Comparison of genome-derived and phenotypic antimicrobial resistance profiles of Shigella species isolated from patients with symptoms of gastrointestinal disease in England, 2015-2020

Increasing antimicrobial resistance (AMR) in Shigellaspecies is a global public health problem. We compared genotypic and phenotypic methods for the detection of AMR in Shigellaspecies to evaluate the use of genome data for surveillance and monitoring of emerging AMR.

Whole genome sequencing (WGS) data from 388 isolates of all four Shigella species were analysed for the presence/absence of specific AMR determinants and selected accordingly. Phenotypic antimicrobial susceptibility testing was performed using in-agar dilution on all viable and pure isolates (n=358). The genotypic and phenotypic AMR profiles were then compared.

There were 335 (93.6%) isolates resistant to at least one antimicrobial and 222 (62%) isolates were multi-drug resistant, of which the majority (77%) were associated with foreign travel. Out of a possible 2864 isolate/antimicrobial class combinations, we identified 119 unexpected results, giving an overall concordance of 96.8% between the two methods. There were 54 samples that had an AMR determinant expected to confer resistance that were phenotypically susceptible, of which 31/54 (57.4%) were associated with tetracycline resistance and trimethoprim-sulfamethoxazole resistance. There were 65 that were phenotypically resistant to a specific antimicrobial class, but no AMR determinant was detected, of which 32/65 (49.2%) were associated with isolates harbouring a single gyrA mutation and exhibiting an unexpectedly high minimum inhibitory concentration (MIC) to ciprofloxacin.

Although comparisons between both methods showed good correlation between the genotypic and phenotypic AMR profiles, phenotypic monitoring is required to identify novel AMR mechanisms and to update reference database used for WGS analysis.