1459. Whole Genome Sequencing Analysis of Enterococcus faecium Clinical Isolates Reveals High Strain Diversity and High Accuracy Prediction of Antimicrobial Resistance

BACKGROUND: Whole genome sequencing (WGS) is a powerful tool to uncover transmission patterns and antimicrobial resistance (AMR) mechanisms of Enterococcus faecium, a major cause of hospital-acquired infections. Most E. faecium genomic studies include isolates from outbreak investigations rather than routine sampling. Additionally, the use of WGS to predict E. faecium AMR has not been tested systematically. Here we use WGS to characterize over 400 E. faecium clinical isolates to assess their strain diversity and AMR mechanisms. METHODS: Clinical E. faecium isolates from the MGH Microbiology Laboratory were collected at random from 1/2016-12/2017 (derivation set; 193 isolates) and with enrichment for more resistant isolates from 1/2018-9/2019 (validation set; 226 isolates). Species identification was performed using the bioMérieux VITEK MS instrument. Susceptibility testing was performed using the AST-GP75 card (bioMérieux VITEK 2), with confirmation by disk diffusion or ETEST when needed. Bacterial DNA from isolates was extracted, purified, sequenced (Illumina NextSeq), and quality filtered. Samples with >20x genome coverage were analyzed with SRST2 and AliView. RESULTS: MLST analysis of the derivation set demonstrated strikingly high diversity compared to previously published studies, with the three most frequent types (ST412, ST18, ST736) comprising fewer than half of samples. We identified and confirmed four novel MLST types comprising 12% of samples. We next analyzed the derivation isolate set to determine which genes and SNPs, if applicable, predicted resistance to seven antibiotics routinely tested at our institution: ampicillin, ciprofloxacin, doxycycline, high-level gentamicin, levofloxacin, tetracycline, and vancomycin. These rules were uniformly applied to the validation isolate set and demonstrated that genotypic AMR prediction has an overall positive predictive value of 97.0% and negative predictive value of 97.1% compared to standard susceptibility methods. Table 1. Summary of validation set predictions of antimicrobial susceptibility based on defined genotypic features. * The intermediate category is considered with the susceptible category. [Image: see text] CONCLUSION: In a diverse and challenging set of clinical E. faecium isolates, known AMR genes and SNPs can be simply applied to predict phenotypic susceptibility with high accuracy for seven routinely tested antibiotics. Further testing will be performed to resolve phenotype-genotype discrepancies. Summary of validation set predictions of antimicrobial susceptibility based on defined genotypic features. * The intermediate category is considered with the susceptible category. [Image: see text] DISCLOSURES: Melis N. Anahtar, MD, PhD, Day Zero Diagnostics (Other Financial or Material Support, Co-founder, consultant, equity holder) Virginia M. Pierce, MD, Selux Diagnostics, Inc. (Grant/Research Support) Douglas S. Kwon, MD, PhD, Day Zero Diagnostics (Consultant, Shareholder, Other Financial or Material Support, co-founder)