Evaluation of the Bruker IR Biotyper for Streptococcus pyogenes emm typing

ABSTRACT Streptococcus pyogenes (Group A Streptococcus [GAS]) is a human pathogen with invasive potential. Investigation of isolate relatedness by emm gene sequencing is time- and cost-intensive, warranting the evaluation of alternative technologies for this purpose. In this study, we evaluated the level of discrimination of predominant GAS isolates that can be achieved by Fourier transform-infrared spectroscopy on the Bruker IR Biotyper platform. Separate emm pattern, emm type, and emm 1 sublineage classifiers were created and cross-validated using spectra from isolates with known emm types. Next, emm pattern and emm type classifiers were combined into a tiered classifier that was externally validated using novel isolates and evaluated in a mock outbreak scenario. Emm pattern and type classifiers demonstrated > 95% accuracy and class precision during cross-validation. The emm 1 sublineage classifier had poor accuracy (<90%) and was excluded from the final tiered classifier. The tiered classifier performed well during external validation, with overall accuracies of 98% and 96% for emm pattern and type, respectively. Individual emm type class recall was variable and impacted by the total number of isolates included. The tiered classifier demonstrated 100% accuracy, class recall, and precision in the simulated outbreak. This study demonstrates that the Bruker IR Biotyper can accurately group GAS isolates to emm type when trained with a tiered classifier model. Moreover, it can be used as a rapid and accurate method to exclude GAS isolates of different emm patterns from outbreak strains and therefore has utility in point prevalence screening and contact tracing during outbreak management. IMPORTANCE Streptococcus pyogenes (Group A Streptococcus [GAS]) is a human pathogen capable of causing localized as well as invasive infection. The latter is associated with significant mortality and, without correct infection control measures, can result in bacterial transmission to close contacts. Management of a GAS outbreak requires the investigator to demonstrate relatedness between bacterial isolates collected from different patients. Typing is based on molecular analysis of part of the M protein gene ( emm ) at centralized reference centers. The turnaround time for testing is often quite prolonged, leading to delays in outbreak recognition and patient care cascades. This study therefore sought to first evaluate the ability of a novel Fourier transform-infrared spectroscopy system—the Bruker IR Biotyper—to accurately identify GAS emm types. Secondarily, we validated a GAS classifier model using both emm types and broader emm patterns as a means by which to accurately, and rapidly, categorize isolates.