In silico and in vitro analyses for the improved diagnosis of bacterial meningitis.

Context: Diagnosing meningitis remains challenging with etiological agents frequently unidentified. Using both in silico and in vitro approaches, this study evaluated published and novel genetic targets for the detection of common bacterial species known to cause meningitis: Neisseria meningitidis, Streptococcus agalactiae, Streptococcus pneumoniae, and Haemophilus influenzae. Methods: A total of 29 genetic targets were investigated for the detection of N. meningitidis, S. agalactiae, S. pneumoniae, and H. influenzae, using the Gene Presence tool and whole genome sequence data (WGS) found in the genomics platform, PubMLST. These targets were further tested in silico by screening WGS using the PCR tool hosted on PubMLST allowing the sensitivity, specificity, Negative Predicted Values (NPV) and Positive Predictive Values (PPV) to be determined. Ten targets were then further evaluated in vitro by real-time PCR against a panel of 44 bacterial isolates representative of the genera evaluated. Results: The best performing in silico genetic determinants targeted: N. meningitidis, sodC (NEIS1339) (sensitivity 99.7%, specificity, 99.4%, PPV, 99.6% and NPV, 99.6%); S. pneumoniae, SP2020 (99.5%, 99.9%, 99.9%, and 81.5%) and H. influenzae, dmsA (HAEM1183) (98%, 100%, 99.6%, and 77.4%). All three of these targets also had the best in vitro sensitivity (100%), specificity [91.7% sodC (NEIS1339), 100% SP2020 and 97.3% dmsA (HAEM1183), PPV (72.7% sodC (NEIS1339), 100% SP2020 and 87.5% dmsA (HAEM1183)] and NPV (100% for all targets). The gene sip (SAG0032) encoding the surface immunogenic protein (sip) exhibited the best sensitivity (99.6%) and NPV (96.9%) for S. agalactiae compared to 99.3% and 94.8% for cfb (SAG2043), respectively in silico. However, in vitro, cfb showed the best sensitivity (100% vs. 85.7%) and NPV (100% vs. 97.4%) when compared to sip. Conclusion: SodC, cfb, SP2020, and dmsA have the potential to enhance the accuracy of molecular diagnostics for the four most common bacterial species causing meningitis. Moreover, a combined in silico and in vitro approach that leverages WGS deposited in databases such as PubMLST, offers an efficient and cost-effective means for the preliminary evaluation of diagnostic targets.