RESUMO
Methicillin-resistant Staphylococcus aureus (MRSA) is a leading cause of health care-associated (HA) and community-associated (CA) infections. USA300 strains are historically CA-MRSA, while USA100 strains are HA-MRSA. Here, we update an antibiotic prediction rule to distinguish these two genotypes based on antibiotic resistance phenotype using whole-genome sequencing (WGS), a more discriminatory methodology than pulsed-field gel electrophoresis (PFGE). MRSA clinical isolates collected from 2007 to 2017 underwent WGS; associated epidemiologic data were ascertained. In developing the rule, we examined MRSA isolates that included a population with a history of incarceration. Performance characteristics of antibiotic susceptibility for predicting USA300 compared to USA100, as defined by WGS, were examined. Phylogenetic analysis was performed to examine resistant USA300 clades. We identified 275 isolates (221 USA300, 54 USA100). Combination susceptibility to clindamycin or levofloxacin performed the best overall (sensitivity 80.7%, specificity 75.9%) to identify USA300. The average number of antibiotic classes with resistance was higher for USA100 (3 versus 2, P < 0.001). Resistance to ≤2 classes was predictive for USA300 (area under the curve (AUC) 0.84, 95% confidence interval 0.78 to 0.90). Phylogenetic analysis identified a cluster of USA300 strains characterized by increased resistance among incarcerated individuals. Using a combination of clindamycin or levofloxacin susceptibility, or resistance to ≤2 antibiotic classes, was predictive of USA300 as defined by WGS. Increased resistance was observed among individuals with incarceration exposure, suggesting circulation of a more resistant USA300 clade among at-risk community networks. Our phenotypic prediction rule could be used as an epidemiologic tool to describe community and nosocomial shifts in USA300 MRSA and quickly identify emergence of lineages with increased resistance. IMPORTANCE Methicillin-resistant Staphylococcus aureus (MRSA) is an important cause of health care-associated (HA) and community-associated (CA) infections, but the epidemiology of these strains (USA100 and USA300, respectively) now overlaps in health care settings. Although sequencing technology has become more available, many health care facilities still lack the capabilities to perform these analyses. In this study, we update a simple prediction rule based on antibiotic resistance phenotype with integration of whole-genome sequencing (WGS) to predict strain type based on antibiotic resistance profiles that can be used in settings without access to molecular strain typing methods. This prediction rule has many potential epidemiologic applications, such as analysis of retrospective data sets, regional monitoring, and ongoing surveillance of CA-MRSA infection trends. We demonstrate application of this rule to identify an emerging USA300 strain with increased antibiotic resistance among incarcerated individuals that deviates from the rule.
