Background
Sepsis is one of the main causes of morbidity and mortality in the intensive care unit (ICU). In the United States, approximately 500,000 people die of sepsis each year [
1,
2]. Because of the prevalence of patients with fever, leukocytosis, and increased heart and respiratory rates in the ICU as well as the lack of sensitivity and specificity of these parameters, it has been a challenge for physicians to diagnose sepsis by these criteria alone [
3,
4]. Blood culture has always been held as the gold standard for sepsis diagnosis [
5]. However, treatments are normally delayed while waiting for lab results. Therefore, it has been a difficult task for ICU physicians to accurately differentiate bacteremia from similar diseases and come to a rapid diagnosis. In addition, clinical studies have shown that early intervention can significantly improve the prognosis of severe sepsis and septic shock and lower the mortality rate [
6,
7]. Thus, it becomes critical to identify any biomarker that can accurately identify infection or noninfection as well as quickly diagnose bacteremia and predict its prognosis. In this study, we evaluated the value of soluble triggering receptor expressed on myeloid cells 1 (sTREM-1), procalcitonin (PCT), and C-reactive protein (CRP) serum levels in terms of their value for sepsis diagnosis, identification of new fever caused by bacteremia in ICU patients, and prediction of prognosis.
Methods
The study was performed from September 2009 to March 2011 in the Respiratory ICU, Surgical ICU, and Emergency ICU of the Chinese People’s Liberation Army (CPLA) General Hospital. This study was approved by the Ethics Committee of the CPLA General Hospital (project No. 20090923–001) and was registered with the U.S. National Institutes of Health Clinical Trials Registry (NCT01410578). Patients or their family members were fully informed and signed informed consent forms.
Inclusion and exclusion criteria
Based on the 1991 ACCP/SCCM Sepsis Directory [
8] and the diagnostic criteria advanced by the 2001 International Sepsis Definition Conference [
9], patients exhibiting two or more of the following signs during their first 24 h in the ICU were eligible for selection: (1) temperature of >38°C or <36°C, (2) pulse rate of >90 beats/min, (3) respiratory rate of >20 breaths/min or hyperventilation with a partial pressure of arterial carbon dioxide (PaCO
2) of <32 mmHg, or (4) white blood cell (WBC) count of >12,000μL
-1 or <4000 μL
-1, or >10% immature cells. Exclusion criteria were: (1) <18 years of age, (2) acquired immunodeficiency syndrome, (3) reduced polymorphonuclear granulocyte count (<500 μL
-1), or (4) died within 24 h after admission into the ICU, refused to participate in the study, or declined treatment during the observation period.
Infection and bacteremia definition
The presence of infection, defined according to the clinical and microbiological criteria of the CDC definitions [
10,
11], was held as a gold standard and determined by three independent experts who were blinded to the CRP, PCT, and WBC results [
12]. Blood culture-positive bacteremia was defined as growth of bacteria with recognized pathogenic capacity in one blood culture or as growth of common skin pathogens (i.e., coagulase-negative
Staphylococcus species, diphtheroids,
Bacillus species,
Propionibacterium species, or micrococci) in two blood cultures [
13].
New fever in ICUs
Fever is a common problem in ICU patients and is mainly traced to multiple infectious or noninfectious causes [
14]. The definition of fever is arbitrary and depends on the purpose for which it is defined. Guidelines for evaluation of new fever in critically ill adult patients define fever in the ICU as a temperature of >38.3°C [
15,
16]. Taking blood cultures from the patient is the initial step in addressing fever [
17]. In this study, “new fever” refers to a new fever (temperature >38.3°C) that occurred more than 48 h after ICU admission.
Clinical data and sample collection
Demographic and disease data of the patients admitted to the ICU included age, gender, chief complaint upon admission, major diagnosis, clinical manifestation, Acute Physiology and Chronic Health Evaluation II (APACHE II) score, Sequential Organ Failure Assessment (SOFA) score, WBC counts, CRP levels, PCT levels, sTREM-1 levels, cause of infection, initial disease status, and survival during the first 28 days. When the patients were admitted to the ICU, 3 mL of intravenous blood was collected within the first 24 h. The blood was centrifuged at 3000 rpm/min at 4°C for 15 min. Samples were collected again if the patient’s temperature reached >38.3°C more than 48 h after admission [
14,
17]. According to the blood culture standards of the Clinical Laboratory Standards Institute (CLSI) [
18], two sets of blood samples were collected from two different sites and were inoculated separately into a Bact/ALERT aerobic bottle and a Bact/ALERT anaerobic bottle (bioMérieux, Marcy-l'Etoile, France).
Serological assays and blood culture procedure
All specimens were renumbered before the experiment. The researchers were relatively blind to each step. sTREM-1 levels were determined in duplicate using a double antibody sandwich enzyme-linked immunosorbent assay (ELISA) (Quantikine Human TREM-1 Immunoassay kit; R&D Systems, Minneapolis, MN, USA). PCT was measured using enzyme-linked fluorescence analysis (ELFA) (VIDAS BRAHMS PCT kit; bioMérieux). CRP was determined using scattering turbidimetry (CardioPhase hsCRP; Siemens Healthcare Diagnostics, Deerfield, IL, USA). A BacT/ALERT 3D automation system (bioMérieux) was used to continuously monitor the blood samples in the blood culture bottles. As soon as bacterial growth was detected, the positive bottle was sent for single-colony subculture. Pathogens were identified manually or using the VITEK II system, and bacterial contaminants were excluded [
19,
20]. All procedures were strictly conducted following the manufacturers’ instructions and standard microbiology guidelines.
Statistical analysis
SPSS statistical software v16.0 (SPSS, Chicago, IL, USA) was used for data analysis. The normally distributed variances were expressed as the mean ± SD. Student’s t-test was used to compare mean values between two groups. The data that were not normally distributed were expressed as medium (interquartile ranges) and analyzed using the rank sum test. Unordered categorical variables were expressed as percentages, and the difference in proportion between two groups was analyzed using the chi-square test. A receiver operating characteristic (ROC) curve was employed to evaluate the effects of sTREM-1, PCT, and CRP levels on sepsis diagnosis, new fever caused by bacteremia, and the prognosis of patients with bacteremia.
Discussion
TREM-1 is a recently discovered member of the immunoglobulin superfamily of receptors that is specifically expressed on the surfaces of neutrophils and monocytes. sTREM-1 is a soluble form of TREM-1 and is released into bodily fluids when TREM-1 is upregulated. Many studies have indicated that sTREM-1 could be a valuable diagnostic biomarker for various infectious diseases [
21‐
28]. Dynamic changes in sTREM-1 levels could predict survival and mortality of patients at the early stage of sepsis [
28‐
30]. PCT expression is related to the severity of bacterial infection; hence, it is one of the biomarkers for infection [
31,
32]. PCT measurements along with other clinical tests for infection might be valuable for determining the prognosis of patients [
33,
34]. CRP is a biomarker involved in variety of inflammatory diseases. Although the majority of hospitals can widely implement CRP analysis in sepsis diagnosis and prognosis [
35,
36], whether CRP is a good biomarker for early diagnosis of sepsis or bacteremia is still controversial [
37,
38]. Our study found that sTREM-1, PCT, and CRP levels indicate infection, while sTREM-1 and PCT levels predict prognosis. However, none of these parameters brings to light the cause of new fever. Moreover, sTREM-1 is the best indicator for diagnosis of sepsis and assessment of prognosis of blood culture-positive bacteremia.
Rivera-Chavaz
et al.[
39] performed a study involving 93 patients in the ICU with SIRS symptoms and suspected infection. The patients were classified as having SIRS (no infection; n = 37) or sepsis (n = 56) according to the diagnosis of the physician in charge and clinical evidence. Patients with sepsis had significantly higher sTREM-1 levels than did those with SIRS. At a cut-off of 30 pg/mL, sTREM-1 correctly identified patients suffering from infection with 96% sensitivity and 91% specificity. Porfyridis
et al. [
40] enrolled 68 patients with acute respiratory illness. A total of 34 patients were diagnosed with community-acquired bacterial pneumonia and 34 with nonbacterial pulmonary disease. sTREM-1 levels were significantly higher in the pneumonia group than in the nonbacterial pulmonary disease group, and this analysis was more sensitive and specific than analysis with CRP levels. Yong J
et al.[
41] performed a meta-analysis of 13 clinical studies that fulfilled the inclusion criteria (980 patients; 557 patients with bacterial and 423 with nonbacterial infections). They found that sTREM-1 level for the diagnosis of infection in the AUC of the summary ROC was 0.86, with a sensitivity of 0.82 and specificity of 0.86. This finding confirmed that sTREM-1 is a reliable biomarker for bacterial infection. However, other studies argue that sTREM-1 is of no value for infection diagnosis [
42‐
44]. Some experts have suggested that CRP and PCT levels are more sensitive than sTREM-1 as biomarkers for the diagnosis of bacterial infection [
45,
46]. Our study, on the other hand, found that on the day of ICU admission, the sepsis group had higher sTREM-1, PCT, and CRP levels and APACHE II scores than did the SIRS group. The indicators above, to different degrees, have values in sepsis identification or diagnosis, of which sTREM-1 proves most efficient. Data obtained through our study are quite similar to those previously reported.
There have been numerous studies on PCT. Stefan
et al.[
47] enrolled 295 patients whose blood culture samples were collected at the emergency department. Based on the blood culture results, the patients were categorized into blood culture-positive, blood culture-negative, and blood culture-contaminated groups. The results indicated that PCT is the most valuable biomarker for the diagnosis of sepsis and bacteremia for patients in the emergency department. Kim
et al.[
48] and Lai
et al. [
38,
49] also arrived at the same conclusion: that PCT levels are the most meaningful parameter for bacteremia diagnosis and for patients with bacteremia and a high fever in the emergency department. On the contrary, CRP levels have no value for diagnosis of patients with bacteremia and a high fever in the emergency department. However, Blijlevens
et al.[
50] questioned the value of PCT levels in sepsis diagnosis. Ruiz-Gonzalez
et al.[
51] recently suggested that sTREM-1 levels are valuable for diagnosing bacteremia with community-acquired pneumonia. Because fewer studies have shown the value of sTREM-1 in the diagnosis of bacteremia, we divided ICU patients with fever into blood culture-positive bacteremia and blood culture-negative groups based on their blood culture results. We found that sTREM-1 and PCT levels in the two groups were very comparable. Interestingly, CRP levels were significantly higher in the blood culture-negative group than in the bacteremia group (
P = 0.033). We speculate that this occurred because of the rise in CRP reactivity rather than because of the bacteremia itself. Our study implies that sTREM-1, CRP, and PCT levels may possess no clinical value in determining whether a given septic patient is complicated with bacteremia on the grounds of a new fever.
Gibot
et al. [
30] studied sTREM-1 levels in 63 patients with severe sepsis and found that decreased sTREM-1 plasma levels were positively correlated with a better prognosis. Thus, sTREM-1 is an excellent biomarker for the prognosis of sepsis. Furthermore, our previous studies and Gibot S
et al. [
28‐
30] found that compared with PCT and CRP levels, dynamic changes in sTREM-1 levels better predict the prognosis of sepsis. However, no studies have reported whether sTREM-1, PCT, or CRP levels are good parameters for the prognosis of bacteremia. We herein showed that sTREM-1 and PCT levels were useful for the prognosis of blood culture-positive bacteremia. For example, within the bacteremia group, sTREM-1 and PCT levels were significantly higher in nonsurvivors than in survivors. However, we failed to find any prognostic value of sTREM-1 or PCT for blood culture-negative sepsis. Moreover, according to the ROC curves of sTREM-1 and PCT levels, sTREM-1 is a more ideal predictor for the prognosis of blood culture-positive bacteremia. Thus, physicians are expected to pay close attention to patients with high levels of sTREM-1.
Our study was, however, limited by the following factors. (1) The study involved a small sample size; only 33 patients with a positive blood culture were enrolled. (2) Only those who developed a new fever (>38.3°C) in the ICU were enrolled. Therefore, we cannot exclude the possibility that patients who did not have a high fever also had bacteremia; unfortunately, however, no such patients were involved in our analyses. (3) The incidence of opportunistic infections is higher in the ICU, and many pathogens circulate in the wards. Consequently, the possibility of false-positive blood cultures could not be excluded. Furthermore, the patients received regular antibiotic treatments, and side effects of such a long-term treatment should not have been ignored. (4) It is well known that more than two-thirds of patients with severe bacterial sepsis have negative blood cultures [
52,
53], but this fact (blood culture negative bacteremia) was not properly addressed in the study.
Conclusions
sTREM-1, PCT, and CRP levels are of substantial value for the early diagnosis of sepsis. In addition, sTREM-1 levels can reflect the infection status more accurately and more specifically than can CRP and PCT levels. However, none of these three parameters could be used to determine whether the fever of patients with sepsis was caused by bacteremia. Finally, sTREM-1 and PCT levels could aid in predicting the prognosis of bacteremic patients. Regrettably, the sample size of this study was small; larger studies are thus needed to further evaluate the value of sTREM-1, PCT, and CRP levels in the diagnosis of bacteremia.
Acknowledgements
Written consent for publication was obtained from the patients or their relatives.
We express our thanks to the staff members of the RICU, SICU, and EICU of the CPLA General Hospital for their assistance in this study.
Funding
This study was supported by the Chinese National Science and Technology Pillar Program (No. 2009BAI86B03). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests
All authors declare that they have no competing interests.
Authors’ contributions
LS and CL designed the study, performed the data analysis, and wrote the first manuscript draft. PY, JD, and YJ carried out the study in the SICU, EICU, and RICU, respectively. BH, LL and CJ performed the experiments. DF performed the statistical work. LX was responsible for protocol revisions, data analysis, and final draft revision. All authors have read and approved the final manuscript.