Abstract
In this report, we examined the (1) activity of ceftazidime–avibactam against clinical isolates Klebsiella pneumoniae, including those harboring blaKPC, (2) potential mechanisms leading to reduced susceptibility, and (3) activity of ceftazidime–avibactam when combined with other agents. Of 802 carbapenem-resistant isolates of K. pneumoniae gathered from New York City from 1999 to 2014, all were susceptible to ceftazidime–avibactam. Minimum inhibitory concentrations (MICs) were higher in isolates with K. pneumoniae, with the carbapenemase (KPC)-3 (compared to KPC-2), and those with a frameshift mutation in ompK35. MICs did not appear to be affected by changes in ompK36 or by expression of acrB. Time-kill experiments demonstrated synergy between either polymyxin B or amikacin and ceftazidime–avibactam in a minority of isolates. In conclusion, ceftazidime–avibactam is active against K. pneumoniae, including blaKPC isolates, in our region, but activity is affected by KPC subtype and by mutations in ompK35. Synergy can occur when combined with polymyxin B or amikacin, but is unpredictable.
Introduction
T
Materials and Methods
Clinical isolates of K. pneumoniae were gathered from periodic surveillance studies conducted from 1999 to 2013–2014 in New York City. 9 Only single-patient isolates were gathered. Isolates that were resistant to ertapenem underwent susceptibility testing to ceftazidime–avibactam using the agar dilution method with Mueller-Hinton agar. For some isolates, MICs were also determined for polymyxin B, amikacin, and tigecycline and interpreted using Clinical and Laboratory Standards Institute breakpoints. 10 The presence of blaKPC was determined by PCR. 11 Select isolates had SHV and TEM β-lactamases and ompK35 and ompK36 identified by PCR using previously identified primers and conditions. 11 In addition, expression of the efflux gene acrB was determined by real-time (RT)-PCR, as previously described. 11 RT-PCR experiments were performed in triplicate using primer and probe concentrations that resulted in amplification efficiencies of 90–110%. The calibrator for acrB expression was K. pneumoniae ATCC 11296. A subset of isolates also underwent genetic fingerprinting by the rep-PCR method with ERIC-2 primer, as previously described. 12 Isolates were considered related if there was a 0–1 band difference.
Time-kill experiments were performed using antibiotic concentrations 0.25 × the MIC in cation-supplemented Mueller-Hinton broth and log phase of growth cultures with a starting inoculum of ∼1 × 106 cfu/ml. 13 The lowest level of detection was 5 × 101 cfu/ml. Synergy was defined as a ≥100-fold increase in killing by a combination at 24 hours compared to either agent alone, and bactericidal activity was defined as a ≥3 log10 cfu/ml decrease at 24 hours.
Statistical comparisons were performed using Student's t test; a two-tailed p value of ≤0.05 was considered significant.
Results
Overall susceptibility results
A total of 802 ertapenem-resistant isolates underwent susceptibility testing. All were susceptible to ceftazidime–avibactam, with MIC50 and MIC90 values of 1 and 2 μg/ml, respectively. Most isolates were from surveillance studies conducted in 2006, 2009, and 2013–2014 and were resistant to other classes of antimicrobial agents (Table 1). For the isolates gathered in the years 2006 (n = 273), 2009 (n = 302), and 2013–2014 (n = 120), there was a gradual creep in the average MIC of ceftazidime–avibactam (0.92 ± 0.69 μg/ml, 1.02 ± 0.90 μg/ml, and 1.15 ± 0.86 μg/ml, respectively; p = 0.01 for 2006 vs. 2013–2014). This was possibly due to the transition from strains carrying predominantly blaKPC-2 to strains with blaKPC-3 (Table 2). For example, in 2004, 95% (59 of 62) KPC-possessing isolates had blaKPC-2. 14 By 2013–2014, 40% (47 of 116) of KPC-possessing isolates had blaKPC-2. For each year, the remainder of isolates that were tested possessed blaKPC-3. Compared to the isolates with KPC-2, the isolates with KPC-3 had higher MIC50 values (1 μg/ml vs. 0.5 μg/ml, respectively). Of the isolates with blaKPC-2, 88% had an MIC of ceftazidime–avibactam of ≤1 μg/ml and 10% had an MIC of 2 μg/ml. In comparison, for the isolates with blaKPC-3, 61% had an MIC of ceftazidime–avibactam of ≤1 μg/ml and 34% had an MIC of 2 μg/ml. Twenty-three of the 116 KPC-producing isolates from 2013–2014, from 11 different hospitals, underwent fingerprinting. Eight rep-PCR types were identified, including 12 isolates that belonged to one clone (data not shown).
KPC, Klebsiella pneumoniae with the carbapenemase.
Susceptibility results of the characterized isolates
Forty-one isolates, including 17 without blaKPC and 24 with blaKPC, had characterization of β-lactamases and outer membrane porins. Compared to the 17 isolates without blaKPC, the 24 isolates with this β-lactamase had significantly higher mean MICs to ceftazidime–avibactam (2.23 ± 2.25 μg/ml vs. 0.24 ± 0.29 μg/ml, p < 0.001). Compared to isolates that did not possess an ESBL, the presence of an ESBL (generally SHV-12) did not appear to affect the MICs of ceftazidime–avibactam. The average MIC for the isolates with an ESBL was 1.68 ± 2.4 μg/ml compared to 1.44 ± 1.52 μg/ml for isolates without an ESBL; this finding was also evident when only isolates without blaKPC were analyzed.
Fifteen of the 41 isolates possessed a frameshift mutation in ompK35 at amino acid 41. These isolates had a higher ceftazidime–avibactam mean MIC compared to isolates without this disruption (2.50 ± 2.76 μg/ml vs. 0.80 ± 0.95 μg/ml, p = 0.04); this difference was confined to isolates that possessed blaKPC (3.36 ± 2.76 μg/ml vs. 1.27 ± 1.10 μg/ml, p = 0.03). Two major phenotypes of OmpK36 were revealed; 14 isolates possessed a Leu-Pro insertion in loop four and an Ala insertion in loop five, and 14 isolates possessed an insertion of Gly-Asp in loop three, Ser-Pro in loop four, and Asn-Gly-Glu-Ser-Asp-Ser insertion, along with substitutions, in loop eight. Nearly all the isolates possessing blaKPC possessed these alterations in OmpK36. For the 17 isolates that did not possess blaKPC, the average MICs for isolates with each of these phenotypes (0.11 ± 0.02 μg/ml and 0.19 ± 0.09 μg/ml, respectively) were not different from isolates lacking either of these changes (0.27 ± 0.36 μg/ml).
The relationship between expression of acrB with the MICs to ceftazidime–avibactam was also assessed. Expression of acrB was similar among KPC-producing isolates with higher and lower MICs.
Results of time-kill experiments
Ten isolates were selected, based on varying fingerprint patterns and susceptibility to ceftazidime–avibactam, for time-kill experiments (Table 3). Fingerprinting of the isolates revealed the three isolates without blaKPC were all unique. Of the seven with blaKPC, four belonged to one clonal group and three were unique strains. All isolates had growth in the presence of one fourth of the MIC of ceftazidime–avibactam, polymyxin B, amikacin, and tigecycline. At 24 hours, bactericidal and synergistic killing was found in three isolates (two with blaKPC) with the combination of ceftazidime–avibactam and polymyxin B and one isolate with ceftazidime–avibactam and amikacin (Fig. 1).
Time-kill studies involving synergistic combinations involving ceftazidime–avibactam, involving Klebsiella pneumoniae isolates.
Discussion
Although the acquisition of KPC raised the MICs to ceftazidime–avibactam, all the isolates remained susceptible to this agent, with all having MICs below the breakpoint of 8 μg/ml. Isolates with KPC-3 had MICs double that of isolates with KPC-2, consistent with findings reported elsewhere and with the increased hydrolytic activity of KPC-3 against ceftazidime. 15 The addition of other β-lactamases common in our region (largely SHV-11 and SHV-12) did not appear to affect susceptibility to ceftazidime–avibactam. Although a membrane diffusion barrier exists in K. pneumoniae for avibactam, one study involving isolates without blaKPC did not suggest that the loss of OmpK35 affects ceftazidime–avibactam susceptibility. 16 However, in our KPC-producing isolates, loss of OmpK35 did reduce susceptibility, likely due to reduced entry of ceftazidime. 17 Consistent with other reports, we could not find any significant contribution to alterations in OmpK36 with reduced susceptibility to ceftazidime–avibactam.15,16 Expression of acrB did not correlate with ceftazidime–avibactam susceptibility, suggesting that efflux of avibactam is not a major factor.7,16
Finally, given that combination therapy results in improved outcomes of infections due to carbapenem-resistant K. pneumoniae, we sought to identify antibiotic combinations that may be synergistic with ceftazidime–avibactam. For a minority of isolates, the combination of ceftazidime–avibactam with a polymyxin or an aminoglycoside resulted in bactericidal and synergistic killing. However, this finding was unpredictable, and individual isolates would require testing before a particular combination could be suggested.
Footnotes
Disclosure Statement
No competing financial interests exist.