Sepsis, defined as “life-threatening organ dysfunction caused by a dysregulated host response to infection,” is counted among the leading causes of critical illness and mortality worldwide (1–3). But, it has proven difficult to develop clinical and laboratory criteria that accurately predict risk for mortality, particularly in the early period of sepsis. A recent large cohort study, performed to validate new clinical criteria for sepsis, found that the recommended criteria, called “the Sequential (Sepsis-Related) Organ Failure Assessment (SOFA) score,” achieved moderate predictive performance for acute mortality, with an area under the receiver operating characteristic curve (AUROC) of 0.74 for in-hospital mortality among ICU encounters and an AUROC of 0.79 for in-hospital mortality among non-ICU encounters. No other sepsis criteria outperformed the SOFA score among ICU encounters; an abbreviated quick SOFA score outperformed SOFA among non-ICU encounters, but the difference was minimal (4).
In the setting of imperfect diagnostic and prognostic tools for sepsis, novel methods for recognizing a dysregulated inflammatory response and predicting associated mortality must be developed. The SOFA score integrates clinical signs and laboratory values, including the PaO2/FIO2 ratio, platelet count, serum bilirubin, mean arterial pressure, Glasgow Coma Scale, serum creatinine, and urine output. The new consensus definition of septic shock incorporates a biomarker, serum lactate, in addition to SOFA criteria (1 , 4 , 5). Procalcitonin, a peptide precursor of the hormone calcitonin, has also been proposed as a tool to improve sepsis diagnosis and prognosis. Procalcitonin, which increases in response to microbial toxins and systemic inflammation, demonstrated value in discriminating infectious and noninfectious causes of sepsis in critical illness (6–8). A recent large randomized controlled trial that de-escalated antibiotics prescribed as empiric sepsis therapy if an 80% decrease in procalcitonin occurred demonstrated a survival benefit at 28 days and a significant reduction in antibiotic use (9). However, the application of procalcitonin and other novel biomarkers requires validation. For this reason, procalcitonin received only a weak recommendation in the Surviving Sepsis Campaign Guidelines, and the recently revised international consensus definitions for sepsis and septic shock exclude procalcitonin, preferring instead serum lactate and the other laboratory values listed above (1 , 4 , 5 , 10).
We sought to identify novel biomarkers that better characterize a dysregulated immune response and predict sepsis-related mortality by comparing predictive validity for acute and total in-hospital mortality among patients with clinically suspected bacterial sepsis. We investigated nine biomarkers associated with a systemic inflammatory response, assayed longitudinally at 24-hour intervals, in a cohort of subjects with suspected bacterial sepsis recruited from both medical ICU (MICU) and surgical ICU (SICU) at an academic medical center.
MATERIALS AND METHODS
Study Design and Setting
Patients admitted to the MICU or SICU at the Hospital of the University of Pennsylvania were prospectively enrolled from January 2012 to May 2014, with the approval of the institutional review board of the University of Pennsylvania. Methods were identical in the MICU and SICU, as described in more detail in prior work (7 , 8). The study protocol was approved by the Institutional Review Board of the University of Pennsylvania. A waiver of informed consent was granted.
Eligibility for study enrollment required presumed bacterial sepsis, defined by meeting two or more systemic inflammatory response syndrome (SIRS) criteria and having new empiric antibiotic therapy initiated and blood cultures ordered within a 4-hour window at ICU admission or at any time during the ICU stay (11 , 12). Two or more SIRS criteria (body temperature > 38°C or < 36°C; heart rate > 90/min; respiratory rate > 20/min; or WBC count > 12,000 or < 4,000 cells/uL) had to be met within 4 hours of the enrollment blood culture. New empiric antibiotic therapy required either the initiation of new antibiotic therapy in a patient previously not on any antibiotics or the broadening of antibiotic therapy in a patient already receiving an antibiotic regimen. Antibiotic review was performed by two physicians trained in infectious diseases (E.L., J.H.H.). Patients were excluded if new or broadened empiric antibiotic therapy had been given for greater than 4 hours prior to the timepoint when baseline biomarkers were measured because of the potential for antibiotic therapy to impact baseline biomarker measures (13).
Exclusion criteria included 1) code status of “do-not-resuscitate (DNR),” 2) cardiopulmonary arrest from which patients had been resuscitated, 3) documented bacterial infection treated with antibacterial therapy in the 5 days prior to enrollment, or 4) evidence of immune compromise (including HIV infection with CD4 cell count < 200 cells/mm3, immunosuppressive therapy after organ transplantation, neutropenia [< 500 neutrophils/mm3], chemotherapy, receipt of ≥ 20 mg/d of prednisone [or the equivalent] for 2 or more weeks in the preceding 3 mo, or cystic fibrosis).
Residual blood samples obtained during the course of routine medical care were stored at –70°C until biomarker assays could be performed (as per assay manufacturer instructions). Nine biomarkers were assayed: α-2 macroglobulin, C-reactive protein (CRP), ferritin, fibrinogen, haptoglobin, procalcitonin, serum amyloid A, serum amyloid P (SAP), and tissue plasminogen activator (TPA). Biomarker measurements were performed at the time a patient met all eligibility criteria and then repeated daily for 3 days (24-, 48-, and 72-hr timepoints). The sample closest to the precise timepoint of interest was chosen in cases of multiple available clinical samples. Samples for each patient over the four timepoints were assayed in the same testing run. Nine biomarkers were measured at each timepoint: procalcitonin using the VIDAS B.R.A.H.M.S procalcitonin assay (bioMérieux, Durham, NC), a one-step immunoassay sandwich method with fluorescent detection, and the remaining eight using the Bio-Plex Pro Human Acute Phase 5- and 4-Plex Panel Complete Kit (Bio-Rad Laboratories, Hercules, CA), a bead-based (xMAP technology; Luminex Corporation, Austin, TX) multiplex assay. Assays were performed per manufacturer’s instructions. The Bio-Plex assay was read using a Luminex 200 reader (Luminex Corporation), with samples from all four timepoints included in the same measurement test run, using a single lot of reagents, and each analyte measured in duplicate (results recorded as the mean of measurements).
At baseline, we recorded relevant demographic information, length of hospital and ICU stay prior to enrollment, and comorbidities with specific attention to hepatic dysfunction (defined as two or more of total bilirubin > 2.5 mg/dL, aspartate aminotransferase or alanine aminotransferase greater than twice the upper limit of normal), solid or hematologic malignancy, diabetes mellitus, chronic kidney disease (with or without requiring hemodialysis), and pulmonary disease (chronic obstructive pulmonary disease or chronic bronchitis). For all subjects with documented variables, Acute Physiology and Chronic Health Evaluation (APACHE) II scores were calculated at enrollment (14).
Definition of Mortality
Fourteen-day mortality was defined by a time of death documented as within 14 days of the time of enrollment during the hospital stay. Total in-hospital mortality was defined by a time of death at any time during the index hospital admission. Fourteen-day mortality was chosen as an outcome measure to capture mortality likely related to the index episode of suspected bacterial sepsis; total in-hospital mortality was chosen to validate associations observed with 14-day mortality.
We first visually explored the temporal trends in biomarker values in aggregate, stratified by 14-day mortality, and stratified by total in-hospital mortality, using LOESS regression with a least-squares estimator and Tukey biweight M-estimator (the latter to limit the impact of outliers). We then compared the median biomarkers values between mortality versus no-mortality groups at each timepoint using the Wilcoxon rank-sum test. We characterized the clinical characteristics of ICU patients with 14-day and in-hospital mortality using the Wilcoxon rank-sum test for continuous variables and the Fisher exact test for categorical variables. For all comparisons, a two-tailed p value of less than 0.05 was considered significant.
AUROCs was calculated for each individual biomarker and timepoint and for logistic regression models incorporating all pairwise combinations of the biomarkers and timepoints (with and without an interaction term) (15–17). We then compared the performance of the best individual biomarkers and pairwise combinations with logistic regression models based on the APACHE II score (14 , 18 , 19) and the criterion of 80% decrease in procalcitonin, which has been used in a recent large randomized trial and found to be a significant predictor of severe sepsis-related mortality (9 , 20). APACHE II score was available for 127 of the 139 enrolled subjects (12 had insufficient data recorded in the electronic medical record to calculate APACHE II score and were excluded from the analysis). Eighty percent decrease in procalcitonin was defined as decrease from maximal value at any point during the 72-hour observation period. Finally, we compared the performance of the best combined biomarker prediction tools across the MICU and SICU groups. Statistical analyses were performed using R v3.2.1 (21). Figures were produced using R’s ggplot2 package v1.0.1 (22).
Clinical Features Associated With 14-Day and Total In-Hospital Mortality in Suspected Sepsis
A total of 485 patients were screened for eligibility, with 139 (28.6%) enrolled. Reasons for exclusion included not meeting SIRS criteria (n = 15), antibiotics broadened more than 4 hours prior to initial biomarker measurement (n = 18), ongoing antibiotic treatment for a known infection (n = 53), immunosuppression (n = 199), DNR status (n = 7), cardiopulmonary arrest (n = 6), and unavailable or incomplete laboratory samples (n = 48).
Of the 139 enrolled subjects, 70 were enrolled from the MICU and 69 from the SICU. Eighteen subjects (12.9%) died within 14 days of enrollment, and 41 subjects (29.5%) died during the index hospital admission.
We evaluated clinical features of the full enrolled cohort, and we compared features between acute mortality (within 14 d of meeting SIRS criteria) and total in-hospital mortality groups. Table 1 shows the clinical features in the entire cohort and features stratified by 14-day and total in-hospital mortality. Older age, higher APACHE II score on enrollment, non-white race, and cirrhosis were significant risks for 14-day mortality. Older age, higher APACHE II score on enrollment, malignancy, and cirrhosis were significantly associated with total in-hospital mortality.
Biomarker Differences to Discriminate 14-Day and Total In-Hospital Mortality in Suspected Sepsis
We began our evaluation of biomarkers to discriminate risk for acute (14 d) and in-hospital mortality after suspected sepsis by visualizing biomarker change over time via LOESS regression (Figs. 1 and 2). For both acute and total in-hospital analyses, SAP and TPA demonstrated the best separation between mortality versus no mortality.
We then compared biomarker values between mortality and no-mortality groups at four categorical timepoints. Supplemental Table 1 (Supplemental Digital Content 1, http://links.lww.com/CCM/D423) presents the median (interquartile range) of each biomarker at the time of meeting SIRS criteria, 24 , 48 , and 72 hours thereafter. For 14-day mortality, Wilcoxon rank-sum testing identified significant differences in ferritin at a single timepoint; SAP at all timepoints; and baseline, 48 , and 72 hours TPA. For total in-hospital mortality, the same approach identified significant differences in SAP and TPA across all timepoints, as well as ferritin and fibrinogen at two timepoints. Concordant with LOESS regression (Figs. 1 and 2), Wilcoxon rank-sum testing highlighted SAP and TPA as the biomarkers that differed most consistently between mortality and no-mortality groups.
To evaluate the ability of individual biomarkers to predict 14-day and total in-hospital mortality, we developed logistic regression models based on each biomarker and compared the model AUROC. The top seven highest AUROC values for 14-day mortality came from SAP or TPA; the top five highest AUROC values for total in-hospital mortality came from SAP or TPA. Seventy-two–hour TPA achieved the highest AUROC for 14-day mortality (0.726). Forty-eight–hour SAP achieved the highest AUROC for total mortality (0.696). Baseline SAP achieved an AUROC of 0.657 for 14-day mortality and of 0.608 for total in-hospital mortality. Baseline TPA achieved an AUROC of 0.651 for 14-day mortality and of 0.680 for total in-hospital mortality. In comparison, a model based on APACHE II score alone achieved an AUROC of 0.694 for 14-day mortality and an AUROC of 0.680 for total in-hospital mortality (14 , 18 , 19). A model based on an 80% decrease in procalcitonin achieved an AUROC for 14-day mortality of 0.666 and an AUROC for total in-hospital mortality of 0.6028 (9 , 20), and baseline procalcitonin achieved an AUROC of 0.574 for 14-day mortality and 0.566 for total in-hospital mortality. Supplemental Table 2 (Supplemental Digital Content 2, http://links.lww.com/CCM/D424) presents the AUROC of logistic regression models for 14-day and total in-hospital mortality based on each biomarker at each timepoint.
Because SAP is produced by the liver and because we observed high mortality in subjects with cirrhosis (Table 1), we performed a subgroup analysis restricted to subjects without cirrhosis to address the possibility that cirrhosis confounds the relationship between SAP and mortality. The restricted cohort, which included 116 of the total 139 subjects, demonstrated significantly lower 48-hour SAP and 72-hour SAP in the mortality versus the no-mortality group (Supplemental Table 3, Supplemental Digital Content 3, http://links.lww.com/CCM/D425). In the restricted cohort, 48-hour SAP remained the second best predictor of total in-hospital mortality, with a model AUROC of 0.672. The findings from the restricted cohort thus recapitulated those from the full cohort.
Biomarker Combinations to Improve 14-Day and Total In-Hospital Mortality Prediction
We further evaluated the performance of logistic regression models based on combined biomarkers to predict both 14-day and total in-hospital mortality. Figure 3 displays the AUROC achieved by each pairwise combination of biomarkers. Each model was evaluated with and without an interaction term. Of 630 unique pairwise combinations, the top 46 AUROCs for 14-day mortality included SAP, TPA, or both. The top 119 AUROCs for total in-hospital mortality included SAP, TPA, or both. The combination of 24- and 72-hour TPA achieved the best performance predicting 14-day mortality (AUROC 0.761 without interaction term, AUROC 0.765 with interaction term). The combination of 48-hour SAP and 72-hour ferritin achieved the best performance predicting total in-hospital mortality (AUROC 0.738 without interaction term, AUROC 0.739 with interaction term).
At baseline (time of meeting two SIRS criteria), SAP and TPA achieved an AUROC of 0.657 (with or without interaction term) for 14-day mortality. For total in-hospital mortality, a model based on the same baseline combination achieved an AUROC of 0.689 without interaction term and an AUROC of 0.693 with interaction term. The combination performed better than baseline SAP or TPA alone, which achieved respective AUROCs of 0.657 and 0.651 for 14-day mortality, 0.608 and 0.680 for total in-hospital mortality.
Differences Between Surgical Intensive Care and Medical Intensive Care
Given the potential for recent surgery to impact biomarkers of inflammation (7 , 8), we compared the performance of biomarker combinations to predict 14-day and total inpatient mortality stratified by whether subjects were admitted to the MICU or SICU (Supplemental Fig. 1, Supplemental Digital Content 4, http://links.lww.com/CCM/D426; legend, Supplemental Digital Content 5, http://links.lww.com/CCM/D427). We found that models incorporating SAP and TPA performed among the best for MICU 14-day mortality, MICU total mortality, and SICU total mortality. However, there was no clear pattern of model performance for SICU 14-day mortality, and ferritin at all timepoints performed well in the prediction of SICU in-hospital mortality, particularly when combined with SAP or TPA.
We evaluated nine biomarkers to predict acute and total in-hospital mortality early in suspected bacterial sepsis, and we found that SAP and TPA had the best predictive performance, both individually and in pairwise combinations. The performance of SAP and TPA matched or exceeded the performance of clinical criteria (APACHE II score) and other, more extensively studied biomarkers (e.g., procalcitonin and CRP) used for diagnosis and mortality prediction in the setting of sepsis.
SAP is a glycoprotein member of the pentraxin superfamily, which serves as a component of the innate immune system, binding to chromatin, apoptotic and necrotic cells, as well as microbes, in a calcium-dependent fashion (23–26). We found that SAP consistently reduced in subjects who would experience acute (within 14 d) mortality or in-hospital mortality, relative to surviving subjects. Other pentraxins (particularly pentraxin-3) have been shown to predict mortality in the setting of infection and sepsis (27 , 28), but to the best of our knowledge, this is the first report of SAP’s utility in predicting mortality associated with sepsis.
TPA is a serine protease produced by endothelial cells to catalyze the conversion of plasminogen to plasmin. We found TPA consistently elevated in subjects who would experience 14-day or in-hospital mortality. TPA has previously been shown to increase in the setting of sepsis (29), and its level has been correlated with clinical outcomes in hemorrhagic fever (30), but it has not previously been evaluated as a biomarker to predict sepsis-associated mortality.
The predictive performance of SAP and TPA was striking in that it exceeded the performance of procalcitonin, which has been studied far more extensively. Single, absolute procalcitonin levels have exhibited variable performance in the prediction of sepsis-associated mortality. Procalcitonin change has shown more promise as a prognostic tool and has been used successfully as the basis for antibiotic stewardship interventions in sepsis (9 , 31–33). As shown above, SAP and TPA alone and in combination outperformed absolute procalcitonin in our cohort. Furthermore, a model based on baseline SAP and TPA values exceeded the performance of procalcitonin change (80% decrease) for the prediction of total in-hospital mortality.
The predictive performance of SAP and TPA also matched the performance of the APACHE II score, which incorporates clinical and common laboratory measures, and is well validated as a tool to predict the mortality of critically ill patients (14 , 18 , 19). In our cohort, the APACHE II score achieved AUROC of 0.680 for total in-hospital mortality, versus the AUROC of 0.693 achieved by baseline SAP and TPA. Our findings suggest that serum SAP and TPA, when combined with the clinical SIRS criteria that defined eligibility for enrollment, may prove a valuable tool to predict mortality in sepsis.
Several limitations must be noted: 1) The study was performed at a single center and 2) enrollment excluded immunocompromised patients, thus limiting its external validity. In addition 3), there was significant heterogeneity among the subjects enrolled, and 4) we did not assess the appropriateness of antibiotic therapy. Specifically, we found that the performance of SAP and TPA was greatest among subjects in the MICU, and ferritin performed particularly well predicting total in-hospital mortality among subjects in the SICU. The difference in the predictive validity of inflammatory biomarkers between patients who have or have not experienced surgery is an important target for future investigation, as is the difference between patients who are or are not prescribed appropriate antibiotic therapy.
Further limitations include 5) the use of SIRS criteria, which are no longer recommended for the diagnosis of sepsis, to determine eligibility for enrollment and 6) the possibility that liver disease confounded that observed association between SAP and mortality. Given that SAP is produced by the liver, the observed association between low SAP and increased mortality risk may have been driven by subjects with cirrhosis, who had high mortality in this cohort (Table 1). As detailed above, we repeated analyses excluding all subjects with cirrhosis and still found significant differences in 48-hour SAP and 72-hour SAP between mortality and no-mortality groups. Furthermore, 48-hour SAP remained the second best predictor of total in-hospital mortality in the restricted cohort. Finally, it must be noted that 7) the combination of SAP and TPA performed comparably with APACHE II and SOFA but was not clearly superior.
Despite the above limitations, the finding that, when assayed in subjects who meet SIRS criteria, SAP and TPA have predictive validity for 14-day and total in-hospital mortality merits close follow-up. Models developed from SAP and TPA performed as well or better than more established clinical and biomarker-based tools in this cohort. It is essential to better characterize the dysregulated immune response that defines sepsis and its attributable mortality. The finding that SAP and TPA predict mortality risk early in the course of suspected bacterial sepsis may contribute to improved models of sepsis-related immune dysregulation, novel risk-stratification strategies, and personalized antibiotic treatment strategies.
SAP and TPA together provide good predictive ability for both 14-day and total in-hospital mortality among subjects with suspected sepsis. The performance of SAP and TPA in this cohort rivals established clinical- and biomarker-based tools for mortality risk-stratification in sepsis. Further study should be directed toward validating SAP and TPA for sepsis prognosis, evaluating SAP and TPA for sepsis diagnosis, better defining the differences in biomarker performance between subjects with and without recent surgery, and assessing the additive value of SAP or TPA in combination with predictive clinical features.
1. Singer M, Deutschman CS, Seymour CW, et alThe third international consensus definitions for sepsis
and septic shock (Sepsis
-3). JAMA 2016; 315:801–810
2. Vincent JL, Marshall JC, Namendys-Silva SA, et alICON investigators: Assessment of the worldwide burden of critical illness: The intensive care over nations (ICON) audit. Lancet Respir Med 2014; 2:380–386
3. Fleischmann C, Scherag A, Adhikari NK, et alInternational Forum of Acute Care Trialists: Assessment of global incidence and mortality of hospital-treated sepsis
. Current estimates and limitations. Am J Respir Crit Care Med 2016; 193:259–272
4. Seymour CW, Liu VX, Iwashyna TJ, et alAssessment of clinical criteria for sepsis
: For the third international consensus definitions for sepsis
and septic shock (Sepsis
-3). JAMA 2016; 315:762–774
5. Shankar-Hari M, Phillips GS, Levy ML, et alSepsis Definitions Task Force: Developing a new definition and assessing new clinical criteria for septic shock: For the third international consensus definitions for sepsis
and septic shock (Sepsis
-3). JAMA 2016; 315:775–787
6. Wacker C, Prkno A, Brunkhorst FM, et alProcalcitonin as a diagnostic marker for sepsis
: A systematic review and meta-analysis. Lancet Infect Dis 2013; 13:426–435
7. Han JH, Nachamkin I, Coffin SE, et alUse of a combination biomarker algorithm to identify medical intensive care unit patients with suspected sepsis
at very low likelihood of bacterial infection. Antimicrob Agents Chemother 2015; 59:6494–6500
8. Kelly BJ, Lautenbach E, Nachamkin I, et alCenters for Disease Control and Prevention (CDC) Prevention Epicenters Program: Combined biomarkers
discriminate a low likelihood of bacterial infection among surgical intensive care unit patients with suspected sepsis
. Diagn Microbiol Infect Dis 2016; 85:109–115
9. de Jong E, van Oers JA, Beishuizen A, et alEfficacy and safety of procalcitonin
guidance in reducing the duration of antibiotic treatment in critically ill patients: A randomised, controlled, open-label trial. Lancet Infect Dis 2016; 16:819–827
10. Dellinger RP, Levy MM, Rhodes A, et alSurviving Sepsis
Campaign Guidelines Committee including The Pediatric Subgroup: Surviving Sepsis
Campaign: International guidelines for management of severe sepsis
and septic shock, 2012. Intensive Care Med 2013; 39:165–228
11. Bone RC, Balk RA, Cerra FB, et alDefinitions for sepsis
and organ failure and guidelines for the use of innovative therapies in sepsis
. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992; 101:1644–1655
12. Levy MM, Fink MP, Marshall JC, et alSCCM/ESICM/ACCP/ATS/SIS: 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis
Definitions Conference. Crit Care Med 2003; 31:1250–1256
13. Meisner MUpdate on procalcitonin
measurements. Ann Lab Med 2014; 34:263–273
14. Knaus WA, Draper EA, Wagner DP, et alAPACHE II: A severity of disease classification system. Crit Care Med 1985; 13:818–829
15. Fawcett TAn introduction to ROC analysis. Pattern Recognit Lett 2006; 27:861–874
16. Kleinbaum DG, Klein MLogistic Regression: A Self-Learning Text. 2010New York, Springer,
17. Austin PC, Steyerberg EWInterpreting the concordance statistic of a logistic regression model: Relation to the variance and odds ratio of a continuous explanatory variable. BMC Med Res Methodol 2012; 12:82
18. Kruse JA, Thill-Baharozian MC, Carlson RWComparison of clinical assessment with APACHE II for predicting mortality risk in patients admitted to a medical intensive care unit. JAMA 1988; 260:1739–1742
19. Knaus WA, Wagner DP, Draper EA, et alThe APACHE III prognostic system. Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991; 100:1619–1636
20. Schuetz P, Birkhahn R, Sherwin R, et alSerial procalcitonin
predicts mortality in severe sepsis
patients: Results from the Multicenter Procalcitonin
(MOSES) Study. Crit Care Med 2017; 45:781–789
21. R Core Team: R: A Language and Environment for Statistical Computing, 2014; Available at: http://www.R-project.org/
. Accessed March 29, 2018
22. Wickham Hggplot2: Elegant Graphics for Data Analysis (Use R!). 2009. First Edition. Corr. Third printing 2010 edition. New York, Springer, New York, Springer, 2010
23. Pepys MB, Dash AC, Markham RE, et alComparative clinical study of protein SAP (amyloid P component) and C-reactive protein in serum. Clin Exp Immunol 1978; 32:119–124
24. Hutchinson WL, Noble GE, Hawkins PN, et alThe pentraxins, C-reactive protein and serum amyloid P
component, are cleared and catabolized by hepatocytes in vivo. J Clin Invest 1994; 94:1390–1396
25. Hutchinson WL, Hohenester E, Pepys MBHuman serum amyloid P
component is a single uncomplexed pentamer in whole serum. Mol Med 2000; 6:482–493
26. Klotz SA, Sobonya RE, Lipke PN, et alSerum amyloid P component and systemic fungal infection: Does it protect the host or is it a trojan horse? Open Forum Infectious Diseases 2016; 3:ofw166
27. Hansen MB, Rasmussen LS, Garred P, et alPentraxin-3 as a marker of disease severity and risk of death in patients with necrotizing soft tissue infections: A nationwide, prospective, observational study. Crit Care 2016; 20:40
28. Liu S, Qu X, Liu F, et alPentraxin 3 as a prognostic biomarker in patients with systemic inflammation or infection. Mediators Inflamm 2014; 2014:421429
29. Hartemink KJ, Hack CE, Groeneveld ABRelation between coagulation/fibrinolysis and lactate in the course of human septic shock. J Clin Pathol 2010; 63:1021–1026
30. McElroy AK, Erickson BR, Flietstra TD, et alEbola hemorrhagic fever: Novel biomarker correlates of clinical outcome. J Infect Dis 2014; 210:558–566
31. Huang MY, Chen CY, Chien JH, et alSerum procalcitonin
clearance as a prognostic biomarker in patients with severe sepsis
and septic shock. Biomed Res Int 2016; 2016:1758501
32. Mat-Nor MB, Md Ralib A, Abdulah NZ, et alThe diagnostic ability of procalcitonin
and interleukin-6 to differentiate infectious from noninfectious systemic inflammatory response syndrome and to predict mortality. J Crit Care 2016; 33:245–251
33. Lipińska-Gediga M, Mierzchała-Pasierb M, Durek GProcalcitonin kinetics - prognostic and diagnostic significance in septic patients. Arch Med Sci 2016; 12:112–119
biomarkers; procalcitonin; sepsis; serum amyloid P; tissue plasminogen activator
Supplemental Digital Content
Copyright © by 2018 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.