DNA concentrations of patients with chronic renal insufficiency did not differ significantly from the values of healthy controls (García Moreira et al. 2006). However, no clinical study has evaluated the impact of liver and renal failure on cell-free plasma DNA concentrations.

Table 4. Studies investigating the prognostic value of plasma DNA in different disease states. StudyDiagnosisSubjectspDNA valueMain pDNA results Comments a median, *mean _____________ _____________________________________________________________________________________________________ Lo trauma n=84 13818-181303 kGE/LaAUC for mortality 0.829 nonsurvivors 200027 controls 3154 kGE/La Rainerstroke n=88 6205 kGE/L nonsurvivorsa AUC 0.89 for 28-day survival, 11 died in 6 months 20031334 kGE/L survivorsapDNA independent predictor for 6-month survival (OR 1.6) Chiuburn injury n=28 1115 kGE/La pDNA increased in patients 200612 controls 287 kGE/La with burn injury vs. controls (p=.0001) Lam stroke n=44 Not notifiedAUC 0.74 for 6 months Only 3 patients had pDNA 2006X controls morbidity or mortalityabove normal range Rainerchest pain n=58 475-1170 kGE/La AUC 0.90 for 2-year 6 nonsurvivors 200621 controls 350 kGE/La mortality Yurgelhead trauma n=41 366485 kGE/L* at admission 24-h pDNA higher in all ICU patients 2007131708 kGE/L* after 24 h ICU nonsurvivors 20 ICU nonsurvivors 13 controls 3031 kGE/L*AUC 0.709 Rainer acute n=287 3450 GE/mL nonsurvivors*AUC 0.80, OR 1.4 for 2008abdominal pain 40 controls 725 GE/mL survivors* 28-day ICU admission or mortality (n=12) ____________________________________________________________________________________________________________________ AUC, area under curve; GE, genome equivalents; ICU, intensive care unit; kGE, kilo genome equivalents; OR, odds ratio; pDNA, plasma DNA

Table 5. Cell-free plasma DNA studies in critically ill ICU patients.

Study Subjects pDNA valuea Results Comments

__________________________________________________________________________________________

Martins 18 patients pDNA is found in septic 11 sepsis patients

2000 11 controls patients

Wijeratne 94 patients 5493 GE/ml pDNA higher in non- ICU mortality=

2004 22 controls 1970 GE/ml survivors univariately hospital mortality (34%)

(p<.001)

AUC for mortality 0.889

Pachl 94 patients In medical patients Own method

2005 86 controls apoptotic pDNA

higher in nonsurvivors (n=15), p<.05, AUC 0.71

Rhodes 52 patients 80 ng/ml pDNA independent predictor 13 ICU nonsurvivors 2006 10 controls 17 ng/ml for ICU mortality (OR 1.002)

AUC for hospital mortality 0.79

__________________________________________________________________________________________

amedian; AUC, area under curve; GE, genome equivalent (1 GE=0.0066 ng); ICU, intensive care unit; OR, odds ratio; pDNA, plasma DNA