Procalcitonin as a prognostic marker for sepsis based on SEPSIS‐3

Abstract Background The revised definition of sepsis is life‐threatening organ dysfunction caused by a dysregulated host response to infection (SEPSIS‐3). The objective of this study was to evaluate procalcitonin (PCT) for the diagnosis and prognosis of sepsis using SEPSIS‐3. Methods We enrolled 248 patients, who were admitted to the emergency department with suspected bacterial infection from June 2016 to February 2017. Definite bacterial infection was defined by proven culture results, and probable bacterial infection was based on diagnostic modalities other than culture. The sequential organ failure assessment (SOFA) score of 2 points or more from the baseline was diagnosed as sepsis. PCT was measured by the AFIAS‐6 immunoassay system (Boditech Med Inc.) using whole blood. White blood cell (WBC), C‐reactive protein (CRP), and erythrocyte sedimentation rate (ERS) were evaluated. Results The final diagnosis was sepsis in 185 patients with infection of respiratory and genitourinary tract constituted 84.6%. The area under the receiver operating characteristic curve (AUROC) with 95% confidence interval (CI) was as follows: PCT, 0.682 (0.589‐0.765); CRP, 0.583 (0.487‐0.673); ESR, 0.540 (0.515‐0.699); and WBC, 0.611 (0.455‐0.633), respectively. In multivariate analysis, age, SOFA, and PCT (log scale) predicted non‐survivors with an odds ratio with 95% confidence interval of 1.055 (1.008‐1.105), 1.303 (1.142‐1.486), and 2.004 (1.240‐3.238), respectively. Among sepsis group, initial PCT was increased in non‐survivor (23.2 ng/dL) compared to survivor group (8.1 ng/dL) with statistical significance (P = .005). Conclusions PCT could support and predict the unfavorable prognosis of sepsis based on SEPSIS‐3, whereas diagnostic potential of PCT requires further evaluations.


| INTRODUC TI ON
The original concept of sepsis, which was defined as a systemic inflammatory response syndrome (SIRS) with documented microbial infection, has been used for more than two decades. 1 This definition was revised in part due to an improved understanding of the pathobiology of sepsis. [2][3][4] Sepsis is now regarded as early activation of both pro-and anti-inflammatory responses with involvement of non-immunologic systems including cardiovascular, endocrine, and coagulation. 5 In addition, the low diagnostic capability of SIRS led to a revision of the definition of sepsis, though SIRS criteria might still be useful for the identification of infection. 3,4 The revised definition of sepsis proposed in 2016 (SEPSIS-3) is life-threatening organ dysfunction caused by a dysregulated host response to infection. 5 In brief, former conditions of sepsis and severe sepsis are now regarded as bacterial infection and sepsis, respectively.
The revised definition emphasizes organ dysfunction, which can be calculated by sequential organ failure assessment (SOFA) score. The SOFA score-based definition of sepsis predicted mortality higher than that of SIRS-based definition. As infection can lead to organ failure, patients with infection should be carefully followed up. 5 For ease of application in clinical environments, the laboratory data included in SOFA score were bilirubin, creatinine, and platelet count.
There have been continuous attempts to diagnosis of SIRS, sepsis, and severe sepsis using biomarkers, especially procalcitonin (PCT), C-reactive protein (CRP), white blood cell (WBC), erythrocyte sedimentation rate (ESR), and various interleukins. [6][7][8] PCT and biomarkers are debated for the usefulness and clinical application, but the previous literatures revealed that PCT could support the clinical diagnosis and treatment of patients. Among these biomarkers, PCT and CRP were included in the diagnostic criteria of inflammatory variables in Surviving Sepsis Campaign 2013. 9 However, in the Surviving Sepsis Campaign 2016, PCT was revised to be a recommended biomarker for sepsis prognosis but not for diagnosis. 10 Both diagnosis and prognosis are important in sepsis, and several biomarkers including PCT, sTREM-1, presepsin, and cytokines have been studied for predicting prognosis under former definition of sepsis. 7,[11][12][13][14] In addition, PCT is related to antimicrobial stewardship, a treatment that encompasses initiation and tapering of antimicrobial treatment. 15,16 The biomarkers evaluated for previously defined severe sepsis might not reflect the performance in revised sepsis due to differences in the details of definitions. In addition, diagnosis of sepsis was revised suing SOFA score, which requires three clinical variables and three laboratory variables, leading to a score range from zero to 24. We hypothesized that the PCT, CRP, WBC, and ESR might result in capabilities to diagnose sepsis and reflect prognosis. Diagnostic capability of biomarkers could be evaluated by the area under the receiver operating characteristic curve (AUROC) value, and prognostic capability of biomarkers could be evaluated by univariate and multivariate analysis. Therefore, in this study, we evaluated accessible biomarkers including PCT, CRP, ESR, and WBC in the clinical setting for their utility in the diagnosis and prognosis of revised sepsis.

| Patients in the study cohort
This was a single-center study performed at a tertiary teaching hospital, and the study protocol was approved by the institutional review board of Incheon St. Mary's Hospital. Individual consent was not required by the institutional review board because the data were obtained during the course of diagnosis and treatment.

| Diagnosis of sepsis
Among enrolled patients, data on the following component of SOFA score were collected and graded: PaO 2 /FiO 2 (mm Hg), platelet count, bilirubin, Glasgow Coma Scale (GCS) score, creatinine or renal output level, and mean arterial pressure. The mean arterial pressure was calculated as follows: diastolic blood pressure -1/3 × (systolic blood pressure -diastolic blood pressure). Enrolled patients with suspected bacterial infection and SOFA score of 2 points or more from the baseline were diagnosed with sepsis. 5

| Laboratory examination
Routine microbiological examination included more than two pair of blood cultures. Samples were cultured using two sets of aerobic and anaerobic bottles (BACTEC plus). BACTEC automated blood culture system (BD Biosciences) was used for incubation for 5 days. 21 Culture and analysis of various body fluids (urine, sputum, broncho-alveolar lavage fluid, cerebrospinal fluid, abscess, and closed wound) were performed. Blood samples for WBC counts, ESR, PCT, CRP, and blood chemistry were drawn immediately after presentation to the emergency department and were analyzed in a central laboratory within 2 hours. Hematologic parameters including WBC were measured by Sysmex XN2000 (Sysmex). High-sensitivity C-reactive protein (CRP) within-laboratory, 14.2%, 12.3%, respectively (Table S2). 22 The limit of blank and limit of detection provided by the manufacturer were 0.044 and 0.066 ng/mL. The claimed analytical measurement range was from 0.1 ng/mL to 100 ng/mL. Cut-off value provided by the manufacturer was 0.5 ng/mL.

| Comparison of non-sepsis and sepsis groups
The cohort consisted of 248 patients diagnosed with suspected bacterial infection who were initially admitted to the emergency department. Of the 248 patients, 63 were classified as the non-sepsis group and 185 as sepsis group. Table 1 shows a comparison of demographic and baseline data between the non-sepsis group and the sepsis group. The mean age, PCT, and WBC of patients in the sepsis group were significantly higher than that of patients in the non-sepsis group. The identified bacteria or other microorganisms are listed in detail in Table S1. Among the identified microbes, Escherichia coli (21.8%) was the most common pathogen, followed by Klebsiella species (13.8%) and Staphylococcus aureus (11.6%).

| Diagnosis of sepsis among patients with suspected bacterial infection
The ROC curves for PCT, CRP, ESR level, and WBC count for diagnosis of sepsis are shown in Figure 1.

| Prediction of non-survivors
Among the age, sex, and biomarkers, age, PCT, WBC, and SOFA score revealed statistical significance in univariate analysis (Table 3A).
With these variables, multivariate analysis was performed. Among them, age, SOFA score, and PCT (log scale) predicted non-survivors with statistical significance in multivariate analysis (Table 3B).

| Comparison of survivor and non-survivor among sepsis
Comparison of survivor (n = 159) and non-survivor (n = 26) was performed among sepsis group (Table 4). PCT was higher in non-survivor group compared to survivor group. Among tested markers, only PCT revealed statistical significance (P = .005). Hemoglobin was lower in non-survivor group. Most of SOFA score components revealed statistical significance except for the platelets and bilirubin. severe sepsis revealed higher sensitivity and specificity of 92.0% and 84.0%, respectively, compared with those of SOFA and qSOFA. 16,24 SIRS criteria might be useful for earlier signs of infection before development of organ dysfunction. 16,24 The diagnosis of sepsis depends on SOFA score that is the re- The revised sepsis definition includes immune dysregulation, which requires to be measured. Some cytokines are suspected to be related to immune dysregulation, and the exact pathobiology must be identified. Cytokines or immune-regulated molecules are complex, and network analysis might reveal pathobiology. Network analysis revealed that the sepsis network was small in size and path length was short, 31 which might reflect immune dysregulation. PCT was one of the molecules that were the hub node among sepsis network, 31 which implies that PCT is expected to play an important role among cytokine network in sepsis and interacted with other molecules.

| D ISCUSS I ON
The limitation of this study was that approximately 80% of patients were diagnosed with respiratory tract and genitourinary tract infection, which might differ from conditions in other hospital intensive care units or emergency departments. Age was increased in sepsis group, and age was one of the prognostic factors that predicted unfavorable prognosis. This was a single-center study, and patient population might have affected the prevalence of sepsis group.
In conclusion, under the revised definition of sepsis, PCT could support prognosis of sepsis and predicted mortality compared with other parameters. Further studies are required to accumulate data on PCT using the revised sepsis definition for patient diagnosis and management.

ACK N OWLED G M ENTS
We thank emergency department and laboratory department staff for supporting this study. We thank Sungyeon Lee, Chankyu Kim, and Se Woon Cho for excellent technical assistance.