A spitting image: molecular diagnostics applied to saliva enhance detection of Streptococcus pneumoniae and pneumococcal serotype carriage.

Abstract

Despite strong historical records on the accuracy of saliva testing, oral fluids are considered poorly suited for pneumococcal carriage detection. We evaluated an approach for carriage surveillance and vaccine studies that increases the sensitivity and specificity of pneumococcus and pneumococcal serotype detection in saliva samples. Quantitative PCR (qPCR)-based methods were applied to detect pneumococcus and pneumococcal serotypes in 971 saliva samples collected from 653 toddlers and 318 adults. Results were compared with culture-based and qPCR-based detection in nasopharyngeal samples collected from children and in nasopharyngeal and oropharyngeal samples collected from adults. Optimal C q cut-offs for positivity in qPCRs were determined via receiver operating characteristic curve analysis and accuracy of different approaches was assessed using a composite reference for pneumococcal and for serotype carriage based on isolation of live pneumococcus from the person or positivity of saliva samples determined with qPCR. To evaluate the inter-laboratory reproducibility of the method, 229 culture-enriched samples were tested independently in the second center. In total, 51.5% of saliva samples from children and 31.8% of saliva samples from adults were positive for pneumococcus. Detection of pneumococcus by qPCR in culture-enriched saliva exhibited enhanced sensitivity and higher agreement with a composite reference compared to diagnostic culture of nasopharyngeal samples in children (Cohen's κ: 0.69-0.79 vs. 0.61-0.73) and in adults (κ: 0.84-0.95 vs. 0.04-0.33) and culture of oropharyngeal samples in adults (κ: 0.84-0.95 vs. -0.12-0.19). Similarly, detection of serotypes with qPCR in culture-enriched saliva exhibited enhanced sensitivity and higher agreement with a composite reference compared to nasopharyngeal culture in children (κ: 0.73-0.82 vs. 0.61-0.73) and adults (κ: 0.90-0.96 vs. 0.00-0.30) and oropharyngeal culture in adults (κ: 0.90-0.96 vs. -0.13 to 0.30). However, results of qPCRs targeting serotype 4, 5, and 17F and serogroups 9, 12, and 35 were excluded due to assays' lack of specificity. We observed excellent quantitative agreement for qPCR-based detection of pneumococcus between laboratories. After exclusion of serotype/serogroup-specific assays with insufficient specificity, moderate agreement (κ 0.68, 95% CI 0.58-0.77) was observed. Molecular testing of culture-enriched saliva samples improves the sensitivity of overall surveillance of pneumococcal carriage in children and adults, but limitations of qPCR-based approaches for pneumococcal serotypes carriage detection should be considered.