Study design
All patients hospitalized in the 4th Department of Internal Medicine of the 'ATTIKON' University Hospital of Athens during the period November 2004 to January 2006 were delegates for the study. The protocol was approved by the Ethics Committee of the hospital and written informed consent was provided by the patients or their relatives.
Inclusion criteria were the concomitant presence of acute pyelonephritis, acute intra-abdominal infection or nosocomial pneumonia within the past 36 hours, and signs of severe sepsis within the past 12 hours. Exclusion criteria were neutropenia (≤500 neutrophils/mm3), HIV infection, oral intake of corticosteroids at a dose equal to or higher than 1 mg/kg equivalent prednisone for a period longer than one month, and administration of drotrecogin alpha prior enrolment.
Diagnosis of acute pyelonephritis was assigned to any patient with the following symptoms [
6]: core temperature >38°C or <36°C, lumbar tenderness or radiological findings consistent with acute pyelonephritis, and pyuria defined as >10 white blood cells/high power field or +3 dipstick of urine for white cells.
Diagnosis of an intra-abdominal infection was made for any patient with the following symptoms [
7]: core temperature >38°C or <36°C, white blood cells >12,000/μl, and pain on palpation or radiological findings consistent with an intra-abdominal infection.
Nosocomial pneumonia was diagnosed by the following criteria: the presence of symptoms at least 48 hours after hospital admission, provided that the infection was not under an incubation period prior to admission; a core temperature >38°C or <36°C; new or persistent consolidation in a lung X-ray scan; a sputum Gram strain with a predominance of Gram-negatives; and a modified clinical pulmonary infection score >5. The modified clinical pulmonary infection score was determined after individual scoring for each of the following parameters [
8]: core temperature, 36.5–38.4°C = 0 points, 38.5–38.9°C = 1 point, and ≤36°C or ≥39°C = 2 points; white blood cell count, 4,000–11,000/μl = 0 points, <4,000 or >11,000/μl = 1 point, and >11,000 points and >10% bands = 2 points; pO
2/FiO
2, ≥240 or the presence of acute respiratory distress syndrome = 0 points, and <240 in the absence of acute respiratory distress syndrome = 2 points; diffuse shadows on lung X-ray scan = 1 point and localized shadow on lung X-ray scan = 2 points; and purulent tracheobronchial secretions = 2 points.
Severe sepsis was determined as the acute dysfunction of at least one organ, indicated by the acute presentation of at least one of the following [
9]: acute respiratory distress syndrome, defined as any pO
2/FiO
2 <200; acute renal failure, defined as the production of <0.5 ml urine/kg body weight per hour for at least two hours provided that the negative fluid balance of the patient was corrected; metabolic acidosis, defined as any pH <7.30 or any base deficit >5 mEq/l and serum lactate at least more than twice the normal value; acute coagulopathy, defined as any platelet count <100,000/μl or an International Normalized Ratio (INR) >1.5; and acute cardiovascular failure, defined as systolic pressure <90 mmHg requiring the administration of inotropic agents for more than one hour provided the negative fluid balance of the patient was corrected.
Fifty-three patients were eligible for the study; four of these patients denied informed consent. A total of 49 patients were therefore enrolled.
Upon enrolment in the study, 10 ml blood was collected. Five milliliters were added to flasks for culture, another 3 ml were collected in an ethylenediamine tetraacetic acid-coated tube (Becton Dickinson, Cockeysville, MD, USA) for immunophenotyping, and the remaining blood was added in a sterile tube. After centrifugation, the serum was kept at -70°C until assayed. Flasks with blood were incubated for seven days. Identification of pathogens was performed by the API20E and the API20NE systems (bioMérieux, Paris, France). Enrolled patients were followed-up on a daily basis for a total of 28 days.
Laboratory techniques
Red blood cells were lysed with ammonium chloride 1.0 mM in the whole-blood sample collected into ethylenediamine tetraacetic acid-coated tubes. White blood cells were washed three times with PBS (pH 7.2) (Merck, Darmstadt, Germany); the cells were subsequently incubated for 15 minutes in the dark with the monoclonal antibodies anti-CD3 and anti-CD19 and the protein ANNEXIN-V with the fluorocolor fluorescein isothiocyanate (emission 520 nm; Immunotech, Marseille, France), and with the monoclonal antibodies anti-CD4, anti-CD8, anti-CD14 and anti-CD(16+56) with the fluorocolor phycoerythrin (emission 550 nm; Immunotech) with or without the monoclonal antibody anti-CD3 with fluorocolor PC5 (emission 600 nm; Immunotech).
The following combinations were applied: anti-CD3/anti-CD4, anti-CD3/anti-CD8, anti-CD3/anti-CD(16+56), ANNEXIN-V/anti-CD4/anti-CD3, ANNEXIN-V/anti-CD8/anti-CD3 and ANNEXIN-V/antiCD14. Anti-CD19 was applied singularly. Cells staining positive for the above antibodies were analyzed after running through the EPICS XL/MSL flow cytometer (Beckman Coulter Co., Miami, FL, USA) with gating for lymphocytes or monocytes based on their characteristic FS/SS scattering. NK cells were defined as CD3-negative and CD(16+56)-positive cells.
Blood was also sampled from six healthy volunteers equally matched for age with the study population. IgG isotypic-negative controls with the fluorocolors fluorescein isothiocyanate and phycoerythrin (Immunotech) were applied before the start of analysis for each patient.
Estimation of sTREM-1 was performed by a home-made enzyme immunoassay. The capture antibody of sTREM-1 (R&D Inc., Minneapolis, MN, USA) was diluted to 4,000 ng/ml and distributed in a 96-well plate at a volume of 0.1 ml/well. After overnight incubation at 25°C, the wells were thoroughly washed with a 0.05% solution of Tween in PBS (Merck) (pH 7.2–7.4). Then 0.1 ml standard concentrations of sTREM-1 (15.1–4,000 pg/ml; R&D Inc.) or serum was added to the wells. After incubation for two hours, the wells were washed three times and 0.1 ml of one 400 ng/ml dilution of sTREM-1 detection antibody (R&D Inc.) was added per well. The plate was then incubated for two hours, and attached antibodies were signaled by streptavidin horseradish peroxidase. The concentrations of sTREM-1 in each well were estimated by the optical density detected at 450 nm after addition of one 1:1 solution of H2O2:tetramethylbenzidine as a substrate (R&D Inc.). All determinations were performed in duplicate; the inter-day variation of the assay was 5.23%.
Isolation of NK cells
NK cells were isolated from two healthy volunteers, as already described [
10]. Briefly, heparinized venous blood was layered over Ficoll Hypaque (Biochrom, Berlin, Germany) and was centrifuged. The buffy coat was washed three times with PBS (pH 7.2) (Merck), and was then diluted at a volume of 2 ml in RPMI 1640 (Biochrom) and incubated with 0.2 ml RosetteSep NK antibody cocktail (StemCell Technologies, Seattle, WA, USA) for one hour at room temperature with thorough mixing every ten minutes. The buffy coats were layered over Ficoll Hypaque and were centrifuged for 20 minutes at 1,200 ×
g. After centrifugation, any cells remaining over Ficoll Hypaque were collected and washed three times in PBS (pH 7.2). These cells were then stained with anti-CD3 fluorescein isothiocyanate and anti-CD(16+56) phycoerythrin, and were analyzed after running through the EPICS XL/MSL flow cytometer with the application of cells stained with IgG isotypic antibodies as negative controls. The collected cells were CD(16+56)-positive and CD3-negative at a purity greater than 90%.
NK cells were distributed into two wells of a 12-well plate with 2.4 ml RPMI 1640 enriched with 10% fetal bovine serum (Biochrom). Lipopolysaccharide of Escherichia coli O111: B4 (Sigma, St Louis, MO, USA) was added to the second well at a concentration of 10 ng/ml. The experiment was run in duplicate for each volunteer. After incubation for 18 hours at 37°C in 5% CO2, the content of each well was collected and centrifuged. sTREM-1 was estimated in the supernatants, as described above.
Statistical analysis
Results are expressed as the median and interquartile range, but those of cell cultures expressed as the means and standard deviation. Comparisons between patients and healthy controls were performed by the Mann–Whitney U test. Statistical correlations were performed after assessment of the nonparametric Spearman coefficient (rs).
The time of survival was estimated after Kaplan-Meier analysis; patients were divided into two groups based on serum levels of sTREM-1 and on the percentage of NK cells. The patients were categorized into those with serum sTREM-1 ≤180 pg/ml and serum sTREM-1 >180 pg/ml, as proposed elsewhere [
11]. After scattering of individual values of NK cells for survivors and nonsurvivors, the patients were also divided into those with NK ≤20% and NK >20%. Comparisons between these subgroups for survival were performed by log-rank tests.
Comparisons of qualitative data were performed according to Fischer's test. P < 0.05 was considered statistically significant.