Non-duplicate GNB isolates were collected consecutively from eight Brazilian medical centers between January 1, 2016, and December 31, 2017. The participating medical centers were located in the Brazilian cities of Rio de Janeiro (one center), Salvador (one center), and São Paulo (six centers). Each participating medical center collected up to 100 consecutive GNB isolated from patients with intra-abdominal (IAI) and respiratory (RTI) infections, and 50 GNB from urinary-tract infections (UTI) per year. GNB were identified at the species level at the respective participant medical center and shipped to a central microbiology laboratory (International Health Management Associates, IHMA, Schaumburg, IL, USA), where confirmation of bacterial species, antimicrobial susceptibility testing, and molecular characterization of β-lactamase encoding genes were carried out. Bacterial identification at the species level was confirmed for all isolates using MALDI-TOF spectrometry (Bruker Daltonics, Billerica, MA, USA).

Antimicrobial susceptibility testing for amikacin, aztreonam, cefepime, cefotaxime, ceftazidime, ceftolozane-tazobactam, ceftriaxone, ciprofloxacin, colistin, ertapenem, imipenem, meropenem, and piperacillin-tazobactam was determined by testing customized MicroScan dehydrated broth microdilution panels (Siemens Medical Solutions Diagnostics, West Sacramento, CA, USA) according to the ISO 227-1 guidelines and interpreted following both CLSI7 and BrCAST/EUCAST8,9 guidelines. Quality control of broth microdilution panels followed the manufacturer’s and CLSI guidelines using the following ATCC strains: E. coli ATCC 25922, P. aeruginosa ATCC 27853, and K. pneumoniae ATCC 700603. Corresponding QC values tested were within the acceptable ranges as specified by CLSI. E. coli and Klebsiella pneumoniae isolates with MICs ≥2μg/mL for ceftazidime, ceftriaxone, or aztreonam were screened as “ESBL phenotype”. Enterobacteriaceae with MIC≥4μg/mL for imipenem and/or meropenem were defined as carbapenem resistant. P. aeruginosa isolates having MICs >8μg/mL and >2μg/mL, respectively, were classified as not susceptible (NS) to ceftazidime and meropenem.

All Enterobacteriaceae with MICs ≥4mg/L for ceftolozane-tazobactam and/or ≥1mg/L for ertapenem (except Proteeae Enterobacteriaceae) and/or ≥2mg/L for imipenem were selected for characterization of β-lactamase content as well as P. aeruginosa isolates displaying ceftolozane-tazobactam MICs ≥8mg/L and/or imipenem MICs ≥4mg/L. For comparison reasons, 50% of the E. coli and K. pneumoniae possessing the ESBL phenotype but showing susceptibility to ertapenem, imipenem, and ceftolozane-tazobactam were also selected for molecular characterization of β-lactamase encoding genes. Qualifying Enterobacteriaceae isolates were screened for the presence of β-lactamase genes (bla) encoding extended spectrum-β-lactamases (ESBLs; TEM, SHV, CTX-M, VEB, PER, GES), AmpC β-lactamases (ACC, ACT, CMY, DHA, FOX, MIR, MOX), and carbapenemases (KPC, OXA-48-like, IMP, VIM, NDM, SPM, and GIM) by multiplex PCR as described previously.8 Limited sequencing was performed on blaTEM and blaSHV to identify genes encoding TEM-type and SHV-type enzymes containing amino acid substitutions common to ESBLs (SHV A146V, G238S, G238A, E240K; TEM E104K, R164S, R164C, R164H, G238S). All detected genes, excluding blaSHV and blaTEM that did not encode ESBLs and the intrinsic, chromosomally coded blaAmpC of Citrobacter spp. (blaCMY-type) and Enterobacter spp. (blaACT-type and blaMIR-type), were amplified and sequenced in their entirety. Qualifying P. aeruginosa isolates were screened for the presence of bla encoding ESBLs (TEM, SHV, CTX-M, VEB, PER, GES), plasmid-encoded AmpC β-lactamases (ACC, ACT, CMY, DHA, FOX, MIR, MOX), and carbapenemases (KPC, OXA-24-like, IMP, VIM, NDM, SPM, and GIM) by multiplex PCR as described. All detected genes, as well as the chromosomally encoded Pseudomonas-derived cephalosporinase blaAmpC (PDC) common to the species, were amplified and sequenced in their entirety.

Antimicrobial susceptibility profiles of the five most relevant pathogens causing infections in the participating Brazilian medical centers is shown in Table 1. For comparison purposes the susceptibility and resistance rates according to the CLSI and EUCAST breakpoints are shown in Table 1; however, since BrCAST/EUCAST criteria have been recommended by the Brazilian Ministry of Health,10 the BrCAST/EUCAST susceptibility/resistance rates were considered when evaluating the antimicrobial activity of the respective antimicrobial agents, when these percentages varied.

Colistin (MIC50/90, ≤1/≤1μg/mL; 99.4% susceptible) and imipenem (MIC50/90, ≤0.5/≤0.5μg/mL; 99.4% susceptible) were the most in vitro active agents tested against the 494 E. coli isolates, followed by meropenem (MIC50/90, ≤0.12/≤0.12μg/mL; 99.0% susceptible), ertapenem (MIC50/90, ≤0.06/≤0.06μg/mL; 97.8% susceptible), and amikacin (MIC50/90, ≤4/8μg/mL; 97.4% susceptible). In contrast, the lowest susceptibility rates were observed for ciprofloxacin (MIC50/90, ≤0.25/>2μg/mL; 55.1% susceptible), followed by aztreonam (MIC50/90, ≤1/>16μg/mL; 71.5% susceptible), ceftriaxone (MIC50/90, ≤1/> 32μg/mL; 71.7% susceptible), cefepime (MIC50/90, ≤1/>32μg/mL, 73.3% susceptible), and ceftazidime (MIC50/90, ≤1/>16μg/mL, 78.1% susceptible). Ceftolozane-tazobactam (MIC50/90, 0.25/0.5μg/mL) remained highly active against E. coli isolates, 94.7% of which were susceptible to this agent, including 87.2% of 102 E. coli isolates exhibiting the ESBL phenotype as shown in Table 2.

Among the 431 K. pneumoniae evaluated in this study, the highest susceptibility rates were observed for amikacin (MIC50/90, ≤4/32μg/mL, 85.9% susceptible) and colistin (MIC50/90, ≤1/>4μg/mL, 85.4% susceptible). Ceftolozane-tazobactam (MIC50/90, ≤4/32μg/mL, 44.1% susceptible) as well as carbapenems (55.2-62.2% susceptible) showed poor activity against isolates of K. pneumoniae as displayed in Table 1. While 22.9% of the 144 K. pneumoniae exhibiting the ESBL phenotype remained susceptible to ceftolozane-tazobactam, 98.8% of the 168 K. pneumoniae non-susceptible to imipenem were resistant to this combination, as shown in Table 2. In fact, against these pathogens, only amikacin and colistin showed susceptibility rates superior to 70%.

A total of 81 E. cloacae were analyzed in this study. Among the extended cephalosporins, ceftolozane-tazobactam (MIC50/90, 2/16μg/mL, 46.9% susceptible) showed the highest susceptibility rate against this species followed by cefepime (MIC50/90, 4/>32μg/mL, 39.5% susceptible), ceftriaxone (MIC50/90, >32/>32μg/mL, 38.3% susceptible), and ceftazidime (MIC50/90, 32/>32μg/mL, 35.8% susceptible). In contrast, imipenem (MIC50/90, ≤0.5/1μg/mL, 98.8% susceptible), meropenem (MIC50/90, ≤0.12/0.25μg/mL, 98.8% susceptible), and amikacin (MIC50/90, ≤4/≤4μg/mL, 98.8% susceptible) were active against most E. cloacae (Table 1). Colistin (MIC50/90, ≤1/>2μg/mL, 95.1% susceptible) showed good in vitro activity against E. cloacae. Against the 52 E. cloacae isolates not susceptible to ceftazidime, only amikacin, imipenem, and meropenem retained good activity against such isolates with identical susceptibility rates (98.0%) followed by colistin (94.2% susceptible) as demonstrated in Table 2.

Ceftolozane-tazobactam (MIC50/90, 0.5/0.5μg/mL; 97.8% susceptible) and meropenem (MIC50/90, ≤0.12/≤0.12μg/mL; 98.9% susceptible) were the most active antimicrobials against the 91 P. mirabilis evaluated followed by piperacillin-tazobactam (MIC50/90, ≤2/≤2μg/mL; 96.7% susceptible), ertapenem (MIC50/90, ≤0.06/≤0.06μg/mL; 95.6% susceptible), and amikacin (MIC50/90, ≤4/8μg/mL; 95.6% susceptible). Although cefepime, ceftazidime, and ceftriaxone showed good in vitro activity with MIC50s ≤1μg/mL, 19.8%, 14.3%, and 22.0% of the isolates tested were resistant to these agents, respectively (Table 1). Ertapenem, imipenem, and meropenem were highly active against the 10 P. mirabilis isolates that exhibited the ESBL phenotype (Table 2). Ceftolozane-tazobactam (MIC50, 0.5μg/mL) and piperacillin-tazobactam (MIC50, ≤2μg/mL) also inhibited nine of these isolates (Table 2).

Ceftolozane-tazobactam was the most potent (MIC50/90, 1/4μg /mL) β-lactam agent tested against 265 P. aeruginosa, inhibiting 90.9% of isolates, while ceftazidime (MIC50/90, 4/>32μg/mL) imipenem (MIC50/90, 1/16μg /mL) and meropenem (MIC50/90, 1/>16μg/mL) inhibited only 66.8%, 70.2%, and 66.0% of these isolates, respectively, as shown in Table 1. Colistin (MIC50/90, ≤1/2μg/mL; 98.9% susceptible) and amikacin (MIC50/90, ≤4/16μg/mL; 87.9% susceptible) were active against P. aeruginosa. Ceftolozane-tazobactam retained moderate activity against P. aeruginosa isolates that were non susceptible to ceftazidime or imipenem exhibiting susceptibility rates of 73.9% and 79.8%, respectively. In contrast, colistin was highly active against isolates displaying both phenotypes, ceftazidime non-susceptible (MIC50/90, ≤1/2μg/mL; 96.6% susceptible) or imipenem non-susceptible (MIC50/90, ≤1/2μg/mL; 98.7%) as shown in Table 2.

According to the resistance profile, the molecular characterization of β-lactamase encoding genes was carried out in 433 isolates; i.e., in 36 of 81 E. cloacae (44.4%), 52 of 494 (10.5%) E. coli, 238 of 431 (55.2%) K. pneumoniae, 3 of 91 (3.3%) P. mirabilis, and 104 of 265 (39.2%) P. aeruginosa that fulfilled the study criteria specified in the material and methods section. The distribution of β-lactamase encoding genes according to bacterial species and medical center location is shown in Table 3. ESBL encoding genes were found in 37 of 52 selected E. coli. These isolates harbored a single blaCTX-M variant, with predominance of blaCTX-M-15 (13 isolates; 35.1%), followed by blaCTX-M-8 (11 isolates; 29.5%), blaCTX-M-2 (8 isolates; 21.6%), blaCTX-M-14 (3 isolates; 8.1%), and blaCTX-M-9 (2 isolates; 5.4%) as displayed in Table 3. While variants like blaCTX-M-14 andblaCTX-M-9 seemed to be restricted to medical centers located, respectively, in São Paulo and Rio de Janeiro, blaCTX-M-2,blaCTX-M-8, and blaCTX-M-15 were widely distributed. Only four E. coli isolates collected from two medical centers located in São Paulo and another located in Salvador (2 isolates) were shown to carry blaKPC-2. One of these isolates also carried blaCTX-M-8. Among the three P. mirabilis exhibiting the ESBL phenotype, blaCTX-M-2, blaCTX-M-8, and blaCTX-M-15 were identified in three distinct medical centers located in São Paulo and Rio de Janeiro. A single E. cloacae isolate harboring blaKPC-2 was collected in Salvador. In addition to blaKPC-2, this isolate also harbored blaCTX-M-15. No other carbapenemase encoding genes were identified among E cloacae isolates evaluated. In contrast, 11 of 36 (30.6%) E. cloacae isolates were shown to carry blaCTX-M-15.

Among 238 K. pneumoniae isolates that were eligible for molecular characterization, 165 (69.3%) were shown to carry blaKPC with almost all harboring blaKPC-2. Single isolates carrying blaKPC-3 or blaKPC-30 were collected from a single medical center located in São Paulo. OXA-48, IMP or VIM encoding genes were not detected among the studied isolates. In contrast, two isolates harboring blaNDM-1 were identified in the medical centers located in Salvador and Rio de Janeiro. While ESBL encoding genes, mainly belonging to the CTX-M family, were very frequently detected in K. pneumoniae isolates, plasmid-mediated AmpCs like blaCMY-2 and blaCMY-141 were very rare. The most frequently detected ESBL encoding genes were blaCTX-M-15, blaCTX-M-2, and blaCTX-M-14 as shown in Table 3. In many occasions, blaKPC2 was associated with blaCTX-M-15 (38 isolates), blaCTX-M-14 (33 isolates), and blaCTX-M-2 (32 isolates).

Among the 10 P. aeruginosa isolates exhibiting MICs ≥ 32μg/mL for ceftolozane-tazobactam, five isolates harboring MBL encoding genes were detected with blaSPM-1 (two isolates), blaIMP-1,blaIMP-74, and blaVIM-2 identified, respectively, in single isolates. One P. aeruginosa isolate harboring blaKPC-2 was found and shown to be resistant to ceftolozane-tazobactam, imipenem, and meropenem with MICs of 16, 32 and 16μg/mL, respectively. A high number of Pseudomonas Derived Cephalosporinase (PDC), also denominated AmpC, encoding genes was detected in our collection with PDC-35, -5, -16, -3, being the most frequently observed as shown in Table 3. In contrast, ESBL encoding genes were rarely found in P. aeruginosa, with four isolates from a single center located in São Paulo carrying blaCTX-M-2 and another isolated from a distinct medical center carrying blaGES-1.

The distribution of beta-lactamases according to the participating Brazilian medical center is shown in Fig. 3. The size of each rectangle is proportional to the frequency of each beta-lactamase encoding gene found in the respective medical center. Although K. pneumoniae isolates harboring KPC-2 were found in all medical centers, its frequency varied among the institutions. The same observation can be extrapolated for various PDC types in P. aeruginosa demonstrating the importance of local epidemiology.

Fig. 3.

Distribution frequency of beta-lactamase encoding genes according to participating Brazilian medical centers. The size and value of each rectangle is proportional to the frequency of each beta-lactamase encoding gene found in that respective medical center.

(0.28MB).