The first nationwide multicenter study of Acinetobacter baumannii recovered in Serbia: emergence of OXA-72, OXA-23 and NDM-1-producing isolates

Background The worldwide emergence and clonal spread of carbapenem-resistant Acinetobacter baumannii (CRAB) is of great concern. The aim of this nationwide study was to investigate the prevalence of CRAB isolates in Serbia and to characterize underlying resistance mechanisms and their genetic relatedness. Methods Non-redundant clinical samples obtained from hospitalized patients throughout Serbia were included in the prospective, observational, multicenter study conducted from January to June 2018. Samples were initially screened for the presence of Acinetobacter baumannii-calcoaceticus (Acb) complex using conventional bacteriological techniques. Acb complexes recovered from clinical samples obtained from inpatients with confirmed bacterial infections were further evaluated for the presence of A. baumannii. Identification to the species level was done by the detection of the blaOXA-51 gene and rpoB gene sequence analysis. Susceptibility testing was done by disk diffusion and broth microdilution method. CRAB isolates were tested for the presence of acquired carbapenemases (blaOXA-24-like, blaOXA-23-like,blaOXA-58-like, blaOXA-143-like, blaIMP, blaVIM, blaGIM, blaSPM, blaSIM, blaNDM) by PCR. Clonal relatedness was assessed by pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST). Results Acb complex was isolated in 280 out of 2401 clinical samples (11.6%). Overall, A. baumannii was identified in 237 out of 280 Acb complex (84.6%). CRAB prevalence was found to be 93.7% (237/222). The MIC50/MIC90 for imipenem and meropenem were 8/> 32 μg/mL and 16/> 32 μg/mL, respectively. Although susceptibility was high for colistin (95.7%; n = 227) and tigecycline (75.1%; n = 178), ten isolates (4.3%) were classified as pandrug-resistant. The following carbapenemases-encoding genes were found: 98 (44.2%) blaOXA-24-like, 76 (34.5%) blaOXA-23-like, and 7 (3.2%) blaNDM-1. PFGE analysis revealed six different clusters. MLST analysis identified three STs: ST2 (n = 13), ST492 (n = 14), and ST636 (n = 10). Obtained results evaluated that circulating CRAB clones in Serbia were as follows: blaOXA66/blaOXA23/ST2 (32.4%), blaOXA66/blaOXA23/blaOXA72/ST2 (2.7%), blaOXA66/blaOXA72/ST492 (37.8%), and blaOXA66/blaOXA72/ST636 (27.1%). Conclusion This study revealed extremely high proportions of carbapenem resistance among A. baumannii clinical isolates due to the emergence of blaOXA-72, blaOXA-23, and blaNDM-1 genes among CRAB isolates in Serbia and their clonal propagation.


Background
In the recent years Acinetobacter baumannii has rapidly emerged as significant opportunistic pathogen and major cause of morbidity and mortality in the healthcare settings. The abilities of the organism to long-term survive in the hospital environment, rapidly develop antibiotic resistance and spread clonally potentiate the persistence and transmission of the species in healthcare settings. Consequently, A. baumannii is able to cause a spectrum of severe nosocomial infections such as ventilator-associated pneumonia, wound infections, bloodstream infections, meningitis, and urinary tract infections. The high risk populations are debilitating patients, those admitted to the intensive care units (ICUs), attached to indwelling foreign devices or mechanically ventilated [1,2].
A. baumannii has a numerous inherent and acquired resistance mechanisms which dramatically reduce the existing therapeutic options [3]. Consequently, A. baumannii is one of the ESKAPE organisms (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species), a group of clinically important pathogens that have the potential for substantial antimicrobial resistance [4]. Carbapenems used to be the mainstay for the treatment of nosocomial infections caused by A. baumannii [3]. However, the resistance to carbapenems has rapidly emerged and resulted in the worldwide dissemination of carbapenem-resistant A. baumannii (CRAB) strains, which usually exhibits a multidrug-resistant (MDR) phenotype, raising serious concerns about available treatment options [5,6]. In addition, CRAB is one of the critical-priority pathogens on the World Health Organization priority list of antibiotic-resistant bacteria for effective drug development [7]. Moreover, in critically ill patients in the ICUs infected with MDR A. baumannii, there is a substantial rise in patient mortality rates [8]. According to the most recent data from Centers for Disease Control and Prevention CRAB is listed as urgent threat in the United States with 8.500 cases and 700 deaths in hospital settings in 2017 [9].
There are numerous reports about the dissemination of CRAB strains, carriers of the OXA and MBL genes, from different geographical areas around the world [5,12,13]. Molecular typing of the CRAB strains from various European hospitals has shown the emergence of three successful clones originally named European clones I to III, which were renamed as international clones (ICs) I to III, after been identified worldwide [14]. In Serbia, there is no data on molecular epidemiology of CRAB in adult patients. So far, there is just one study, reporting the distribution of carbapenemase encoding genes among clinical isolates of A. baumannii from a single pediatric hospital [15]. However, there are several reports on OXA and MBL-positive CRAB strains, recovered from patients migrating from Serbia to Western Europe [16,17].
Therefore, the aims of this nationwide multicenter study were: i) to evaluate the prevalence and antimicrobial resistance patterns of A. baumannii circulating in Serbia, ii) to determine the prevalence of OXAs and MBLs among CRAB isolates, and iii) to assess their genetic relatedness and identify disseminated CRAB clones.

Study design and setting
The study was designed as a prospective, observational multicenter study. Serbia is Southeastern European country with roughly seven million inhabitants. It could be arbitrarily divided into three regions: Belgrade (the capital, two million), South Serbia (three million) and Vojvodina (two million). In the present study, regional microbiological laboratories from seven cities across Serbia evaluated 2401 randomly selected clinical specimens recovered from inpatients with confirmed bacterial infections admitted at nine hospitals throughout Serbia during the period January-June 2018. Microbiological laboratories from each region of the country evaluated the following number of hospital specimens: Vojvodina, n = 800 (General hospital Subotica, General hospital Pancevo, General hospital Sombor and Institute for pulmonary diseases of Vojvodina), Belgrade, n = 801 (University hospital medical center Bezanijska kosa, University hospital center dr Dragisa Misovic, Institute for cardiovascular diseases Dedinje), and Southern Serbia n = 800 (Clinical center Kragujevac, Clinical center Nis). All isolated Acinetobacter calcoaceticus-baumannii (Acb) complexes were sent to the coordinating laboratory at the Institute of Microbiology and Immunology of Medical Faculty University in Belgrade for further investigation. The inclusion criteria was the isolation of the Acb complex from non-redundant clinical samples (one per infected patient) obtained during the routine laboratory work. Infection was diagnosed using corresponding clinical and laboratory criteria. The exclusion criteria were detection of the non-Acb complex microbial agents as a possible source of infection and/or isolation of Acb complex without any clinical manifestations related to infection or without positive laboratory inflammatory markers. We addressed the potential bias of misdiagnosis of infectious disease by having the diagnostic criteria validated by an independent physician. This physician was blinded in terms of medical records.

Bacterial isolation and identification
Bacterial isolation and identification of Acb complex was done during the routine work in regional clinical laboratories by VITEK®2 system (bioMérieux, Marcy-l'Étoile, France) and conventional bacteriological techniques, such as colony appearance on sheep blood agar and MacConkey agar, Gram staining, motility test and biochemical characteristics.
All isolated Acb complexes together with the following data: type of specimen, ward of admission, patients' age, gender, clinical diagnosis, underlying conditions and other risk factors were sent to the coordinating laboratory for further investigation. All patients were deidentified in regional clinical laboratories and re-coded, before the bacterial strains were sent to the coordinating laboratory. The study was approved by the ethical committee of the Medical Faculty, University of Belgrade (1550/II-4).
Species identification was done in the coordinating laboratory. The DNA from an overnight culture on Columbia blood agar was extracted using a QIAamp DNA Mini Kit (QIAGEN GmbH, Hilden, Germany) according to the manufacturer's instructions. All isolates of Acb complex were initially screened for the presence of A. baumannii by PCR detection of the intrinsic bla OXA-51 gene, as previously reported [18]. The primers are shown in Table 1. The species identification was then confirmed by rpoB gene sequence analysis, as previously described [23]. All identified A. baumannii strains were stored at − 80°C until further analysis.

Detection of carbapenemase encoding genes and ISAba1 element
All CRAB isolates were screened for the presence of bla OXA-23-like , bla OXA-24-like , bla OXA-58-like , bla OXA-143-like , bla IMP , bla VIM , bla GIM , bla SPM , bla SIM and bla NDM with primers published elsewhere (Table 1) [19][20][21]. Multiplex PCRs were performed with separate groups of 4-6 primers and following the same thermal cycling conditions: initial denaturation at 94°C for 10 min and 36 cycles of amplification consisting of 30 s at 94°C, 40 s at 52°C, and 50 s at 72°C, with 5 min at 72°C for the final extension. Deionized distilled water was used as the negative control and isolates whose sequences were previously confirmed were used as the positive control. ISAba1 was detected performing uniplex PCR, as previously reported [22]. The primers are shown in Table 1.
Randomly selected OXA and all MBL amplicons of CRAB isolates were purified using the QIAquick PCR Purification Kit (QIAGEN GmbH, Hilden, Germany) following manufacturer's instructions and sequenced using conventional Sanger sequencing method. The obtained sequences were searched for homology with sequenced genes in the GenBank database using the National Center for Biotechnology Information's BLAST search program 2.7.0 for nucleotides (http://www.ncbi.nlm.nih. gov). All gene sequences (manually checked) and reference strain sequences from GenBank database were aligned using CLUSTAL W implemented in BioEdit ver. 7.1.3 software.

Molecular typing
Pulsed-field gel electrophoresis (PFGE) analysis PFGE included 60 CRAB isolates randomly selected from all participating hospitals, with respect to the specimen types and gene content. PFGE was performed with a 2015 Pulsafor unit (LKB Instruments, Broma, Sweden), as previously described [28]. DNA restriction was done with ApaI enzyme (Thermo Scientific, Lithuania). The Lambda Ladder 48.5-727.5 kb PFG Marker (New England Biolabs, US) was used as DNA size marker. The stained gels were scanned using the Diversity Database software image-capturing system (Bio-Rad Laboratories Ltd., UK). The Dice coefficient was used to calculate similarities of the banding patterns with a tolerance of 1.5%, and the unweighted-pair group method using average linkages (UPGMA) was used for cluster analysis with BioNumerics software, version 4.0 (Applied Maths, St-Martens-Latem, Belgium). The isolates with more than 80% similarity in their DNA patterns were defined as genetically related and part of the same cluster.

Statistical analysis
Statistical analysis was done using SPSS version 14.0 (Chicago, Illinois USA). Statistical significance was assessed using the χ 2 test or Fisher's exact test, as appropriate. A P ≤ 0.05 value was considered to be significant.

Bacterial isolates
During the study period, 2401 clinical samples were screened for inclusion (Fig. 1) 6%). Bacterial isolates were predominantly recovered from lower respiratory tract specimens (37.6%), wound exudates (36.7%) and blood samples (11.8%). Majority of the patients were elderly men (median age 66; within the range 14-87). Diabetes mellitus (23.2%) and surgical procedures (46.8%) were the most commonly reported underlying conditions. Detail information concerning specimen types, patient characteristics, underlying conditions, and risk factors are listed in Table 2.

Antimicrobial resistance
The antibiotic resistance patterns of 237 A. baumannii isolates are shown in Table 3. CRAB prevalence was   between the hospitals and regions regarding the proportion of isolates carrying different acquired OXA genes. However, bla NDM-1 positive isolates were detected only in two cities: Belgrade (the capital) and Nis (the city in Southern Serbia).

Molecular typing
The results of PFGE and MLST analysis of the tested CRABs, along with the information on the hospital locations, intrinsic and acquired OXA gene content are summarized in Fig. 3. The clonal relatedness of 60 CRAB isolates obtained from all participating hospitals was initially studied by PFGE. The similarity of the tested CRAB isolates ranged from 70 to 99%. Overall, PFGE analysis revealed six different clusters (A-F) and three singletons (Fig. 3). Largest clusters B, C and D which contained 5 (8.3%), 33 (55%), and 12 (20%) isolates, respectively, were comprised of CRAB isolates from Vojvodina, Belgrade and Southern Serbia. Isolates in cluster B and C were associated with the carriage of both bla OXA-24-like and bla OXA-23-like carbapenemases. Isolates of the remaining four clusters [cluster A (n = 3, 5%), cluster D (n = 12, 20%), cluster E (n = 2, 3.3%), and cluster F (n = 2, 3.3%)] were more homogenous and carried only bla OXA-24-like genes. Based on the PFGE clustering, OXA genes content, and hospital origin, out of 60 CRAB isolates analyzed by PFGE, 37 were selected for further characterization by MLST and OXA gene sequencing.

Discussion
A. baumannii poses a significant challenge to clinicians due to the increased incidence of hospital-acquired infections and its emergence of drug-resistance. This is the first Serbian nationwide multicenter study, providing data about the carbapenem resistance and clonality assessments of A. baumannii recovered from different hospitals throughout the country.
As expected, the vast majority of the Acb complexes tested in the present study was identified as A. baumannii (84.6%). Indeed, A. baumannii is the species of Acb complex most frequently involved in nosocomial outbreaks or sporadic infections [13]. Results of this study indicated that the prevalence of bloodstream infections caused by A. baumannii in Serbia (7.9%) is comparable to those found in some African and Asian countries (Morocco: 9.2%; India: 7.5%) [30,31]. More than onethird of the tested A. baumannii recovered from clinical specimens were of respiratory origin (37.6%). The respiratory tract has already been confirmed to be the most common isolation site of A. baumannii [13]. Furthermore, nearly 50% of the isolated A. baumannii in Mexican tertiary care hospital were of respiratory origin while a 42.6% were found in Chinese university hospital, within a 3-year study period [32,33]. Within our study, the second most common isolation site of A. baumannii was wound tissue (36.7%). Slightly lower isolation rates were recorded in China (burn wards: 25.2%; general surgery wards: 32.2%) [33] and in Mexico (23.2%) [32]. It is necessary to emphasize that the high rates of our A. baumannii isolated from wound exudates, point out the necessity of the strict following infection control measures, especially those focusing on the minimizing the transmission rate. In addition, various risk factors noticed in the present study, have already been significantly associated with infections caused by A. baumannii: mechanical ventilation, indwelling catheters, and admission to the ICUs [1]. In this study, the overall CRAB prevalence was higher than 90%, with IMP being a slightly more effective than MER (92.4% vs. 93.7%). This finding is similar to other reports of discordant susceptibilities to carbapenems, in which an isolate susceptible to imipenem but resistant to meropenem was reported [3,34]. Polymyxins and TYG are the only antimicrobial options for XDR Acinetobacter. Our results revealed that COL was the most active antibiotic against A. baumannii. However, the detection of 10 COL resistant isolates in this study is worrisome. Although still uncommon, A. baumannii resistance to COL has been reported from different geographic regions [35][36][37]. Currently, for PDR A. baumannii, limited therapeutic options exist and are usually reflected in combination therapies of two or more antibiotics [38]. Among the tested CRAB isolates, bla OXA-24 was the most frequent carbapenemase gene (44.2%), followed by bla OXA-23 (34.5%), and bla NDM-1 (3.2%). Furthermore, all sequenced bla OXA-24 were identified as bla OXA-72 , its single-amino-acid variant. In contrast to bla OXA-23 , which is the most common acquired carbapenemase in A. baumannii worldwide [5], bla OXA-72 producing isolates have been reported in only several countries, including Croatia, Lithuania, and Brazil [39][40][41]. Results obtained in this research are in concordance with the study conducted in one pediatric hospital in Serbia, where bla OXA-24 and bla OXA-23 were also the most prevalent acquired OXAs among pediatric CRAB isolates [15]. Furthermore, bla OXA-24 was also dominant in Bulgaria (46.7%) and Croatia (97.1%), while bla OXA-23 is globally distributed, as well as in some regional countries such as Greece (96.9%) [5,35,39,42]. However, contrary to our results, Novovic et al. reported bla OXA-58 in more than onethird of the tested CRAB isolated from children, indicating potential hospital dissemination of this clone [15]. Although, bla OXA-58 was also detected in all three aforementioned studies from Balkan region, its prevalence was less than 3% of CRABs [35,39,42]. Our results are in line with other studies Fig. 3 Dendrogram of the pulsed-field gel electrophoresis patterns of the A. baumannii strains isolated from nine regional hospitals in Serbia showing clusters (A-F); carriage of intrinsic (iOXA) and acquired bla OXA variants (aOXA); sequence type (ST) and corresponding international clone (IC), and hospital location showing international shift from bla OXA-58 towards bla OXA-23 or bla OXA-24 [32,[43][44][45][46].
The bla NDM-1 carrying A. baumannii has already emerged in Europe, Asia, Africa and South America [12,44,47,48]. Interestingly, two reports, from France and Germany, characterized bla NDM-1 positive A. baumannii isolates obtained from patients repatriated to these countries from Serbia [16,17]. In concordance with the above mentioned reports this study confirms that Serbia might be an endemic region for CRAB isolates carrying bla NDM-1 genes.
In the present evaluation, MLST results correlated well with PFGE clustering. The possibility that interhospital transmission of CRAB clones have occurred in Serbia is reinforced by the detection of the majority of the PFGE patterns almost uniformly distributed throughout the country. Nevertheless, with results obtained under the Pasteur MLST scheme few genetic lineages were shown to circulate in Serbia: ST636, as a singleton along with ICII representatives (ST2 and ST492), all carrying the bla OXA-66 allele of the bla OXA-51-like gene. Similar to our findings, numerous reports highlighted that ICII is widely distributed in all continents, and notably in the Mediterranean area [14,49]. Hence, various studies indicated the worldwide distribution of the A. baumannii ST2/ICII having bla OXA-23 gene [5]. Contrastingly to the aforementioned strain, the A. baumannii ST636 harboring bla OXA-72 gene was isolated sporadically, from patients in Sweden, Lebanon, and one Russian patient from Germany [50][51][52]. Furthermore, obtained results revealed that Serbia seems to remain the reservoir for the A. baumannii pertained to ST492/ICII, carrying bla OXA-72 and bla NDM-1 genes. To the best of our knowledge, besides this study, there is only one case report of infection caused by bla OXA-72 /bla NDM-1 producing A. baumannii, belonging to ICII/ST492, recovered from a urine sample of a Serbian patient hospitalized in France [16].
The limitation of the present study could be the small sample size of the randomly selected isolates for MLST and sequencing of the genes encoding carbapenemases. Nonetheless, despite these limitations, this nationwide research provides an important insight into the molecular epidemiology of the A. baumannii circulating in this region, especially comparing with the majority of the available European reports on A. baumannii which are focused on one health-care facility and particular hospital outbreak. Moreover, obtained results would serve as the base for the future studies and trend assessment of infections caused by Acinetobacter. Consequently, these data can be used to support regional initiatives to enhance infection control practices and antimicrobial stewardship, by facilitating available and inexpensive tools and resources to tackle this problem effectively and systematically.

Conclusion
In summary, this first nationwide study highlights the molecular epidemiology of the clinical isolates of A. baumannii. Alarming proportions of CRAB isolates are the consequences of the emergence of bla OXA-72 , bla OXA-23 , and bla NDM-1 among the tested isolates. Therefore, the extremely high antimicrobial resistance rates and the emergence of certain CRAB clones point out the critical need for the implementation of continuous surveillance of infections caused by A. baumannii and developing accurate prevention strategies in Serbia.