Analysis of Paradoxical Efficacy of Carbapenems against Carbapenemase-Producing Escherichia coli in a Murine Model of Lethal Peritonitis

Abstract

The clinical benefit of carbapenems against carbapenemase-producing Enterobacteriaceae (CPE) remains in question. MICs of imipenem (IMP) and ertapenem (ERT) against isogenic derivatives of the wild-type strain Escherichia coli CFT073 producing KPC-3, OXA-48, or NDM-1 were 0.25, 2, 16, and 64 mg/liter for IMP and 0.008, 0.5, 8, and 64 mg/liter for ERT, respectively. Swiss ICR-strain mice with peritonitis were treated for 24 h with IMP or ERT. Despite a limited duration of time during which free antibiotic concentrations were above the MIC (down to 0% for the NDM-1-producing strain), IMP and ERT significantly reduced bacterial counts in spleen and peritoneal fluid at 24 h (P < 0.005) and prevented mortality. Several possible explanations were investigated. Addition of 4% albumin or 50% normal human serum did not modify IMP activity. Bacterial fitness of resistant strains was not altered and virulence did not decrease with resistance. In the presence of subinhibitory concentrations of ERT, growth rates of OXA-48, KPC-3, and NDM-1 strains were significantly decreased and filamentation of the NDM-1 strain was observed. The expression of bla_NDM-1 was not decreased in vivo compared to in vitro No zinc depletion was observed in infected mice compared with Mueller-Hinton broth. In conclusion, a paradoxical in vivo efficacy of IMP and ERT against highly resistant carbapenemase-producing E. coli was confirmed. Alternative mechanisms of antibacterial effects of subinhibitory concentrations of carbapenems may be involved to explain in vivo activity. These results are in agreement with a potential clinical benefit of carbapenems to treat CPE infections, despite high carbapenem MICs.

Keywords: Escherichia coli; antibiotic resistance; carbapenemase; carbapenems; peritonitis.

FIG 1

Bacterial counts in peritoneal fluid before treatment and after 24 h treatment with IMP or ERT for each strain. Results are expressed in log₁₀ CFU/ml.

FIG 2

Normalized expression of bla_NDM-1 (compared to rrsA) in E. coli CFT073-NDM1 cells grown in vitro or in vivo conditions. ns, not significant.

FIG 3

Time-kill curves of IMP at the MIC alone or in combination with or without 50% pooled human serum against susceptible E. coli CFT073 (A) or resistant E. coli CFT073-NDM-1 (B).

FIG 4

Time to achieve maximal growth rate (T_max × 10⁴ s) of the study strains in the presence of subinhibitory concentrations of ERT (left) and IMP (right).

FIG 5

Photographs of rod-shaped NDM-1-producing E. coli after 24 h growth in MH medium with no antibiotic (A), ERT at 8 mg/liter (0.125 × MIC) (B), or IMP at 8 mg/liter (0.125 × MIC) (C). Photographs were taken using a light microscope at magnification of ×1,000 before (upper) and after (lower) Gram staining.