Clusters of imipenem-resistant Acinetobacter baumannii clones producing different carbapenemases in an intensive care unit


Corresponding author and reprint requests: A. Tsakris, Department of Microbiology, Medical School, University of Athens, Athens, Greece


During a 2-year period (April 2005–March 2007), 31 intensive care unit (ICU) patients in a Greek hospital were infected or colonised with imipenem-resistant isolates of Acinetobacter baumannii. Twelve patients died, with imipenem-resistant A. baumannii infection contributing to the death of seven patients. The 31 representative A. baumannii isolates were multidrug-resistant and clustered in four distinct clones, each of which contained different carbapenemase genes: clone I was predominant and contained blaVIM-1, blaOXA-58 and the intrinsic blaOXA-66 gene; clone II contained blaVIM-4, blaOXA-58 and the intrinsic blaOXA-69 gene; clone III contained blaOXA-58 and the intrinsic blaOXA-69 gene; and clone IV contained only the intrinsic blaOXA-66 gene. ISAba1 was not associated with the intrinsic blaOXA-51-like alleles, whereas ISAba3 was found upstream and downstream of blaOXA-58 in isolates of clone I, and upstream of blaOXA-58 in isolates of clone III, but was not detected in isolates of clone II. PCR, curing and hybridisation experiments indicated that the blaVIM alleles were chromosomally located, whereas the blaOXA-58 alleles were plasmid-located. This study provides the first description of the clonal spread of multidrug-resistant A. baumannii isolates carrying blaVIM-1 and blaVIM-4 metallo-β-lactamase genes, and revealed that distinct carbapenem-resistant A. baumannii clusters bearing different carbapenemase genes may emerge and cause severe infections, even in a well-defined regional hospital setting.


Acinetobacter baumanni is an opportunistic pathogen of increasing relevance in hospital infections during the last 15 years. The organism causes a wide range of infections, including pneumonia, sepsis, wound infection, urinary tract infection and post-neurosurgical meningitis, especially among critically-ill patients in intensive care units (ICUs) [1]. Extensive use of antimicrobial agents within hospitals has contributed to the emergence of multidrug-resistant A. baumannii strains that exhibit resistance to a wide range of antibiotics, including new-generation broad-spectrum β-lactams, aminoglycosides and fluoroquinolones [2].

Carbapenems have potent activity against acinetobacters and, until recently, were often used to treat infections caused by multiresistant A. baumannii isolates. However, acinetobacters may develop resistance to carbapenems by various mechanisms, including decreased permeability, overexpression of efflux pumps and production of carbapenemases [3]. In recent years, carbapenem resistance has been attributed increasingly to the production of carbapenemases, which may be class D carbapenem-hydrolysing oxacillinases or, less frequently, class B metallo-β-lactamases (MBLs) [3]. The class D carbapenemases cluster into four groups, represented by OXA-23, OXA-24, OXA-58 and the intrinsic OXA-51-like enzymes [3–7], while MBLs of the IMP, VIM and SIM types have also been recognised among A. baumannii isolates with reduced susceptibility to carbapenems [3,8–10].

Outbreaks of carbapenem-resistant A. baumannii strains have been documented in diverse geographical areas, including America, Europe and the Far East [2–5,11,12]. In Greek hospitals, imipenem-non-susceptible A. baumannii strains are isolated with increasing frequency from clinical sources and have been associated with outbreaks of infection in hospitals [13–15]. In our own laboratory, susceptibility data have revealed that a number of infections among hospitalised patients were caused by A. baumannii isolates that were resistant to carbapenems and almost all alternative antimicrobial agents. As it was evident that most isolates came from the medical–surgical ICU of the hospital, a 2-year study was initiated to study the characteristics of this outbreak and the molecular epidemiology of the carbapenem resistance.

Materials and methods

Setting, bacterial strains and surveillance procedures

The General Hospital of Serres is an acute-care regional hospital in northern Greece, with 400 beds and a medical–surgical ICU with a capacity of five patients, serving a population of c. 200 000 inhabitants. The present study included all imipenem-resistant A. baumannii isolates recovered consecutively from the clinical specimens of separate patients in the ICU of the hospital during the period April 2005 to March 2007. Isolates with an intermediate level of resistance to imipenem (MIC 8 mg/L) were also included. Only the first patient isolate (one isolate per patient) obtained from cultures taken >48 h after ICU admission were investigated. Clinical and epidemiological histories were taken from each patient colonised or infected with imipenem-resistant A. baumannii. Data requested included demographical details, reason for ICU admission, sources of isolation, colonised or infected sites, and outcome in relation to A. baumannii infection [16]. Nosocomial infections were defined according to standard CDC definitions [17].

Following initiation of the study, the hospital infection control team suggested restriction of carbapenem usage and the use of stringent antiseptic techniques. The latter included disinfection of the unit and the rigorous use of antiseptic solutions before and between patient and equipment contact, as well as before leaving the unit. In addition, environmental surveillance was instituted, focusing on environmental sites and equipment adjacent to the patients that should be free of contamination. The samples were collected using pre-moistened swabs that were then cultured on MacConkey agar plates; any imipenem-resistant A. baumannii isolates recovered from the environment were included in the study.

Identification and antimicrobial susceptibilities

Initial identification and susceptibility testing was performed using the Microscan system (Dade Behring Inc., West Sacramento, CA, USA). The isolates were also identified using the API  20NE system (bioMérieux, Marcy l’Étoile, France), with identification as A. baumannii confirmed by a biochemical identification scheme and the analysis of partial rpoB gene sequences [18].

Imipenem and meropenem MICs were determined by the agar dilution method using CLSI interpretative criteria [19]. Imipenem-resistant isolates were also tested for resistance to amikacin, ampicillin–sulbactam, aztreonam, ciprofloxacin, cefepime, ceftazidime, colistin sulphate, gentamicin, piperacillin–tazobactam, tigecycline and tobramycin using the agar dilution method according to CLSI guidelines [19] (when available). For tigecycline, the US Food and Drug Administration recommendation was used (i.e., susceptible, ≤2 mg/L; resistant, ≥8 mg/L). MBL production was tested phenotypically using MBL Etests (AB Biodisk, Solna, Sweden). Pseudomonas aeruginosa ATCC  27853 was used as a control in susceptibility tests, and a VIM-type carbapenemase-producing A. baumannii strain [10] was used as a control for the phenotypic MBL assay.

Pulsed-field gel electrophoresis (PFGE) and cluster analysis

PFGE of ApaI-digested genomic DNA of A. baumannii isolates was performed using a CHEF-DRIII system (Bio-Rad, Hemel Hempstead, UK) according to previously described methods [20], with the band patterns interpreted according to the criteria of Tenover et al. [21]. ApaI macrorestriction patterns were digitised and analysed using Quantity One Software (Bio-Rad Laboratories Inc., Hercules, CA, USA) for calculating Dice correlation coefficients and for cluster analysis using the unweighted pair-group method with arithmetic averages.

PCR assays and DNA sequencing

Carbapenemase-encoding genes (blaIMP, blaVIM, blaSIM, blaOXA-23-like, blaOXA-24-like, blaOXA-58-like) were sought by PCR using consensus primers that were specific for each enzyme group [9,10]. The isolates were also screened for intrinsic blaOXA-51-like carbapenemase genes using partially degenerate primers (sense, 5′-TGAACATTAAAICACTCTT; antisense, 5′-CTATAAAATACCTAATTGTT) that were designed to amplify an 825-bp product from all blaOXA-51-like alleles [14]. For integron mapping, PCR assays combining primers specific for the conserved 5′-CS and 3′-CS sequences with primers specific for blaVIM genes were performed as described previously [10]. For sequencing purposes, specific primers that amplify the entire blaVIM gene were used [22]. PCR amplicons were purified using ExoSAP-IT reagent (USB Corporation, Cleveland, OH, USA), and both strands were sequenced using the standard dideoxynucleotide method in an ABI Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA, USA). Sequence similarity searches were carried out with the BLAST program (

Insertion sequences ISAba1 and ISAba3, which have been associated previously with expression of blaOXA-51-like and blaOXA-58-like genes, respectively, were amplified as described previously [6,7]. The genetic elements carrying the blaOXA-51-like genes were investigated by PCR mapping using ISAba1 forward/OXA-51-like reverse and OXA-51-like forward/ISAba1 reverse primers, while the genetic elements carrying the blaOXA-58-like genes were investigated using ISAba3 forward/OXA-58 reverse and OXA-58 forward/ISAba3 reverse primers.

Plasmid analysis, curing and hybridisation experiments

Plasmid analysis was performed using a ChargeSwitch Plasmid ER Mini Kit (Invitrogen, Carlsbad, CA, USA). Chromosomal and plasmid bands were extracted separately from agarose 0.8% w/v gels using a PureLink Quick Gel Extraction Kit (Invitrogen); plasmid DNA was further purified using a ChargeSwitch Pro Plasmid Miniprep Kit (Invitrogen) to exclude chromosomal contamination. Eluted chromosomal and plasmid DNA extracts were subjected to PCRs specific for blaVIM and blaOXA-58 genes to check the location of each gene. Plasmid DNA extracts were also used as templates in PCRs specific for the 16S rDNA gene to confirm the exclusion of chromosomal fragments.

Plasmid curing experiments were performed using ethidium bromide at the maximum concentration (400 mg/L) that allowed the growth of isolates. The chromosomal location of blaVIM alleles was also demonstrated by Southern blotting of unsheared genomic DNA, followed by gene-specific hybridisation with a blaVIM digoxigenin-labelled probe [23].


During the study period, 47 patients, who had each been hospitalised for >2 days, became infected or colonised with A. baumannii, with imipenem-resistant A. baumannii isolates being recovered from one or more clinical samples submitted to the clinical laboratory from 31 of these patients. The characteristics of these 31 patients and the properties of their imipenem-resistant isolates are summarised in Table 1. The age of the patients ranged from 22 to 83 years (median, 68 years); 17 (54.8%) patients were male and 14 (45.2%) were female. The main reasons for admission to the ICU were pulmonary infection, cerebrovascular incident, poly-trauma, heart disease and post-surgical complications (Table 1). All patients were mechanically ventilated. Fifteen (48.4%) patients were infected and 16 (51.6%) were colonised, according to CDC definitions. The sites of infection were the bloodstream (seven patients), the respiratory tract (six patients), and a surgical wound (two patients).

Table 1.   Case histories and properties of the isolates for intensive care unit (ICU) patients colonised or infected with imipenem-resistant Acinetobacter baumannii
No.Date of isolation (month/year)Age (years)/ genderReason for ICU admissionIsolate sourceStatus/ infectionOutcomePFGE typeAntibiotic MIC (mg/L)a
  1. aAll isolates were resistant to ceftazidime, cefepime, ciprofloxacin, piperacillin–tazobactam, amikacin, gentamicin, netilmicin and tobramycin.

  2. bDeath was not related to A. baumannii infection.

  3. AZT, aztreonam; COL, colistin; IMP, imipenem; MEM, meropenem; SAM, ampicillin–sulbactam; TGC, tigecycline; M, male; F, female; PFGE, pulsed-field gel electrophoresis; IV, intravenous.

 204/0564/MSeptic shockBronchial aspirate/bloodSepsisDiedIa1616>256160.251
 304/0577/MPoly-trauma/pneumoniaBronchial aspirateColonisedDischargedIa168>2561610.25
 404/0567/FSeptic shock/pneumoniaBronchial aspiratePneumoniaDischargedIa1616>2561620.5
 505/0568/FAbdominal surgery/pneumoniaBronchial aspirate/bloodSepsisDiedIa3216>256160.50.5
 605/0572/MStroke/respiratory infectionUrineColonisedDischargedII168163280.5
 805/0572/MCerebral haemorrhage/pneumoniaBronchial aspirateColonisedDischargedIa1616>2561611
 909/0558/FStroke/nephrotic syndromeBronchial aspirateColonisedDiedbIV88>2564320.25
1009/0565/MPoly-traumaBronchial aspirateColonisedDischargedIV816>2561640.12
1109/0568/FHeart failure/pneumoniaBronchial aspiratePneumoniaDiedIV3232>2566421
1310/0568/FPneumoniaBronchial aspiratePneumoniaDeathIb3232>256>1280.50.5
1404/0668/FStrokeBronchial aspirateColonisedDischargedIb6432>2563211
1504/0664/MStroke/pneumoniaBronchial aspirateColonisedDischargedIb6464>2563211
1607/0675/FSeptic shockBronchial aspiratePneumoniaDiedIa3232>2561680.5
1807/0675/MCerebral haemorrhage/ pneumoniaBlood/bronchial aspirateSepsisDiedII8416480.5
1909/0665/FStroke/pneumoniaBronchial aspirateColonisedDischargedIb3232>2563210.5
2010/0683/MFracture/respiratory infectionBloodSepsisDiedIV3232>2561621
2112/0667/MRespiratory failureBronchial aspirateColonisedDischargedIb3232>25612820.5
2212/0665/MAbdominal surgeryWoundWound infectionDiedbIa1616>256320.50.5
2312/0642/FPoly-traumaBronchial aspirateColonisedDischargedIII168323210.5
2412/0670/MIntestinal bleeding/febrilityBloodSepsisDischargedIb3216>2561610.5
2501/0724/MPoly-traumaBronchial aspirateColonisedDischargedΙΙ1681281621
2602/0763/FRespiratory failureBronchial aspirateColonisedDischargedΙa3216>2561611
2702/0778/MPancreatitis/respiratory failureBronchial aspirateColonisedDischargedΙΙ161664160.50.5
2802/0776/MSeptic shockBronchial aspiratePneumoniaDiedbΙa168>256810.5
2903/0771/FHeart failure/cerebral haemorrhageBloodBacteraemiaDiedbΙa168>256810.5
3003/0772/FAbdominal surgery/pneumoniaIV catheter/bronchial aspiratePneumoniaDischargedΙV168>2561620.5
3103/0758/MPoly-traumaBronchial aspirate/woundWound infectionDiedbΙa1616>256160.50.5

For infections caused by ampicillin–sulbactam-susceptible isolates, the patients received intravenous colistin in combination with ampicillin–sulbactam for bloodstream and wound infections, while nebulised colistin in combination with intravenous ampicillin–sulbactam was administered for respiratory tract infections. In cases where infections were caused by ampicillin–sulbactam-resistant isolates, ampicillin–sulbactam was replaced with high intravenous doses of imipenem. Twelve of the 31 patients died, with A. baumannii infection contributing to death in seven cases (Table 1). Analysis of 45 environmental samples obtained during the study period yielded three imipenem-resistant A. baumannii isolates; one from the floor of the ICU, one from the bed of an ICU patient, and one from the external site of the ventilator tube of an infected patient.

Imipenem and meropenem MICs for the 31 clinical isolates ranged between 4 and 64 mg/L (Table 1). The isolates were multidrug-resistant, exhibiting resistance to ciprofloxacin, aminoglycosides and β-lactams, but showed some susceptibility to tigecycline, ampicillin–sulbactam and aztreonam, for which 27, 21 and three isolates, respectively, were defined as susceptible or intermediate. All isolates were susceptible to colistin.

PFGE of the imipenem-resistant clinical isolates revealed four distinct clonal types (Table 1). Nineteen isolates belonged to a predominant clone (type I), two clones (types II–IV) contained five isolates, and one clone (type III) contained two isolates. The 19 isolates in the major clone were divided into two clonal subtypes (Ia and Ib; 83% similar), which contained 12 and seven isolates, respectively. Two of the three environmental isolates belonged to subtype Ia, while the third belonged to type II.

Following initiation of the prospective study in April 2005, the reduction in usage of carbapenems, combined with the other infection control measures that were instituted, resulted in an absence of further infections caused by imipenem-resistant A. baumannii in June–August 2005 and in November 2005 to March 2006. However, carbapenem usage subsequently returned almost to previous levels, and this may have contributed to the reappearance in the ICU of the predominant clone I and the three other clones (Fig. 1).

Figure 1.

 Occurrence of patients infected or colonised with different clones of imipenem-resistant Acinetobacter baumannii in the medical–surgical intensive care unit between April 2005 and March 2007.

MBL Etests were positive for 21 of the 31 imipenem-resistant clinical isolates; however, PCRs revealed that 24 isolates were positive for a blaVIM gene (Table 2). All isolates were negative for the remaining MBL genes (blaIMP and blaSIM), as well as for blaOXA-23-like and blaOXA-24-like genes. A blaOXA-58-like gene was amplified from 26 isolates, while an intrinsic blaOXA-51-like gene was amplified from all 31 clinical isolates. Sequencing of blaVIM amplicons identified a blaVIM-1 allele in 19 isolates (all were MBL-positive according to Etests) and a blaVIM-4 allele in five isolates (two were positive and three were negative according to MBL Etests). In all cases, the blaVIM alleles were inserted in class 1 integron variable structures (data not shown).

Table 2.   Results of molecular analysis of 31 clinical and three environmental imipenem-resistant isolates of Acinetobacter baumannii
PFGE typeNo. of clinical isolatesNo. of environmental isolatesPCR
ISAba1 unrelatedblaOXA-51-likeISAba3 upstreamISAba3 downstreamblaOXA-58-likeblaOXA-23-likeblaOXA-24-likeblaVIM
  1. PFGE, pulsed-field gel electrophoresis.


Sequencing of the blaOXA-51-like amplicons revealed blaOXA-66 (24 isolates) and blaOXA-69 (seven isolates), while all 26 blaOXA-58-like alleles were the classic blaOXA-58 allele. It was of interest that isolates from each one of the four clones harboured a different pattern of carbapenemase genes: isolates of clone I contained blaVIM-1, blaOXA-58 and the intrinsic blaOXA-66 gene; isolates of clone II contained blaVIM-4, blaOXA-58 and the intrinsic blaOXA-69 gene; isolates of clone III contained blaOXA-58 and the intrinsic blaOXA-69 gene; and isolates of clone IV contained only the intrinsic blaOXA-66 gene (Table 2). The three environmental isolates contained the carbapenemases that were detected in clinical isolates of clone I (two isolates) and clone II (one isolate).

Three imipenem-susceptible (MIC ≤4 mg/L) A. baumannii isolates obtained during the study period from other patients hospitalised in the ICU were negative for all carbapenemase genes except the intrinsic blaOXA-51-like gene. These three isolates belonged to two clonal types that were distinct from those detected among imipenem-resistant isolates (data not shown).

PCR for the insertion sequence ISAba1 was positive for all 34 imipenem-resistant isolates, but PCR mapping revealed that it was not associated with the blaOXA-51-like alleles. PCR for the insertion sequence ISAba3 was positive only for imipenem-resistant isolates of clones I and III. ISAba3 was found upstream and downstream of the blaOXA-58 gene in blaVIM-1- and blaOXA-58-producing isolates (clone I), but only upstream of the blaOXA-58 gene in isolates belonging to clone III (Table 2).

Three isolates, representative of each of the three clonal types that produced blaVIM and/or blaOXA-58 genes, were analysed for their plasmid content; all three isolates carried a large plasmid of 45–70 MDa, but small plasmid bands were also detected (data not shown). The large plasmids and the chromosomal DNA band of each isolate were extracted from the agarose gel and used as templates in PCRs for the blaVIM and/or blaOXA-58 genes. The results suggested that the blaVIM alleles were located on the chromosome, whereas the blaOXA-58 alleles were located on the plasmid DNA. Consistent with these findings, the blaOXA-58 gene was cured from three isolates harbouring blaOXA-58, while hybridisation of intact genomic DNA with the blaVIM probe indicated a chromosomal location for blaVIM in two isolates.


The impact of imipenem-resistant A. baumannii on ICU-acquired infection may be substantial. The occurrence of a predominant multiresistant A. baumannii antibiotype in a well-defined clinical setting may lead to the assumption that an outbreak is caused by a single disseminated clone [2,5,11,13]. However, the present study revealed the emergence of four imipenem-resistant A. baumannii clones producing different carbapenemases in the ICU of a Greek regional hospital. During the first period of the investigation (April–May 2005), all the isolates that infected or colonised the patients belonged to clonal type I, except for two isolates that belonged to clonal types II and III, respectively. After reinforcement of the barrier methods and cleaning protocols in April 2005, no patients infected or colonised with imipenem-resistant isolates were detected for the next 3 months (June–August 2005). However, a new clone (type IV) was introduced and disseminated in September 2005, followed in October 2005 by the reappearance of the predominant clones I and II. All clonal types also reappeared after the second period (November 2005 to March 2006), in which imipenem-resistant isolates were not recovered from patients’ clinical samples. It is also of interest that the infection control measures were reinforced in the first trimester of 2007, which resulted in an absence of imipenem-resistant A. baumannii infections after the study period ended in March 2007 (data not shown).

Carbapenemase-producing strains of A. baumannii have been involved in outbreaks in Europe, Asia and both North and South America [2,3]. The carbapenemases produced by such strains are mostly class D oxacillinases [3–5,13,14], although metallo-enzymes of the IMP, VIM and SIM families have also been detected in some well-defined geographical regions [3,12]. Among the latter enzymes, the IMP-1 and VIM-2 types have been associated with clusters of carbapenem-resistant A. baumannii isolates [8,24,25], and the sporadic production of VIM-1 in clonally distinct A. baumannii isolates from Greece has been reported previously [10]. The present study is the first to describe the clonal dissemination of VIM-1-producing A. baumannii isolates in a hospital environment. The isolates carrying blaVIM-1 also had an MBL gene and 59-bp elements that were identical to those described originally in P. aeruginosa [26], and subsequently among other Gram-negative bacteria in Greece [27]. This apparent spread could have been facilitated by the extensive use of carbapenems against the multidrug-resistant Gram-negative bacteria that have been isolated frequently from patients in the ICU studied. In addition, the detection of common clones among environmental and clinical imipenem-resistant isolates of A. baumannii suggests that environmental contamination might have contributed to the difficulty in restricting the spread of these organisms in the unit. It is possible that some environmental reservoirs may have remained undetected using the swab technique [28]. Nevertheless, environmental surveillance and strict antiseptic techniques may have contributed to the reduced spread of these bacteria for periods during and after the end of the survey.

Three additional imipenem-resistant clones were detected during this prospective study, one of which carried blaVIM-4, which is a point mutant of blaVIM-1 and was first identified in a carbapenem-resistant clinical isolate of P. aeruginosa in Greece [29]. This variant has now been detected in several P. aeruginosa clones in Greece [30], as well as among pseudomonads and enterobacteria in several other European countries, Tunisia and Australia [31–34]. However, to our knowledge, the present study is the first report of A. baumannii isolates carrying blaVIM-4. Nevertheless, these clonal isolates exhibited relatively low levels of resistance to carbapenems, perhaps attributable to the fact that the companion blaOXA-58 gene was not associated with ISAba3 promoter sequences. It has been shown previously that OXA-58 has weak carbapenemase activity and plays a role in carbapenem resistance in A. baumannii only when blaOXA-58 is up-regulated and highly expressed [6].

Much of the resistance in the isolates studied could also depend on impermeability or other combined mechanisms that simultaneously affect resistance to other antibiotics, e.g., expanded-spectrum cephalosporinases [3]. Thus, no acquired carbapenemases were detected in five clinical isolates that exhibited resistance to carbapenems, and expression of the intrinsic blaOXA-51-like alleles in these isolates was not activated by the presence of ISAba1 upstream of the carbapenemase gene [7]. The incidence and spread of carbapenem-resistant clones in this ICU, which was associated with a fatal outcome in several cases, remains a concern and merits continuation of the surveillance programme and the infection control measures.


The authors declare that they have no conflicting interests in relation to this research.