Validation of whole genome sequencing analyses to predict antimicrobial MICs to replace phenotypic antimicrobial susceptibility testing for surveillance of Streptococcus pyogenes and Streptococcus agalactiae

Funding period: 2023–2025
Leads: Alyssa Golden and Irene Martin
Total GRDI funding: $290,300

Invasive Streptococcus pyogenes (group A Streptococcus, group A strep) is responsible for a wide range of human diseases, the most serious of which include bacteraemia, streptococcal toxic shock syndrome, necrotizing fasciitis and endocarditis.

Invasive Streptococcus agalactiae (group B streptococcus, GBS) is commonly associated with neonatal disease, manifesting in bacteraemia, pneumonia and meningitis; it is also increasingly associated with disease in adults, particularly bacteraemia and skin/soft tissue infections. The National Microbiology Laboratory (NML) conducts antimicrobial susceptibility testing (AST) of GAS and GBS submitted by provincial and territorial (P/T) public health laboratories for surveillance and outbreak response purposes. Currently, AST is performed using the gold-standard Kirby-Bauer disc diffusion method for seven antimicrobials. While this method is cost effective and easy to perform, it has the critical drawback of only producing qualitative results (i.e., a categorical interpretation of susceptible, intermediate or resistant to an antimicrobial). NML also performs broth microdilution testing with a large panel of antimicrobials, however this done for a limited number of GAS and GBS isolates as there is minimal funding to routinely perform this test.

Recent reductions in the cost of whole genome sequencing (WGS) reagents and improved methodologies now provide a possibility of routinely sequencing all isolates received for characterization at less cost than acquiring the same information phenotypically. Identifying molecular antimicrobial resistance determinants using WGS will allow us to recognize and characterize emerging resistance mechanisms, as well as monitor resistance patterns, which are both critical to protecting public health by informing treatment and interventions. Replacing traditional phenotypic testing with validated minimum inhibitory concentration (MIC) predictions derived using machine learning methods will result in cost savings and improve the efficiency of laboratory workflows while increasing analytical capacity; and the validation data will form a diverse reference sample set to monitor shifts in molecular epidemiology. By developing and sharing methodologies to advance the field of surveillance with the close collaboration of our frontline P/T laboratories, our mandate to provide leadership, partnership and innovation will be fulfilled.

We propose to validate the methodology for routine prediction of GAS and GBS MICs using WGS, which will replace our existing phenotypic workflows. Our aims are to: i) perform quantitative antimicrobial susceptibility testing by E-test to obtain accurate and precise MIC values for GAS and Full Project - GRDI Round 8 GBS isolates, ii) complete the corresponding WGS for these isolates, and iii) use machine learning techniques to design, develop and validate the methodology to predict MICs from WGS data.