Biomarkers of Pediatric Pnuemonia

Student Researcher:
Jack Underschultz

Supervisor / Principle Investigator:
Dr. Michael Hawkes

Additional Authors:
Michael Hawkes
Bob Opoka
Andrea Conroy
Sophie Namasopo
Jeremy Soo
Ravi Bhargava

MD Class of 2021


Introduction: Pneumonia is the leading cause of childhood infectious death globally. Chest x-ray (CXR) is commonly used as a diagnostic tool, but is not available in many resource-limited areas of the world. We hypothesized that serum biomarkers of inflammation, endothelial activation, and lung injury would be associated with CXR consolidation and could be used as a point-of-care diagnostic test to replace the CXR.

Methods: We conducted a cross-sectional study of 108 Ugandan children hospitalized with clinical pneumonia. All children received a CXR. Enzyme-linked immunosorbent assays were performed to measure serum host response biomarkers: C-reactive protein (CRP), Chitinase 3-like 1 (CHI3L1), Lipocalin-2 (LCN2), Tie-2, Intercellular adhesion molecule-1 (ICAM1), endoglin, Tissue Inhibitor of Metalloprotease-1 (TIMP1), and surfactant protein-D (SP-D).

Results: A total of 108 children were included and categorized by CXR findings as primary end-point pneumonia (n=24), other infiltrates (n=49), or normal CXR (n=35). Clinical variables such as respiratory rate and chest indrawing were not predictive of CXR consolidation. Compared to children with normal x-ray, children with end-point pneumonia had significantly higher levels of CRP, CHI3L1, LCN2, and SP-D. CRP had the best discriminatory power (AUROC 0.73) with a sensitivity of 71% and a specificity of 79%. Compared to children with other infiltrates, children with end-point pneumonia had elevated CRP, CHI3L1, LCN2, ICAM1, and TIMP1. CRP had the best discriminatory power (AUROC 0.75) with a sensitivity of 63% and a specificity of 89%. Biomarkers did not differ statistically between children with other infiltrates and those with normal chest x-rays.

Conclusions: In Ugandan children with clinical pneumonia, host biomarkers distinguished between end-point pneumonia, other infiltrates, and normal CXR; however, the sensitivity and specificity of any individual biomarker may not be sufficient to have clinical utility. Further work exploring the enhanced test performance characteristics of combinations of biomarkers is underway.