Dalbavancin exposure in vitro selects for dalbavancin-non-susceptible and vancomycin-intermediate strains of methicillin-resistant Staphylococcus aureus

Abstract

Objectives: Dalbavancin is a lipoglycopeptide active against methicillin-resistant Staphylococcus aureus (MRSA). Its long half-life (8.5-16 days) allows for once-weekly or single-dose treatments but could prolong the mutant selection window, promoting resistance and cross-resistance to related antimicrobials such as vancomycin. The objective of this study was to evaluate the capacity of post-distributional pharmacokinetic exposures of dalbavancin to select for resistance and cross-resistance in MRSA.

Methods: We simulated average, post-distributional exposures of single-dose (1500 mg) dalbavancin (fCmax 9.9 μg/mL, β-elimination t_1/2 204 h) in an in vitro pharmacokinetic/pharmacodynamic (PK/PD) model for 28 days (672 h) against five MRSA strains and one methicillin-susceptible strain (MSSA). Samples were collected at least daily, and surviving colonies were enumerated and screened for resistance on drug-free and dalbavancin-supplemented medium respectively. Isolates from resistance screening plates were subjected to whole-genome sequencing (WGS) and susceptibly testing against dalbavancin, vancomycin, daptomycin, and six β-lactams with varying penicillin-binding protein (PBP) affinities.

Results: Dalbavancin was bactericidal against most strains for days 1-4 before regrowth of less susceptible subpopulations occurred. Isolates with eight-fold increases in dalbavancin MIC were detected as early as day 4 but increased 64-128-fold in all models by day 28. Vancomycin and daptomycin MICs increased 4-16-fold, exceeding the susceptibly breakpoints for both antibiotics; β-lactam MICs generally decreased by two-to eight-fold, suggesting a dalbavancin-β-lactam seesaw effect, but increased by eight-fold or more in certain isolates. Resistant isolates carried mutations in a variety of genes, most commonly walKR, apt, stp1, and atl.

Conclusions: In our in vitro system, post-distributional dalbavancin exposures selected for stable mutants with reduced susceptibility to dalbavancin, vancomycin, and daptomycin, and generally increased susceptibility to β-lactams in all strains of MRSA tested. The clinical significance of these findings remains unclear, but created an opportunity to genotype a unique collection of dalbavancin-resistant strains for the first time. Mutations involved genes previously associated with vancomycin intermediate susceptibility and daptomycin non-susceptibility, most commonly walKR-associated genes.

Keywords: Cross-resistance; Daptomycin; Lipoglycopeptide; MRSA; PK/PD; Seesaw effect; VISA; walK; walR; β-lactam.

Figure 1.

Mean bacterial survival +/− standard deviation (error bars) of 6 S. aureus strains over 28-days. Time windows when MICs first exceeded susceptibility break points shown in red (vancomycin, MIC≥ 4 mg/L), yellow (daptomycin MIC≥ 2 mg/L), and blue (dalbavancin, MIC≥ 0.25 mg/L); all strains tested became resistant to each drug. Resistance line is the average growth on resistance screening plates +/− standard deviation (error bars) containing 0.25 mg/L of dalbavancin. Only undiluted samples were plated on screening plates so 5.5 log₁₀CFU/mL is the maximum countable. Dotted line is the lower detection limit of filtered samples.

Figure 2.

Median log₂-fold change in MIC (x-axis) of S. aureus isolates recovered from the models over time (y-axis) to each study drug. Error bars indicate range. Beta-lactam MICs for the MSSA strain (W304) were excluded from this analysis. This figure represents over 1100 duplicate MIC experiments on >200 isolates but not all susceptibilities were performed for all isolates recovered for all timepoints. Overall the beta-lactam MICs declined over time on average indicating the seesaw effect. Conversely, the vancomycin and daptomycin MICs increased with increasing dalbavancin MICs over time indicating cross resistance.

Figure 3.

Hierarchical clustering of a selection of 52 strains generated from dalbavancin PK/PD models based on genotype. Red indicates the presence of any genetic variant in the genes across the bottom in the strains listed on the right. Mutants are named by using the following format: [parent strain]-[timepoint in model when isolated], [replicate of the model (A or B)]

Figure 4.

(A-C) Partial least squares regression analysis of contribution of gene mutations to variations in MIC of the indicated drugs. The larger the loading, the larger the contribution of the mutation to the MIC variation. Red rectangle: walK or walR; green rectangle: apt. (D-E) Correlation between dalbavancin and vancomycin (D) and daptomycin (E) fold-changes in MIC of resistant strains relative to parent strains.