Colistin heteroresistance (HR) refers to a bacterial population comprised of several subpopulations with different levels of resistance to colistin. In this study, we discuss the classic form of HR, in which a resistant subpopulation exists within a predominantly susceptible population. We investigated the prevalence of colistin HR and its evolution into full resistance among 173 clinical carbapenem-resistant Acinetobacter baumannii isolates and examined the effect of HR on clinical outcomes. To determine HR, we performed population analysis profiling. Our results showed a high prevalence of HR (67.1%). To examine evolution of HR strains into full resistance, the HR strains were grown in colistin-containing broth, transferred onto colistin-containing plates, and colonies on these plates were transferred into colistin-free broth. Many of the HR strains (80.2%) evolved into full resistance, 17.2% reverted to HR, and 2.6% were borderline. We used logistic regression to compare 14-day clinical failure and 14-day mortality between patients infected by HR versus susceptible non-HR carbapenem-resistant A. baumannii. In the subgroup of patients with bacteremia, HR was significantly associated with 14-day mortality. IMPORTANCE To our knowledge, this is the first large-scale study to report on HR in Gram-negative bacteria. We described the prevalence of colistin HR in a large sample of carbapenem-resistant A. baumannii isolates, the evolution of many colistin HR isolates to a resistant phenotype following colistin exposure and withdrawal, and the clinical consequences of colistin HR. We found a high prevalence of HR among clinical carbapenem-resistant A. baumannii isolates; most evolved into a resistant phenotype following colistin exposure and withdrawal. In patients treated with colistin, evolution of HR A. baumannii into full resistance could lead to higher rates of treatment failure and contribute to the reservoir of colistin-resistant pathogens in health care settings.
- KEYWORDS Acinetobacter baumannii
- population analysis profiling