[ODP-208] Evaluation of genotype-based antimicrobial resistance prediction in Serratia marcescens
Serratia marcescens has emerged as an important opportunistic nosocomial pathogen in the last few decades. Although its high level of antimicrobial resistance (AMR) is known, the correlation of resistance gene repertoires and actual resistance phenotypes is not well characterized. The purpose of this study is to evaluate the accuracy of AMR prediction based on genotypes for S. marcescens. A collection of 227 strains was tested for their susceptibilities to 41 antimicrobials. Acquired AMR genes were identified by whole genome sequencing and searched in the ARG-ANNOT database. Genotype-based prediction was summarized based on literature information related to S. marcescens and other species. False-negative rates (resistant but genotyped as susceptible) were notably high (e.g., 42.7% for chloramphenicol), whilst the highest false-positive rate (susceptible but genotyped as resistant) was 14.5% for ampicillin/sulbactam. Even though several genotype-based prediction tools are available, their predictions were not as accurate in S. marcescens as expected for Escherichia coli. This inaccuracy may be due to uncharacterized AMR mechanisms. Moreover, its recent re-classification as Yersiniaceae reinforces that parameters used for Enterobacteriaceae may not be suitable. Therefore, more detailed analysis of genotype-phenotype correlation in S. marcescens is necessary.