Diagnostic value of sTREM-1 and procalcitonin levels in the early diagnosis of sepsis.

OBJECTIVE: Sepsis is still major cause of morbidity and mortality, despite improvements in diagnosis and treatment in modern medicine. Therefore, laboratory examinations that provide correct and rapid results are needed to support the diagnosis. This study was conducted to investigate value of immunological indicators procalcitonin (PCT) and soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) in differential diagnosis of patients with sepsis and systemic inflammatory response syndrome (SIRS), as well as to assess their importance in determining prognosis of patients with sepsis.
METHODS: Total of 90 patients, 38 with SIRS and 52 with sepsis, who were between the ages 20 to 92, were included in this prospectively planned study. Blood sample was collected from the patients during hospitalization and again in follow-up visit. Enzyme-linked immunosorbent assay (MyBioSource, Inc., San Diego, CA, USA) was used to measure sTREM-1, and PCT was measured using mini VIDAS B.R.A.H.M.S PCT assay (Biomerieux, S.A., Marcy-l'Étoile, France). In addition, patients were clinically assessed using Acute Physiology and Chronic Health Evaluation (APACHE) II scoring system.
RESULTS: On day of intensive care unit admission, sTREM-1 and PCT levels, as well as APACHE II score were significantly higher in sepsis group than SIRS group (p=0.001, p=0.01, p=0.001, respectively). Values of sTREM-1 and APACHE II score were higher in the patients with positive blood cultures than those with negative culture results (p=0.002, p=0.006, respectively). PCT, C-reactive protein, and sTREM-1 levels were significantly higher in nonsurviving group. In differentiation of SIRS from sepsis, sTREM-1 cut-off value ≥133 pg/mL and PCT cut-off value of 1.57 ng/mL yielded sensitivity of 71.1% and 67.33%, and specificity of 73.3% and 65.79%, respectively.
CONCLUSION: In patients with suspected sepsis, sTREM-1 and PCT can be used as indicators, in addition to scoring systems such as APACHE II and Sepsis-related Organ Failure Assessment score. However, it would be appropriate to support present findings with studies of larger series.

Sepsis is a common cause of morbidity and mortality in critically ill patients [1, 2]. Lack of specific clinical signs and symptoms for sepsis frequently leads to delay in treatment. However, accurate and timely diagnosis could allow for early treatment, reduce misuse of antibiotics and costs, limit morbidity, and improve patient outcomes. Microbiological cultures are still considered optimal for sepsis diagnosis, but this method has low sensitivity and specificity [3, 4]. In this context, it is imperative to identify ideal markers that would prove useful in differentiating sepsis from noninfectious diseases. Therefore, evaluation of sepsis biomarkers in different patient groups is still of great consequence.

C-reactive protein (CRP) has been widely used as an indicator of infection, but it is also elevated by other inflammatory conditions at early stages. Among recent potentially useful sepsis markers, procalcitonin (PCT) has been considered one of the most promising, and cumulative published reports support use of PCT measurement [3-5]. Serum triggering receptor expressed on myeloid cells-1 (TREM-1) is a recently discovered member of the immunoglobulin superfamily, expression of which on neutrophils and monocytes was upregulated by exposure to bacteria and fungi. sTREM-1 is soluble form of TREM-1 and is released from activated phagocytes into body fluids, such as plasma, pleural fluid, bronchoalveolar lavage fluid, urine, and cerebrospinal fluid. Many studies and several meta-analyses have indicated that sTREM-1 could be a valuable diagnostic biomarker for various infectious diseases [5-9].

In this study, sTREM-1, PCT, CRP serum levels and a scoring system, Acute Physiology and Chronic Health Evaluation (APACHE) II, were assessed in terms of their value for sepsis diagnosis and prediction of prognosis for hospitalized patients.

MATERIALS AND METHODS

This prospective observational study was performed in the adult medical-surgical intensive care units (ICUs) of a Training and Research Hospital, which has 725 beds, between May 2013 and January 2014. Approval of the hospital ethics committee and signed informed consent forms were obtained from patients or their families before inclusion in the study.

Patients

Consecutive hospitalized patients who had 2 or more of the following signs during their first 24 hours in the units were diagnosed as SIRS and were included in the study: temperature of >38°C or <36°C, pulse rate of >90 beats/min, respiratory rate of >20 breaths/min or hyperventilation with partial pressure of arterial carbon dioxide of <32 mmHg, or white blood cell (WBC) count of >12 000µL or <4000 µL, or >10% immature cells. Patients exhibiting 2 or more signs of SIRS with proven infection were diagnosed as sepsis. Exclusion criteria were: <18 years of age, acquired immunodeficiency syndrome, neutropenia (polymorphonuclear granulocyte count <500 µL), or died within 24 hours after admission to the hospital, elected not to participate in the study, or declined treatment during the observation period.

All medical records of the patients were retrospectively evaluated and the patients were classified as sepsis or SIRS at the time of submission by 2 clinicians blinded to biomarker results.

Data collection

Upon admission to the units, the following data were recorded for each patient: age, gender, main complaints for admission, symptoms, body temperature, leukocyte count, APACHE II score, etiological factors, and underlying diseases. Survival or death in hospital was evaluated during 28-day follow-up period.

When the patients were admitted, first blood sample was drawn within the first 24 hours and second was taken during treatment when patient had fever or mental state disorder (consistent with sepsis criteria). Blood samples were centrifuged at 3000 rpm for 15 minutes and supernatant was stored at -80ºC.

Assays

BacT/ALERT 3D automated microbial detection system (Biomerieux, S.A., Marcy-l’Étoile, France) was used for blood cultures and pathogens were identified conventionally or using automated (VITEK 2 Compact; Biomerieux, S.A., Marcy-l’Étoile, France) system. WBC was determined using CELL-DYN 3700 hematology analyzer (Abbott Laboratories, Abbott Park, IL, USA). CRP was determined using nephelometric assay (Beckman Coulter, Inc., Brea, CA, USA and Siemens Healthcare Diagnostics Products GmbH, Marburg, Germany). PCT was measured using mini VIDAS B.R.A.H.M.S PCT assay (Biomerieux, S.A., Marcy-l’ Étoile, France), sTREM-1 level was determined in duplicate using an enzyme-linked immunosorbent assay (MyBioSource, Inc., San Diego, CA, USA). All procedures were performed in strict accordance with manufacturers’ instructions and standard microbiology guidelines.

Statistical analysis

Descriptive results of continuous variables with normal distribution are presented as mean± SD or median (interquartile range). Mann-Whitney U test was used for comparison of variables and to evaluate differences between groups. Wilcoxon signed-rank test was used to analyze relationship between second and first measurements, and Yates continuity correction was applied to compare qualitative data. Receiver operating characteristic (ROC) curve was employed to evaluate effects of sTREM-1, PCT, CRP levels, and APACHE II scores on sepsis diagnosis. NCSS 2007 and PASS 2008 (NCSS, LLC, Kaysville, UT, USA) statistical software programs were used for statistical analysis. Two-tailed p<0.05 was considered significant.

RESULTS

A total of 90 consecutive patients admitted to ICUs of the hospital were enrolled in the study. Final diagnosis of 52 patients was established to be sepsis and 38 were diagnosed as SIRS. Forty-seven patients (52.2%) were male and 43 (47.8%) were female. Mean age of the patients was 64.27±15.54 years (range: 20–92 years). Comorbidities were frequent (92.2%). In terms of predisposing factors, only rate of cerebrovascular occlusion was statistically higher in sepsis group than in SIRS group (p=0.03). Blood cultures were positive in 38 (73.3%) patients with sepsis. Most frequently isolated microorganisms were methicillin-resistant coagulase-negative Staphylococcus (36%), Escherichia coli (13%), and Acinetobacter baumannii (13%), respectively. Main source of infection was the lungs (44.2%), followed by the blood (21%). Length of stay in ICU was 14±12 (SD) days. Twenty-two patients (24.4%) died. Mortality rate was much higher in sepsis group than in SIRS group (p=0.01) (Table 1).

Table 1

Baseline characteristics of the patients

	SIRS n=38		Sepsis n=52		p
	n	%	n	%
Age	65.71±15.23		63.21±15.83		NS
Gender
Male	18	47.4	29	55.8	NS
Female	20	52.6	23	44.2
Positive blood culture	0	0	38	73.1*	<0.001
APACHE II	23.84±5.26		27.83±5.76		0.001
Etiological factors
Respiratory system infection	–	–	21	40	–
Bacteremia	–	–	11	21	–
Urinary tract infection	–	–	5	9	–
Unknown	–	–	4	8	–
Catheter-related bloodstream infection	–	–	4	8	–
Intraabdominal infection	–	–	4	8	–
Surgical site infection	–	–	2	4	–
Endocarditis	–	–	1	2	–
Predisposal factors
CVO	4	10	11	21	0.03
Malignancy	5	15	5	9	NS
Respiratory disorder	4	10	3	6	NS
CKD	4	10	3	6	NS
Trauma	1	3	2	4	NS
COPD	2	5	2	4	NS
Neurological disorder	2	5	2	4	NS
CAD	1	3	2	4	NS
HT	2	5	1	2	NS
Length of hospital stay (days)	25.05±14.19		32.7±21.37		NS
Mortality	5	13.2	17	32.7	0.01

APACHE: Acute Physiology and Chronic Health Evaluation; CAD: Chronic arterial disease; CKD: Chronic kidney disease; COPD: Chronic obstructive pulmonary disease; CVO: Chronic vascular occlusion; HT: Hypertension; SIRS: Systemic inflammatory response syndrome.

sTREM-1, PCT level, and APACHE II score were significantly elevated in sepsis group compared with SIRS group (p=0.001, p=0.01, p=0.001, respectively), on first day of ICU admission. There were no significant differences in median CRP value or WBC count (p=0.498, p=0.180, respectively). Both first and second measurements of sTREM-1 level were significantly higher in sepsis group than in SIRS group (p=0.001, p=0.001, p<0.01, respectively). First and second measurement of PCT level (p=0.01, p=0.01, p<0.05) and APACHE II score (p=0.001, p<0.01) were significantly higher in sepsis group. In terms of differentiation between sepsis and SIRS, there was no significant difference between first and second measurement of sTREM-1, PCT and CRP levels, or WBC count (Table 2).

Table 2

PCT, sTREM-1, CRP, WBC levels and APACHE II score of patients with sepsis and SIRS

		Sepsis (n=52)	SIRS (n=38)	[a]p
PCT
1st measurement	Min-Max (Median)	0.05–95.03 (3.86)	0.05–71.95 (1.14)	0.010*
	Mean±SD	11.33±19.31	5.86±13.57
PCT
2nd measurement	Min-Max (Median)	0.06–106.70 (2.62)	0.05–13.34 (0.91)	0.010*
	Mean±SD	11.53±18.96	2.48±3.38
[b]p		0.820	0.516	0.958
Difference; Min-Max (Median)		-66.02/105.45 (-0.23)	-61.15/12.21 (-0.05)
sTREM-1
1st measurement	Min-Max (Median)	67.49–2000 (164.66)	18.33–371.62 (96.78)	0.001**
	Mean±SD	389.82±524.44	110.86±57.66
sTREM-1
2nd measurement	Min-Max (Median)	70.86–2000.00 (171.28)	37.08–289.36 (99.85)	0.001**
	Mean±SD	397.59±494.75	111.45±45.84
[b]p		0.426	0.602	0.947
Difference; Min-Max (Median)		-829.12/797.78 (2.85)	-82.26/49.58 (1.89)
CRP
1st measurement	Min-Max (Median)	3.0–42.8 (13.7)	4.3–32.5 (11.75)	0.498
	Mean±SD	14.55±7.00	13.91±7.18
CRP
2nd measurement	Min-Max (Median)	1.93–42.90 (11.0)	1.98–30.00 (9.61)	0.367
	Mean±SD	13.24±8.14	11.63±7.21
[b]p		0.201	0.213	0.922
Difference; Min-Max (Median)		-19.80/33.85 (-3.22)	-24.85/16.39 (-1.52)
WBC
1st measurement	Min-Max (Median)	6.0–46.7 (14.1)	2.4–46.2 (13.05)	0.180
	Mean±SD	15.86±7.97	14.10±8.64
WBC
2nd measurement	Min-Max (Median)	2.4–55.9 (12.2)	3.1–45.1 (10.9)	0.170
	Mean±SD	14.83±8.95	13.14±8.45
[b]p		0.473	0.293	0.927
Difference; Min-Max (Median)		-37.70/38.90 (-1.00)	-24.10/20.39 (-0.86)
APACHE II	Min-Max (Median)	10–40 (28.5)	10–32 (23.5)	0.001**
	Mean±SD	27.83±5.76	23.84±5.26

APACHE: Acute Physiology and Chronic Health Evaluation; CRP: C-reactive protein; PCT: Procalcitonin; SIRS: Systemic inflammatory response syndrome; sTREM-1: Soluble triggering receptor expressed on myeloid cells-1; WBC: White blood cell;

Mann-Whitney U test;

Wilcoxon signed-rank test;

p<0.05;

p<0.01

sTREM-1 value and APACHE II score were higher in the patients with positive blood cultures than negative ones (177.9 pg/mL vs 113.96 pg/mL, p=0.002; 26 vs 29.5, p= 0.006). PCT, CRP, and WBC levels were found to be higher in the patients with positive blood cultures, but the difference between 2 groups was not statistically meaningful.

PCT and CRP levels were significantly higher (1.07 ng/mL vs 3.11 ng/mL, p=0.01; 9.15 mg/dL vs 17.80 mg/dL, p=0.02) in second measurement, and initial APACHE II score was statistically higher (26 vs 31; p=0.001), in nonsurviving patients. sTREM-1 level increased in second measurement of nonsurviving group (104.37 pg/mL vs 160.74 pg/mL; p <0.01). First and second sTREM-1 levels were higher in nonsurviving group (135.83 pg/mL vs 104.37 pg/mL; 140 pg/mL vs 160.74 pg/mL), but the difference was not statistically significant (p=0.840, p=0.512) between surviving and nonsurviving groups.

ROC was used to determine cut-off values of sTREM-1, PCT, and APACHE II. Area under the ROC curve (AUROC) was 0.78 (95% CI, 0.69–0.86) for sTREM-1, 0.65 (95% CI, 0.53–0.76) for PCT, and 0.71 (95% CI, 0.60-0.81) for APACHE II score. sTREM-1 cut-off value of ≥133 pg/mL yielded sensitivity of 71.15%, specificity of 76.32%, positive predictive value (PPV) of 80.43, and negative predictive value (NPV) of 65.91 for differentiating patients with SIRS from those with infection. When 1.57 ng/mL was set as cut-off value for PCT, sensitivity was 67.31%, specificity established was 65.79%, PPV was 72.92, and NPV was determined to be 59.52. APACHE II score of greater than 25 produced sensitivity of 80.77%, specificity of 52.63%, PPV of 70.00, and NPV of 66.67 (Figure 1).

Figure 1

Receiver operating characteristic curve of PCT, sTREM-1, CRP, WBC, and APACHE II values.

DISCUSSION

Sepsis has high mortality rate, with estimated range of 10% to 52%. Early identification of sepsis and appropriate initial treatment has a major effect on clinical course and outcomes of patients [10-14]. Currently, physicians use a variety of tools, such as WBC, CRP, PCT, and severity scores, such as APACHE II, to help discriminate between infectious and noninfectious conditions, as well as to predict clinical outcomes. Many sepsis-related biomarkers have been identified for this purpose; however, it is still controversial which biomarker is best for early sepsis diagnosis in daily practice [15, 16].

Although CRP level has been widely proposed as a useful screening test for sepsis, it is doubtful that it is a good indicator for early diagnosis of sepsis [17]. PCT complies with most of the desirable preanalytical, analytical, and postanalytical features for an ideal laboratory biomarker. PCT level increases in bacterial infection, especially Gram-negative bacterial infections [18]. Compared to other biomarkers of sepsis, PCT has superior diagnostic accuracy and remains unaffected even in the presence of concomitant immunosuppressive therapy. PCT level was found to be the most valuable biomarker for diagnosis of sepsis and bacteremia for patients in the emergency department and ICU [19-21]. Endo et al. [22] reported that if they used serum PCT cutoff point of 2.0 ng/mL to discriminate between sepsis and severe sepsis, overall diagnostic efficiency was 84.3%. On the other hand, several studies have also reported less convincing data, making use of this biomarker controversial [23, 24].

Many studies have indicated that sTREM-1 could be valuable diagnostic biomarker for various infectious diseases [25-31]. Dynamic changes in sTREM-1 level could predict outcome of patients at early stage of sepsis. Rivera-Chavez et al. [26] evaluated 93 patients in a surgical ICU with SIRS symptoms and suspected infection. The patients were classified as having SIRS (n=37) or sepsis (n=56) according to decision of the treating physician and bacteriological evidence. The authors reported that the patients with sepsis had significantly higher sTREM-1 level than those with SIRS. sTREM-1 cut-off value of 230 pg/mL yielded sensitivity of 98% and specificity of 91% in differentiating patients with SIRS from those with infection. Su et al. [27] performed a study involving 144 patients in ICU (60 patients with SIRS and 84 patients with sepsis complicated by new onset of fever). They found that sepsis group had higher serum sTREM-1, PCT, and CRP levels compared with SIRS group (p<0.05) on first day of ICU admission. In study conducted by Li et al. [28], 52 consecutive patients hospitalized in surgical ICU with suspicion of infection included 14 patients with SIRS, 9 patients with sepsis, 14 patients with severe sepsis, and 15 patients with septic shock. They found that in postoperative patients, plasma level of sTREM-1 was higher in patients with sepsis than in patients with SIRS (111.7 pg/mL vs 64.1 pg/mL; p<0.05), with sensitivity, specificity, and predictive values higher than those of PCT and tumor necrosis factor alpha.

In meta-analysis of 13 clinical studies that fulfilled inclusion criteria (Total 980 patients, 557 patients with bacterial and 423 with nonbacterial infection), Jiyong et al. [9] reported that sTREM-1 level for diagnosis of infection in AUC of summary ROC was 0.86, with sensitivity of 0.82, and specificity of 0.86. These findings indicated that sTREM-1 is reliable biomarker for bacterial infection. Wu et al. [16] evaluated 11 studies with total of 1795 patients in a recent meta-analysis. AUC of summary ROC was 0.87, pooled sensitivity and specificity were 79% and 80%, respectively. The authors concluded that plasma sTREM-1 had moderate diagnostic performance in differentiating sepsis from SIRS. Moreover, some researchers have suggested that PCT and CRP are more sensitive biomarkers than sTREM-1 for diagnosis of bacterial infection [32, 33]. In the present study, we found that plasma sTREM-1 and PCT values, as well as APACHE II score, were significantly higher in patients with sepsis than in patients with SIRS. There was no significant difference between median CRP value and WBC count of the 2 groups. Additionally, AUROC was 0.78 for sTREM-1, 0.65 for PCT, and 0.71 for APACHE II score. sTREM-1 cut-off value of ≥133 pg/mL yielded higher sensitivity for differentiating patients with SIRS from those with infection.

Although many studies on predictive value of PCT in diagnosis of bacteremia exist, there are only a few studies about using sTREM-1for the same purpose [19–21, 27, 34]. Ruiz-Gonzales et al. [34] suggested that sTREM-1 is useful biomarker in diagnosis of secondary bacteremia due to community-acquired pneumonia. Su et al. [27] recently reported that there were no significant differences in serum sTREM-1 or PCT levels between blood culture-positive and negative groups with ICU-acquired new onset fever. However, we found that initial sTREM-1level and APACHE II score were significantly higher in patients with positive blood culture than negative ones (p=0.02, p=0.06, respectively).

Gibot et al. [25] evaluated alterations in serum sTREM-1, PCT, and CRP levels in sepsis, severe sepsis, and septic shock in a medical ICU and reported that sTREM-1 level was significantly lower in the survival group. Su et al. [27] found that within bacteremia group, sTREM-1 and PCT levels were significantly higher in nonsurvivors, and they reported that sTREM-1 and PCT levels were useful for predicting prognosis of bacteremia. In this study, we found that APACHE II score of the patients was high, and serum plasma levels of CRP and PCT were elevated in second measurement of nonsurviving group. Interestingly, although sTREM-1 level was higher in nonsurvivors than survivors, the difference between the 2 groups was not statistically meaningful.

Main limitations of this study must also be acknowledged. First, number of enrolled patients was relatively small. Second, severe bacterial sepsis patients with negative blood cultures were not investigated in this study.

In conclusion, present study demonstrated that measurement of sTREM-1 in plasma may be valuable tool for early distinction between sepsis and SIRS in adult patients in daily practice. Our data also suggest that plasma sTREM-1 at admission is the most useful biomarker for prediction of bacterial culture positivity. Initial high APACHE-II score and elevation of CRP and PCT levels are important parameters to predict outcome of ICU patients with sepsis.