Soluble CD14 subtype (sCD14-ST) as biomarker in neonatal early-onset sepsis and late-onset sepsis: a systematic review and meta-analysis

Iris van MaldeghemEmail author View ORCID ID profile,
Charlotte M. Nusman and
Douwe H. Visser

BMC Immunology201920:17

https://doi.org/10.1186/s12865-019-0298-8

Received: 23 August 2018
Accepted: 21 May 2019
Published: 3 June 2019

Abstract

Background

Early diagnosis of bacterial sepsis in neonates is hampered by non-specific symptoms and the lack of rapid responding laboratory measures. The biomarker soluble CD14 subtype (sCD14-ST) seems promising in the diagnostic process of neonatal sepsis. In order to evaluate the differences in diagnostic accuracy of sCD14-ST between early onset sepsis (EOS) and late onset sepsis (LOS) we assessed this systematic review and meta-analysis.

Results

Twelve articles were included in the systematic review and 10 in the meta-analysis. There was a high risk of bias on patient selection, index test and/or flow and timing. The overall quality of the included studies was moderate.

At sepsis onset a consequently higher level of sCD14-ST was found in septic neonates compared to healthy controls with significant higher levels in LOS compared to EOS. In the first 24 h after sepsis onset a significant increase in pooled means of plasma sCD14-ST levels was seen in EOS (t(71.6) = 7.3, p < .0001) while this was not seen in LOS or healthy controls. Optimal cut-off values ranged from 305 to 672 ng/l for EOS cases versus healthy controls. The pooled sensitivity was 81% (95%CI: 0.76–0.85), the pooled specificity was 86% (0.81–0.89) with an AUC of 0.9412 (SE 0.1178). In LOS optimal cut-off values ranged from 801 to 885 ng/l with a pooled sensitivity of 81% (0.74–0.86) and a pooled specificity of 100% (0.98–1.00). An AUC and SROC was not estimable in LOS because of the low number of studies.

Conclusions

sCD14-ST is a promising and rapid-responding diagnostic biomarker for EOS and LOS. The difference in pooled means between EOS and LOS underlines the importance to consider EOS and LOS as two different disease entities, requiring separate analysis in original articles and systematic reviews.