The Acinetobacter calcoaceticus–Acinetobacter baumannii (Acb) species complex comprises four species. A. calcoaceticus is clinically unimportant whereas A. baumannii is a well-established pathogen. The significance of Acinetobacter genomic species (gen. sp.) 3 and Acinetobacter gen. sp. 13TU is uncertain because identification to species level is not routine. A recent paper has proposed the names ‘Acinetobacter pittii’ and ‘Acinetobacter nosocomialis’ for these respective species [Reference Nemec1] and these names will be validly published by citation in Validation list 140 of the July 2011 issue of the International Journal of Systematic and Evolutionary Microbiology (J. Euzeby, personal communication), and are used here.

This study was performed to determine the prevalence, distribution of specimen sources, and antimicrobial susceptibility of the Acb species complex in Singapore. One hundred and ninety-three non-replicate Acb species complex clinical isolates were collected from six hospitals over a 1-month period in 2006. Identification of A. baumannii was carried out by a one-tube multiplex PCR [Reference Chen2]. Intergenic spacer (ITS) sequencing and genomic fingerprint analysis based on selective amplification of restriction fragments (AFLP™) were used for identification of other members of the Acb species complex [Reference Chang3, Reference van4]. AFLP analysis was also used to type nine isolates of carbapenem-resistant A. baumannii in the present study, representing clusters defined by RAPD–PCR using the primers DAF4 and M13 (data not shown), and six archived isolates from Hospital S that had previously been characterized [Reference Koh5]. AFLP profiles generated were compared with each other and to a library of >2000 reference strains of all Acinetobacter spp. including taxonomically and epidemiologically defined strains. Isolates were identified as the same species, European clone or strain based on percentage similarities of ⩾50%, ⩾80% or ⩾90%, respectively. Multilocus sequence typing (MLST) using the Institut Pasteur scheme was performed on five isolates representative of the main AFLP defined clusters at 80% similarity (http://www.pasteur.fr/recherche/genopole/PF8/mlst/Abaumannii.html).

Minimal inhibitory concentrations (MICs) of sulbactam-ampicillin (SAM), piperacillin-tazobactam (TZP), ceftazidime (CAZ), cefepime (FEP), imipenem (IPM), meropenem (MEM), amikacin (AMK), gentamicin (GEN), ciprofloxacin (CIP), polymixin B (POL), and tigecycline (TGC) were determined by microbroth dilution using custom Sensititre plates (Trek Diagnostic Systems Ltd, UK). Antimicrobial susceptibilities were interpreted in accordance with the guidelines of the Clinical Laboratory Standards Institute [6], except for tigecycline where the manufacturer's breakpoint for Enterobacteriaceae was used.

Genes encoding bla _OXA-23-like, bla _OXA-24-like, bla _OXA-51-like, bla _OXA-58-like, and bla _OXA-143 carbapenemases were detected by multiplex PCR [Reference Higgins, Lehmann and Seifert7]. The presence of insertion sequences preceding the bla _OXA genes in carbapenem-resistant (meropenem or imipenem MIC ⩾8 mg/l) isolates was detected using the forward primers ISAba1B, ISAba2A, ISAba3C, and ISAba4B in combination with reverse primers for bla _OXA genes [Reference Poirel and Nordmann8, Reference Corvec9]. Metallo-β-lactamase genes were sought using a multiplex method [Reference Ellington10].

One hundred and fifty-two (78·7%) isolates were identified as A. baumannii, 18 (9·3%) as A. pittii, and 23 (11·9%) as A. nosocomialis. The distribution of isolates according to the species and type of specimen showed that most (63·2% A. baumannii, 55·6% A. pittii, 60·9% A. nosocomialis) were recovered from respiratory and wound specimens and the respective proportions from blood for these species were 7·2%, 16·7% and 26·1%; A. baumannii and A. pittii were of similar frequency from urine specimens (25·7% and 27·8%, respectively).

The antimicrobial resistance profiles of the different species are shown in Figure 1. A high proportion of A. baumannii (110 isolates, 72·4%), but also five (27·8%) A. pittii and eight (34·8%) A. nosocomialis isolates were resistant to carbapenems. Overall, 150 (77·7%) isolates were multidrug resistant, defined as resistant to three or more antimicrobial agents. This comprised 127 (83·6%) of A. baumannii isolates, 11 (61·1%) of A. pittii, and 12 (52·2%) of A. nosocomialis.

Fig. 1.

Antimicrobial resistance profiles of A. baumannii (▪), A. pittii (), and A. nosocomialis (□). For abbreviations of antimicrobials see main text.

One hundred and sixteen isolates (108 A. baumannii, eight A. nosocomialis) were positive for bla _OXA-23-like. Of the isolates that were resistant to imipenem, ISAba1 was located upstream of this OXA gene (ISAba1-bla _OXA-23-like) in 70 A. baumannii and seven A. nosocomialis. Only two imipenem-susceptible A. baumannii had bla _OXA-23-like. In both cases, there was no IS element upstream of the bla _OXA-23-like gene.

All A. baumanii isolates and one A. nosocomialis were positive for the bla _OXA-51-like gene. Of the imipenem-resistant A. baumannii, ISAba1 was upstream of the OXA-51-like gene in only 12 isolates (ISAbaI-bla _OXA-51-like) and in only three of these was ISAbaI-bla _OXA-51-like likely to be the major contributor to imipenem resistance as the remainder also possessed ISAbaI-bla _OXA-23-like concurrently. This is in contrast to Taiwan where A. baumannii carbapenem resistance was mostly associated with ISAbaI-bla _OXA-51-like [Reference Lee11]. It has been suggested that the presence of bla _OXA-51-like genes can serve to identify A. baumannii [Reference Turton12]. The presence of bla _OXA-51-like in A. nosocomialis in this study, and as recently described in Taiwan [Reference Lee13], suggests that this may not be a sufficiently specific marker.

Thirteen isolates were positive for bla _OXA-58-like (one A. baumannii, four A. nosocomialis, eight A. pittii). Of the imipenem-resistant isolates, this gene was preceded by ISAba3 (ISAba3-bla _OXA-58-like) in an isolate of A. nosocomialis and three A. pittii isolates. Only one imipenem-resistant A. baumannii isolate (positive for bla _OXA-51-like and bla _OXA-23-like. both not with ISAbaI located upstream) and one imipenem-resistant A. pittii (positive for ISAba3-bla _OXA-58-like) had the bla _IMP-like gene. None of the isolates tested was positive with primers for bla _OXA-24-like, bla _OXA-143, ISAba2A, or ISAba4B.

The AFLP profiles of two isolates including a bla _OXA-69 containing outbreak strain isolated from Hospital S in 2001 [Reference Koh5], clustered with clone I profiles at 80%. Both had MLST sequence type (ST)1, clonal complex (CC)1. Seven isolates were linked with clone II isolates at 78%; of these, four were isolated from hospital S in 2001 and 2006, and included the predominant outbreak strain in 2001 that contained bla _OXA-66. One of these isolates was identified by MLST to ST2, CC2 [Reference Koh5]. No isolate belonged to clone III. With one exception, all isolates that clustered with clones I and II were positive for bla _OXA-23-like. Altogether, assignment of isolates by AFLP to clones I and II correlated with assignment by MLST to ST1 and ST2, respectively, emphasizing the global spread of these two clones which appear to be associated with bla _OXA-23-like genes [Reference Mugnier14].

Six isolates, including the predominant outbreak strains from Hospital S in 1996 that contained bla _OXA-64 had AFLP profiles that were unrelated to European clones I–III. Two isolates from 1996 with bla _OXA-64 and bla _OXA-88 were found to have the MLST ST25 and the novel type ST111, respectively. ST25 has been associated with A. baumannii in Greece, Italy and Turkey [Reference Di Popolo15].

In our survey, the relative prevalence of A. nosocomialis seems to be greater than that reported in other studies. In Ireland, clinical isolates of A. pittii exceeded that of A. baumannii by a factor of 1·8, while carbapenem resistance in these A. pittii isolates (22%) also exceeded that of A. baumannii (4%) [Reference Boo16]. A. nosocomialis made up only 5·4% of isolates in the Czech Republic whereas A. baumannii (mostly clone II but also clone I) and A. pittii accounted for 73·5% and 20·4%, respectively [Reference Nemec17]. In that study, A. pittii and A. nosocomialis isolates were susceptible to most antimicrobials tested including the carbapenems. Further, an 8-year survey in a university hospital in The Netherlands found A. pittii (40·3% of strains belonging to the Acb complex) was second to A. baumannii (55·8%) whereas A. nosocomialis (3·9%) was much less common [Reference van4]. The prevalence of multidrug resistance in A. pittii ranged from 0% to 22% over the course of the study and no carbapenem-resistant isolates were detected.

The present situation in Singapore is therefore similar to that in China and Korea where spread of international clones with bla _OXA-23-like is responsible for most of the carbapenem-resistant A. baumannii [Reference Fu18, Reference Park19]. The relatively high rates of occurrence of A. nosocomialis and A. pittii, their presence in bloodstream infections, and the multidrug and carbapenem resistance in these species underscore their potential clinical significance.