Molecular pneumococcal capsular typing using whole genome sequencing: moving the <i>Streptococcus pneumoniae </i>reference service into the genomic era

Streptococcus pneumoniae isolates typically express one of 92 immunologically distinct polysaccharide capsules (serotypes). Some of these serotypes are difficult to distinguish from each other when using commercially available typing antisera and are defined as members of a serogroup.  Additional factor antisera are then required to definitively allocate serotypes within a serogroup, based on patterns of reactions. However, this method requires extensive experience and the reading of the results can be subjective. We sequenced the genomes of 926 clinical S. pneumoniae isolates and, based on percentage sequence similarities within the polysaccharide capsule operon locus, we determined that, of the 92 serotypes, 32 could be unambiguously identified. The remaining 60 could only be allocated to one of 20 ‘genogroups’ that largely correspond to the immunologically defined serogroups. By comparing reference polysaccharide capsular locus sequences for the 92 serotypes (downloaded from NCBI), unique molecular differences were determined for serotypes within 18 of the 20 genogroups and verified using our set of 926 isolates. These differences were then used in a ‘capsular typing’ bioinformatics tool to predict serotype for 90/94 (92 + 2 molecular types/subtypes) from WGS data and up to serogroup level for types 24B, 24F and serogroup 32. Our ‘capsular typing’ tool has been validated to-date with a set of 426 isolates covering 63/94 capsular types. In 82.5% of the cases the predicted capsular type was concordant with the immunologically derived serotype. Only 2.5% of cases comprise the 4 serotypes the tool can only type to serogroup level. The remaining 15% of cases have given discordant results; in 48% of these our tools predicted to serotype level when the immunological method gave ambiguous results. In summary, we have developed a WGS-based serotyping method that can predict capsular type to serotype level for 90/94 serotypes and to serogroup level for the remaining 4. This approach could be integrated into routine diagnostic workflows, phasing out phenotypic immunological testing. In PHE we plan on implementing this as part of the S. pneumoniae Reference Service by January 2016 following a side-by-side validation period where WGS and the standard phenotypic method are run in parallel.