Lack of recovery in monocyte human leukocyte antigen-DR expression is independently associated with the development of sepsis after major trauma

This prospective observational study was carried out over a 15-month period (July 2008 to September 2009). The protocol was reviewed by the institutional ethics committee, which waived the need for informed consent because the study was observational and involved sampling of very small quantities of blood (100 μL). The purpose of the study was explained to the patients or members of their families. Samples were collected from residual blood after completion of routine follow-up.

Inclusion criteria were an Injury Severity Score (ISS) [21, 22] of more than 25 and admission to the intensive care unit (ICU). Clinical exclusion criteria were age of less than 18 years, ISS of less than 25, chronic corticosteroid therapy, and death in the first 48 hours after admission. Patients admitted on a Saturday were excluded because mHLA-DR cannot be measured on day 1 or 2 (blood samples were not collected on Saturdays or Sundays, when the laboratory did not operate).

All patients admitted were followed up with prospectively until day 14 by daily clinical examination and biological tests. During follow-up, clinical and biological data were collected. The data collection comprised demographic characteristics (age and gender), infection characteristics (source, microorganisms identified, delay between trauma, and onset of sepsis), and outcome at 28 days (death or survival). Therapeutic data were also collected (a) on admission to the trauma room (the need for inotropic or vasoactive support and blood products [red blood cells, fresh frozen plasma, platelets, and albumin] and their quantities used to sustain a mean arterial pressure [MAP] up to 70 mm Hg [or 90 mm Hg in the case of cranial trauma], and the type and quantity of prophylactic antibiotics) and (b) during support (number of ventilator days, quantity and type of vasoactive support and of blood products, and use of massive transfusion, which was defined as more than 10 units of blood [23] or the replacement of the patient's total blood volume [24] over a 24-hour period). Creatinine, lactate concentration, and abnormal biphasic pulse transmittance waveform (BPW) were measured daily. Three clinical scores were recorded: ISS on admission (range of 0 to 75), initial severity of disease as assessed by the new Simplified Acute Physiology Score II (SAPS II) (range of 0 to 164) [25], and the Sepsis-related Organ Failure Assessment (SOFA) score (range of 0 to 24) on admission and every day during follow-up [26]. Severe brain and thoracic injury, which are well established as risk factors for sepsis development, were also taken into account [22].

The American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference [27] definition of sepsis was used for this study, namely the presence of an identifiable site of infection and evidence of a systemic inflammatory response on the basis of at least two of the following criteria: (a) body temperature of greater than 38°C or of less than 36°C, (b) heart rate of greater than 90 beats per minute, (c) respiratory rate of greater than 20 breaths per minute or hyperventilation as indicated by an arterial partial pressure of carbon dioxide (PaCO₂) of less than 32 mm Hg (less than 4.3 kPa), and (d) a white blood cell count of greater than 12,000 cells/mm³ or of less than 4,000 cells/mm³ or the presence of more than 10% immature neutrophils. The onset of sepsis was defined, as recommended by the Consensus Conference [27], as the day on which the site of infection was identified. The final diagnosis of sepsis was retrospectively established by two experts assessing the complete medical data and not involved in case management. Diagnoses of pneumonia and urinary infection were established according to the guidelines of the American Thoracic Society and the Infectious Diseases Society of America [28] and of the Centers for Disease Control [29], respectively. Physicians were not informed of mHLA-DR results. BPW was also determined as it may be used as an indicator of sepsis development [30‐32].

Ethylenediaminetetraacetic acid (EDTA)-anticoagulated blood samples were collected at 8 a.m. every 2 days after injury (on Mondays, Wednesdays, and Fridays) (that is, at days 1 and 2, days 3 and 4, days 5 and 6, days 7 and 8, days 9 and 10, and days 11 and 12). Flow cytometric (EPICS XL; Beckman Coulter, Inc., Hialeah, FL, USA) expression of monocyte HLA-DR was assessed on arterial, venous, or capillary blood. Blood samples were stored immediately at 4°C and stained within 2 hours after collection, in accordance with the standardization recommendations for mHLA-DR measurement [33, 34]. Staining and cell acquisition were undertaken as described in the European standardized protocol. Monoclonal antibodies and their respective isotype controls were used according to the manufacturers' recommendations: fluorescein isothiocyanate (FITC)-labeled anti-CD14 (10 μL; Immunotech, Marseille France) and phycoerythrin (PE)-labeled anti-HLA-DR (20 μL; BD Pharmingen, San Diego, CA, USA) per 100 μL of whole blood. Monocytes were characterized on the basis of their CD14 expression. Results were expressed as the number of anti-HLA-DR antibodies per cell (AB/C) (normal >15,000), which is correlated with the number of HLA-DR molecules expressed on each monocyte [33].

Because sepsis alone can amplify a drop in mHLA-DR expression, mHLA-DR expression data were excluded from the analysis after the onset of sepsis, thereby precluding calculation of a difference in mHLA-DR expression between septic and non-septic patients at days 7 and 8, 9 and 10, and 11 and 12 (because of insufficient numbers of values for statistical analysis).

The Kolmogorov-Smirnov test was used to verify all data for normality. Baseline characteristics were described by frequency, median and interquartile range (IQR), or (where appropriate) mean ± standard deviation. Patients were separated into two groups: those who developed sepsis and those who did not. The groups were compared using the Mann-Whitney U test for continuous non-parametric variables, the independent paired t test for continuous parametric variables, and the chi-square test for categorical data. mHLA-DR expression was stratified according to the best threshold chosen using the Youden index. Receiver operating characteristic (ROC) curves and the areas under the curve were calculated for the slope in mHLA-DR between days 1 and 2 and days 3 and 4. Univariate and multivariate logistic regression analyses were used to identify variables associated with the risk of infection and assessed by odds ratios (ORs) and 95% confidence intervals (CIs). A P value of less than 0.05 was taken as the significance level. The Bonferroni correction was used to avoid spurious results from the multiple statistical tests performed simultaneously. The alpha values for three or six tests were 0.016 and 0.008, respectively. MedCalc software version 9.6.4.0 (MedCalc Software bvba, Mariakerke, Belgium) was used to perform the statistical analyses.

A total of 536 consecutive patients in the early stages of trauma were admitted to the trauma room between July 2008 and May 2010. One hundred five of these patients met the inclusion criteria of the study (Figure 1). One hundred thirty patients were excluded because they had been rapidly transferred to another hospital for different reasons: no available rooms in our unit or the need for specific care such as aortic rupture isthmus or brain surgery (following severe brain injury). Table 1 shows the baseline characteristics on these 105 patients. SAPS II was significantly higher in septic patients (P < 0.05) than in non-septic patients as were the SOFA scores every day during follow-up and the incidence of severe brain injury. There were no statistical differences of the ISS or the incidence of severe thoracic injury between the two groups. In the emergency room, administration of vasoactive drug to maintain an MAP of up to 65 mm Hg and administration of prophylactic antibiotics were not different. Frequency of massive transfusion and the overall quantity of transfused blood were not different for the sepsis and non-sepsis groups over the first 2 post-trauma days. There was a higher proportion of patients under vasoactive drug during the first 2 days in the septic group (P = 0.0004). During follow-up, no difference in renal function (assessed by plasma creatinine concentration) or in lactate concentration was observed. Septic patients required mechanical ventilation more often and for longer periods of time than non-septic patients did (P < 0.0001). Six patients died (three from septic shock and three from cardiogenic shock), and there were no statistical differences between the two groups.

Thirty-seven patients developed sepsis during follow-up. Pneumonia was the more frequent infection (n = 30), followed by urinary tract infection (n = 7). Causative bacteria were fairly evenly distributed between Gram-positive (n = 14) and Gram-negative (n = 21) organisms. Two patients had a mixed bacterial infection (Gram-positive and -negative). The median interval between trauma and onset of sepsis was 4 days (3 to 6.25).

At day 2, mHLA-DR expression was diminished in all 105 patients (Figure 2 and 3a). At days 1 and 2, mHLA-DR expression showed no statistically significant difference between septic and non-septic patients (Table 1 and Figure 3b). At days 3 and 4, mHLA-DR expression had risen in non-septic patients but remained low in septic patients (13,723 ± 7,766 AB/C versus 9,271 ± 6,029 AB/C; P = 0.004) (Figure 3b). At days 5 and 6, septic patients still tended to exhibit lower mHLA-DR expression than non-septic patients, but the difference failed to reach statistical significance.

Given the wide fluctuation of mHLA-DR expression data, ratios were calculated between values for two points in time, namely days 3 and 4/days 1 and 2 and days 5 and 6/days 3 and 4. The slope of mHLA-DR expression at days 3 and 4 showed a highly significant statistical difference between non-septic and septic patients (1.44 ± 0.53 versus 0.83 ± 0.43, respectively; P < 0.0001) (Table 1). We next established an ROC analysis (Figure 4). The area under the curve was 0.80 (P = 0.05, 95% CI 0.69 to 0.88). ROC curve analysis for delta mHLA-DR provided a 1.2 variation in mHLA-DR expression (days 3 and 4/days 1 and 2) as the best cutoff value to discriminate between septic and non-septic patients. At that threshold, the test had an 83% sensitivity, a 61% specificity, a 42% positive predictive value, and an 87% negative predictive value. At days 5 and 6, no significant difference was observed in mHLA-DR expression or in the mHLA-DR slope (days 5 and 6/days 3 and 4) between the two patient groups.

Multivariate logistic regression analysis indicated, after adjustment for other confounding factors, that an mHLA-DR ratio of days 3 and 4/days 1 and 2 less than or equal to 1.2 was associated with sepsis to a highly significant degree (adjusted OR 5.41, 95% CI 1.42 to 20.52) (Table 2).