SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

ACADEMIC EMERGENCY MEDICINE 2011; 18:807–815 © 2011 by the Society for Academic Emergency Medicine

Abstract

Objectives:  The objectives were to evaluate the diagnostic accuracy for sepsis in an emergency department (ED) population of the cluster of differentiation-64 (CD64) glycoprotein expression on the surface of neutrophils (nCD64), serum levels of soluble triggering receptor expressed on myeloid cells-1 (s-TREM-1), and high-mobility group box-1 protein (HMGB-1).

Methods:  Patients with any of the following as admission diagnosis were enrolled: 1) suspected infection, 2) fever, 3) delirium, or 4) acute hypotension of unexplained origin within 24 hours of ED presentation. Levels of nCD64, HMGB-1, and s-TREM-1 were measured within the first 24 hours of the first ED evaluation. Baseline clinical data, Sepsis-related Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation (APACHE II) score, daily clinical and microbiologic information, and 28-day mortality rate were collected. Because there is not a definitive criterion standard for sepsis, the authors used expert consensus based on clinical, microbiologic, laboratory, and radiologic data collected for each patient during the first 7 days of hospitalization. This expert consensus defined the primary outcome of sepsis, and the primary data analysis was based in the comparison of sepsis versus nonsepsis patients. The cut points to define sensitivity and specificity values, as well as positive and negative likelihood ratios (LRs) for the markers related to sepsis diagnosis, were determined using receiver operative characteristics (ROC) curves. The patients in this study were a prespecified nested subsample population of a larger study.

Results:  Of 631 patients included in the study, 66% (95% confidence interval [CI] = 62% to 67%, n = 416) had sepsis according with the expert consensus diagnosis. Among these sepsis patients, SOFA score defined 67% (95% CI = 62% to 71%, n = 277) in severe sepsis and 1% (95% CI = 0.3% to 3%, n = 6) in septic shock. The sensitivities for sepsis diagnosis were CD64, 65.8% (95% CI = 61.1% to 70.3%); HMGB-1, 57.5% (95% CI = 52.7% to 62.3%); and s-TREM-1, 60% (95% CI = 55.2% to 64.7%). The specificities were CD64, 64.6% (95% CI = 57.8% to 70.8%), HMGB-1, 57.8% (95% CI = 51.1% to 64.3%), and s-TREM-1, 59.2% (95% CI = 52.5% to 65.6%). The positive LR (LR+) for CD64 was 1.85 (95% CI = 1.52 to 2.26) and the negative LR (LR–) was 0.52 (95% CI = 0.44 to 0.62]; for HMGB-1 the LR+ was 1.36 (95% CI = 1.14 to 1.63) and LR– was 0.73 (95% CI = 0.62 to 0.86); and for s-TREM-1 the LR+ was 1.47 (95% CI = 1.22 to 1.76) and the LR– was 0.67 (95% CI = 0.57 to 0.79).

Conclusions:  In this cohort of patients suspected of having any infection in the ED, the accuracy of nCD64, s-TREM-1, and HMGB-1 was not significantly sensitive or specific for diagnosis of sepsis.

Sepsis is an important cause of morbidity and mortality worldwide. In the United States it occurs in 0.3% of the general population, with approximately 750,000 cases per year.1,2 The overall mortality is around 30%, rising to 50% in patients with severe sepsis and septic shock.3 Previous reports have shown that early therapy improves survival and clinical outcomes.4–6 However, the early identification of septic patients in the ED remains a challenge and is based on the original definition of sepsis stated in 2001 at the Society of Critical Care Medicine/European Society of Intensive Care Medicine/American College of Chest Physicians/American Thoracic Society/Surgical Infection Society International Sepsis Definitions Conference.7 These symptoms and signs are not specific, and most of them appear in advanced stages of sepsis, which urgently calls for research to develop tools to diagnose sepsis early in the first encounter with patients in the ED.

Three promising diagnosis biomarkers for the systemic inflammatory response syndrome (SIRS) that helps define sepsis are the cluster of differentiation-64 (CD64), soluble triggering receptor expressed on myeloid cells-1 (s-TREM-1), and high-mobility group box-1 (HMGB-1), which have been studied mainly in the intensive care unit (ICU) setting.8–10 The expression of CD64 is almost exclusively restricted to monocytes and macrophages and is virtually undetectable on mature neutrophils from healthy individuals. It increases within hours both in vitro and in vivo in the presence of mediators of inflammation such as interferon (IFN)-γ and granulocyte–colony-stimulating factor (G-CSF), and it also increases rapidly in infected patients.11–13 TREM-1 can be released as a soluble molecule in response to microbial products, before the occurrence of the classical clinical findings of sepsis. Some reports have shown high levels of s-TREM-1 in biologic fluids of patients with infections.14–17 Furthermore, Gibot et al.18 found that s-TREM-1 levels were elevated in all nonsurviving patients, whereas s-TREM-1 levels rapidly decreased in survivors during the first week after study admission. HMGB-1, a cytokine-like mediator of inflammation, is released later in the inflammatory process, and clinical reports reveal that its levels are increased significantly in critically ill patients with sepsis.19,20

The combined characterization of these biomarkers that overlap across the sepsis spectrum from early to late disease could help to identify the septic patient who will benefit from early treatment in the ED. Consequently, we conducted a clinical study to determine the diagnostic accuracy of CD64, s-TREM-1, and HMGB-1 in the diagnosis of sepsis in the emergency department (ED).

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

Study Design

This was a cross-sectional study with prospective data collection to determine the operative characteristics of diagnostic tests. The patients in this study were a prespecified nested subsample of the population of a larger study, “Toward an Operative Diagnosis in Sepsis: A Latent Class Approach.”21 The local scientific ethical committee approved sample collection on the basis of verbal informed consent.

Study Setting and Population

The study was conducted at the ED of the Hospital Universitario San Vicente de Paúl, Colombia. This is a 550-bed, tertiary care university hospital with an admission rate of approximately 1,800 patients per month through the ED and is a reference institution for a region including approximately 3 million habitants.

Enrollment began on July 2007 and concluded on September 2008. All patients were 18 years or older and were recruitment to the study within 24 hours of ED admission. We considered patients to have possible sepsis syndrome if they had at least one of the following criteria as the admission diagnosis to the hospital: 1) suspected or confirmed infection, 2) fever of unknown origin, 3) delirium or any type of encephalopathy of unknown origin, or 4) acute hypotension not explained by hemorrhage, myocardial infarction, stroke, or heart failure. We selected these relatively broad criteria according with the last consensus conference on sepsis definitions,7 addressed to provide a wide framework to define the systemic response to infection, expanding the list of signs and symptoms of sepsis to reflect clinical bedside experience. Exclusion criteria were: 1) refusal by the patients, their families, or the attending physician to be part of the study; 2) antimicrobial treatment received at another medical institution immediately before admission to the study; 3) medical decision to treat the patient ambulatory or in a different institution within 24 hours after admission; 4) homeless or inability of the patient to follow up; and 5) previous participation in the same study.

Study Protocol

Three physicians (FJ, GDLR, or MLV) and two trained nurses recruited patients by checking admission lists and clinical records and collected data daily from Monday to Saturday of each week. The general protocol for each patient after recruitment was: collection of baseline clinical data,21 calculation of entrance Sepsis-related Organ Failure Assessment (SOFA) score22 and Acute Physiology and Chronic Health Evaluation (APACHE II) score,23 and blood sampling (two ethylenediaminetetraacetate [EDTA] and two anticoagulant-free tubes with separation gel), all within 24 hours of the first ED evaluation. During the first 7 days of hospital stay, the patients were monitored with daily recording of any relevant data registered in medical or nurse records, using a standardized case report form.

The serum samples, in anticoagulant-free tubes with separation gel, were separated by centrifugation at 450 g for 20 minutes within 2 hours after collection to avoid falsely elevated levels of HMGB-1 secondary to passive release from the intracellular compartment.24 They were then stored at –80°C for later analysis of s-TREM-1 and HMGB-1. The measurement of CD64 from EDTA tubes was made less than 15 hours after blood extraction. Blood specimens remain acceptable for up to 8 hours when held at room temperature (18 to 22°C) or for 48 hours when refrigerated (2 to 8°C).25 Laboratory technicians unaware of any clinical data performed all the procedures. The handling of samples and the techniques for biomarker measurement described below were similar to those used in previous reports.10,26,27

HMGB-1 and s-TREM-1 Serum Concentrations.  Serum HMGB-1 levels were determined using commercially available enzyme-linked immunosorbent assay (ELISA) following the manufacturer’s instructions (human HMGB-1 kit Cat. No. 326052202, Shino-Test Corporation, Kanagawa, Japan). The lower limit of detection of the test is 1 ng/mL. The cross-reaction to HMGB-2 is less than 2% and the coefficient of variation was 10%. s-TREM-1 serum concentration was determined by ELISA (Quantikine kit, human TREM-1 immunoassay, Cat. No. DTRM10, R&D Systems, Inc., Minneapolis, MN). The lower limit of detection of the test is 3.88 pg/mL. ELISA results were quantified using a microplate reader set PowerWave X (Biotek Instruments Inc, Winooski VT). The inter- and intraassay coefficients of variation for the s-TREM-1 and HMGB-1 tests were <10%.

CD64 on the Surface of Neutrophils.  Neutrophil CD64 (nCD64) levels were measured with the Leuko 64 kit (LK64-250, Trillium Diagnosis, LLC, Brewer, ME) using a Coulter Epic XL flow cytometer (Coulter Corp., Miami FL). The data were analyzed in the Leuko64 software included in the kit. The results were expressed in molecules of equivalent soluble fluorochrome (MESF) units28 and then compared with the expression of positive control (monocytes) and negative control (lymphocytes).

Outcome Measures

Because there is not a definitive criterion standard for sepsis, we used an expert consensus based on clinical, microbiologic, laboratory, and radiologic data collected for each patient during the first 7 days of hospitalization. The experts also took into account the definitions stated in 2001 at the International Sepsis Definitions Conference,7 as well as the formal Centers for Disease Control and Prevention (CDC) definitions for infection.29 The consensus was formed by a panel of three physicians with certified training and expertise in intensive care (AG), internal medicine (CMA), and infectious diseases (CIG). First, each physician established a diagnosis individually, in which they agreed on 65% of the cases. The remaining 35% of the patients were fully discussed to determine a final diagnosis. All the experts were blinded to the results of s-TREM-1, HMGB-1, and CD64 glycoprotein expression on the surface of neutrophils (nCD64). The consensus classified the admitted patients into sepsis and nonsepsis groups. Among the sepsis patients, severe sepsis was defined as a SOFA score higher than 1 in any domain of organ dysfunction, and septic shock was defined as a score of at least 2 in the cardiovascular domain.30 Determination of mortality for patients not at the hospital at 28 days was done either by outpatient control or phone calls to each patient or their first relatives. The outpatient control was defined for some patients according to their clinical condition at discharge. In those patients, either the attending physician or one of the physician investigators did a common clinical interview and a physical examination. For the rest of the participants, phone calls were made by research nurses assigned to the study, asking to talk with each patient to ask about his or her general health condition. When they were not available, the nurses asked other family members about the patient’s health status and vital condition.

Data Analysis

The primary analysis was sepsis versus nonsepsis condition according to consensus diagnosis. The cut points for the study tests were determined using receiver operative characteristics (ROC) curves,31 searching for the best sensitivity without a significant decrease in specificity. The standard method based in Bayes theorem32 was used to determine the operative characteristics of the tests against the consensus definition. According to the method proposed by Buderer,33 and assuming a prevalence of disease of 60%, a sample size of 600 patients has a 95% power to detect both sensitivity and specificity higher than 90%. Exact 95% confidence intervals (CIs) were estimated for values of sensitivity, specificity, predictive values, and likelihood ratios (LRs). Additional analyses combining pairs of tests (HMGB-1/s-TREM-1, HMGB-1/CD64, and s-TREM-1/CD64) were done. These pairs were calculated trying to combine appropriately “prognosis” (HMGB-1) and “inflammation” (s-TREM-1 and CD64) markers. For these analyses, a pair was considered positive if both biomarkers were above the cut point and negative otherwise. Furthermore, as a sensitivity analysis, alternative reference standards for sepsis patients were considered: 1) only those who had positive blood culture, 2) only those who had any microbiologically confirmed infection, and 3) only those who were diagnosed as sepsis patients independently by one of the experts (65% among the total sepsis population). Analyses were performed with STATA SE (Version 10, StataCorp, College Station, TX).

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

During the study period, a total of 1,340 patients met the inclusion criteria, and 709 were excluded (Figure 1). Among 631 patients included, 580 (92%) had an infection as admission diagnosis, 31 (5%) fever of unknown cause, 16 (2.5%) delirium or encephalopathy of unknown origin, and 4 (0.5%) unexplained hypotension.

image

Figure 1.  Flow chart of recruitment, patients’ classification and biomarkers values. Data are presented as samples available, median (IQR). *Due to logistic or technical reasons, 15 individuals did not have samples available for analysis of any marker, and six individuals had serum available for analysis but no samples for nCD64 analysis. CD64 = cluster of differentiation-64; HMGB-1 = high-mobility group box-1 protein; IQR = interquartile range; MESF = molecules of equivalent soluble fluorochrome; nCD64 = CD64 glycoprotein expression on the surface of neutrophils; s-TREM-1 = soluble triggering receptor expressed on myeloid cells-1.

Download figure to PowerPoint

There were 323 females (51%), the median age was 51 years (interquartile range [IQR] = 36 to 68 years), and the median time of symptoms before consultation was 72 hours (IQR = 24 to 192 hours). There was no comorbidity in 333 (53%) of the participants, and the most frequent previous diseases were diabetes mellitus (n = 116, 18%), chronic obstructive pulmonary disease (n = 69, 11%), chronic renal failure (n = 69, 11%), history of cancer during the past year (n = 54, 8.5%), and use of corticosteroids or chemotherapy during the past 3 months (n = 50, 8%), some of these in the same patient. The median hospital length of stay was 9 days (IQR = 5 to 17 days), 50 patients (8%) required treatment in ICU, and the overall 28-day mortality rate was 12% (95% CI = 9% to 14%, n = 74). The main clinical and laboratory characteristics by group are shown in Table 1.

Table 1.    Clinical Characteristics According to the Groups Defined by the Expert Consensus (Reference Standard)
Clinical CharacteristicsSepsis (n = 416)No Sepsis (n = 215)
  1. Data are presented as median (IQR, observations available).

  2. APACHE II = Acute Physiology and Chronic Health Evaluation; SOFA = Sepsis-related Organ Failure Assessment score; MAP = mean arterial pressure; WBC = white blood cell.

  3. *Overall 28-day mortality rate was determined by outpatient control or phone calls.

SOFA score3 (2–4, 416)2 (1–4, 215)
APACHE II score11 (6–16, 416)8 (4–12, 215)
28-day mortality rate*56 (14%)18 (8%)
Temperature (°C)37 (36–38, 387)37 (36.5–37, 195)
Heart rate (beats/min)100 (88–115, 405)85 (77–101, 208)
Respiratory rate (breaths/min)22 (20–28, 97)20 (17–24, 33)
MAP (mm Hg) 104 (91–120, 405)110 (99–128, 208)
WBC count (×106 cells/L)12,850 (9250–17900, 412)10,000 (7600–11800, 214)
Neutrophils (%)82 (74–89, 412)75 (65–83, 213)
Hemoglobin (g/dL)12 (10–14, 412)13 (11–14, 213)
Creatinine (mg/dL)1 (0.8–1.8, 412)0.9 (0.8–1.3, 212)
Lactic acid (mg/dL)17.4 (11–25.8, 406)14.2 (9.5–19.1, 204)
Bilirubin (mg/dL)0.7 (0.5–1.1, 405)0.7 (0.4–1.1, 209)
PaO2/FiO2 (Torr)299 (211–360, 404)335 (253–395, 203)

A total of 416 patients were defined by expert consensus as having sepsis. The kappa-statistic measure of inter-rater agreement was 0.69 between experts 1 and 2, 0.64 between experts 2 and 3, and 0.69 between experts 1 and 3. Among sepsis patients, 67% (95% CI = 62% to 71%, n = 277) had severe sepsis according to the SOFA score, and six patients (1%, 95% CI = 0.3% to 3%) had septic shock. The main admission diagnosis in the sepsis group was community-acquired pneumonia (n = 93, 22%), followed by urinary tract infection (n = 67, 16%), soft tissue infection (n = 65, 16%), clinical sepsis defined according to the CDC definitions of nosocomial infections29 (n = 55, 13%), and intraabdominal infection (n = 35, 8%). A microbiologic diagnosis was confirmed in 185 (44%) sepsis patients. The values of each biomarker according to the clinical diagnosis in septic patients are shown in Table 2.

Table 2.    Main Admission Diagnosis in Sepsis Patients and Its Respective Biomarkers Value
Main Diagnosis at Admission (n, %)CD64 (MESF)HMGB-1 (ng/mL)sTREM-1 (pg/mL)
  1. Data are presented as median (IQR, samples available).

  2. CDC = Centers for Disease Control and Prevention; HMGB-1 = high-mobility group box-1 protein; MESF = molecular equivalent soluble fluorochrome;28 s-TREM-1 = soluble triggering receptor expressed on myeloid cells.

  3. *Clinical sepsis was defined according to the CDC definitions of nosocomial infections:29 patient has at least one of the following clinical signs or symptoms with no other recognized cause: fever (38°C), hypotension (systolic pressure of 90 mm Hg), or oliguria (20 cm3/hr) and blood culture not done or no organisms or antigen detected in blood and no apparent infection at another site and physician institutes treatment for sepsis.7

  4. †Other: catheter-associated infection, endometritis, other lower respiratory tract infections, tracheobronchitis, other gastrointestinal infections, upper respiratory tract infection, otitis and mastoiditis, endocarditis, meningitis or ventriculitis, arthritis, osteomyelitis, bloodstream infection, organ/space surgical site infection.

Community-acquired pneumonia (93, 22)2.1 (1.4–3.2, 92)8.2 (1.2–19, 90)185.4 (57.2–454.2,90)
Urinary tract infection (67, 16)2.8 (1.6–4.2, 65)8.4 (4.1–16, 65)220.5 (77.1–409.2, 65)
Soft tissue infection (65, 16)1.7 (1.-2.7, 62)9.6 (4.7–19, 64)212.3 (83.4–476.7, 64)
Clinical sepsis* (55, 13)3.2 (2.3–4.5, 52)7.8 (3.3–16.3, 53)411.7 (153.6–694.7, 53)
Intraabdominal infection (35,8)2.6 (1.8–3.5, 35)11.1 (5.3–23, 35)192 (81–465, 35)
Superficial surgical site infection (14,3)1.7 (1.4–2.3, 14)9.6 (6.4–15.4, 14)174 (101.2–218, 14)
Gastroenteritis (11, 3)2.7 (1.7–4.1, 11)8.7 (2.3–12, 11)111.2 (0–282.7, 11)
Reproductive tract infection (7, 2)3.6 (3–5, 7)19.7 (9.8–25.8, 7)264.3 (55.3–354.3, 7)
Deep surgical site infection (6, 1)1.5 (0.9–2.3, 6)8.7 (16.2–4.7, 6)115 (0–190, 6)
Skin infection (5, 1)1.9 (0.9–2.3, 5)4.2 (1.9–9.8, 5)74.7 (15.-116.4, 5)
Catheter-associated urinary tract infection (5, 1)3.4 (2.1–4.7, 5)9.7 (4.4–10, 5540.8 (253–758.3, 5)
Other† (53, 14)2.87 (2.1–3.4, 50)12.5 (9.4–14.9, 50)219.4 (168–395.3, 50)

Due to logistic or technical reasons, 15 individuals did not have samples available for analysis of any marker, and six individuals had serum available for analysis but no samples for nCD64 analysis. The expression of CD64 on neutrophil membranes in the samples from 610 patients had a mean (±SD) of 2.45 (±1.75) MESF units. The serum concentrations of HMGB-1 and s-TREM-1, which were determined in 616 patients, had means (±SDs) of 11.9 (±15.7) and 290.4 (±391.9) pg/mL, respectively. The ROC curve analyses for these biomarkers showed areas under the curve of 0.706 (95% CI = 0.664 to 0.748) for CD64, 0.599 (95% CI = 0.552 to 0.647) for HMGB-1, and 0.614 (95%CI = 0.567 to 0.661) for s-TREM-1 (Figure 2). According to the ROC curve analysis, the cutoffs points with the best sensitivity and specificity for CD64, HMGB-1, and s-TREM-1 were 1.7 MESF units, 7 ng/mL, and 135 pg/mL, respectively. Their operating characteristics are shown in Table 3. Analyses combining pairs of tests (HMGB-1/s-TREM-1, HMGB-1/CD64, and s-TREM-1/CD64) did not show any improvement in diagnostic accuracy. Similar results were seen using other end points or alternative reference standards (data not shown).

image

Figure 2.   ROC curve analysis for CD64, HMGB-1, and s-TREM-1 in patients with suspected infection with sepsis and nonsepsis according to the experts’ consensus as criterion standard. CD64 = cluster of differentiation-64; HMGB-1 = high-mobility group box-1 protein; ROC = receiver operating characteristics; s-TREM-1 = soluble triggering receptor expressed on myeloid cells.

Download figure to PowerPoint

Table 3.    Diagnostic Accuracy of Biomarkers for Sepsis
Operative Characteristics (95% CI)CD64 (1.7 MESF)HMGB-1 (7 ng/mL)sTREM-1 (135 pg/mL)
  1. CD64 = cluster of differentiation-64; HMGB-1 = high-mobility group box-1 protein; LR+ = likelihood ratio of a “positive” test; LR– = likelihood ratio of a “negative” test; MESF = molecules of equivalent soluble fluorochrome;28 NPV = negative predictive value; PPV = positive predictive value; s-TREM-1 = soluble triggering receptor expressed on myeloid cells.

Sensitivity65.8% (61.1–70.3)57.5% (52.7–62.3)60% (55.2–64.7)]
Specificity64.6% (57.8–70.8)57.8% (51.1–64.3)59.2% (52.5–65.6)
PPV78.5% (73.8–82.5)72.4% (67.2–77)73.9% (68.9–78.3)
NPV49.1% (43.2–55.6)41.5% (36–47.2)43.6% (37.9–49.3)
LR+1.85 (1.52–2.26)1.36 (1.14–1.63)1.47 (1.22–1.76)
LR–0.52 (0.44–0.62)0.73 (0.62–0.86)0.67 (0.57–0.79)

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

The biomarkers nCD64, s-TREM-1, and HMGB-1 have been previously studied in the diagnosis of sepsis, mainly in the ICU setting.34–36 Although some studies have been performed in recently admitted patients,26 a definitive conclusion about their accuracy in the ED for early diagnosis of sepsis would require further research in different populations. In our investigation, none of the markers was able to discriminate septic from nonseptic patients admitted through the ED. These findings are not explained by differences in preanalytical variables, the experimental setup, or the accuracy of test results, as we used the same procedures previously stated in the literature.10,26,27

The soluble triggering receptor expressed on myeloid cells (s-TREM-1) is a recently identified molecule involved in the inflammatory response.37,38 Although the membrane-bound TREM is substantially up-regulated during infection,34,37 the soluble form seems specifically released in pneumonia and sepsis, where high levels of s-TREM are observed in serum samples from septic shock patients but not in controls. Additionally, s-TREM-1 has been found to be more sensitive and specific than C-reactive protein (CRP) and procalcitonin (PCT) in sepsis diagnosis in the ICU.9,15 However, these results are controversial, because recently Bianchi et al.39 reported that CRP and erythrocyte sedimentation rate performed better than s-TREM-1 in diagnosing infection in the ICU, with specificity (60%) and sensitivity (70%) similar to those obtained in our study. In addition, in critical patients admitted with SIRS, s-TREM-1 had poor discriminative power to identify patients with infection, and s-TREM-1 levels did not add diagnostic information to that provided by other routinely available clinical tests.40 Finally, Bopp et al.41 reported patients with SIRS, severe sepsis, or septic shock where plasma s-TREM-1 levels were not elevated, compared with healthy controls, and these levels were not able to distinguish between survivors and nonsurvivors. Our results, in addition to those reports, suggest that the role of s-TREM as diagnostic marker in sepsis should be carefully verified.

Since neutrophils are activated early in both infectious and inflammatory conditions, determining the neutrophil response to cytokines by measuring the cell surface expression of molecules like CD64 is a way to get an in vivo report of immunological status. Accordingly, CD64 was found to be increased in bacterial infections when cultures for bacteria were positive in comparison with negative cultures,11,42,43 and it also has proved to be useful in other situations like autoimmune diseases to distinguish infections from autoimmune flares.12,36 Two studies showed that in neutrophils, but not in monocytes, an elevated CD64 level can distinguish infected versus not infected adult patients.44,45 Further studies have shown that CD64 in neutrophils is higher in patients from the ICU with septic shock than in controls8,46,47 and that those patients also had elevated levels of IFN-γ and G-CSF in serum.46 Furthermore, there were even higher levels in sepsis patients with acute respiratory distress syndrome, which suggests a positive correlation between CD64 and the severity of sepsis.48 Davis and Bigelow49 assessed the relative ability of flow cytometry to measure the expression of CD64, neutrophil counts, and myeloid immaturity differential counts on an automated hematology analyzer, correlating with the presence of infection determined by a retrospective clinical scoring system of infection or sepsis. Relative to the other laboratory parameters, the neutrophil CD64 had the best statistical performance in diagnosis and also provided the best separation on the basis of likelihood of infection, sepsis, or severe tissue injury. As is the case with the other biomarkers, however, we found a lower sensitivity and specificity than what was previously reported. Nevertheless, there are no studies addressing the possible difference in the expression of CD64 between Colombian and other populations. A recent meta-analysis found a good pooled sensitivity (79%) and specificity (91%) for CD64 expression as a marker of bacterial infection.50 Interestingly, they observed that the best performance was in adult patients, studies that analyzed CD64 expression by flow cytometry, and studies that diagnosed infection with a positive culture. However, we did not find any improvement in the diagnostic performance analyzing the subgroup of patients with positive blood culture as the criterion standard.

HMGB-1 was originally identified as a nuclear DNA-binding protein critical for proper transcriptional regulation.39 More recently, HMGB-1 has been found to act as a “late” inflammatory cytokine that contributes to the pathologic progression of sepsis and other inflammatory disorders.20 HMGB-1 has been proven to be a successful therapeutic target in experimental models of diverse infectious and inflammatory diseases, and it is currently a focus of attention in this field.51 Angus et al.52 showed that HMGB-1 is elevated in almost all patients with community-acquired pneumonia, and higher circulating HMGB-1 levels were associated with mortality. Extending its potential clinical role, in a clinical study of 185 individuals it was shown that levels of HMGB-1 were higher in infected patients compared with a healthy control group.53 Levels of HMGB-1 also were higher in the severe sepsis group compared with the sepsis group and in bacteremic patients compared with nonbacteremic patients. However, no differences were observed in fatal cases compared with survivors. In contrast, the same authors, in another study with 194 patients with suspected community-acquired infections, showed that there was no statistically significant difference in serum levels of HMGB-1 between the infected and the noninfected patients, using the same commercial kit we used.54 Gibot et al.10 in a recent study concluded that the measurement of HMGB-1 on Day 1 is probably debatable, because this molecule parallels other severity markers and provides no further information regarding outcome. Furthermore, in patients with severe sepsis, the kinetics of circulating HMGB-1 may differ from the classic mouse model findings, depending on the primary source of infection.55 Our study, with a higher number of individuals, confirms that serum HMGB-1 measurement does not have any role in the diagnosis of sepsis.

Recently, it has been reported that combining information from several biomarkers may significantly improve the clinician’s ability to differentiate patients with bacterial infections from those with systemic inflammation of nonbacterial origin.56 In our study, however, a combined analysis of biomarkers did not show any advantage in diagnostic accuracy compared to the individual characteristics of each one.

Limitations

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

Although this study has one of the larger samples in sepsis diagnosis worldwide, and the largest in Latin America, it also has some potential limitations. Compared to the majority of populations analyzed in sepsis trials, we have a relatively heterogeneous and young population, with less comorbidity and less severity of illness according to APACHE II and SOFA scores. These conditions suggest that our study may be subject to spectrum bias. However, this patient profile seems very close to the one encountered everyday in a common ED in our region.57 Additionally, the study was intended to recruit patients at the critical moment of the first contact in the ED and not at later and probably more severe stages as those presented at ICU admission. Furthermore, the production and release of these molecules may vary at different times according to the clinical course of the patients, and their concentrations also may be different depending on when they are measured. Although we recruited only patients within the first 24 hours of ED admission, patients often present during various time courses of their disease. Thus, the peak of these markers may have been missed based on the time outside the ED and delayed consultation. Consequently, only one determination of each biomarker could not be enough to reflect such dynamic molecules. Finally, there is a recognized lack of a consistent sepsis definition for clinical research, which makes the studies difficult to compare.58 In the current study, shortcomings in the ability to use a perfect criterion standard diagnosis of sepsis, which was necessarily subject to judgment even in the context of an expert panel, could make it difficult to determine the true sensitivity and specificity of the mediators of interest.

Conclusions

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

Our results suggest that in the context of an ED of a tertiary care hospital in a typical Latin American city, the three biomarkers evaluated do not offer any advantage over other general criteria used currently in sepsis diagnosis. The search for more reliable diagnostic markers in the early stages of sepsis is still open. The strengths and weaknesses of biomarkers must be recognized to use them rationally and in a cost-efficient way.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

The authors acknowledge Dr. Bruce Davis for the reduced-cost nCD64 Flow cytometry kit and Dr. Janet Arno for the language review of the manuscript.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References