New immunological serum markers in bacteraemia: anti-inflammatory soluble CD163, but not proinflammatory high mobility group-box 1 protein, is related to prognosis

Authors


S. Gaïni, Department of Infectious Diseases, Odense University Hospital, Søndre Boulevard 29, DK-5000 Odense C, Denmark.
E-mail: shahin.gaini@ouh.fyns-amt.dk

Summary

High mobility group-box 1 protein (HMGB1) is a late-onset proinflammatory cytokine. Soluble haemoglobin scavenger receptor (sCD163) is a specific marker of anti-inflammatory macrophages. The study purpose was to relate the levels of these new markers in bactaeremic patients to levels of well-known pro- and anti-inflammatory markers [procalcitonin, lipopolysaccharide (LPS)-binding protein, interleukin (IL)-6, IL-10] and to evaluate the levels in relation to disease severity and aetiology. A total of 110 patients with bacteraemia were included in a prospective manner from the medical department at a large Danish university hospital. Levels of HMGB1 and sCD163 were higher in patients with bacteraemia compared to controls (P < 0·001). HMGB1 correlated with proinflammatory molecules [procalcitonin (PCT)] and traditional infectious parameters [C-reactive proteins (CRP), white blood cells (WBC) and neutrophils], whereas sCD163 correlated with levels of IL-6, IL-10 but not to lipopolysaccharide-binding protein (LBP), PCT or CRP. Levels of sCD163 and IL-6 were significantly higher among non-survivors compared to survivors (P < 0·05). Neither HMGB1 nor any of the proinflammatory markers were elevated in fatal cases compared to survivors. There was no statistically significant difference in HMGB1 and sCD163 levels in Gram-negative versus Gram-positive bacteraemia. HMGB1 reflects proinflammatory processes, whereas sCD163 reflects anti-inflammatory processes as judged by correlations with traditional marker molecules. sCD163 and IL-6, but not HMGB1, were prognostic markers in this cohort pointing to an anti-inflammatory predominance in patients with fatal disease outcome.

Introduction

Sepsis is a serious condition with a significant morbidity and mortality. The prevalence of sepsis among in-patients varies between 2% and 11% [1]. Since 1992, sepsis has been defined as a combination of the presence of infection and the systemic inflammatory response syndrome (SIRS) [2]. Only a minority of sepsis patients will have bacteraemia when they present with their disease. Bacteraemia is strong evidence for the presence of infection in patients with symptoms of infection or SIRS. Clinicians treating patients with sepsis are in need of better diagnostic and prognostic markers and new knowledge in the immunopathogenesis of severe infections and sepsis would have the potential to generate new options for diagnosis and treatment of these patients.

High mobility group-box 1 protein (HMGB1) is a protein that has been known for more than 30 years as a nuclear chromosomal protein [3,4]. In an attempt to identify late-onset proinflammatory mediators/cytokines in sepsis, HMGB1 was identified as a proinflammatory cytokine in cultured macrophages stimulated with endotoxin [3]. The background for these attempts was that disappointing results had been found in several clinical studies focusing upon blockage of early mediators/proinflammatory cytokines in sepsis [5]. It was hypothesized that early-onset cytokines such as tumour necrosis factor (TNF)-α and interleukin (IL)-1 were not driving the proinflammatory response in late phases of sepsis [3]. Animal models showed that administration of exogenous HMGB1 to septic animals increased mortality, whereas administration of antibodies against HMGB1 could ameliorate sepsis in these models [3].

Haemoglobin scavenger receptor (CD163) is a membrane-bound protein involved in endocytosis of haptoglobin–haemoglobin complexes [6]. It is produced almost exclusively by macrophages and monocytes [7,8]. The expression of the membrane-bound CD163 is up-regulated by dexamethasone, IL-6 and IL-10 [9–11]. A soluble form of CD163 (sCD163) is shed from the monocyte–macrophage membrane upon Toll-like receptor activation and other inflammatory stimuli [12,13]. Macrophages and monocytes have an important role in the innate immune system in sepsis when the immune system is challenged by an invading pathogen [14]. sCD163 is considered to be a marker of alternatively activated (anti-inflammatory) macrophages [15]. It has been shown that the levels of sCD163 are elevated in Gram-positive bacteraemia and in community-acquired severe sepsis [16,17].

Procalcitonin (PCT) is a protein that has been proposed as a sensitive and specific marker of bacterial infection and sepsis [18]. PCT levels are correlated to the severity of sepsis [19–21]. Lipopolysaccharide-binding protein (LBP) is an acute-phase protein involved in the innate immune response in serious infections [22]. LBP has an important role in early phases of the innate immune response both in Gram-positive and Gram-negative infections [22].

The study purpose was to relate the levels of these proteins (HMGB1, sCD163) in bacteraemic patients to levels of well-known pro- and anti-inflammatory markers (PCT, LBP, IL-6, IL-10), and to evaluate the levels in relation to disease severity and aetiology.

Patients and methods

Patients

Patients were included in a prospective manner in the period November 2003–July 2005. The hospital setting was a large medical department at a university hospital serving a local population of approximately 185 000 inhabitants. Inclusion criteria were culture-positive bacteraemia verified by the department of clinical microbiology. The exclusion criteria were age < 18 years, earlier participation in the study or growth of a bacteriae considered to be non-pathogenic. The department of clinical microbiology contacted the research staff immediately when a case of bacteraemia was identified. Informed consent was obtained from the patient or their relatives and blood (plasma and serum) was sampled immediately afterwards. The samples were processed and frozen at −80°C within 1·5 h. The patients received standard of care according to departmental guidelines. The project protocol was approved by the Ethics Committee of Fyns and Vejle Counties.

Baseline characteristics, demographic data, biochemical parameters, SIRS criteria and severity score were obtained at the time of inclusion. Severity was assessed with the Sequential Organ Failure Assessment Score (SOFA) [23]. Co-morbidity was assessed with the Charlson Index [24]. Patients were classified at the time of inclusion according to the SIRS criteria [2]. Severe sepsis was defined as the presence of sepsis and one or several of the following indices of organ dysfunction: Glasgow coma scale ≤ 14; PaO2 ≤  9·75 kPa; oxygen saturation ≤ 92%, PaO2/FIO2 ≤ 250, systolic blood pressure ≤ 90 mmHg; systolic blood pressure fall ≥ 40 mmHg from baseline; pH ≤ 7·3; lactate ≥  2·5 mmol/l, creatinine ≥ 177 μmol/l; 100% increase of creatinine in patients with known kidney disease; oliguria ≤ 30 ml/h in >3 h or ≤0·7 l/24 h; prothrombin time ≤ 0·6 (reference: 0·70–1·30); platelets ≤ 100*10E9/l; bilirubin ≥ 43 μmol/l; paralytic ileus. Septic shock was defined as hypotension persisting despite adequate fluid resuscitation for at least 1 h. If a patient had any comorbidity that could more probably explain one or more of the criteria for organ dysfunction stated above the patient could not be categorized as having severe sepsis.

Controls

Levels of HMGB1, PCT, LBP, IL-6 and IL-10 were measured in a control group consisting of 32 healthy hospital workers (mean age 44·4 years with range 24–62 years). This control group consisted of 20 women and 12 men. These controls have been used in previous publications [25,26]. In this control group the following levels were observed: HMGB1 median 0·77 ng/ml (IQR: 0·6–1·5); PCT median 0·05 ng/ml (IQR: 0·04–0·06); LBP median 12·7 μg/ml (IQR: 9·8–16·8); IL-6 median 3·4 pg/ml (IQR: 3–3·7), IL-10 ≤ 5 pg/ml in 31 of the controls (range: undetectable, 7·3 pg/ml). Levels of sCD163 were measured previously in a control group consisting of 130 healthy blood donors (mean age 43·9 years with range 23–63 years). This control group consisted of 65 women and 65 men. The median of sCD163 in this control group was 1·9 mg/l and the IQR 0·8–4·4.

Laboratory assays

HMGB1 was measured in serum with a commercially available enzyme-linked immunosorbent assay (ELISA) (HMGB1 ELISA kit; Shino-Test Corporation, Kanagawa, Japan). The measuring range was 0·6–93·8 ng/ml. The coefficient of variation was 5% for samples above 10 ng/ml and 10% for samples between 2 and 5 ng/ml. The range could be broadened by dilution of high samples. Recovery of HMGB1 in this ELISA has been reported to be 92% to 111% [27]. The detection limit of HMGB1 was 0·6 ng/ml. sCD163 was measured in serum with an in-house ELISA, as described previously [28]. The detection limit of sCD163 was 0·00625 mg/l. PCT was measured in plasma with a time-resolved amplified cryptate emission (TRACE) technology assay (Kryptor PCT®; Brahms, Hennigsdorf, Germany). The functional detection limit was 0·06 ng/ml. LBP and IL-6 were measured in plasma with a chemiluminiscent immunometric assay (Immulite-1000®; DPC, Los Angeles, CA, USA). IL-10 was measured in serum with a chemiluminiscent immunometric assay (Immulite-1000®; DPC). The detection limit of LBP was 0·2 μg/ml. The detection limit of IL-6 was 2 pg/ml. The detection limit of IL-10 was 5 pg/ml. CRP was measured with an immunoturbidimetric principle (Modular P®; Roche, Mannheim, Germany). White blood cells (WBC) and neutrophils were counted on a Sysmex SE 9000 (TOA®; TOA Corporation, Kobe, Japan).

Statistical analyses

Data are presented as medians, interquartile ranges (IQR) and means ± standard deviation. Significance testing was performed using the Kruskal–Wallis test and Wilcoxon's two-sample test. A two-tailed P-value < 0·05 was considered statistically significant. The Spearman's rank correlation test was used to determine correlations. The Bonferroni correction was used when multiple comparison were made for a singled marker. HMGB1 levels below 0·6 ng/ml were assigned a value of 0·6 ng/ml for calculations. IL-6 measurements below 2 pg/ml were assigned a value of 2 pg/ml for calculations. IL-10 measurements below 5 pg/ml were assigned a value of 5 pg/ml for calculations. All statistical calculations were performed using the stata 8® statistical software package (STATA Corporation, College Station, TX, USA).

Results

Patient characteristics

One hundred and ten adult patients were included in our study. The patients were divided into the following groups for analyses of the levels of inflammatory markers: bacteraemia without SIRS (n = 24), sepsis (n = 41), severe sepsis (n = 39) and septic shock (n = 6). The patients were also divided into survivors (n = 99) and fatal cases (n = 11) for analysis of the prognostic abilities of the examined inflammatory markers. Finally, patients were divided according to the microbiological aetiology of their bacteraemia. Patients with the three most common pathogens represented in our study were used for this purpose. The most common pathogens were Escherichia coli (n = 37), Streptococcus pneumoniae (n = 29) and Staphylococcus aureus (n = 12). The majority of the cases were mono-microbial bacteraemia, with Gram-positive infections being the most common. As expected, pneumonia and urinary tract infections were the most common diagnoses. The baseline characteristics and the outcome are presented in Table 1. The microbiological and infection characteristics are presented in Table 2. Twelve (10·9%) of the included patients were treated with immunosuppressive drugs (all with prednisolone): three patients with bacteraemia without SIRS, eight patients with sepsis and one patient with severe sepsis. Sixty-five patients were sampled 24–48 h after blood cultures were drawn, 37 patients were sampled 48–72 h after blood cultures were drawn and eight patients were sampled 72–96 h after blood cultures were drawn.

Table 1.  Baseline characteristics and outcome of the patients with bacteraemia.
VariableBacteraemia without
SIRS (n = 24)
Sepsis
(n = 41)
Severe sepsis
(n = 39)
Septic shock
(n = 6)
  1. Data are presented as the absolute number (%) or the mean ± standard deviation. SIRS, systemic inflammatory response syndrome; SOFA, sepsis-related organ failure assessment.

Male16 (66·7)22 (53·7)21 (53·8)2 (33·3)
Female8 (33·3)19 (46·3)18 (46·2)4 (66·7)
Age70·1 ± 15·566·1 ± 16·069·0 ± 19·261·0 ± 7·8
Hospitalization (days)16·2 ± 8·412·4 ± 15·116·9 ± 11·229·2 ± 31·6
Mortality on day 281 (4·2)2 (4·9)5 (12·8)3 (50)
SOFA score2·1 ± 1·61·9 ± 1·33·9 ± 2·43·7 ± 1·2
Charlson Index of comorbidity2·7 ± 2·22·3 ± 2·61·5 ± 2·33·0 ± 3·9
Hb (mmol/l)7·4 ± 1·07·6 ± 1·38·5 ± 1·18·4 ± 1·4
Platelet count (10E9/l)244·2 ± 75·9244·9 ± 143·9261·6 ± 435·9172·3 ± 74·5
Bilirubin (μmol/l)13·0 ± 5·917·5 ± 29·225·3 ± 25·210·8 ± 2·3
Prothrombin (time)0·8 ± 0·30·9 ± 0·20·7 ± 0·30·7 ± 0·2
Creatinine (μmol/l)136·9 ± 80·6125·5 ± 89·3160·4 ± 95·0146·0 ± 21·5
Table 2.  Microbiological and infection characteristics of the patients with bacteraemia.
VariableBacteraemia without
SIRS (n = 24)
Sepsis
(n = 41)
Severe sepsis
(n = 39)
Septic shock
(n = 6)
  1. SIRS, systemic inflammatory response syndrome.

Microbiological findings (n)
 Gram-positive bacteria1316244
 Gram-negative bacteria825122
 Polymicrobial bacteraemia3030
Focus of infection (n)
 Meningitis1030
 Pneumonia34115
 Endocarditis0320
 Pyelonephritis1230
 Cystitis81771
 Skin/soft tissue infection0310
 Bone/joint infection3220
 Other810100

Levels of HMGB1 and sCD163 in bacteraemic patients

Levels of HMGB1 and sCD163 were higher in patients with bacteraemia compared to healthy controls (P < 0·0001), which were also seen for LBP, PCT, IL-6 and IL-10 (Table 3).

Table 3.  Levels of inflammatory markers in bacteraemia.
VariableHealthy
controls
Infection without
SIRS (n = 24)
Sepsis
(n = 41)
Severe sepsis
(n = 39)
Septic shock
(n = 6)
P-value
(Kruskal–Wallis)
  1. Data are presented with medians and interquartile ranges. SIRS, systemic inflammatory response syndrome; HMGB1, high mobility group box-1 protein; sCD163, soluble haemoglobin scavenger receptor; LBP, lipopolysaccharide-binding protein; PCT, procalcitonin; IL-6, interleukin-6; IL-10, interleukin-10; CRP, C-reactive protein; WBC, white blood cells. *Compared to the previous group in the table (Wilcoxon's two sample test). The nine P-values from the Kruskall–Wallis test and the 3–4 P-values in each row were subjected to a Bonferroni-correction.; n.s.: not significant.

HMGB1 (ng/ml)     < 0·001
 Median0·772·53·74·44·9 
 IQR0·6–1·51·4–3·31·9–6·12·1–8·11·7–9·0 
 P-value* < 0·001n.s.n.s.n.s. 
sCD163 (mg/l)     < 0·001
 Median1·93·34·96·25·5 
 IQR0·8–4·42·6–4·92·6–9·13·7–11·13·5–12·3 
 P-value* < 0·001n.s.n.s.n.s. 
LBP (μg/ml)     < 0·001
 Median12·735·458·961·875·9 
 IQR9·8–16·828·2–55·846·1–86·742·9–11045·6–88·6 
 P-value* < 0·001< 0·01n.s.n.s. 
PCT (ng/ml)     < 0·001
 Median0·050·83·29·244·9 
 IQR0·04–0·060·2–2·20·8–14·23·8–16·220·6–66 
 P-value* < 0·001< 0·05n.s.n.s. 
IL-6 (pg/ml)     < 0·001
 Median3·436·965·096·2706·3 
 IQR3–3·722·3–68·831·9–11955·9–274127–2445 
 P-value* < 0·001n.s.< 0·05n.s. 
IL-10 (pg/ml)     < 0·001
 Median56·66·56·238·9 
 IQR5–55–7·95–16·15–13·79·4–219·5 
 P-value* < 0·001n.s.n.s.n.s. 
CRP (mg/l)     < 0·001
 Median 176·0200·5317·0362·0 
 IQR 105–239141·5–285199–394349–371 
 P-value*  n.s.< 0·05n.s. 
WBC (10E9/l)     < 0·001
 Median 8·711·813·813·4 
 IQR 7·5–108·4–14·410·8–19·110·2–25·8 
 P-value*  < 0·01< 0·05n.s. 
Neutrophils (10E9/l)     < 0·001
 Median 6·89·8312·312·4 
 IQR 6·1–7·97·3–12·18·8–169·8–21·2 
 P-value*  < 0·01< 0·05n.s. 

Levels of HMGB1 and sCD163 in bacteraemia with E. coli, bacteraemia with S. pneumoniae and bacteraemia with Staph. aureus

The HMGB1 median in patients with E. coli bacteraemia was 4·0 ng/ml (IQR: 2·4–6·1), 5·6 ng/ml (IQR: 1·9–7·9) in patients with S. pneumoniae and 3·3 ng/ml (IQR: 1·9–4·9) in patients with Staph. aureus. The sCD163 median in patients with E. coli bacteraemia was 5·7 mg/l (IQR: 3·0–9·1), 3·7 mg/l (IQR: 2·6–8·3) in patients with S. pneumoniae and 7·3 mg/l (IQR: 4·2–12·8) in patients with Staph. aureus. There was no statistically significant difference regarding HMGB1 or sCD163 levels between Gram-negative and Gram-positive bacteraemia.

Levels of examined inflammatory markers in relation to survival

Of all examined markers, only sCD163 and IL-6 showed statistically significantly higher levels in the group with fatal cases compared to the survivor group (P < 0·05) (Table 4).

Table 4.  Inflammatory markers in survivors and in fatal cases.
VariableSurvivors (n = 99)Fatal cases (n = 11)P-value*
  1. Data are presented with medians and interquartile ranges. HMGB1, high mobility group-box 1 protein; sCD163, soluble haemoglobin scavenger receptor; PCT, procalcitonin; LBP, lipopolysaccharide-binding protein; IL-6, interleukin-6; IL-10, interleukin-10; CRP, C-reactive protein, WBC, white blood cells. *Wilcoxon's two sample test; n.s.: not significant.

HMGB1 (ng/ml)3·4 (2·0–6·2)1·9 (1·3–4·4)n.s.
sCD163 (mg/l)4·8 (2·9–8·3)12·3 (4·8–18·9)P < 0·05
PCT (ng/ml)4·7 (0·7–12·7)4·4 (2·7–43·7)n.s.
LBP (μg/ml)58 (42·4–91)46·4 (36·8–72)n.s.
IL-6 (pg/ml)68·8 (32·8–142·5)236·3 (88·5–1203·5)P < 0·01
IL-10 (pg/ml)6·5 (5–12·1)11·9 (5·7–61·5)n.s.
CRP (mg/l)232 (141–336)232 (151–395)n.s.
WBC (10E9/l)11·2 (8-4-15·2)13 (7·5–14·9)n.s.
Neutrophils (10E9/l)9·7 (6·8–13·3)12 (7·1–13·3)n.s.

Correlations between the examined markers

Moderate, but highly significant, correlations were found between HMGB1 versus PCT, CRP, WBC and neutrophils (Table 5). Moderate but highly significant correlations between sCD163 versus IL-6, IL-10 were found, whereas sCD163 did not correlate with either PCT, LBP, CRP or WBC (Table 5).

Table 5.  Correlations between HMGB1/sCD163 and the examined inflammatory markers.
 r*P-value r*P-value
  1. HMGB1, high mobility group-box 1 protein; sCD163, soluble haemoglobin scavenger receptor; PCT, procalcitonin; LBP, lipopolysaccharide-binding protein; IL-6, interleukin-6; IL-10, interleukin-10; CRP, C-reactive protein, WBC, white blood cells. *Spearman r. The 8 P-values in each column were subjected to a Bonferroni-correction; n.s.: not significant.

HMGB1 versus  sCD163 versus  
sCD163−0·06n.s.HMGB1−0·06n.s.
PCT0·41< 0·001PCT0·15n.s.
LBP0·23n.s.LBP0·06n.s.
IL-60·15n.s.IL-60·39< 0·0001
IL-100·03n.s.IL-100·39< 0·0001
CRP0·36< 0·001CRP0·17n.s.
WBC0·45< 0·0001WBC0·18n.s.
Neutrophils0·44< 0·0001Neutrophils0·23n.s.

Discussion

HMGB1 is a nuclear chromosomal protein that has been shown to be highly conserved among different species [4,29]. During recent years attention has come to HMGB1 regarding a possible role as a proinflammatory cytokine. To our knowledge, eight studies have been published regarding HMGB1 levels in infected patients [3,25,26,30–34]. Different methods were used for measuring HMGB1 in these studies. Five of the studies used ELISA [25,26,31,32,34] and the other three studies used blotting methods [3,30,33]. The first published study reported the highest HMGB1 levels to be found among sepsis patients with fatal outcome (median 84 ng/ml) [3]. Surviving sepsis patients in this study had HMGB1 median of 25 ng/ml [3]. In a prospective observational study examining HMGB1 levels among patients with different degrees of sepsis admitted to intensive care units, levels of HMGB1 were found to be elevated up to a week after inclusion [30]. In another study a median HMGB1 level of 4·54 ng/ml was observed in infected patients [31]. In a large prospective study focusing on HMGB1 levels in community-acquired pneumonia higher levels were observed among patients (HMGB1 median of 190 ng/ml) compared to healthy controls [33]. In two previous studies by our group focusing on mild and severe community-acquired infections and sepsis low levels of HMGB1 were observed [25,26]. Our present study data show slightly higher HMGB1 levels among patients with bacteraemia. In our study data HMGB1 correlate with other proinflammatory markers, but do not correlate with IL-6, IL-10 and sCD163. These observations point toward a proinflammatory role for HMGB1 in human sepsis/bacteraemia. The levels in our study and the previous studies using ELISA for HMGB1 measurements have shown much lower levels of HMGB1 compared to studies using blotting techniques. The presence of interfering inhibitory factors/autoantibodies to HMGB1 in human serum could affect results of HMGB1 measurements with ELISA techniques [35]. It is still unknown if the currently used assays detect biologically active HMGB1. This is an important issue in studies focusing upon HMGB1 levels and disease activity.

The role of sCD163 has not yet been elucidated fully [15]. sCD163 is considered to be a marker of alternatively activated (anti-inflammatory) macrophages [15]. To our knowledge only six studies have focused upon sCD163 levels in relation to infectious conditions [16,17,36–39] One study focusing upon mild community-acquired infections/sepsis reported that levels of sCD163 were elevated only in patients with severe sepsis (median 3·63 mg/l) and in patients with bacteraemia (median 4·9 mg/l) [17]. In another study focusing upon patients with S. pneumoniae bacteraemia the following median levels of sCD163 were observed: 5·4 mg/l in patients < 75 years of age and 3·5 mg/l in patients ≥ 75 years of age [16]. In that study sCD163 levels were associated to the prognosis with the following median levels: 4·8 mg/l in survivors, 14·3 mg/l in fatal cases (P < 0·0001) [16]. In the present study we confirm that sCD163 is a prognostic marker in bacteraemia, sCD163 correlates with IL-6, sCD163 correlates with the anti-inflammatory marker IL-10 and levels of sCD163 are higher among bacteraemic patients compared to controls. sCD163 does not correlate to disease severity in the present study. The observed correlations between the above-mentioned markers point towards an anti-inflammatory role for sCD163 in human sepsis/bacteraemia. At present IL-6 is considered as having both pro- and anti-inflammatory qualities as a cytokine. However, several studies point towards an anti-inflammatory role for this cytokine [40–43]. The correlation between IL-6 and sCD163 is unsurprising, as IL-6 is known to up-regulate membrane-bound CD163 [9]. Another biomarker of immunodeficiency related to monocytes/macrophages in critical illness is the human leucocyte antigen D-related (HLA-DR) expression on monocytes [44–46]. To our knowledge, no studies have examined the correlation between HLA-DR expression on monocytes and sCD163 levels in serum. An inverse relation between monocyte surface CD163 expression and sCD163 was observed in a single study [47]. The effect of blood transfusion or haemolysis on levels of sCD163 is, to our knowledge, not known. Increased levels of sCD163 have been observed in patients treated with steroids undergoing major surgery [48].

Since 1993 PCT has been associated with severe bacterial infections among children and adults [18]. Several studies have reported data suggesting PCT to have an important role in the immunopathogenesis in severe infections. Animal models have shown that administration of exogenous PCT increased mortality and administration of anti-serum to PCT-protected infected animals [49]. Several studies have focused upon PCT levels and diagnostic test abilities in patients with bacteraemia [50]. Our present study confirms that PCT levels are high in patients with bacteraemia. Our data also show that PCT is a marker of severity of infection in bacteraemia. LBP is an acute-phase protein with an important role in the innate immune response both in Gram-positive and Gram-negative infection [22,51]. Two previous studies have examined levels of LBP in bacteraemia [52,53]. One of the studies showed the following median levels of LBP: 54·2 mg/l in Gram-negative bacteraemia, 21·1 mg/l in Gram-positive bacteraemia and 21·2 mg/l in febrile patients without bacteraemia [52]. Another study found the following mean levels of LBP: 40·8 μg/ml in Gram-negative bacteraemia and 35·6 μg/ml in Gram-positive bacteraemia [53]. The present study shows that LBP is a severity marker with higher levels among sepsis patients with bacteraemia compared to bacteraemic patients without SIRS. The present study data reports higher levels of LBP among patients with Gram-positive bacteraemia compared to the two above-mentioned previous studies.

During recent years there has been an increasing focus upon the anti-inflammatory aspects of sepsis [54]. It has been suggested that the current view on sepsis as primarily a proinflammatory state has been too simplistic [54]. The failures of several attempts to improve the outcome of sepsis patients by dampening the proinflammatory response could maybe be explained by this simplistic view on the immunopathogenesis of sepsis. A more complex understanding of the immunopathogenesis of sepsis, with the possibility of the presence of a proinflammatory state and an immunosuppressed state either consecutively or in different body compartments/infectious focus at the same time, could perhaps generate new individualized treatment options for each single sepsis patient. At present there are no good methods in determining the immune status of the sepsis patient seen in the clinical setting. There is therefore a need to develop assays that could monitor the immune status of each single sepsis patient. The perspective could be the possibility to tailor an individual immunological treatment strategy to each single patient with sepsis, depending upon their immune status. In this context, HMGB1 and sCD163 could perhaps be possible candidate markers for proinflammation and anti-inflammation, respectively. Our current observations of a correlation between HMGB1 and the studied proinflammatory markers (PCT, CRP, WBC, neutrophils) and a correlation between sCD163 versus IL-6 and versus IL-10 are interesting in this context.

The patients included in this prospective study were representative of patients admitted to departments of internal medicine. They were elderly, with a burden of comorbidity and 10% mortality. Strengths in our study were the prospective design and patients were well characterized. We avoided work-up bias in the classification of the infectious status of the patients by blinding the research physician and the laboratory technicians. We avoided imperfect gold standard bias by requiring good evidence for the presence of infection. The drawbacks of the study design, as with most other clinical sepsis studies, were heterogeneity among included patients, different lengths of diseases prior to hospitalization and a heavy burden of comorbidity. Another drawback of the study was that some of the patients were on immunosuppressive treatment. This could have an influence on the levels of the inflammatory markers in these patients. Another drawback was that the control groups were healthy hospital workers and blood donors and they were not age- and gender-matched. However, the study had an exploratory purpose and did not elucidate any diagnostic test abilities, which would have demanded a matched control group. Another drawback due to the time-consuming process of culturing/processing in the laboratory of clinical microbiology was the delay and variation of sampling (research blood samples) time. Different bacterial species grow at a different pace, and this is the reason for the variation in sampling time. However, 92·7% of the patients were sampled within 72 h after blood cultures were drawn.

Conclusions

This is the first study reporting on levels of HMGB1 in a prospectively included cohort of patients with bacteraemia. The studied markers (HMGB1, sCD163) were elevated in patients with bacteraemia compared to a healthy control group. There was no statistically significant difference regarding levels of HMGB1 and sCD163 in Gram-negative bacteraemia versus Gram-positive bacteraemia. Levels of HMGB1 correlated to levels of PCT, CRP, WBC and neutrophils. Levels of sCD163 correlated with levels of IL-6 and IL-10. HMGB1 reflects proinflammatory processes, whereas sCD163 reflects anti-inflammatory processes as judged by correlations to traditional marker molecules. sCD163 and IL-6, but not HMGB1, were prognostic markers in this cohort, pointing to an anti-inflammatory predominance in patients with fatal disease outcome. Further studies are needed to unravel the role of HMGB1 and sCD163 in the immunopathogenesis of sepsis. Studies focusing upon consecutive measurements of HMGB1 and sCD163 should also be encouraged.

Acknowledgements

The study was supported financially by the University of Southern Denmark and the Danish Medical Research Council (22-03-0355 HJM). The authors thank the doctors at the Department of Clinical Microbiology for excellent microbiological assistance, and the study nurses L. Hergens, A. Nymark, N. Bülow and H. Møller for excellent clinical assistance. We also thank J. Clausen, H. Madsen and K.B. Petersen for excellent technical assistance.

Ancillary