Molecular epidemiology of Acinetobacter baumannii and Acinetobacter nosocomialis in Germany over a 5-year period (2005–2009)

Authors


Corresponding author: H. Seifert, Institute for Medical Microbiology, Immunology, and Hygiene, University of Cologne, Goldenfelsstraße 19-21, 50935, Cologne, Germany.
E-mail: harald.seifert@uni-koeln.de

Abstract

To investigate the species distribution within the Acinetobacter calcoaceticus–Acinetobacter baumannii complex and the molecular epidemiology of A. baumannii and Acinetobacter nosocomialis, 376 Acinetobacter isolates were collected prospectively from hospitalized patients at 15 medical centres in Germany during three surveillance studies conducted over a 5-year period. Species identification was performed by molecular methods. Imipenem minimum inhibitory concentrations (MIC) were determined by broth microdilution. The prevalence of the most common carbapenemase-encoding genes was investigated by oxacillinase (OXA) -multiplex polymerase chain reaction (PCR). The molecular epidemiology was investigated by repetitive sequence-based PCR (rep-PCR; DiversiLab™). Acinetobacter pittii was the most prevalent Acinetobacter species (n = 193), followed by A. baumannii (n = 140), A. calcoaceticus (n = 10) and A. nosocomialis (n = 8). The majority of A. baumannii was represented by sporadic isolates (n = 70, 50%) that showed unique rep-PCR patterns, 25 isolates (18%) clustered with one or two other isolates, and only 45 isolates (32%) belonged to one of the previously described international clonal lineages. The most prevalent clonal lineage was international clone (IC) 2 (n = 34) and IC 1 (n = 6). According to CLSI, 25 A. baumannii isolates were non-susceptible to imipenem (MIC ≥ 8 mg/L), all of which produced an OXA-58-like or OXA-23-like carbapenemase. The rate of imipenem susceptibility among A. baumannii isolates decreased from 96% in 2005 to 76% in 2009. All other Acinetobacter isolates were susceptible to imipenem. The population structure of carbapenem-susceptible A. baumannii in Germany is highly diverse. Imipenem non-susceptibility was strongly associated with the clonal lineages IC 2 and IC 1. These data underscore the high clonality of carbapenem-resistant A. baumannii isolates.

Introduction

The genus Acinetobacter is composed of 25 distinct species with valid species names and a number of as yet unnamed species including both environmental and clinically relevant species [1]. Of these, Acinetobacter baumannii and Acinetobacter nosocomialis are the most important pathogens causing mainly nosocomial infections, especially in intensive-care units (ICUs) [1–3]. Because of their phenotypic and genotypic similarity, A. baumannii, Acinetobacter pittii (previously Acinetobacter genomic species 3), A. nosocomialis (previously Acinetobacter gen. sp. 13TU), and A. calcoaceticus were combined into the A. calcoaceticus–A. baumannii complex (Acb complex) [1,2]. From a clinical point of view this is problematic because A. calcoaceticus, in contrast to the other three species, is considered an environmental species that has not been implicated in clinical disease [1,2]. Furthermore, this similarity makes correct identification to species level difficult for routine microbiology laboratories [1,2].

Over the last decade the resistance of A. baumannii to the carbapenems as well as the increasing number of hospital outbreaks have become matters of concern [3–5]. Carbapenem resistance was first described in A. baumannii in the early 1990s [1]. It is most often mediated through carbapenem hydrolysing enzymes although permeability may also play a role. The most common carbapenemases in A. baumannii are class D oxacillinases (OXA) which encompass the intrinsic OXA-51-like, and the acquired OXA-23-like, OXA-58-like, OXA-40-like and OXA-143 enzymes [6,7]. Overexpression of OXA enzymes is often associated with insertion elements.

To date, studies on the molecular epidemiology of A. baumannii have identified eight worldwide clonal lineages (WW1–WW8) hereafter referred to as international clones (IC) 1–8 [8]. IC 1–IC 3 correspond to the European clones I–III [8]. IC 2 is the most widespread clonal lineage and it is often associated with carbapenem resistance [8–10].

The aim of this study was to investigate the species distribution among clinical Acinetobacter isolates and the molecular epidemiology of carbapenem-susceptible and non-susceptible A. baumannii and A. nosocomialis isolates recovered in Germany over a 5-year period (2005–2009).

Materials and Methods

Bacterial strains

In all, 388 Acinetobacter isolates were collected as part of the German Tigecycline Evaluation Surveillance Study (G-TEST) [11]. Fifteen microbiological diagnostic laboratories, most of them from large referral medical centres, provided prospectively collected clinical isolates from hospitalized patients with community-acquired or nosocomial infections during three study periods in 2005, 2007 and 2009. Each collecting centre was requested to provide ten A. baumannii isolates from different patients per study period.

Species identification

Species identification was initially performed using semi-automated identification systems at the different participating centres. Identification at the species level of isolates of the Acb complex was performed by gyrB multiplex PCR [12,13]. For isolates not belonging to the Acb complex, ARDRA and rpoB-sequencing were performed [14,15].

Molecular typing

The molecular epidemiology of A. baumannii and A. nosocomialis isolates was investigated using the DiversiLab™ system (bioMérieux, Nürtingen, Germany) [16]. Results were analysed with the DiversiLab software using the Kullback–Leibler statistical method and a cluster was defined as at least two isolates with a similarity index ≥95% [8]. Isolates having a similarity index of ≥99% were considered identical [17]. In addition, A. baumannii isolates were investigated by the sequence-type multiplex PCR as described by Turton et al. [4].

Antimicrobial susceptibility testing and detection of carbapenemase genes

Imipenem MICs were determined by broth microdilution according to the standard of the German DIN (Deutsches Institut für Normung) 58940 guidelines [11,18]. For the purpose of this study, CLSI breakpoints were used to interpret imipenem MICs [19]. Isolates with an imipenem MIC ≥8 mg/L were considered non-susceptible to imipenem. OXA-multiplex PCR was used to detect Acinetobacter carbapenemase-encoding genes as previously described [20,21].

Results

Species identification

The species distribution of isolates is summarized in Tables 1 and 2. Among 388 isolates submitted as A. baumannii by the collecting centres, 11 isolates were found not to belong to the genus Acinetobacter and one could not unambiguously be identified to species level. These isolates were excluded from further analysis. Among the remaining 376 Acinetobacter isolates, the most prevalent species was A. pittii (n = 193; 51%); A. baumannii accounted for 140 isolates (37%); ten isolates were A. calcoaceticus (3%); eight isolates were A. nosocomialis (2%); and 25 isolates were other Acinetobacter spp. (7%). Isolates were recovered from respiratory tract specimens (34%), wound swabs (29%), blood cultures (21%) and intra-abdominal specimens (8%). Of these isolates, 68% were from patients cared for on general wards and 32% were from ICU patients.

Table 1. Species distribution of Acinetobacter isolates (n = 376) from the German Tigecycline Evaluation Study (G-TEST) by study period
SpeciesNo. of isolates in each study periodTotal
200520072009
A. pittii 734971193
A. baumannii 465242140
A. calcoaceticus 32510
A. nosocomialis 5128
A. lwoffii 1416
A. bereziniae  145
A. haemolyticus 1 23
A. ursingii  213
Acinetobacter close to 13TU1  1
A. gyllenbergii 1  1
A. beijerinckii 1  1
A. johnsonii 1  1
A. radioresistens  1 1
A. junii  1 1
Acinetobacter gen. sp. 13BJ 1 1
Acinetobacter gen. sp. 14BJ  11
Total133114129376
Table 2. Species distribution of Acinetobacter isolates on intensive-care unit (ICU; n = 124) and non-ICU (n = 242) wards*
WardSpecies
A. baumannii A. pittii A. nosocomialis A. calcoaceticus Other Acinetobacter spp.
No. (percent) of isolates
  1. *No data are available on the origin for 10 Acinetobacter isolates and are excluded from this table.

ICU59 (48)54 (44)3 (2) 8 (6)
Non-ICU74 (31)136 (56)5 (2)10 (4)17 (7)

Although A. baumannii (48%) was the most prevalent Acinetobacter species recovered from ICU patients followed by A. pittii (44%), A. pittii was the predominant species obtained from patients in general wards (56%) whereas A. baumannii accounted for only 31% of those isolates (p < 0.005; Table 2). Of ten A. calcoaceticus isolates, five were obtained from the respiratory tract, four from wound swabs and one from urine from patients on general wards (internal medicine (three isolates), surgery (two isolates), haematology/oncology (two isolates), dermatology, orthopaedics and paediatrics, one isolate each) from seven different centres. Their true clinical significance, however, remains uncertain.

Molecular typing

The results of the repetitive sequence-based PCR (rep-PCR) analysis of 140 A. baumannii isolates, along with the number of centres where these isolates were recovered are shown in Table 3. Seventy A. baumannii isolates were considered sporadic isolates showing unique rep-PCR patterns. Twenty-five isolates clustered with one or two other strains and represented ten small clusters that did not correspond to the major international clonal lineages. Whereas four of the small clusters contained similar but not identical isolates, three of the small clusters consisted of two identical isolates each. Isolates from these clusters were obtained from the same centre and study period and therefore probably represented transmission from patient to patient. The remaining three small clusters included both similar and identical strains. Among these, two identical isolates obtained from one centre but in different years may indicate persistence or reintroduction of this strain.

Table 3. Molecular epidemiology of Acinetobacter baumannii as determined by repetitive sequence-based polymerase chain reaction
Molecular typing resultsYear of isolation
200520072009
No. of isolates (no. of centres) and percentage of isolates representing clustered and unclustered isolates
  1. IC, international clone.

Unclustered28 (14)61%23 (12)44%19 (12)45%
Part of a small cluster9 (4)20%10 (4)19%6 (6)14%
IC 26 (6)13%18 (6)35%10 (7)24%
IC 11 (1)2%  5 (1)12%
IC 42 (1)4%1 (1)2%1 (1)2%
IC 7    1 (1)2%
Total46 52 42 

Forty-five A. baumannii isolates (32%) clustered with one of the previously described international clonal lineages IC 1–8. The most prevalent clonal lineage was IC 2 (34 of 140 isolates; 24.3%). The high prevalence in 2007 was most probably a result of hospital outbreaks that occurred at five centres comprising up to five isolates. The IC 2 isolates were geographically widespread; strains were obtained from 12 different centres, i.e. from six centres in 2005 and 2007 and from seven centres in 2009. In addition, we identified six IC 1 isolates, four IC 4 isolates and one IC 7 isolate. In 2009, five IC 1 isolates were obtained from one centre within 3 weeks. Four of these isolates were identical as determined by rep-PCR.

For comparison, sequence-type multiplex PCR was performed for all A. baumannii isolates. Only two of 6 IC 1 (33.3%) isolates and 22 of 34 IC 2 isolates (64.7%) showed the expected electrophoresis pattern, indicating that this method although being a convenient first approach to molecular typing of A. baumannii in the outbreak setting needs to be complemented by other typing methods to reliably identify the major clonal lineages if A. baumannii isolates representing a more diverse epidemiological background are to be investigated.

The rep-PCR analysis of A. nosocomialis showed that six of eight isolates had unique banding patterns. Two isolates recovered in 2005 from one centre had similar (98.6%) but not identical banding patterns.

Antimicrobial susceptibility testing

Over the 5-year period, carbapenem non-susceptible A. baumannii increased from 4.3% in 2005, to 25% and 23.8% in 2007 and 2009, respectively. The proportion of isolates that were non-susceptible to imipenem was higher among isolates belonging to IC 2 and IC 1 than among other A. baumannii isolates. Among IC 2 (n = 34) and IC 1 (n = 6) isolates, 22 (55%) were non-susceptible to imipenem, whereas only three (3%) of the remaining A. baumannii isolates were non-susceptible to imipenem. All Acinetobacter isolates other than A. baumannii were susceptible to imipenem including isolates that harboured an acquired blaOXA gene (Table 4). Susceptibility data of other antimicrobial compounds have been published previously [11].

Table 4. Prevalence of acquired blaOXA genes in Acinetobacter spp. by study period. For A. baumannii, data of molecular typing by repetitive sequence-based polymerase chain reaction is included
SpeciesYear of isolation
200520072009
blaOXA-like genes detected (no. of isolates)
  1. IC, international clone.

  2. *OXA-23 is intrinsic to A. radioresistens

A. baumannii (n = 25)
UnclusteredOXA-58 (n = 2)  
IC 1  OXA-23 (n = 4)
IC 2 OXA-58 (n = 8)OXA-23 (n = 6)
  OXA-23 (n = 4) 
IC 4 OXA-58 (n = 1) 
A. pittii (n = 7)OXA-23 (n = 1)OXA-23 (n = 1)OXA-23 (n = 1)
 OXA-58 (n = 1) OXA-58 (n = 3)
A. calcoaceticus   OXA-40 (n = 1)
Acinetobacter gen. sp. 13BJ OXA-23 (n = 1) 
Acinetobacter gen. sp. 14BJ  OXA-58 (n = 1)
A. radioresistens*  OXA-23 (n = 1) 
A. junii  OXA-58 (n = 1) 

Prevalence of blaOXA genes

Table 4 summarizes the presence of acquired blaOXA genes in Acinetobacter species. All imipenem-non-susceptible A. baumannii were found to harbour blaOXA genes. In 2005, the only acquired blaOXA gene found in A. baumannii was blaOXA-58-like (n = 2). In 2007, four isolates harbouring blaOXA-23-like genes and nine isolates harbouring blaOXA-58-like genes were detected. Whereas the number of blaOXA-23-like genes increased in 2009 (n = 10), strains harbouring OXA-58-like were no longer found. Acquired blaOXA genes were also detected in 12 imipenem-susceptible Acinetobacter isolates other than A. baumannii, including seven A. pittii isolates.

Discussion

This is the first study providing representative data on the molecular epidemiology of A. baumannii in Germany, as it includes isolates collected prospectively during comparable time periods in 2005, 2007 and 2009 from different German medical centres.

The distribution of Acinetobacter species in countries other than Germany has been investigated in several studies showing very different results. In this study, A. pittii (51%) and A. baumannii (37%) were the two predominant Acinetobacter species. Acinetobacter pittii was also the most prevalent species among 114 Acinetobacter isolates recovered in an Irish hospital [22]. In contrast, A. nosocomialis (46.9%) was the most frequent species followed by A. pittii (19.5%) in a recent study of Acinetobacter bloodstream isolates from Norway [23]. In another study from the UK, A. baumannii (78%) was followed in frequency by Acinetobacter lwoffii (8.8%) and Acinetobacter ursingii (4%), whereas A. pittii made up only 1.7% of Acinetobacter isolates [24]. Wisplinghoff et al. [25] investigated 295 Acinetobacter isolates from patients with bloodstream infections from the USA. Of these isolates, 63% were identified as A. baumannii, 21% as A. nosocomialis and 8% as A. pittii. Of interest, clinical outcome was less favourable in patients with bloodstream infections caused by A. baumannii [25]. Chuang et al. [26] presented similar results from Taiwan, showing a predominance of A. baumannii in bacteraemia cases that were associated with a higher mortality and resistance to antibiotics.

Our results confirm previous work indicating the importance of using genetic methods for identification of Acinetobacter spp. [2]. In this study, of 376 Acinetobacter isolates identified as A. baumannii by the collecting centres, further testing revealed that 351 (90%) belonged to the Acb complex, of which 140 isolates were confirmed as A. baumannii and 193 isolates were A. pittii. This finding underscores the fact that the phenotypic identification methods based on semi-automated identification systems that are usually employed in large international surveillance studies are insufficient and may largely overestimate the prevalence of A. baumannii while underestimating the prevalence of other Acinetobacter species and their potential clinical relevance. Furthermore, these data suggest that carbapenem resistance in A. baumannii may have been underestimated in previous studies by inclusion of Acinetobacter isolates potentially misidentified as A. baumannii [26–28].

It may be speculated that multi-drug resistance and in particular carbapenem-resistance of A. baumannii could explain its success in the ICU environment where exposure to antibiotics is high and where other, more susceptible Acinetobacter species may not survive [2]. In fact, in our study 25% of A. baumannii isolates recovered from ICU patients were non-susceptible to imipenem whereas this was true for only 12% of isolates from general ward patients. On general wards where there is less antibiotic selection pressure, A. pittii might predominate over A. baumannii, in part because A. pittii is part of the human skin flora and can be introduced from the community [29]. Similarly, the prevalence of methicillin-resistant Staphylococcus aureus on ICUs is usually higher than on general wards [30].

The molecular epidemiology of A. baumannii in Germany confirms the previously described predominance of IC 2 among clinical A. baumannii isolates and the association between epidemic clonal lineages, in particular of IC 2, and non-susceptibility to carbapenems [9,10]. The majority of A. baumannii isolates, however, were sporadic isolates with unique rep-PCR patterns and isolates representing small clusters. Over the 5-year-period the percentage of isolates representing IC 1 and IC 2 increased from 15.2% to 35.7%, and of the 25 imipenem non-susceptible A. baumannii isolates, 18 isolates were IC 2 and four were IC 1. In addition, one IC 4 isolate and two sporadic isolates were non-susceptible to imipenem. All isolates had either OXA-58-like or OXA-23-like carbapenemases. Of note, we observed a small outbreak involving four identical isolates belonging to IC 1 at one centre in 2009, the isolates were non-susceptible to imipenem and harboured a blaOXA-23-like gene. The predominance of IC 1 and IC 2 among carbapenem-resistant A. baumannii isolates has been shown previously [9,10]. Importantly, an excellent correlation of rep-PCR and multilocus sequence typing to identify the major international clonal lineages was recently shown among A. baumannii bloodstream isolates [31].

The prevalence of blaOXA genes among A. baumannii changed over the 5-year period with a decrease of isolates harbouring blaOXA-58-like and an increase in isolates harbouring blaOXA-23-like genes. Our data support previous findings from the USA and worldwide indicating OXA-23 being the most frequent acquired carbapenemase found in A. baumannii [9,32]. The increasing prevalence of blaOXA-23-like genes is probably the result of the increasingly widespread presence of AbaR4 islands, which contain it [33].

Our study has several limitations. First, mainly isolates that were identified by semi-automated systems as A. baumannii or Acb complex were included, and isolates not belonging to the Acb complex are therefore underrepresented. Second, although A. pittii was the most prevalent species among our isolates, these isolates were not subjected to molecular typing. However, high heterogeneity among A. pittii bloodstream isolates was shown previously by plasmid profiling [34], and it was therefore considered unnecessary to repeat these analyses. Third, the sampling strategy employed (i.e. ten isolates per centre and year) and the epidemiological patient data available do not allow assessment of the role of nosocomial cross-transmission. Also, the impact of Acinetobacter hospital outbreaks in Germany, although they appear to be infrequent, could not be reliably explored.

In conclusion, this nationwide study showed a high prevalence of A. pittii both among patients hospitalized on general wards and in patients in the ICU. Our data also confirm that carbapenem resistance is mainly found in A. baumannii and may contribute to its increasing clonal spread, most notably that of IC 2. More accurate identification of Acinetobacter species in routine microbiology laboratories might clarify the true clinical impact and expansion of carbapenem-resistant A. baumannii clonal lineages in relation to other A. baumannii strains. Of note, we identified common blaOXA genes in Acinetobacter spp. other than A. baumannii, though these isolates were still susceptible to imipenem. Our data suggest that currently carbapenem-resistant Acinetobacter in Germany mainly involves A. baumannii. However, as carbapenem resistance has been described in other Acinetobacter species including A. nosocomialis and A. pittii [35,36], it is important to monitor carbapenem-resistance and the prevalence of resistance genes in these species.

Acknowledgements

The technical assistance of Danuta Stefanik is gratefully acknowledged.

Transparency Declaration

The contribution of PGH and HS was supported by a grant from Bundesministerium für Bildung und Forschung (BMBF), Germany, Klinische Forschergruppe Infektiologie (grant number 01KI0771). The G-TEST study was supported by Pfizer Pharma, Germany. The authors declare that they have no conflict of interest.

Ancillary