Genomic analysis of antibiotic resistance for Acinetobacter baumannii in a critical care center

Abstract

Aim

Acinetobacter baumannii is commonly associated with outbreaks and antibiotic‐resistant nosocomial infection. This study aimed to determine the relationship between antibiotic resistance and genotypes of A. baumannii.

Methods

A study was undertaken in the critical care center (CCC) of Juntendo University Urayasu Hospital (Urayasu, Japan) between January 2012 and September 2015. Antimicrobial susceptibility tests were carried out according to the Clinical and Laboratory Standards Institute guidelines. All A. baumannii isolates were verified to carry carbapenemase genes and the ISA ba1 element using polymerase chain reaction. The genetic relationship of all A. baumannii isolates was determined by pulsed‐field gel electrophoresis and multilocus sequence typing.

Results

During the study period, 1634 patients were admitted to the CCC. Acinetobacter baumannii was detected in 43 patients (average age, 58 ± 19 years; 67.4% men). Six patients were determined to be extensively drug‐resistant A. baumannii and 21 patients determined to be multidrug‐resistant A. baumannii. Antimicrobial susceptibility linked genotypes of A. baumannii. Molecular characterization by pulsed‐field gel electrophoresis and multilocus sequence typing showed that closely related clones of A. baumannii had spread in the CCC.

Conclusion

Resistance to antimicrobial drugs was significantly associated with certain A. baumannii genotypic types and molecular types. Thus, we might be able to predict whether the genotype has spread in the CCC or not when the susceptibility is examined, facilitating the appropriate isolation of patients.

In this retrospective study, we undertook genotypic determination of Acinetobacter baumannii isolates from patients in the critical care center between January 2012 and September 2015 by polymerase chain reaction. We also determined the genetic relationship of all A. baumannii isolates in the study using pulsed‐field gel electrophoresis and multilocus sequence typing, and evaluated their antimicrobial sensitivity/resistance patterns. Our results showed that closely related A. baumannii clones were spreading in the critical care center.

Introduction

Acinetobacter baumannii is an aerobic Gram‐negative bacillus that is ubiquitous in the natural environment. Although A. baumannii is typically harmless and colonizes the skin and respiratory tracts of patients, as well as medical instruments,1 it is an opportunistic pathogen that can cause bacteremia, ventilator‐associated pneumonia, and urinary tract infection in the hospital setting.2 Since the early 1980s, it has been associated with antibiotic‐resistant nosocomial infections, especially in intensive care units in Europe.

The intercontinental spread of multidrug‐resistant A. baumannii (MDR‐AB) between Europe and other continents has been described since the 1990s.2 Molecular epidemiological typing has been used to investigate intercontinental outbreaks of A. baumannii and to determine the genotypes of A. baumannii isolates from various locations worldwide, mainly in Europe. In addition to clones identified from these international outbreaks, two A. baumannii clones from Europe that have spread globally were identified. These have been designated as global clone (GC) I and II (previously known as the European clone or international clone I and II). The GCI and II lineages were first reported 23 years ago.3

In 2008, a university hospital in Japan reported that 23 patients in the critical care center (CCC) were infected with MDR‐AB, and four patients died.4 Subsequently, several CCCs in Japan also reported continuous MDR‐AB outbreaks in 2009, 2010, and 2011. Each CCC serves a predetermined population and accepts ambulances from the surrounding area. Thus, a CCC closure due to an outbreak can disrupt the local emergency care system. In 2010, four academic associations related to infectious diseases in Japan suggested the necessity of effective Acinetobacter surveillance, promotion of the development of new antibiotics, and their accelerated approval for domestic use (http://www.kansensho.or.jp/modules/guidelines/index.php?content_xml:id=26).

Acinetobacter baumannii can easily acquire multidrug resistance, including resistance to carbapenem.5 The most common mechanism of carbapenem resistance in A. baumannii is through the action of carbapenem‐hydrolyzing class D β‐lactamase, also known as an oxacillinase (OXA). To date, four main phylogenetic subgroups of OXA‐type carbapenemase (OTC) have been identified in A. baumannii: OXA‐23‐like, OXA‐40‐like, OXA‐51‐like, and OXA‐58‐like. The OXA‐51‐like enzymes are chromosomally encoded naturally occurring OTCs in A. baumannii. However, OXA‐51‐like enzymes are not typically potent against antibiotics and were reported to confer carbapenem resistance when an ISAba1 element is inserted upstream of the gene. Genes encoding acquired OXAs, including OXA‐23‐like, OXA‐40‐like, and OXA‐58‐like, have been found mostly on plasmids. An insertion sequence linked to OXA‐23 and OXA‐58 genes were also reported to regulate the expression of the carbapenemase.2

The expression of carbapenem‐hydrolyzing class B β‐lactamases, also known as metallo‐β‐lactamases (MBLs), is a powerful mechanism for carbapenem resistance carried by plasmids. To date, three groups of acquired MBLs have been identified in A. baumannii: IMP‐like, VIM‐like, and SIM‐1. However, MBL rarely mediates carbapenem resistance in A. baumannii.6

The CCC at Juntendo University Urayasu Hospital (Urayasu, Japan) has been experiencing an MDR‐AB outbreak since 2012. The majority of MDR‐AB outbreaks were found in our hospital, and there was no transfer to other hospitals in the same medical area. On average, 51 cases of Acinetobacter spp., including A. baumannii, colonization were detected yearly from 2003 to 2011 prior to the outbreak. However, the number of Acinetobacter spp. cases increased to 115 in 2012, 80 in 2013, 97 in 2014, and 100 in 2015. Since 2012, significant efforts, included clarification of the clinical features between antibiograms and polymerase chain reaction (PCR) patterns of A. baumannii isolates. The present study investigated the relationship between A. baumannii genotypes and their resistance to antibiotics.

Methods

Study design, setting, and participants

This study was carried out at Juntendo University Urayasu Hospital, a 653‐bed facility that includes 15 and in the CCC. The CCC serves approximately 1.6 million people residing in the east side of Chiba prefecture, which is located adjacent to Tokyo, Japan. Approximately 22,000 patients, including approximately 5000 patients in critical condition transported by ambulance, are treated in the emergency center annually. These include trauma, burn, poisoning, cardiopulmonary arrest, shock, and other critical internal medical conditions.

We screened a sputum or throat swab, using selective medium for isolation of Acinetobacter spp., at the time of admission to the CCC and weekly thereafter as a surveillance. Bacterial cultures obtained from the surveillance undertaken in the CCC between January 2012 and September 2015 included A. baumannii. Exclusion criteria were death within 24 h of admission, and pediatric patients <15 years old.

Definitions of MDR‐AB and extensively drug‐resistant A. baumannii

According to the new definition for drug‐resistant A. baumannii provided by the European Centre for Disease Prevention and Control and the Centers for Disease Control and Prevention, MDR‐AB is defined as A. baumannii with non‐susceptibility to one or more agents in at least three or more of the following categories: combinations of penicillins and β‐lactam inhibitors, extended‐spectrum cephalosporins, aminoglycosides, fluoroquinolones, carbapenems, polymyxins, and tigecycline. Extensively drug‐resistant A. baumannii (XDR‐AB) is defined as non‐susceptibility to one or more agents in the standard categories except polymyxins or tigecycline.7

Antimicrobial susceptibility testing

Once bacteria were cultured from the sputum or throat swab sample, antimicrobial susceptibility tests for common bacteria were undertaken to identify the causative agent according to the routine protocol of our bacteriology laboratory. Minimal inhibitory concentrations against piperacillin (PIPC), ceftazidime (CAZ), meropenem (MEPM), levofloxacin (LVFX), and amikacin (AMK) were determined using the broth microdilution method according to guidelines from the Clinical and Laboratory Standards Institute.8

Amplification of carbapenem resistance genes and IS by PCR

For samples with successful A. baumannii culture and isolation, the presence of carbapenemase genes (OXA‐23‐like, OXA‐24‐like, OXA‐51‐like, and OXA‐58‐like) were evaluated by multiplex PCR using four pairs of previously described primers9, 10 using the LightCycler 96 system (Roche, Tokyo, Japan). Additionally, two MBL genes (IMP‐ and VIM‐like) were also evaluated by PCR as previously described.11 The presence of ISAba1 element in all A. baumannii isolates was assessed by PCR, using the HFR and HRR primers12 and the LightCycler 96 system.

Pulsed‐field gel electrophoresis and multilocus sequence typing

The genetic relationship of all A. baumannii isolate was determined by pulsed‐field gel electrophoresis (PFGE) using the Sma1 restriction enzyme (Takara Bio, Kusatsu, Japan). Pulsed‐field gel electrophoresis was carried out using a Fingerprinting II software version 3.0 with support from SRL (Japan). Relatedness determinations were carried out as previously described.13 Sequence types (STs) for A. baumannii were determined by multilocus sequence typing (MLST), specifically for clonal lineages using the Pasteur scheme,14, 15 by amplification and sequence analysis of fragments of seven internal housekeeping genes (cpn60, fusA, gltA, pyrG, recA, rplB, and rpoB). Determination of the STs was carried out using the Pasteur MLST database website (http://pubmlst.org/abaumannii/).

Results

Patient background and bacteria isolation

The enrollment criteria for the study are provided in Figure 1. During the observation period, 1634 patients were admitted to the CCC. Of these, 281 patients had positive cultures from sputum or throat swab. Nine patients were excluded: five were pediatric patients (<15 years old), and four patients died within 24 h of admission. After the exclusion, samples from 272 patients were further studied.

Figure 1.

Summary of study design and participants. After implementing the exclusion criteria, Acinetobacter baumannii was detected in 43 patients treated in the critical care center (CCC) at Juntendo University Urayasu Hospital (Urayasu, Japan) between January 2012 and September 2015.

Acinetobacter baumannii was detected in 43 patients with an average age of 58 ± 19 years. The majority of patients were men (29/43, 67.4%). Three patients were identified as having acquired nosocomial infection: two patients were diagnosed with ventilator associated pneumonia due to A. baumannii, and one patient was positive for A. baumannii from blood cultures. Forty patients were colonized with A. baumannii.

Genotypic determination and antimicrobial susceptibility testing

The carbapenemase genotypes of A. baumannii are listed in Table 1. Multiplex PCR identified the same A. baumannii genotype (possessing OXA‐51‐like, ISAba1, and IMP, but not OXA‐23‐like, OXA‐40‐like, OXA‐58‐like, or VIM) in six of the 43 patients. The average age of patients with this genotype of A. baumannii was 51 ± 24 years, and 33.3% (2/6) were men. Their average length of stay in the CCC was 26 ± 5 days (Table 2). Acinetobacter baumannii isolated from these patients were resistant to PIPC, CAZ, MEPM, LVFX, and AMK (Fig. 2A). These isolates were determined to be XDR‐AB.

Table 1.

Genotypic determination of carbapenemase genes of Acinetobacter baumannii

Patient no.	Carbapenemase genotype
	OXA‐23	OXA‐40	OXA‐51	OXA‐58	ISAba1	IMP	VIM
25, 34, 36, 37, 38, 39	−	−	+	−	+	+	−
1, 3, 4, 5, 6, 8, 10, 11, 12, 13, 14, 15, 22, 23, 24, 26, 29, 30, 31, 35, 41	−	−	+	−	+	−	−
2, 7, 9, 16, 17, 18, 19, 20, 21, 27, 28, 32, 33, 40, 42, 43	−	−	+	−	−	−	−

Table 2.

Patient background and Acinetobacter baumannii genotypes

Genotype	OXA‐51‐like(+) ISA‐ba1(+) IMP(+)	OXA‐51‐like(+) ISA‐ba1(+) IMP(−)	OXA‐51‐like(+) ISA‐ba1(−) IMP(−)
Number of patients	6	21	16
Age (years)	51 ± 24	53 ± 16	68 ± 13
Sex (% male)	33.3	76.2	68.7
Time in CCC (days)	26 ± 5	25 ± 13	14 ± 10
Drug resistance	Resistant to PIPC, CAZ, MEPM, LVFX, and AMK (XDR‐AB)	Not susceptible to PIPC, CAZ, LVFX or AMK; susceptible to MEPM (MDR‐AB)	Sensitive to virtually all drugs: PIPC (6.3%), CAZ (0%), MEPM (0%), LVFX (6.3%), and AMK (6.3%)
PFGE type	X‐1	X‐1 or X‐2	Y1‐14
Spreading in CCC	Yes	Yes	No

AMK, amikacin; CAZ, ceftazidime; CCC, critical care center; LVFX, levofloxacin; MDR‐AB, multidrug‐resistant A. baumannii; MEPM, meropenem; PFGE, pulsed‐field gel electrophoresis; PIPC, piperacillin; XDR‐AB, extensively drug‐resistant A. baumannii.

Figure 2.

Antimicrobial susceptibility of Acinetobacter baumannii according to genotype. Susceptibility to piperacillin (PIPC), ceftazidime (CAZ), meropenem (MEPM), levofloxacin (LVFX), and amikacin (AMK) was evaluated according to Clinical and Laboratory Standards Institute guidelines. A, Antimicrobial susceptibility of the OXA‐51‐like(+) ISA ba1(+) IMP(+) genotype. B, Antimicrobial susceptibility of the OXA‐51‐like(+) ISA ba1(+) IMP(−) genotype. C, Antimicrobial susceptibility of the OXA‐51‐like(+) ISA ba1(−) IMP(−) genotype. White bars, sensitive (S); gray bars, intermediate‐sensitive (I); black bars, resistant (R).

Acinetobacter baumannii from 21 of the 43 patients had OXA‐51‐like and ISAba1, but not IMP, OXA‐23‐like, OXA‐40‐like, OXA‐58‐like, or VIM. The average age of these patients was 53 ± 16 years, and 76.2% (16/21) were male. Their average length of stay at the CCC was 25 ± 13 days (Table 2). These isolates were not susceptible to PIPC, CAZ, LVFX, or AMK, and were susceptible to MEPM (Fig. 2B). They were determined to be MDR‐AB.

The other 16 patients had A. baumannii possessing only the OXA‐51‐like genotype, and not ISAba1, OXA‐23‐like, OXA‐40‐like, OXA‐58‐like, IMP, or VIM. The average age of these patients was 68 ± 13 years and 68.7% (11/16) were men. Their average CCC stay was 14 ± 10 days (Table 2). These isolates were sensitive to virtually all drugs: PIPC (6.3%), CAZ (0%), MEPM (0%), LVFX (6.3%), and AMK (6.3%) (Fig. 2C).

Based on these results, the antibiotic susceptibility or resistance was significantly associated with certain A. baumannii genotypes. Therefore, antimicrobial susceptibility could indicate the genotype of A. baumannii.

Pulsed field gel electrophoresis and MLST findings

Pulsed field gel electrophoresis was undertaken for samples from all 43 patients with A. baumannii including those with XDR‐AB and MDR‐AB. Isolates from three patients (nos. 10, 33, and 40) were excluded due to the potential for contamination with other bacteria during long‐term sample storage at −80°C. The PFGE results showed that isolates from 26 patients with A. baumannii genotypes carrying OXA‐51‐like and ISAba1, regardless of the presence of the IMP gene, belonged to two closely related types: PFGE type X‐1 and X‐2 (Fig. 3). The similarity coefficient for isolates from these two PFGE types was 93%. Based on this similarity, we hypothesized that closely related A. baumannii genotypes had been spreading in the CCC for 4 years.

Figure 3.

Results of pulsed‐field gel electrophoresis (PFGE) analysis of Acinetobacter baumannii carrying OXA‐51‐like and ISA ba1 from 26 patients, which belong to two closely related PFGE types, X‐1 and X‐2.

In contrast, A. baumannii isolated from the other 14 patients (OXA‐51‐like(+) ISAba1(−) IMP(−) genotype) were detected as various clones belonging to PFGE type Y‐1 to Y‐14 (Fig. 3). The lineage clonality of these A. baumannii isolates was low, and the similarity coefficient was approximately 55–75%.

Results of the MLST showed that the PFGE type X‐1 and X‐2 isolates carrying OXA‐51‐like and ISAba1 belonged to ST2 according to the protocol of the Pasteur Institute (two alleles each for cpn60, fusA, gltA, pyrG, recA, rplB, and rpoB). Those belonging to GCII were the only strains that could cause outbreaks in our CCC and potentially could have spread from the CCC throughout the hospital.

Discussion

To date, most of the previous studies about MDR‐AB were international spread or multicenter studies, because MDR‐AB is rarely detected.2 The present study shows the relationship of A. baumannii from 43 patients in terms of the susceptibility, genotype, and molecular type at a single center. This study shows that genotypic determination of A. baumannii can predict the susceptibility to antimicrobial drugs. We found that two closely related strains, as determined by molecular characterization by PFGE and MLST, had spread in our CCC. Therefore, we could predict whether the genotype spread in the CCC or not when the susceptibility was examined, facilitating the appropriate isolation of patients. This isolation method can be applied not only to A. baumannii but also to Gram‐negative bacillus when susceptibility is consistent with the genotype of bacteria.

Specifically, A. baumannii genotypes carrying OXA‐51‐like and ISAba1 had spread in our CCC. Four OXA enzymes of A. baumannii have been reported to cause outbreaks globally, with OXA‐23‐like and OXA‐51‐like as the major oxacillinases associated with outbreaks in Asia.2 The strains isolated in our CCC also carried OXA‐51‐like; however, they did not possess the OXA‐23‐like gene. Based on the MLST results, the strains found to be spreading in the CCC belonged to GCII, which has been associated with the presence of OXA‐51 and ISAba1, OXA‐23, OXA‐40, or OXA‐58.3 In our CCC, the GCII strains possessed OXA‐51‐like and ISAba1.

To date, the IMP gene has been identified in A. baumannii in a variety of geographic regions, including Italy, Spain, Portugal, Greece, Australia, Japan, Singapore, Korea, and Hong Kong.2, 6, 16 The present PCR and PFGE analyses revealed that six patients harbored A. baumannii with the IMP gene and the PFGE type X‐1 detected two genotype strains: IMP(+) and IMP(−). Plasmids carrying the IMP gene were more than 60 kbp.17 The PFGE analysis we used was capable of detecting differences of approximately 50–300 kbp. However, plasmids containing the IMP gene could not be detected by PFGE, and both the IMP(+) and IMP(−) genotypes would be assigned to the same molecular type.

Some limitations exist in this study. First, the present outbreak of A. baumannii is characterized by the specific genotype, OXA‐51‐like(+) and ISAba1(+), in the CCC. However, other molecularly characterized strains could be spread in other hospitals and must be detected independently by PCR and PFGE. Second, A. baumannii can easily acquire multidrug resistance owing to plasmid‐mediated gene transformation.5 However, as has been stated, the PFGE type X‐1 belonged to OXA‐51‐like(+) and ISAba1(+) regardless of the presence of the IMP gene, and the PFGE technique was unable to detect minor genetic differences, such as the IMP gene.

In conclusion, susceptibility to antimicrobial drugs can predict A. baumannii genotypic types in the CCC.

Disclosure

Approval of the research protocol: This study was approved by the institutional review board of the National Human Genome Research Institute and was carried out according to the principles of the Declaration of Helsinki. This study was also approved by the Ethics Committee of the Faculty of Medicine, Juntendo University Urayasu Hospital (approval no. 30‐004).

Informed consent: Informed consent was obtained from the guardians by opt‐out in publicity documents.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

Conflict of interest: None declared.