Lipocalin-2 is a Sensitive and Specific Marker of Bacterial Infection in Children

Jethro Herberg; Honglei Huang; Marie L. Thezenas; Victoria Janes; Michael Carter; Stuart Gormley; Melisa S. Hamilton; Benedikt Kessler; Michael Levin; Climent Casals-Pascual

LCN2 is a 21-kD glycoprotein secreted by neutrophils, hepatocytes and renal tubular cells and it is usually found at low concentration ( 20 ng/mL) in biological fluids (8). The role of LCN2 in the innate defence against bacterial infection has been attributed to its ability to interfere with bacterial iron uptake through competition with the bacterial siderophore enterobactin (9, 10). The role of LCN2 as a potential marker of bacterial infection as described in previous studies (5, 11). Similarly, LCN2 gene expression has been shown to discriminate bacterial from viral infection, with 22-fold increased expression in bacterial patients compared to controls (12) . Un- like CRP, the diagnostic performance of LCN2 is unaffected by age: LCN2 gene expression is central to an immunemetabolic network implicated in neonatal bacterial infection (13) . In comparison with CRP, LCN2 has a much shorter half-life (<20 minutes) (14) , rising more rapidly than CRP in a study in neonates (15) and responding more rapidly to successful antibiotic treatment (11) . Although a fast turnover is a desirable quality for a biomarker to monitor infection and treatment, it is also a diagnostic limitation when patients have received prior antibiotic treatment at the time of diagnosis. LCN2 has recently emerged as a biomarker of acute kidney injury (7). We investigated if the association of LCN2 with SBI was confounded by impaired renal function. Adjusting the analysis for creatinine concentration did not affect the diagnostic performance of LCN2 to predict SBI, which suggests that neutrophils are the likely source of LCN2. The diagnostic performance of LCN2 exceeded that of MMP-8, previously proposed as a paediatric sepsis biomarker (16) . In this study we have defined patient groups by a gradient from low-to high likelihood of SBI. This has the advantage to capture the real world situation in which the majority of patients have no definitively confirmed or excluded infection. The clinical group most likely to have SBI was defined by a positive blood culture (gold standard) and we used this variable as our predicted outcome to evaluate the diagnostic performance of LCN2 and MMP-8. Thus, it was not possible to compare diagnostic value of LCN2 to the gold standard; instead, we used CRP for comparison as the measurement of this biomarker to guide antibiotic treatment is commonly used in secondary and tertiary care. We have shown that the biomarker concentration of LCN2 and MMP-8 was proportional to the estimated likelihood of SBI, with unconfirmed patients tending to have intermediate levels (47% of PB group had LCN2 > 175.3 ng/mL threshold). We did not compare the levels of LCN2 or MMP-8 to CRP in the groups with diagnostic uncertainty, as this comparison would be confounded by the use of the CRP result to allocate the patients to clinical groups (PB, U, DV). Procalcitonin is a promising biomarker of bacterial infection with kinetics comparable to those of LCN2 and it shown a modest advantage over CRP in the prediction of SBI (17), but we did not have these data for comparison. Unless a biomarker is identified which discriminates SBI and non-SBI patients at non-overlapping concentrations, children with intermediate, 'grey' clinical markers of bacterial infection will remain as the group in whom the biomarkers are least effective, but most needed. The potential for LCN2 alone or in combination with CRP to improve the rapid identification of children with SBI at an earl