Clinical microbiology has long relied on growing bacteria in culture to determine antimicrobial susceptibility profiles, but the use of whole-genome sequencing for antibiotic susceptibility testing (WGS-AST) is now a powerful alternative. This review discusses the technologies that made this possible and presents results from recent studies to predict resistance based on genome sequences. We examine differences between calling antibiotic resistance profiles by the simple presence or absence of previously known genes and single-nucleotide polymorphisms (SNPs) against approaches that deploy machine learning and statistical models. Often, the limitations to genome-based prediction arise from limitations of accuracy of culture-based AST in addition to an incomplete knowledge of the genetic basis of resistance. However, we need to maintain phenotypic testing even as genome-based prediction becomes more widespread to ensure that the results do not diverge over time. We argue that standardization of WGS-AST by challenge with consistently phenotyped strain sets of defined genetic diversity is necessary to compare the efficacy of methods of prediction of antibiotic resistance based on genome sequences.

No abstract

Staphylococcus aureus clinical isolates with vancomycin MICs of 2 g/ml have been associated with vancomycin therapeutic failure and the heteroresistant vancomycin-intermediate S. aureus (hVISA) phenotype. A population analysis profile (PAP) with an area under the curve (AUC) ratio of >0.9 for the AUC of the clinical isolate versus the AUC for hVISA strain Mu3 is most often used for determining hVISA, but it is timeconsuming and labor-intensive. A collection of 140 MRSA blood isolates with vancomycin MICs of 2 g/ml by reference broth microdilution and screened for hVISA using PAP-AUC (21/140 [15%] hVISA) were tested by additional methods to detect hVISA. The methods included (i) Etest macromethod using vancomycin and teicoplanin test strips, brain heart infusion (BHI) agar, and a 2.0 McFarland inoculum; (ii) Etest glycopeptide resistance detection (GRD) using vancomycin-teicoplanin double-sided gradient test strips on Mueller-Hinton agar (MHA) with 5% sheep blood and a 0.5 McFarland inoculum; and (iii) BHI screen agar plates containing 4 g/ml vancomycin and 16 g/liter casein using 0.5 and 2.0 McFarland inocula. Each method was evaluated using PAP-AUC as the reference method. The sensitivity of each method for detecting hVISA was higher when the results were read at 48 h. The Etest macromethod was 57% sensitive and 96% specific, Etest GRD was 57% sensitive and 97% specific, and BHI screen agar was 90% sensitive and 95% specific with a 0.5 McFarland inoculum and 100% sensitive and 68% specific with a 2.0 McFarland inoculum. BHI screen agar with 4 g/ml vancomycin and casein and a 0.5 McFarland inoculum had the best sensitivity and specificity combination, was easy to perform, and may be useful for clinical detection of hVISA.

Vancomycin-intermediate Staphylococcus aureus (VISA) is currently defined as having minimal inhibitory concentration (MIC) of 4–8 µg/ml. VISA evolves through changes in multiple genetic loci with at least 16 candidate genes identified in clinical and in vitro-selected VISA strains. We report a whole-genome comparative analysis of 49 vancomycin-sensitive S. aureus and 26 VISA strains. Resistance to vancomycin was determined by broth microdilution, Etest, and population analysis profile-area under the curve (PAP-AUC). Genome-wide association studies (GWAS) of 55,977 single-nucleotide polymorphisms identified in one or more strains found one highly significant association (P = 8.78E-08) between a nonsynonymous mutation at codon 481 (H481) of the rpoB gene and increased vancomycin MIC. Additionally, we used a database of public S. aureus genome sequences to identify rare mutations in candidate genes associated with VISA. On the basis of these data, we proposed a preliminary model called ECM+RMCG for the VISA phenotype as a benchmark for future efforts. The model predicted VISA based on the presence of a rare mutation in a set of candidate genes (walKR, vraSR, graSR, and agrA) and/or three previously experimentally verified mutations (including the rpoB H481 locus) with an accuracy of 81% and a sensitivity of 73%. Further, the level of resistance measured by both Etest and PAP-AUC regressed positively with the number of mutations present in a strain. This study demonstrated 1) the power of GWAS for identifying common genetic variants associated with antibiotic resistance in bacteria and 2) that rare mutations in candidate gene, identified using large genomic data sets, can also be associated with resistance phenotypes.