Diagnostic accuracy of lipopolysaccharide-binding protein for predicting bacteremia/clinical sepsis in children with febrile neutropenia: comparison with interleukin-6, procalcitonin, and C-reactive protein

In febrile neutropenia (FN), no reliable marker has been identified to discriminate between severe infection and other causes of fever early in the clinical course. Since lipopolysaccharide-binding protein (LBP) has proven to be an accurate biomarker of bacteremia/clinical sepsis in critically ill non-immunocompromised infants and children, we performed a prospective study to determine the diagnostic accuracy of LBP in children with FN. Concentrations of LBP, procalcitonin (PCT), interleukin-6 (IL-6), and C-reactive protein (CRP) were prospectively measured on two consecutive days in 90 FN episodes experienced by 47 children. Receiver operating characteristic curve analysis was performed for each biomarker to predict bacteremia/clinical sepsis and severe sepsis. Eighteen of the 90 episodes were classified as bacteremia/clinical sepsis. On both days 1 and 2, all biomarkers had a low to intermediate diagnostic accuracy for sepsis, and no significant differences were found between them (area under the curve (AUC) for LBP, 0.648 and 0.714; for PCT, 0.665 and 0.744; for IL-6, 0.775 and 0.775; and for CRP, 0.695 and 0.828). Comparison of their AUCs to the AUC of maximum body temperature on admission (AUC = 0.668) also failed to show any significant differences. In severe sepsis, however, the best diagnostic accuracies were found for IL-6 and PCT (AUC 0.892 and 0.752, respectively), and these were significantly higher than those for LBP (AUC 0.566) on admission. On admission and 24 h later, the LBP concentration is less accurate for predicting bacteremia/clinical sepsis compared to IL-6, PCT, and CRP.