ABSTRACT

TEXT

Pseudomonas aeruginosa is a nosocomial pathogen associated with significant morbidity and mortality rates (1). Ceftolozane-tazobactam is a cephalosporin-β-lactamase inhibitor combination antibiotic that has better outer membrane permeability and improved stability against chromosomal AmpC β-lactamase than do other β-lactam antibiotics, resulting in potent in vitro activity against P. aeruginosa, including multidrug-resistant strains (2, 3). In vitro selection for ceftolozane-tazobactam resistance in P. aeruginosa requires multiple mutations leading to overexpression and structural modifications of AmpC (4). The emergence of P. aeruginosa ceftolozane-tazobactam resistance due to AmpC overexpression and structural modifications was recently reported for two patients during prolonged courses of ceftolozane-tazobactam treatment, albeit involving different amino acid positions than in the in vitro selection study (5). Here we investigated the mechanism(s) leading to in vivo ceftolozane-tazobactam resistance development in sequential clinical P. aeruginosa isolates following 6 weeks of ceftolozane-tazobactam treatment.

In April 2015, a 75-year-old man presented to a tertiary-care hospital in South Carolina with left neck wound dehiscence suggesting infection. Following resection and X-ray therapy, the patient had experienced recurrent wound infections with mixed Gram-positive and Gram-negative organisms, including Pseudomonas aeruginosa (pan-susceptible), dating back to December 2014. The patient had continuously received antimicrobial therapy since December 2014, due to inadequate closure of the fistula, including courses of vancomycin plus piperacillin-tazobactam (and subsequently meropenem), followed by trimethoprim-sulfamethoxazole and ciprofloxacin.

An operating room (OR) wound culture in April 2015 revealed heavy growth of P. aeruginosa (two morphotypes, i.e., PA-105A [spready] and PA-105B [round]) and scant growth of mixed Gram-positive organisms. The patient began vancomycin and cefepime therapy. Antimicrobial susceptibility testing (AST) results (Table 1) revealed nonsusceptibility of PA-105B to all routinely tested antipseudomonal β-lactams (6), but PA-105A and PA-105B were susceptible and intermediate to ceftolozane-tazobactam according to the FDA breakpoint (MICs of 1/4 μg/ml and 8/4 μg/ml, respectively) (7). All MICs were determined by ARUP Laboratories (Salt Lake City, UT) using broth microdilution (Table 1). Cefepime was then transitioned to ceftolozane-tazobactam (1.5 g intravenously [i.v.] every 8 h) and i.v. vancomycin treatment was continued. Therapy was planned to continue until surgery. Approximately 6 weeks later, the patient was taken to the OR for debridement and pectoral flap closure of the wound. OR cultures grew P. aeruginosa of two morphotypes (PA-147A [spready] and PA-147B [round]), methicillin-resistant Staphylococcus aureus (MRSA), and Candida tropicalis. PA-147A and PA-147B were resistant to ceftolozane-tazobactam (MICs of ≥32/4 μg/ml and ≥32/4 μg/ml, respectively). Ceftolozane-tazobactam therapy was discontinued, and treatment with imipenem-cilastatin (1 g i.v. every 8 h [extended infusion of 3 h]) and tobramycin (7 mg/kg i.v.) was started. Tobramycin therapy was stopped after 8 weeks, and the patient remained on vancomycin, imipenem, and micafungin therapy indefinitely. Subsequent P. aeruginosa isolates recovered in March 2016 were imipenem resistant and ceftolozane-tazobactam susceptible but were not available for further testing. The patient ultimately was transitioned to palliative care and died 8 months later, after several additional courses of antibiotics.

Multilocus sequence typing showed that all four P. aeruginosa isolates were the same sequence type (ST), ST-316 (https://pubmlst.org/paeruginosa) (8, 9). PCR and Sanger sequencing of the full-length β-lactamase genes (AmpC gene bla_PDC and bla_OXA-50, including the promoter region) were then performed. The results showed that an aspartic acid-to-glycine substitution at Ambler amino acid position 183 (G183D) was encoded by the AmpC gene (bla_PDC) of PA-147A and PA-147B, recovered after 42 days of ceftolozane-tazobactam treatment. In comparison with PA-105A and PA-105B, PA-147A and PA-147B were resistant to ceftolozane-tazobactam, with ≥3-fold increases in MICs (Table 1) (10). No mutations were found in the promoter region of the mutant bla_PDC (PA-147A and PA-147B), compared with the baseline bla_PDC (PA-105A and PA-105B), or in bla_OXA-50.

In order to examine whether the ceftolozane-tazobactam and ceftazidime-avibactam resistance in PA-147A and PA-147B was due to the G183D substitution encoded by the AmpC gene, we cloned the full-length bla_PDC gene from PA-105A and PA-147A, along with its native promoter region (using the primers PDC-F and PDC-R) (Table 2), into the pGlow vector (Invitrogen) in Escherichia coli TOP10. The resultant pGlow vectors carrying the wild-type gene and the gene encoding the G183D substitution were subsequently electroporated into PA-105A (yielding PA105A-WT and PA105A-MT, respectively). Further susceptibility testing of the PA-105A transconjugates showed that the mutant bla_PDC encoding the G183D variant increased the ceftolozane-tazobactam MIC ≥6-fold, compared to the corresponding wild-type bla_PDC-isogenic strain (Table 1), providing good evidence that the mutant bla_PDC is responsible for the ceftolozane-tazobactam resistance observed in the clinical isolates.

We also investigated, by quantitative reverse transcription-PCR (RT-qPCR), whether the ceftolozane-tazobactam resistance was associated with overexpression of the two β-lactamase genes. Expression of the bla_PDC and bla_OXA-50 genes in the four P. aeruginosa strains (PA-105A, PA-105B, PA-147A, and PA-147B) was tested using the primers listed in Table 2. RT-qPCR revealed no significant changes in bla_PDC expression between ceftolozane-tazobactam-susceptible and -resistant isolates from the same morphotypes (between PA-105A and PA-147A or between PA-105B and PA-147B); however, the bla_PDC expression levels in round isolates (PA-105B and PA-147B) were significantly higher (~150-fold greater than those in P. aeruginosa PAO1) than were those in spready isolates (PA-105A and PA-147A) (~1.2-fold greater than those in PAO1). We suspect that isolates of different morphotypes may have different genetic signatures involving bla_PDC expression regulation. However, the G183D substitution encoded by bla_PDC, instead of the high levels of bla_PDC expression, primarily contributes to the ceftolozane-tazobactam resistance. No differences in bla_OXA-50 expression were observed among these four isolates (PA-105A, PA-105B, PA-147A, and PA-147B).

Interestingly, compared with PA105A-WT, PA105A-MT had 4- to 8-fold MIC increases for aztreonam, cefepime, piperacillin-tazobactam, and ceftazidime-avibactam but a MIC decrease for imipenem (≥4-fold doubling dilution) (Table 1), suggesting that the G183D substitution encoded by bla_PDC also contributes to the changes in resistance to these β-lactam antibiotics in P. aeruginosa. Of note, similar increases in ceftazidime-avibactam resistance but restoration of carbapenem susceptibility due to point mutations in the bla_KPC-3 gene were described for Klebsiella pneumoniae (11). However, the molecular mechanisms underlying the multidrug resistance in PA-105B, which does not harbor the G183D substitution, remain unclear. Multiple mechanisms, including high levels of bla_PDC expression as well as other mechanisms (such as efflux or porins), may be involved. Further whole-genome sequencing and transcriptome analysis may help to decipher the molecular mechanisms of diverse resistance profiles between the parent and subsequent strains and between isolates of different morphotypes (Table 1).

Previous studies with in vitro selection and characterization of ceftolozane-tazobactam-resistant mutants among P. aeruginosa strains associated the development of high-level resistance with structural modifications in the conserved residues of AmpC (F147L, Q157R, G183D, E247K, or V356I) (4). Similar in vitro studies with ceftazidime-avibactam-resistant P. aeruginosa mutants also found the G183D substitution, which is less effectively inhibited by avibactam (12). However, in vivo development of ceftolozane-tazobactam resistance among clinical patients was only recently observed following 8 days of treatment; resistance was mediated by AmpC overexpression and was associated with substitutions within the AmpC Ω-loop (5). To our knowledge, the current case report is the first report of the clinical emergence of P. aeruginosa ceftolozane-tazobactam resistance mediated by the G183D substitution in AmpC. More importantly, we have proved that this substitution is the cause of the ceftolozane-tazobactam resistance. The development of resistance to ceftolozane-tazobactam occurred after several weeks of therapy. Notably, our patient received 1.5 g ceftolozane-tazobactam i.v. every 8 h, which is the dosing regimen approved for complicated urinary tract and intra-abdominal infections (7). A higher, pharmacokinetic/pharmacodynamic-derived, dose of 3 g i.v. every 8 h is currently being investigated in nosocomial pneumonia clinical trials (ClinicalTrials registration no. NCT02070757). Our finding of the emergence of coresistance to ceftazidime-avibactam is concerning. Further studies on the implications of this mutation for the susceptibility to other β-lactam antibiotics are warranted.

ACKNOWLEDGMENTS

REFERENCES