O0177Interleukin-6 as a diagnostic biomarker for sepsis in neonates, children and pregnant women: a real-world cohort study

02. Bacterial infection & disease
02b. Severe sepsis, bacteraemia & endocarditis (incl epidemiology, diagnosis, host biomarkers, treatment, and outcome prediction)
S.O. Whelan1, 2, 3, M. Eogan2, 4, N. Mccallion2, 4, A. Walsh1, F. Moriarty4, M. Elsammak1, 2, 5, R. Drew2, 3, 4.
1CHI at Temple Street - Dublin (Ireland), 2Rotunda Hospital - Dublin (Ireland), 3Irish meningitis and sepsis reference laboratory - Dublin (Ireland), 4RCSI University of Medicine & Health Sciences - Dublin (Ireland), 5National Newborn Bloodspot Screening Laboratory - Dublin (Ireland)

Background

Biomarkers may be useful adjuncts to clinical assessment in the diagnosis of paediatric, maternal and neonatal sepsis. Traditional biomarkers like C-reactive protein (CRP) and procalcitonin (PCT) are sub-optimally sensitive and limited by delayed response to infectious stimuli. Interleukin-6 (IL-6) has a favourable dose-response relationship, and therefore may have greater diagnostic promise in the early detection and risk stratification of patients with suspected sepsis. There have been few real-world studies performed to date in a broad paediatric population, and none amongst pregnant women. Our aim was to assess diagnostic performance of serum IL-6 compared to in-use biomarkers (PCT, CRP, neutrophil-lymphocyte ratio [NLR]) in these populations.


Methods

This was a retrospective cohort study performed in a specialist children’s hospital and a maternity hospital. All patients in whom IL-6 was measured to investigate suspected sepsis over one-year were included. Serial sampling was used to describe kinetics. Case categorisation was performed using pre-specified criteria along aetiological (bacterial, viral or no infection) and physiological (normal, systemic inflammatory response syndrome, sepsis and septic shock) axes. Biomarker performance was assessed by area under the receiver operating characteristic curve (AUROC) and measures of diagnostic accuracy at optimal cut-points.


Results

252 patients were included (111 paediatric, 72 maternity and 69 neonatal). IL-6 was the only marker with significantly different median values across all categories for both axes in all populations. IL-6 fell rapidly on repeated sampling, compared to further increases in other markers. IL-6 had statistically significantly higher AUROCs versus other markers for primary physiological and aetiological sepsis end-points in the paediatric population, for primary maternal physiological and aetiological end-points (apart from NLR), and for the neonatal sepsis end-point (apart from PCT). It performed particularly well at discriminating bacterial from non-bacterial infection, with sensitivities/specificities >80% in the paediatric (AUROC 0.91, cut-point 43.6pg/ml) and maternity (AUROC 0.94, cut-point 40.2pg/ml) populations.


Conclusions

This study of the real-world performance of IL-6 in the diagnosis of sepsis suggests diagnostic promise in these populations compared to conventional biomarkers. Further prospective study of the impact on patient outcomes is merited.


Conclusions

Case(s) description

Discussion

References

Figure 1. Fractional polynomials showing maternity biomarker trajectories of individual patients in grey, and superimposed best-fit line, in red.

Figure 2. Graphic summarising IL-6 profiles for each population and axis. Green denotes statistically significant differences between each category within the relevant Axis. Stairs figure indicates stepwise progression across axes (e.g. increasing incrementally from normal to septic shock).

Figure 3. Summary of the diagnostic performance of each biomarker for the primary aetiological (top) and physiological (bottom) outcomes in the paediatric population, with area under the receiver operator characteristic curve (left), and table summarising diagnostic accuracy measures,

Keyword 1
Bacteria and bacterial infections
Keyword 2
Paediatric ID
Keyword 3 (Please provide your suggestion)
Sepsis

Conflicts of interest


Do any of the authors have conflicts of interest related to the studies presented in this abstract?
No