Genetic Basis of Multidrug Resistance in Acinetobacter baumannii Clinical Isolates at a Tertiary Medical Center in Pennsylvania

Abstract

A total of 49 unique clinical isolates of multidrug-resistant (MDR) Acinetobacter baumannii identified at a tertiary medical center in Pittsburgh, Pennsylvania, between August 2006 and September 2007 were studied for the genetic basis of their MDR phenotype. Approximately half of all A. baumannii clinical isolates identified during this period qualified as MDR, defined by nonsusceptibility to three or more of the antimicrobials routinely tested in the clinical microbiology laboratory. Among the MDR isolates, 18.4% were resistant to imipenem. The frequencies of resistance to amikacin and ciprofloxacin were high at 36.7% and 95.9%, respectively. None of the isolates was resistant to colistin or tigecycline. The presence of the carbapenemase gene bla_OXA-23 and the 16S rRNA methylase gene armA predicted high-level resistance to imipenem and amikacin, respectively. bla_OXA-23 was preceded by insertion sequence ISAba1, which likely provided a potent promoter activity for the expression of the carbapenemase gene. The structure of the transposon defined by ISAba1 differed from those reported in Europe, suggesting that ISAba1-mediated acquisition of bla_OXA-23 may occur as an independent event. Typical substitutions in the quinolone resistance-determining regions of the gyrA and parC genes were observed in the ciprofloxacin-resistant isolates. Plasmid-mediated quinolone resistance genes, including the qnr genes, were not identified. Fifty-nine percent of the MDR isolates belonged to a single clonal group over the course of the study period, as demonstrated by pulsed-field gel electrophoresis.

Acinetobacter baumannii is a gram-negative, non-lactose-fermenting organism that is increasingly recognized as a major pathogen causing nosocomial infections including bacteremia and ventilator-associated pneumonia, particularly in patients admitted to intensive care units (23, 25). The organism is characterized by its tendency to acquire resistance to multiple classes of antimicrobials (3). Of note, increasing resistance to carbapenems has been observed worldwide in the past decade, frequently mediated by production of Ambler's class D β-lactamases, which possess carbapenemase activity (26). Several outbreaks caused by multidrug-resistant (MDR) A. baumannii have been reported from the United States (21, 22, 28). Additionally, infections due to MDR A. baumannii have been observed in military personnel returning from Iraq and Afghanistan (18, 29). The Infectious Diseases Society of America recently identified A. baumannii as one of the six particularly problematic pathogens in terms of antimicrobial availability issues arising from resistance (32).

The emergence of A. baumannii clinical isolates with resistance to multiple classes of antimicrobials, including carbapenems, aminoglycosides, and fluoroquinolones, was observed at our medical center in the latter half of 2006. For these patients, therapeutic options were limited to salvage agents such as colistin and tigecycline. In the present study, we conducted a detailed investigation of the molecular epidemiology and genetic basis of multidrug resistance among A. baumannii clinical isolates identified at our medical center over a 1-year period, with a focus on the mechanisms of carbapenem, aminoglycoside, and fluoroquinolone resistance.

MATERIALS AND METHODS

Clinical isolates and definition of MDR.

A. baumannii isolates recovered from patient specimens at the University of Pittsburgh Medical Center (UPMC) Presbyterian Campus between August 2006 and September 2007 were included in this study. A. baumannii was identified in the clinical microbiology laboratory by using the Gram-Negative Identification Panel (Microscan, Dade Behring Inc., Sacramento, CA). For automated identification in the electronic medical records, MDR was defined as nonsusceptibility to three or more of the antimicrobials that are routinely tested in the clinical laboratory and to which A. baumannii would have been expected to be susceptible. These included ampicillin-sulbactam, piperacillin-tazobactam, cefepime, ciprofloxacin, trimethoprim-sulfamethoxazole, a carbapenem (imipenem or meropenem), and an aminoglycoside (amikacin, tobramycin, or gentamicin). As a result, 65 MDR isolates from different patients were identified from this period. Forty-nine of the 65 isolates were available for further analysis in the research laboratory. All the study isolates were aliquoted and stored at −80°C until further use.

Susceptibility testing.

The susceptibilities of the isolates to ampicillin-sulbactam, ceftazidime, cefepime, meropenem, tobramycin, gentamicin, ciprofloxacin, and tetracycline were tested using the standard disk diffusion method on Mueller-Hinton (MH) agar plates (BD Microbiology Systems, Sparks, MD) and using the breakpoints defined by the Clinical and Laboratory Standards Institute (CLSI) (5). MICs of imipenem, amikacin, colistin, and tigecycline were determined by use of Etest strips (AB Biodisk, Solna, Sweden). They were also interpreted according to the CLSI breakpoints, except for tigecycline, for which the breakpoints endorsed by the British Society of Antimicrobial Chemotherapy (BSAC) (MICs, ≤1 μg/ml for susceptibility and >2 μg/ml for resistance) were used. BSAC is the only organization that has defined tigecycline breakpoints for A. baumannii. For non-imipenem-susceptible isolates, a phenotypic screening test for metallo-β-lactamase production using sodium mercaptoacetic acid disks was also conducted (2).

PFGE.

For pulsed-field gel electrophoresis (PFGE), the genomic DNA was digested with ApaI (New England Biolabs, Beverly, MA). The resultant fragments were then separated by PFGE using a temperature-controlled CHEF DR III system (Bio-Rad, Hercules, CA) as described previously (31). For PFGE pattern analysis, Bionumerics software, version 4.0 (Applied Maths, Sint-Martens-Latem, Belgium), with the unweighted-pair group method using average linkages and the Dice setting for clustering analysis was applied. The genetic relatedness of isolates was determined by the criteria of Tenover et al. (33).

PCR analyses for detection of resistance genes.

PCR analyses were performed for detection of various resistance genes in all isolates. A loopful of bacteria was taken from each fresh overnight culture on MH agar plates (BD Microbiology Systems), suspended in 1 ml of sterile water, and boiled for 10 min. After centrifugation, the supernatant was used as the template. Amplification was performed using a 9700 GeneAmp thermocycler (Applied Biosystems, Foster City, CA). The genes investigated included the carbapenemase genes bla_OXA-23, bla_OXA-40, bla_OXA-51, and bla_OXA-58; the cephalosporinase gene bla_ADC; potential extended-spectrum β-lactamase (ESBL) genes bla_TEM, bla_SHV, and bla_CTX-M; the 16S rRNA methylase gene armA; the aac(6′)-Ib, aac(6′)-Iad, and aph(3′)-VIa genes, encoding amikacin-modifying enzymes; and the plasmid-mediated quinolone resistance genes qnrA, qnrB, and qnrS. Select PCR products were sequenced by use of an ABI 3100 instrument (Applied Biosystems). For genes with negative results in PCRs in which no positive control was used [bla_OXA-40, bla_OXA-58, and aac(6′)-Iad], PCR amplifications were repeated at least twice. A negative control was run with every PCR. The quinolone resistance-determining regions (QRDRs) of the gyrA and parC genes of 13 isolates (1 representative isolate for each of the 10 pulsotypes identified in this study and 3 ciprofloxacin-susceptible control isolates from outside the study) were also amplified by PCR and subjected to sequencing. The primers used for the PCR analyses are listed in Table 1.

TABLE 1.

Primers used for amplification of resistance genes

Primers	Sequences (5′ to 3′)a	Target gene(s)	Reference
OXA Set A FOR	ATGAAAAAATTTATACTTCC	blaOXA-24, blaOXA-25, blaOXA-26, blaOXA-33, blaOXA-40, blaOXA-72	18
OXA Set A REV	TTAAATGATTCCAAGATTTTC
OXA Set C FOR	ACAGAARTATTTAAGTGGG	blaOXA-51, blaOXA-58, blaOXA-64, blaOXA-69, blaOXA-70, blaOXA-71, blaOXA-75, blaOXA-78	18
OXA Set C REV	GGTCTACAKCCMWTCCCCA
OXA 23 FOR	GATGTGTCATAGTATTCGTCG	blaOXA-23, blaOXA-27, blaOXA-49	1
OXA 23 REV	TCACAACAACTAAAAGCACTG
OXA 58 5′	ATGAAATTATTAAAAATATTGAGTTTAG	blaOXA-58, blaOXA-96	18
OXA58 3′	TTATAAATAATGAAAAACACCCAAC
ADC-7 FOR	ATGCGATTTAAAAAAATTTCTTGT	blaADC-1, blaADC-2, blaADC-3, blaADC4, blaADC-5, blaADC-6, blaADC-7	18
ADC-7 REV	TTATTTCTTTATTGCATTCAG
SHV S1	ATTTGTCGCTTCTTTACTCGC	blaSHV	37
SHV S2	TTTATGGCGTTACCTTTGACC
TEM1	ATGAGTATTCAACATTTCCGTG	blaTEM	11
TEM4	TTACCAATGCTTAATCAGTGAG
CTX-M/F	TTTGCGATGTGCAGTACCAGTAA	blaCTX-M	10
CTX-M/R	CGATATCGTTGGTGGTGCCATA
CTXM 2 FOR	AAATGTGCTGCTCCTTTCGTGAGC	blaCTX-M-2	This study
CTXM 2 REV	AGGGTTCGTTGCAAGACAAGACTG
ArmAf	TGCATCAAATATGGGGGTCT	armA	This study
ArmAr	GGATTGAAGCCACAACCAAA
aac(6′)-1b f	TTGCGATGCTCTATGAGTGGCTA	aac(6′)-Ib	24
aac(6′)-1b r	CTCGAATGCCTGGCGTGTTT
ABA-F	TTTGGCTATGATCCTATG	aac(6′)-Iad	9
ABA-R	CATGTCGAACAAGTACGC
APH F	ATACAGAGACCACCATACAGT	aph(3′)-VI	36
APH R	GGACAATCAATAATAGCAAT
gyrA-1	AAATCTGCCCGTGTCGTTGGT	gyrA	18
gyrA-2	GCCATACCTACGGCGATACC
parC-1	AAACCTGTTCAGCGCCGCATT	parC	18
parC-2	AAAGTTGTCTTGCCATTCACT
qnrA f	ATTTCTCACGCCAGGATTTG	qnrA	28a
qnrA r	GATCGGCAAAGGTTAGGTCA
qnrB f	GATCGTGAAAGCCAGAAAGG	qnrB	28a
qnrB r	ACGATGCCTGGTAGTTGTCC
qnrS f	ACGACATTCGTCAACTGCAA	qnrS	28a
qnrS r	TAAATTGGCACCCTGTAGGC

^aM stands for A or T; R stands for A or G; W stands for A or T; K stands for G or T.

Transfer of bla_OXA-23.

A. baumannii HE130 (a clinical strain susceptible to carbapenems) and Escherichia coli DH10B were used as the recipients for transformation experiments. AB017 and AB026, two bla_OXA-23-positive isolates of pulsotype A, were used as the donor strains. The competent cells of the recipient strains were prepared and transformed by electroporation with plasmid DNA extracted from the donor strains by the standard alkaline lysis method. The transformants were selected on Luria-Bertani (LB) agar plates containing 2 μg/ml of meropenem.

Cloning and sequencing of bla_OXA-23.

It has been demonstrated that OXA-23 confers high-level carbapenem resistance on A. baumannii when it is expressed under the control of a potent promoter provided by insertion sequence ISAba1 (7, 16). To examine if this is the case for our isolates, the genetic environment of bla_OXA-23 was investigated. Genomic DNA of A. baumannii AB017 was digested with XbaI and ligated with pBC-SK(−) (Stratagene, La Jolla, CA). Escherichia coli DH10B was then transformed with the ligated products by electroporation. Transformants that possessed recombinant plasmids carrying bla_OXA-23 were selected on LB agar plates containing chloramphenicol (25 μg/ml) and ampicillin (50 μg/ml). The DNA insert obtained was sequenced on both strands using custom sequencing primers.

Nucleotide sequence accession numbers.

The nucleotide sequences reported in this paper have been submitted to the GenBank/EMBL/DDBJ database under accession no. EU594641.

RESULTS

Antimicrobial susceptibility of MDR A. baumannii.

A total of 142 unique A. baumannii isolates were identified at the clinical microbiology laboratory at UPMC during the study period. Of those, 65 isolates (45.8%) met the MDR criteria used in the present study. Among the 49 MDR isolates that were available for the study, rates of full resistance were as follows: 95.9% for ciprofloxacin, 87.8% for ceftazidime, 79.6% for cefepime, 40.8% for ampicillin-sulbactam, 18.4% for imipenem, 22.4% for meropenem, 36.7% for amikacin, 61.2% for tobramycin, 77.6% for gentamicin, and 79.6% for tetracycline. Overall, 8 out of the 49 isolates (16.3%) were resistant to six classes of antibiotics tested (ampicillin-sulbactam, ciprofloxacin, a cephalosporin, a carbapenem, tetracycline, and an aminoglycoside). None of the isolates was resistant to colistin or tigecycline. However, 55.1% showed tigecycline MICs between 1.5 and 2 μg/ml, which are interpreted as intermediate according to the BSAC breakpoints.

None of the non-imipenem-susceptible isolates gave a positive result with the phenotypic screening test for metallo-β-lactamase production.

Molecular typing and clonal detection through PFGE.

A total of 10 pulsotypes that comprised genetically indistinguishable or closely related isolates were observed by PFGE. Pulsotype A was predominant, comprising 29 isolates identified at different time points during the study period. Pulsotypes B, C, and D were possibly related to pulsotype A. The other six pulsotypes (E through J) were all different from each other as well as from pulsotype A (Table 2; Fig. 1).

TABLE 2.

PCR results based on pulsotypes

Pulsotype	Total no. of strains	No. (%) of strains positive for the indicated:
		β-Lactamase gene								Amikacin resistance gene
		blaOXA-23	blaOXA-40	blaOXA-51	blaOXA-58	blaADC	blaCTX-M-2	blaTEM	blaSHV	armA	aph(3′)-VIa	aac(6′)-Ib	aac(6′)-Iad
A	29	7	0	29	0	29	2	27	0	16	2	9	0
B	3	0	0	3	0	3	0	3	0	0	0	0	0
C	4	0	0	4	0	4	0	3	0	0	0	0	0
D	7	0	0	7	0	0	3	2	0	0	0	6	0
E	1	0	0	1	0	1	0	1	0	1	0	0	0
F	1	0	0	1	0	0	0	0	0	0	0	1	0
G	1	0	0	1	0	1	0	0	0	0	0	0	0
H	1	1	0	1	0	1	0	0	0	0	1	0	0
I	1	0	0	1	0	1	0	1	0	0	0	0	0
J	1	0	0	0	0	0	0	0	0	0	0	0	0
Total	49	8 (16.3)	0 (0)	48 (97.9)	0 (0)	40 (81.6)	5 (10.2)	36 (73.5)	0 (0)	17 (34.7)	3 (6.1)	16 (32.6)	0 (0)

FIG. 1.

PFGE patterns of all pulsotypes identified in the study.

β-Lactamase genes.

The carbapenemase gene bla_OXA-23 was identified in all eight isolates for which imipenem MICs were >32 μg/ml. In contrast, none of the isolates without bla_OXA-23 had imipenem MICs of >32 μg/ml (Tables 2 and 3). bla_OXA-40 and bla_OXA-58, the other frequently reported carbapenemase genes in A. baumannii, were not detected in any of the isolates. A bla_OXA-51-like gene was identified in most isolates. OXA-51-like oxacillinases constitute a group of β-lactamases with low-level catalytic efficiency for carbapenems (15). bla_ADC genes, a group of cephalosporinase genes commonly found in A. baumannii and related species (17), were detected in 81.6% of the isolates. Only one isolate was susceptible to both ceftazidime and cefepime. Therefore, it was not possible to correlate cephalosporin resistance with the presence of bla_ADC.

TABLE 3.

MICs of imipenem in the presence or absence of bla_OXA-23

blaOXA-23 status (no. of isolates)	No. of isolates with a MIC (μg/ml) of:
	>32	32	24	16	8	6	4	3	2	1.5	1	0.75	0.5	<0.38
Positive (8)	8
Negative (41)			1		3	1		1	10	9	4	5	5	2

In the investigation of ESBL genes, five isolates were found to encode bla_CTX-M, which was identified as bla_CTX-M-2, upon the sequencing of the entire structural gene (Table 2). These isolates were all resistant to cefepime but not necessarily to ceftazidime or ampicillin-sulbactam. bla_TEM was detected in 36 isolates. Sequencing of select amplicons revealed that they encoded TEM-1. bla_SHV was not identified in any of the isolates.

Aminoglycoside resistance genes.

Eighteen isolates were resistant to amikacin. Among these, 17 were positive for armA, a 16S rRNA methylase gene (Table 2). All of the armA-positive isolates had amikacin MICs of >256 μg/ml (Table 4). The only amikacin-resistant, armA-negative isolate had an amikacin MIC of 128 μg/ml. This isolate did not yield positive PCR results for any of the three amikacin-modifying enzyme genes investigated. aac(6′)-Ib, an aminoglycoside acetyltransferase gene, was present in 16 isolates. The deduced amino acid sequences of select amplicons were consistent with AAC(6′)-Ib, but not with AAC(6′)-Ib-cr, which has been implicated in low-level resistance to fluoroquinolones. MICs of amikacin for aac(6′)-Ib-positive isolates ranged from 2 μg/ml to 12 μg/ml, except for eight isolates that were armA positive as well. Three isolates were positive for aph(3′)-VIa, an aminoglycoside phosphotransferase gene. Two of them were also armA positive. The only aph(3′)-VIa-positive, armA-negative isolate had an amikacin MIC of 12 μg/ml. None of the study isolates was positive for aac(6′)-Iad, the other aminoglycoside acetyltransferase gene implicated in amikacin resistance in A. baumannii.

TABLE 4.

MICs of amikacin in the presence or absence of armA

armA status (no. of isolates)	No. of isolates with MICs (μg/ml) of:
	>256	256	128	64	48	32	24	16	12	8	6	4	2	<1
Positive (17)	17
Negative (32)			1		1		2	3	5	9	4	4	3

QRDRs and resistance genes.

The sequencing results for the QRDRs of gyrA and parC, encoding DNA gyrase and DNA topoisomerase IV, respectively, revealed the presence of S83L and S80L substitutions in the respective enzymes for all eight ciprofloxacin-resistant isolates sequenced. The QRDRs of the two ciprofloxacin-intermediate isolates in the study and of the three susceptible control isolates did not possess substitutions implicated in fluoroquinolone resistance in the amino acid sequence of either gene. In addition, the V101I substitution in gyrA was observed in all of the susceptible, intermediate, and resistant isolates; this likely represented a polymorphism that did not affect susceptibility to fluoroquinolones.

None of the isolates gave positive PCR results for the plasmid-mediated quinolone resistance gene qnrA, qnrB, or qnrS.

Transfer of bla_OXA-23.

Both AB017 and AB026 yielded bla_OXA-23-positive transformants with A. baumannii HE130 as the recipient. Imipenem and meropenem MICs were 4 to 8 μg/ml and 6 to 12 μg/ml for the transformants compared with 0.5 μg/ml and 0.125 μg/ml for the recipient, respectively. No transformants could be obtained with E. coli DH10B as the recipient.

Genetic environment of bla_OXA-23.

A cloning experiment yielded a recombinant plasmid with a 4.0-kb insert carrying bla_OXA-23. A schematic representation of its genetic environment is given in Fig. 2. As has been reported earlier for strains from Europe and East Asia, bla_OXA-23 was preceded by insertion sequence ISAba1, encoding a transposase in the opposite orientation. Promoter sequences consisting of the −35 sequence (TTAGAA) and the −10 sequence (TTATTT), known to be responsible for the overexpression of β-lactamase genes located downstream of them (7, 16), were identified between 87 and 60 bp upstream of bla_OXA-23. However, the left inverted repeat of ISAba1 (i.e., the 5′ end of ISAba1) was located closer to the 5′ end of bla_OXA-23 due to the presence of a 7-bp deletion compared with the corresponding sequence that was initially characterized in Tn2006 from France (7). This particular deletion is also observed in sequences that have been submitted from several other countries (nucleotide accession no. AJ132105, EF120622, and EF016357). Furthermore, unlike Tn2006, the transposon identified in this study, tentatively designated Tn2008, was not part of a composite transposon. The sequences flanking Tn2008 were distinct from those flanking Tn2006. An open reading frame with moderate identity to a putative DNA binding protein described in several A. baumannii genome sequences was identified downstream of Tn2008 (Fig. 2).

FIG. 2.

Schematic representation of the genetic environments of bla_OXA-23 from the United States, France, and Turkey. (A) Strain AB017 (investigated in the present study); (B) strain AB13 (7); (C) strain AcKOU1 (GenBank accession no. EF120622). The boundaries of Tn2006 and Tn2008 are indicated, along with the target site duplications (underlined). The 7-bp difference in the site of insertion of ISAba1 for strain AB13 is double underlined. Target site duplication was not identified within the available sequence for strain AcKOU1.

DISCUSSION

MDR A. baumannii has emerged as a substantial clinical problem worldwide (23, 25, 26). This trend has paralleled the overall increase in the prevalence of Acinetobacter spp., including A. baumannii, as nosocomial pathogens. For example, data from the National Nosocomial Infections Surveillance System indicate that the proportion of Acinetobacter spp. associated with pneumonia in intensive-care units increased from 4% in 1986 to 7% in 2003 in the United States (13). We recently reported the emergence of two MDR A. baumannii isolates that were highly resistant to both carbapenems and aminoglycosides due to production of both the OXA-23 carbapenemase and the ArmA 16S rRNA methylase, respectively (8). The present study was conducted to define the genetic basis of multidrug resistance in A. baumannii by using a larger set of isolates.

Approximately 20% of the MDR isolates were resistant to carbapenems. We observed a clear correlation between the presence of the OXA-23 gene and high-level carbapenem resistance (Table 3). Indeed, transfer of bla_OXA-23 to a susceptible strain led to an 8- to 96-fold increase in carbapenem MICs. This is in contrast to the other reports of carbapenem-resistant A. baumannii outbreak investigations in the United States, where carbapenem resistance was attributed to the production of the OXA-40 carbapenemase or reduced expression of outer membrane proteins in the absence of carbapenemase activities (21, 28). Production of OXA-23 is the most frequently encountered mechanism of carbapenem resistance in A. baumannii worldwide (26). The degree of resistance in our isolates was likely accentuated by the presence of strong promoter sequences provided by ISAba1, leading to overproduction of the enzyme, as has been demonstrated with European strains (7). Of note, the origin of bla_OXA-23 was recently identified as the chromosome of Acinetobacter radioresistens, a commensal species of the human skin (27). Taking this together with the fact that ISAba1 is commonly found in various Acinetobacter spp. (30), we may speculate that mobilization of bla_OXA-23 from A. radioresistens to A. baumannii occurs concurrently under selective pressure from carbapenems in different geographic areas. The diversity observed in the structures of transposons carrying ISAba1 and bla_OXA-23 in the United States and Europe supports this hypothesis. The likelihood that high-level carbapenem resistance may be acquired by A. baumannii through transposon-mediated gene transfer from a commensal organism underscores the importance of continued efforts to limit carbapenem use in order to retain susceptibility to these agents.

Resistance to amikacin was seen in nearly 40% of the study isolates. Most of the resistant isolates were highly resistant (MIC, >256 μg/ml), and this resistance coincided with the presence of the 16S rRNA methylase gene armA (Table 4). Among the series of acquired 16S rRNA methylases, ArmA appears to be the most common enzyme worldwide to date. Although armA was initially identified in 2002 (12), it has been shown that it was present in clinical isolates as early as 1997 (19). armA has been found mostly in Enterobacteriaceae, but its presence has been documented in A. baumannii as well, mostly in East Asia (20, 38). It is somewhat puzzling that we are observing an increasing number of A. baumannii isolates that produce ArmA, while the overall systemic use of aminoglycosides has decreased at our facility in the past several years (data not shown). One possibility is the presence of collateral selective pressure from other resistance genes that are located near armA. armA has been shown to be borne on an IS26-based composite transposon in E. coli, which likely plays a role in its mobilization (14). However, the only resistance genes contained in this transposon were ant(3[dprime])-9, sul1, and dfrXII, which confer resistance to streptomycin, sulfonamides, and trimethoprim, respectively. Preliminary sequencing of the genetic environment of our isolates revealed an identical structure at least in proximity to armA (8). We are currently conducting further investigations of the genetic environment of armA in order to clarify the mode of its mobilization in A. baumannii.

In contrast to the strong association observed between the presence of armA and high-level aminoglycoside resistance, including amikacin resistance, the roles of the amikacin-modifying enzyme genes aac(6′)-Ib and aph(3′)-VIa in amikacin resistance were less clear. One possible explanation, at least for aac(6′)-Ib, is that the gene is typically carried on integrons as a gene cassette lacking its own promoters, and thus its expression may be suboptimal depending on its distance from the common promoter sequences located at the 5′ conserved segment of the integrons (6).

All of the isolates were resistant or intermediate to ciprofloxacin. Resistance was associated with the typical substitutions in the QRDRs of DNA gyrase and DNA topoisomerase IV in all ciprofloxacin-resistant pulsotypes (34, 35). On the other hand, no plasmid-mediated quinolone resistance genes were detected in any of the isolates. These findings suggest that resistance to fluoroquinolone in MDR A. baumannii depends on target modification conferred by substitutions in the QRDRs. Given the very high rate of resistance, however, this class is unlikely to have any clinical role in the treatment of MDR A. baumannii at our medical center.

More than half of the MDR isolates in the study belonged to the same clonal type, i.e., pulsotype A, as evidenced by PFGE (Table 2). Indeed, seven of eight isolates that were resistant to six classes of antimicrobials belonged to this pulsotype and carried both bla_OXA-23 and armA. However, bla_OXA-23 and armA were also detected in different pulsotypes, indicating that these genes are likely disseminating among A. baumannii strains by means of horizontal transfer as well as clonal spread.

In conclusion, we have described the genetic basis of resistance in MDR A. baumannii at a tertiary medical center in Pennsylvania. Multidrug resistance was conferred predominantly by the production of OXA-23 carbapenemase, ArmA 16S rRNA methylase, and resistance substitutions in the QRDRs of DNA gyrase and DNA topoisomerase IV. As the use of salvage agents such as colistin and tigecycline to treat infections caused by these MDR organisms increases, close monitoring of susceptibility to these agents is also warranted.