This was an observational retrospective study performed with 107 carbapenem-resistant A. baumannii isolates recovered from not-repeated patients with bacteraemia attending at Botucatu Medical School Hospital/UNESP (BMSH/UNESP), from 2008 to 2014. The study was approved as a retrospective study by the Local Research Ethics Committee (Process CAAE 49985115.5.0000.0059). We were granted an exemption from the requirement to obtain written informed-consent from the participants and/or their legal guardians because the isolates included in the study had already been stored, on an ongoing basis, in the Culture Collection of the Department of Microbiology and Immunology, UNESP, Botucatu, São Paulo, Brazil.

BMSH/UNESP is a 415-bed (52 intensive care unit beds) tertiary regional reference hospital, located in the inner of the State of Sao Paulo, Brazil. For this study, frozen isolates stocked in deep-freezer were recovered in Brain-Heart Infusion (BHI) broth and streaked onto BHI Agar plates. Acinetobacter isolates were initially identified by morphological and biochemical characteristics (Gram stain, oxidase-negative, catalase-positive, glucose oxidation, ability to grow at 42° and 44°C) (Vaneechoutte et al., 2015). A. baumannii species was screened by PCR detection of bla_{OXA−51−like} (Woodford et al., 2006) and gltA genes (Wong et al., 2014). A subset of randomly selected isolates was submitted to ITS and/or rpoB gene sequencing (Chang et al., 2005; La Scola et al., 2006). As MLST was carried out for each pulsotype (see below), A. baumannii identification was also confirmed by this method.

PCR for CHDL oxacillinases-encoding genes (bla_{OXA−23−like}, bla_{OXA−24/40−like}, bla_{OXA−143−like} and bla_{OXA−58−like}) and other carbapenemases (bla_KPC, bla_NDM, bla_SPM, bla_IMP, bla_VIM, bla_OXA−48) was performed in all isolates as previously described (Woodford et al., 2006; Higgins et al., 2010; Poirel et al., 2011). Full sequencing of the bla_OXA−24-like and the bla_{OXA−143−like} was carried out for allele determination (Héritier et al., 2005; Higgins et al., 2009; Cayô et al., 2014). Presence of the ISAba1 upstream the bla_OXA genes was also investigated by PCR mapping, employing the ISAba1 forward primer (Segal et al., 2005) and the bla_OXA reverse primers (Woodford et al., 2006).

Minimum inhibitory concentration (MIC) values were determined for ampicillin-sulbactam, minocycline, and tetracycline (broth microdilution method), imipenem, meropenem, tigecycline, and polymyxin B (E-test, BioMeriéux, Marci l'Etoile, France). The susceptibility profile of the isolates was completed by disc-diffusion method to amikacin, cefepime, ceftazidime, cefotaxime, ciprofloxacin, gentamicin, levofloxacin, piperacillin-tazobactam, trimethoprim-sulfamethoxazole, ticarcillin-clavulanate, and tobramycin. Breakpoints employed to define susceptibility, intermediate or resistance followed CLSI recommendations (Clinical Laboratory Standards Institute, 2015), except for tigecycline, for which the US Food and Drug Administration breakpoints were applied (susceptible: ≤ 2 μg ml⁻¹; resistant: ≥8 μg ml−1). Isolates were categorized as multidrug- (MDR, non-susceptibility to ≥1 agent in ≥3 antimicrobial categories among aminoglycosides, antipseudomonal carbapenems, antipseudomonal fluoroquinolones, antipseudomonal penicillins + β-lactamase inhibitors, extended-spectrum cephalosporins, folate pathway inhibitors, penicillins + β-lactamase inhibitors, polymyxins, tetracyclines) or extensively-drug-resistant (XDR, non-susceptibility to ≥1 agent in all but ≤ 2 categories, as described earlier) (Magiorakos et al., 2012).

Genetic diversity among all the 107 A. baumannii isolates were investigated by PFGE (Seifert et al., 2005). Macrorestriction was performed with ApaI (Promega) and DNA digested fragments were resolved using a CHEF-DR-III (Bio-Rad). Dendrogram was generated with BioNumerics v.7.6.2 (Applied Maths, Sint-Martens-Latem, Belgium) based on the Dice similarity using the UPGMA method, with tolerance and optimization parameters set at 1.5%. Clusters were defined as isolates with similarity ≥87% and named with capital letters (A to K) while pulsotypes were defined as each electrophoretic pattern with 100% similarity (named with capital letters and numbers, from A1 to K4).

MLST was performed in a representative isolate of each pulsotype as per the Institute Pasteur protocol (https://pubmlst.org/abaumannii/info/primers_Pasteur.shtml), in order to sequence the internal region of the genes gltA, fusA, recA, cpn60, pyrG, rplB and rpoB. PCR products were purified with enzyme ExoSAP-IT (Affymetrix) according to manufacturer's instructions. Nucleotide sequences were obtained using an Applied 3730 Automatic Sequencer (Applied Biosystems). The results were analyzed on BioNumerics v.7.6.2 (Applied Maths, Sint-Martens-Latem, Belgium), and compared with the Institute Pasteur database (https://pubmlst.org/abaumannii/); clonal complexes were determined by the eBurst algorithm (http://eburst.mlst.net/).

The 107 isolates were recovered from non-repetitive patients with bacteremia attending a teaching hospital in the State of São Paulo, Brazil, between 2008 and 2014. These isolates were recovered from blood (96.3%) or vascular catheter (3.7%).

All isolates were confirmed as A. baumannii species by PCR detection of bla_{OXA−51−like} and/or sequencing of the rpoB and ITS genes, as well as by the MLST analysis. Out of these isolates, 104 (97.2%) carried the bla_{OXA−23−like} with ISAba1 upstream; remaining strains carried bla_OXA−231 (bla_{OXA−143−like}; n = 2; 1.9%) or bla_OXA−72 (bla_{OXA−24−like}; n = 1; 0.9%). The bla_{OXA−58−like} gene was not detected, as well as the additional carbapenemases investigated.

According to the susceptibility test, 39.3% of isolates were considered MDR, and 60.7% XDR. The entire population evaluated confirmed resistance to imipenem, meropenem, ciprofloxacin, piperacillin-tazobactam and levofloxacin, while susceptibility to minocycline was observed in all the isolates. Resistance rates to other antimicrobials tested by disc-diffusion method were: amikacin (89.7%), cefepime (98.1%), ceftazidime (96.2%), cefotaxime (99.0%), gentamicin (76.6%), trimethoprim-sulfamethoxazole (68.2%), ticarcillin-clavulanate (99.0%), and tobramycin (68.2%). The MIC₅₀/MIC₉₀ (μg/ml) values were calculated for each antimicrobial agent, as follows: imipenem (>32/>32), meropenem (>32/>32), tetracycline (8/16), ampicillin-sulbactam (16:8/32:16), minocycline (0.25/0.5) and tigecycline (2/3). Although resistance to tigecycline was not detected, 38.3% of isolates presented intermediate susceptibility to this drug (according to FDA breakpoints).

Distribution of susceptibility rates among the clones (Table 1) evidenced high-rates of non-susceptibility to cephalosporins, carbapenems, quinolones and ticarcillin-clavulanic acid. For the aminoglycosides, the highest susceptibility rate to amikacin was detected for isolates belonging to CC1 (54.5%), while gentamycin and tobramycin were more active against isolates belonging to CC317 (60.7% and 53.6% of susceptibility, respectively). Furthermore, isolates belonging to CC317 presented the highest susceptibility rates to sulfamethoxazole-trimethoprim (92.9%), tigecycline (75%), and ampicillin-sulbactam (53.6%).

The 107 CRAB isolates were distributed into 53 PFGE pulsotypes belonging to eleven clusters (A, B, C, D, E, F, G, H, I, J, K) and eight additional pulsotypes with a single isolate (Figure 1). By extrapolating the results of MLST to isolates presenting the same pulsotypes, we observed the frequencies of 49.5% of strains belonging to clonal complexes (and sequence types) CC79 (ST79, 8.4%; ST175, 0.9% and ST730, 40.2%); 26.2% for CC317 (ST317, 100%), 10.3% for CC1 (ST1, 9.3%; ST986, 0.9%); 9.3% for CC15 (ST15, 9.3%), 1.9% for CC25 (ST25, 0.9%; ST945, 0.9%), 1.9% for CC107 (ST107, 100%), and 0.9% for CC22 (ST22 0.9%). With the exception of the PFGE cluster I, which presented the bla_OXA−231 allele, the bla_{OXA−23−like} gene associated with upstream ISAba1 was present in all clusters (Figure 1).

Distribution of clones over the time evidenced the emergence of A. baumannii pulsotypes belonging to clonal complex 79 in 2010, which became endemic in the institution until 2014 (Figure 2). The CC79 strains (n = 53) were distributed into 29 pulsotypes (Figure 1), recovered from 2010 to 2014. Although the pulsotype A1 was the most numerous (9 isolates), it was distributed in 2011 (4 isolates), 2013 (4 isolates) and 2014 (1 isolate). The remaining 28 pulsotypes were represented by 1, 2, or 3 isolates, each, and were recovered from the period comprised between 2010 over 2014. The 2014 isolates (the year with the largest number of isolates (13) belonging to CC79) were represented by 9 different pulsotypes with only 1 or 2 isolates in each electrophoretic pattern.

On the other hand, reduction in the frequency of CC317 was remarkable (Figure 2). The CC317 isolates (n = 28) were typed into 12 pulsotypes. A major clone (pulsotype E2) was identified comprising 12 isolates recovered from 2008 (6), 2009 (2), 2010 (2), 2011 (1), 2012 (1). Conversely, the remaining 11 pulsotypes were represented by 1, 2, or 3 isolates, each, recovered over the period of 2008–2013.

CC1 strains were detected in 2013 at frequency lower than 10% (2 isolates) but in 2014 this clonal complex represented more than 30% of the CRAB isolated from BSI (9 isolates belonging to 8 different pulsotypes), same year that CC25 strains were firstly detected.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.