Bacterial Identification using 16S rRNA Gene Sequencing in a University Teaching Hospital

Abstract

Background. 16S rRNA gene sequencing can identify bacteria that are not identified using manual and automated systems, and characterize previously undescribed species. We describe our experience with 16S sequencing at the 1500-bed Yale New Haven Hospital (YNHH) in New Haven, CT, USA. Methods. The microbiology laboratory at YNHH performs 16S sequencing weekly on bacterial isolates that cannot be identified with biochemical tests (API, RapID, Vitek) or MALDI-TOF MS. Bacterial colonies are processed using MicroSeq sequencing kits and a validated protocol. Sequence data are analyzed using MicroSeq and BLAST databases, and identification is assigned using Clinical and Laboratory Standards Institute (CLSI) guidelines MM-18. We reviewed 16S rRNA sequencing results obtained between April 2011 and July 2016. Results. 586 isolates from 581 clinical specimens belonging to 513 patients were sequenced. Most isolates were from the blood (34%), respiratory (33%) and wound (25%) benches (Figure 1). Anaerobes (32%), mycobacteria (23%), and Gram-negative rods (21%), accounted for 76% of isolates. For 566 (97%) of isolates, an identification was made to the species (62%), complex (14%) or genus (24%) (Figure 2). These 566 isolates were distributed among 108 unique genera (Figure 3). The five most common genera were Mycobacteria (24%), Actinomyces (13%), Nocardia (6%), Clostridium (5%), and Aerococcus (2%) species. 49 genera were represented by a single isolate, 17 genera by 2 isolates each, and 11 genera by 3 isolates each. Of the remaining 20 isolates (3%), 11 were identified at the taxonomical level of family or order. For 5 isolates, no identification was possible, while 4 isolates were potentially novel species. Following introduction of MALDI-TOF MS there was a decline in isolates needing 16S sequenc-ing for identification. Conclusion. Our experience with 16S sequencing demonstrates its utility in clinical microbiology. We identified organisms from a wide variety of specimen types with an identification at the level of genus or below in 97% of isolates. 566 isolates belonged to 108 unique genera, with 4 potential novel species, suggesting conventional identification methods may underestimate the diversity of bacteria in clinical specimens.