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

to the spread of antibiotic resistance (4). There is an urgent need for new biomarkers that can reliably identify SBI. We identified lipocalin-2 (LCN2, neutrophil gelatinaseassociated lipocalin, NGAL) and neutrophil collagenase (matrix metalloproteinase-8, MMP-8) as potential protein candidates to distinguish bacterial pneumonia from viral pneumonia in African children (5). Here, we validate the diagnostic performance of these markers in a cohort of UK children with confirmed, probable or possible SBI versus viral infection and healthy controls. tests. The study had approval of the St Mary's Research Ethics Committee (REC 09/H0712/58). Written, informed consent was obtained. Pathogen diagnosis. Bacterial diagnostics included culture of blood and (when clinically indicated) cerebrospinal and pleural fluid, and pneumococcal antigen detection in blood or urine. Respiratory or nasopharyngeal secretions were screened for viruses by nested PCR, including RSV, coronavirus, adenovirus, parainfluenza 1-4, influenza A+B, bocavirus, metapneumovirus and rhinovirus. Patient categorisation and selection. Patients were assigned to clinical categories corresponding to likelihood of infection after evaluation of all clinical, radiological and laboratory data ( Figure 1 ). Patient categories were 'Definite Bacterial' (DB), 'Probable Bacterial' (PB), 'Uncertain Bacterial or Viral' (U), 'Probable Viral' (PV) or 'Definite Viral' (DV). DB children had microbiologically confirmed bacterial infection identified at a sterile-site, irrespective of the CRP value. As there were only 4 patients in the PV category, analysis was not carried out on this group. Controls (C) were recruited in out-patients. After categorising, patients were prioritised for biomarker estimation based on the quantity of available plasma, and the chronological order of recruitment to give 5 groups as follows: 42 (DB), 34 (PB), 33 (U), 42 (V), 40 (C). Biomarker measurement. Heparinised blood samples were kept at 4C pending plasma recovery by centrifugation and storage at -80C. LCN2 and MMP-8 were measured using commercially available immunoassays (RD Systems, UK) following manufacturer's instructions, by laboratory personnel blinded to the diagnostic group of each sample. Presence of visible haemolysis and the number of previous plasma freeze-thaw cycles prior to EIA had no systematic effect on measurement. The coefficient of variation for these measurements was 8.26% (calculated from 28 samples with 2 replicates). CRP and creatinine values contemporary with research blood sampling were obtained from hospital data for 156 children. CRP was measured by ELISA in controls (RD Systems, UK). Statistical analyses. Diagnostic performance of clinical markers was assessed using the area under the receiver operating characteristic (AUROC) curves to compare the sensitivity and specificity of selected markers. Cut-off values were chosen based on highest sensitivity and specificity to predict outcome (Supplementary Table 1 ). The diagnostic performance of biomarkers was assessed by logistic regression, using LCN2, CRP and MMP-8 as independent variables and the condition to diagnose as the dependent variable. A Chi-square statistic was used to test for equality of the AU-ROC (Stata 11, Stata Corporation) (6). Laboratory measures and ages in each patient group were compared using Kruskal-Wallis tests and Dunn's pairwise comparisons. Frequency variables including sex, ethnicity and severity of illness were