Diagnostic and prognostic value of presepsin for sepsis / 中华检验医学杂志

ABSTRACT

Objective To investigate the diagnostic and prognostic value of presepsin for sepsis.Methods Diagnostic accuracy test.The plasma presepsin levels of 57 sepsis patients,64 systemic inflammatory response syndrome (SIRS) patients and 120 healthy individuals admitted to the 263 Clinical Branch,General Hospital of Beijing Military Region between January 2012 and December 2013were detected by PATHFAST system.Receiver operating characteristic (ROC) curve analysis was used to assess and compare the diagnostic value of presepsin and procalitonin (PCT).Logistic regression model was used to estimate the association between presepsin and sepsis.In addition,the correlations between presepsin and the clinical characteristics were analyzed in sepsis patients.Results Sepsis patients [1 266 (754-2 181) pg/ml] had higher presepsin level than SIRS[517 (349-939)pg/ml] and healthy individual controls [(182 ± 56)pg/ml] (Z value was 5.94 and 10.71,respectively,P value was ＜ 0.01 for all).The areas under curve (AUCs) of presepsin and PCT were 0.81 (95％ CI:0.74-0.89) and 0.78 (95％ CI:0.71-0.86),respectively,with no statistical significance (x2 =0.60,P =0.47).After adjusted for PCT,presepsin ＞ 1 060 pg/ml was independently associated with sepsis,with odds ratio (OR) of 7.80 (95 ％ CI:3.07-20.32).Severe sepsis patients [2 723 (2 002-4 234) pg/ml] had higher presepsin than sepsis patients[1 145 (656-1 436) pg/ml] (Z =4.00,P ＜0.01).The patients with inhosptal mortality [2 365 (1 256-3 567)pg/ml] had higher presepsin than survival ones[1 146 (660-1 452) pg/ml] (Z =2.99,P =0.003).Presepsin was positively correlated with PCT (r =0.75,P ＜ 0.01).The reference for presepsin was 72 to 292 pg/ml.Conclusions Presepsin was an useful biomarker for sepsisdiagnosis.The diagnostic value of presepsin and PCT was not completely overlap,and combinational using of these two biomarkers may improve the diagnostic accuracy of sepsis.In addition,presepsin had potential value for prognosis estimation.