Abstract
Streptococcus pneumoniae is the most common cause of community-acquired pneumonia (CAP). Despite the low prevalence of CAP caused by methicillin-resistant Staphylococcus aureus (MRSA), CAP patients often receive empirical antibiotic therapy providing coverage for MRSA such as vancomycin or linezolid. An early differentiation between S. pneumoniae and S. aureus pneumonia can help to reduce the use of unnecessary antibiotics. The objective of this study was to identify candidate biomarkers that can discriminate pneumococcal from staphylococcal pneumonia. A genome-wide transcriptional analysis of lung and peripheral blood performed in murine models of S. pneumoniae and S. aureus lung infection identified an interferon signature specifically associated with S. pneumoniae infection. Prediction models built using a support vector machine and Monte Carlo cross-validation, identified the combination of the interferon-induced chemokines CXCL9 and CXCL10 serum concentrations as the set of biomarkers with best sensitivity, specificity, and predictive power that enabled an accurate discrimination between S. pneumoniae and S. aureus pneumonia. The predictive performance of these biomarkers was further validated in an independent cohort of mice. This study highlights the potential of serum CXCL9 and CXCL10 biomarkers as an adjunctive diagnostic tool that could facilitate prompt and correct pathogen-targeted therapy in CAP patients.