Evaluation of whole-genome sequencing-based subtyping methods for the surveillance of Shigella spp. and the confounding effect of mobile genetic elements in long-term outbreaks

Abstract

Many public health laboratories across the world have implemented whole-genome sequencing (WGS) for the surveillance and outbreak detection of foodborne pathogens. PulseNet-affiliated laboratories have determined that most single-strain food-borne outbreaks are contained within 0–10 multi-locus sequence typing (MLST)-based allele differences and/or core genome single-nucleotide variants (SNVs). In addition to being a food-and travel-associated outbreak pathogen, most Shigella spp. cases occur through continuous person-to-person transmission, predominantly involving men who have sex with men (MSM), leading to long-term and recurrent outbreaks. Continuous transmission patterns coupled to genetic evolution under antibiotic treatment pressure require an assessment of existing WGS-based subtyping methods and interpretation criteria for cluster inclusion/exclusion. An evaluation of 4 WGS-based subtyping methods [SNVPhyl, coreMLST, core genome MLST (cgMLST) and whole-genome MLST (wgMLST)] was performed on 9 foodborne-, travel-and MSM-related retrospective outbreaks from a collection of 91 Shigella flexneri and 232 Shigella sonnei isolates to determine the methods’ epidemiological concordance, discriminatory power, robustness and ability to generate stable interpretation criteria. The discriminatory powers were ranked as follows: coreMLST