Acinetobacter baumannii and Acinetobacter genospecies 13TU and 3 bacteraemia: comparison of clinical features, prognostic factors and outcomes

Abstract

Objectives

To investigate the clinical impact of different genospecies of the Acinetobacter calcoaceticus–Acinetobacter baumannii complex (ACB complex; A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3) on the severity of bacteraemia.

Methods

We retrospectively compared the clinical features and outcomes of patients with bacteraemia caused by A. baumannii, Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3. The genospecies were identified using oligonucleotide array sequence analysis (interspacer sequence), and the clonality of Acinetobacter gen. sp. 13TU and 3 isolates was determined by PFGE analysis.

Results

A total of 215 patients with bacteraemia due to ACB complex were evaluated. Among them, 117 (54.4%) had A. baumannii bacteraemia, 77 (35.8%) had Acinetobacter gen. sp. 13TU bacteraemia and 21 (9.8%) had Acinetobacter gen. sp. 3 bacteraemia. A. baumannii bacteraemia was associated with a higher 14 day mortality rate (P < 0.001), a higher 30 day mortality rate (P < 0.001) and a higher in-hospital mortality rate than bacteraemia due to Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3. Independent prognostic factors for the 30 day mortality included the Charlson co-morbidity index (P < 0.001) and Pitt bacteraemia score (P < 0.001). Bloodstream infection caused by a multidrug-resistant A. baumannii isolate appeared to be associated with a poor outcome (P = 0.069). There was no clonal spread of Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3 during the study period.

Conclusions

Bacteraemia due to multidrug-resistant strains but not A. baumannii per se appears to be associated with poor outcome.

Acinetobacter genospecies 13TU, Acinetobacter genospecies 3, 16S–23S rRNA gene, bacteraemia, multidrug resistance, mortality

Introduction

The genus Acinetobacter is currently defined as Gram-negative, strictly aerobic, glucose-non-fermenting, non-motile, catalase-positive and oxidase-negative bacteria with a DNA G + C content of 39%–47%.¹ At least 32 named and unnamed Acinetobacter spp. have been described.^2,3 Genospecies 1 (Acinetobacter calcoaceticus), genospecies 2 (Acinetobacter baumannii), genospecies 3 and genospecies 13 are genetically closely related, and are difficult to distinguish phenotypically using routine laboratory methods. Therefore, it has been proposed that these species be referred to as a group, the A. calcoaceticus–A. baumannii complex (ACB complex).^1–4A. calcoaceticus is a soil and water organism, and is rarely isolated from clinical specimens. The latter three species (A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3) of this complex have been implicated as causes of both community-acquired and nosocomial infections.¹

In recent years, bacteria of the ACB complex have become one of the most common causes of nosocomial infections. The emergence and spread of multidrug-resistant (MDR) Acinetobacter spp. poses an important challenge to hospitals.² Studies have shown that there is significant variation in the susceptibility to antimicrobial agents among A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3, with A. baumannii isolates being less susceptible to most antimicrobial agents than Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3 isolates.^3,4 The correlation between antimicrobial susceptibility among different genospecies and clinical outcome is uncertain.

Apart from DNA–DNA hybridization (the gold standard for the identification of different genospecies of the ACB complex), other molecular methods have been developed and validated, such as amplified 16S rRNA gene restriction analysis,⁵ high-resolution fingerprint analysis by amplified fragment length polymorphism analysis,⁶ ribotyping,⁷ tRNA spacer fingerprinting,⁸ restriction analysis of the 16S–23S rRNA intergenic spacer (ITS) sequences,⁹ sequence analysis of the 16S–23S rRNA gene spacer region,¹⁰ and sequencing of the rpoB gene.¹¹ ITS sequences have been suggested to be a good candidate for identification of the species that comprise the ACB complex.¹⁰

There is little clinical information about bacteraemia caused by different genospecies of the ACB complex. In this retrospective cohort study, we compared the clinical characteristics, the risk factors and the prognostic factors associated with bacteraemia caused by A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3 that had been identified using oligonucleotide array sequence analysis of ITS sequences. The antimicrobial susceptibility profiles of the isolates were also compared.

Methods

Setting and bacterial isolates

This study was retrospectively conducted at the National Taiwan University Hospital (NTUH), a 2200 bed tertiary care centre in northern Taiwan. Patients who had fever, sepsis syndrome and blood cultures (at least two sets) positive for the ACB complex at NTUH during the period October 2007 to December 2008 were included. The clinical features, prognostic factors and mortality rates were compared among patients with bacteraemia caused by different genospecies of the ACB complex.

Genospecies identification

ACB complex isolates were presumptively identified by colony morphology, Gram staining, growth at 37°C, negative oxidase test and oxidation of glucose. The Phoenix bacterial identification system (Becton-Dickinson Diagnostic Instrument Systems, Sparks, MD, USA) was then used to confirm the identity of the isolates. An array with six oligonucleotide probes based on 16S–23S rRNA gene ITS regions was used for species and genospecies identification.¹⁰ The identified genospecies were classified as A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3. Patients with bacteraemia caused by mixed genospecies and unidentified species were excluded.

Antimicrobial susceptibility testing

Susceptibility to and MICs of 12 agents (ampicillin/sulbactam, ticarcillin/clavulanate, piperacillin/tazobactam, ceftazidime, cefepime, meropenem, imipenem, doripenem, amikacin, gentamicin, ciprofloxacin and levofloxacin) were determined using the agar dilution method according to CLSI recommendations.¹² MICs of colistin and tigecycline were determined by the broth microdilution method according to the CLSI guidelines.¹² Susceptibility to tigecycline was determined by US FDA breakpoints for Enterobacteriaceae.¹³ Isolates with tigecycline MIC values ≤2 mg/L were considered susceptible.

PFGE

Extraction and purification of DNA from the 77 Acinetobacter gen. sp. 13TU and 21 Acinetobacter gen. sp. 3 isolates were carried out as described previously.¹⁴ DNA was digested by the restriction enzyme SmaI and the restriction fragments were separated in a CHEF-DR III unit (Bio-Rad Laboratories, Hercules, CA, USA) at 200 V for 27 h. The PFGE results were analysed with Gelcompar for Windows Version 3.1. Isolates with a pattern similarity >87% were considered to be derived from a cluster (closely related strains), indicating the possible presence of an outbreak.^15,16

Clinical characteristics and outcome analysis

We reviewed the medical records of all patients to collect clinical information, including baseline demographic data to calculate the Charlson co-morbidity index,¹⁷ Pitt bacteraemia score,¹⁸ length of stay before bacteraemia, place of acquisition, source of bacteraemia, antimicrobial susceptibilities, initial and definitive antimicrobial therapy, appropriateness of initial and definitive antimicrobial therapy, as well as 14 day, 30 day and in-hospital mortality.

Definitions

Co-morbidity at diagnosis was classified using the Charlson co-morbidity index.¹² The severity of bacteraemia was evaluated by the Pitt bacteraemia score.¹³ The episode of bacteraemia was considered to be community acquired if the bacteraemia occurred before admission and hospital acquired if the bacteraemia was noted after hospitalization. The source of bacteraemia was defined according to the Centers for Disease Control and Prevention.¹⁹ If no infectious focus was identified, the bacteraemia was counted as primary. Polymicrobial bacteraemia was defined as bacteraemia caused by microorganisms other than Acinetobacter spp. Multidrug resistance was defined as resistance to three or more antimicrobial classes, including ampicillin/sulbactam, antipseudomonal cephalosporins (ceftazidime or cefepime), carbapenem (meropenem or imipenem), fluoroquinolones (ciprofloxacin or levofloxacin) and aminoglycosides (amikacin or gentamicin). Carbapenem resistance was defined as resistance to meropenem or imipenem. Appropriate initial antimicrobial therapy was defined as treatment with in vitro active agents before the availability of susceptibility test results. Appropriate definitive antimicrobial therapy was defined as treatment with in vitro active agents after the availability of susceptibility test results.

Statistics

Statistical analyses were performed using the Statistical Package for the Social Sciences for Windows (Version 18.0, SPSS Inc., Chicago, IL, USA). Continuous variables are reported as the mean ± SD and were compared with one-way analysis of variance. The Scheffe procedure was used for post hoc analysis. Categorical variables are expressed as percentages of the total number of patients analysed and were compared with the χ² test or Fisher's exact test, as appropriate. Post hoc analysis using the χ² test or Fisher's exact test used a modified Bonferroni-adjusted α for pair-wise comparisons if the result of the χ² test or Fisher's exact test was statistically significant. For prognostic factors associated with 30 day mortality, variables with a P value <0.2 in the univariate analysis were analysed by the multivariate logistic regression method. A P value <0.05 was considered to indicate statistical significance.

Results

Of the 247 non-duplicate blood isolates identified as belonging to the ACB complex by routine laboratory procedures during the study period, 118 isolates (47.8%) were A. baumannii, 79 isolates (32.0%) were Acinetobacter gen. sp. 13TU and 22 isolates (8.9%) were Acinetobacter gen. sp. 3. Four patients (one with bacteraemia due to A. baumannii, two due to Acinetobacter gen. sp. 13TU and one caused by Acinetobacter gen. sp. 3) were excluded from the analysis because they had not been admitted to the hospital. A total of 215 patients with fever and sepsis syndrome were included in the final analysis (117 with A. baumannii, 77 with Acinetobacter gen. sp. 13TU and 21 with Acinetobacter gen. sp. 3 bacteraemia).

The demographic characteristics of the patients with bacteraemia are shown in Table 1. There was no significant difference in gender (P = 0.163) or Charlson co-morbidity index (P = 0.251) among the three groups of patients. The patients with bacteraemia caused by Acinetobacter gen. sp. 3 tended to be younger than those with bacteraemia caused by Acinetobacter gen. sp. 13TU (P = 0.046). There was no significant difference in individual items of the Charlson co-morbidity index, except for a lower prevalence of leukaemia in patients with bacteraemia caused by Acinetobacter gen. sp. 13TU (P = 0.006).

Table 1.

Comparison of demographic characteristics, Charlson co-morbidity index and underlying diseases of patients with bacteraemia caused by different genospecies of A. calcoaceticus–A. baumannii complex treated at National Taiwan University Hospital from October 2007 to December 2008

Demographic characteristics	A. baumannii, n = 117, n (%)	Acinetobacter genospecies 13TU, n = 77, n (%)	Acinetobacter genospecies 3, n = 21, n (%)	P value
Age, years (mean ± SD)	57.79 ± 19.505	62.34 ± 17.862	50.71 ± 18.762	0.033
Gender
male	74 (63.2)	42 (54.5)	16 (76.2)	0.163
female	43 (36.8)	35 (45.5)	5 (23.8)
Charlson co-morbidity index (mean ± SD)	3.96 ± 2.408	4.58 ± 2.706	4.43 ± 3.429	0.251
Co-morbidities
myocardial infarction	3 (2.6)	3 (3.9)	0 (0.0)	0.615
congestive heart failure	6 (5.1)	6 (7.8)	3 (14.3)	0.298
peripheral artery occlusive disease	2 (1.7)	2 (2.6)	0 (0.0)	0.726
cerebral vascular accident	1 (0.9)	3 (3.9)	0 (0.0)	0.247
dementia	5 (4.3)	1 (1.3)	0 (0.0)	0.336
chronic lung disease	11 (9.4)	3 (3.9)	0 (0.0)	0.14
connective tissue disease	7 (6.0)	3 (3.9)	0 (0.0)	0.451
peptic ulcer disease	9 (7.7)	8 (10.4)	4 (19.0)	0.265
mild liver disease	13 (11.1)	12 (15.6)	3 (14.3)	0.653
diabetes mellitus	20 (17.1)	14 (18.2)	4 (19.0)	0.967
hemiplegia	14 (12.0)	8 (10.4)	2 (9.5)	0.914
renal disease, moderate to severea	26 (22.2)	15 (19.5)	5 (23.8)	0.866
DM with end-organ disease	8 (6.8)	10 (13.0)	1 (4.8)	0.265
tumour without metastasis	17 (14.5)	16 (20.8)	5 (23.8)	0.397
leukaemia	16 (13.7)	1 (1.3)	4 (19.0)	0.006
lymphoma	4 (3.4)	5 (6.5)	3 (14.3)	0.124
liver disease, moderate to severeb	10 (8.5)	4 (5.2)	3 (14.3)	0.365
tumour with metastasis	28 (23.9)	29 (37.7)	5 (23.8)	0.103
AIDS	3 (2.6)	0 (0.0)	0 (0.0)	0.28

DM, diabetes mellitus.

P values marked in bold if <0.05.

^aSevere renal disease included patients on dialysis, those who had a transplant and those with uraemia. Moderate renal disease included patients with serum creatinine >3 mg/dL.

^bSevere liver disease was defined as cirrhosis, portal hypertension and a history of variceal bleeding. Moderate liver disease was defined as cirrhosis with portal hypertension, but without bleeding. Mild liver disease was defined as cirrhosis without portal hypertension or chronic hepatitis.

Table 2 shows the clinical manifestations and treatment outcomes of patients with bacteraemia due to each of the three different genospecies of the ACB complex. There was no significant difference in the length of hospital stay before bacteraemia (P = 0.265), prevalence of community- or hospital-acquired infection (P = 0.301 for both), central venous catheter use (P = 0.255) or polymicrobial bacteraemia (P = 0.409). A. baumannii bacteraemia occurred more frequently in the intensive care unit (ICU) (P = 0.001). In addition, patients with bacteraemia caused by A. baumannii had higher Pitt bacteraemia scores (P < 0.001), were more likely to be receiving mechanical ventilation (P < 0.001) and were more likely to have bacteraemic pneumonia (P < 0.001). In contrast, the prevalence of primary bacteraemia was significantly lower in patients with A. baumannii bacteraemia (P = 0.004).

Table 2.

Comparison of length of hospital stay before onset, place of acquisition, Pitt bacteraemia score, source of bacteraemia, treatment and outcomes of patients with bacteraemia caused by different genospecies of A. calcoaceticus–A. baumannii complex treated at National Taiwan University Hospital from October 2007 to December 2008

				Statistical analysis (P value)
Clinical manifestations	A. baumannii, n = 117, n (%)	Acinetobacter genospecies 13TU, n = 77, n (%)	Acinetobacter genospecies 3, n = 21, n (%)	AB versus 13TU	13TU versus G3	AB versus G3	Overall
Length of stay before bacteraemia (mean ± SD)	22.85 ± 30.128	19.99 ± 18.979	13.33 ± 12.031				0.265
Acquisition place
community	17 (14.5)	7 (9.1)	1 (4.8)				0.301
hospital	100 (85.5)	70 (90.9)	20 (95.2)				0.301
ICU onset	57 (48.7)	25 (32.5)	2 (9.5)	0.027	0.052	0.001	0.001
Pitt bacteraemia score	3.98 ± 3.081	2.21 ± 2.086	1.21 ± 1.988	<0.001	0.348	<0.001	<0.001
hypotension	58/116 (50.0)	28/77 (36.4)	4/19 (21.1)	0.076	0.28	0.025	0.024
use of mechanical ventilator	<0.001	<0.001	1/19 (5.3)	<0.001	0.18	<0.001	<0.001
Use of central venous catheter	92 (78.6)	58 (75.3)	13 (61.9)				0.255
Source of bacteraemia
primary	11 (9.4)	18 (23.4)	7 (33.3)	0.012	0.401	0.008	0.004
CRBSI	69 (59.0)	47 (61.0)	12 (57.1)				0.934
pneumonia	41 (35.0)	10 (13.0)	0 (0.0)	<0.001	0.113	<0.001	<0.001
urinary tract infection	4 (3.4)	0 (0.0)	0 (0.0)				0.181
intra-abdominal infection	4 (3.4)	1 (1.3)	0 (0.0)				0.479
wound	3 (2.6)	1 (1.3)	0 (0.0)				0.654
Polymicrobial bacteraemia	47 (40.2)	25 (32.5)	6 (28.6)				0.409
Treatment
appropriate initial antimicrobials	42/117 (35.9)	56/77 (72.7)	12/20 (60.0)	<0.001	0.284	0.050	<0.001
appropriate definitive antimicrobial therapy	61/89 (68.5)	60/68 (88.2)	14/20 (70.0)	0.004	0.078	1.000	0.013
time to appropriate antibiotic therapy  (mean ± SD)	1.37 ± 1.779	0.59 ± 1.162	0.53 ± 1.060	0.009	0.991	0.007	0.005
Outcome—mortality
14 day	41 (35.0)	11 (14.3)	1 (4.8)	0.002	0.452	0.004	<0.001
30 day	50 (42.7)	12 (15.6)	3 (14.3)	<0.001	1.000	0.015	<0.001
in-hospital	58 (49.6)	18 (23.4)	5 (23.8)	<0.001	1.000	0.034	<0.001

AB, A. baumannii; 13TU, Acinetobacter gen. sp. 13TU; G3, Acinetobacter gen. sp. 3; CRBSI, catheter-related bloodstream infection.

P values marked in bold if <0.05.

The antimicrobial susceptibility test results are listed in Table 3. A. baumannii isolates had significantly lower susceptibility rates than the other two genospecies isolates for the antimicrobials tested, including tigecycline (P ≤ 0.001), but not colistin (P < 0.001). All A. baumannii isolates were susceptible to colistin (MIC ≤ 2 mg/L), but isolates of Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3 were less susceptible to that agent [93.5%, (MIC range 1–32 mg/L) and 95.2% (MIC range 1–16 mg/L), respectively]. In contrast, A. baumannii isolates were less susceptible to tigecycline (MIC range 0.06–32 mg/L) than Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3 isolates (P ≤ 0.001). The proportion of multidrug resistance and carbapenem resistance was higher among A. baumannii isolates than among Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3 isolates (P < 0.001).

Table 3.

Comparison of antimicrobial susceptibility of A. calcoaceticus–A. baumannii complex isolates of different genospecies from patients with bacteraemia at National Taiwan University Hospital from October 2007 to December 2008

	Susceptibility, n (%)			Statistical analysis (P value)
	A. baumannii, n = 115,an (%)	Acinetobacter genospecies 13TU, n = 77, n (%)	Acinetobacter genospecies 3, n = 21, n (%)	AB versus 13TU	13TU versus G3	AB versus G3	Overall
Ampicillin/sulbactam	43 (37.4)	63 (81.8)	21 (100.0)	<0.001	0.036	<0.001	<0.001
Ticarcillin/clavulanate	38 (33.0)	68 (88.3)	16 (76.2)	<0.001	0.172	<0.001	<0.001
Piperacillin/tazobactam	44 (38.3)	64 (83.1)	17 (81.0)	<0.001	0.756	0.001	<0.001
Ceftazidime	42 (36.5)	67 (87.0)	18 (85.7)	<0.001	1.000	<0.001	<0.001
Cefepime	41 (35.7)	69 (89.6)	17 (81.0)	<0.001	0.280	<0.001	<0.001
Meropenem	65 (56.5)	73 (94.8)	21 (100.0)	<0.001	0.575	<0.001	<0.001
Imipenem	65 (56.5)	73 (94.8)	21 (100.0)	<0.001	0.575	<0.001	<0.001
Doripenem	64 (55.7)	73 (94.8)	21 (100.0)	<0.001	0.575	<0.001	<0.001
Amikacin	46 (40.0)	73 (94.8)	20 (95.2)	<0.001	1.000	<0.001	<0.001
Gentamicin	43 (37.4)	70 (90.9)	16 (76.2)	<0.001	0.125	0.001	<0.001
Ciprofloxacin	41 (35.7)	72 (93.5)	20 (95.2)	<0.001	1.000	<0.001	<0.001
Levofloxacin	42 (36.5)	73 (94.8)	20 (95.2)	<0.001	1.000	<0.001	<0.001
Colistin	115 (100.0)	72 (93.5)	20 (95.2)	<0.001	1.000	0.023	<0.001
Tigecycline	79 (68.7)	75 (97.4)	21 (100.0)	<0.001	1.000	0.001	<0.001
Resistance profiles
multidrug resistanceb	73 (63.5)	7 (9.1)	1 (4.8)	<0.001	1.000	<0.001	<0.001
carbapenem resistance	51 (44.3)	4 (5.2)	0 (0.0)	<0.001	0.575	<0.001	<0.001

AB, A. baumannii; 13TU, Acinetobacter gen. sp. 13TU; G3, Acinetobacter gen. sp. 3.

P values marked in bold if <0.05.

^aTotal 115 isolates of A. baumannii were available for susceptibility test.

^bMultidrug resistance was defined as resistance to three or more of the following classes of antimicrobials: antipseudomonal cephalosporins, antipseudomonal carbapenems, ampicillin/sulbactam, fluoroquinolones and aminoglycosides.

The treatment and outcomes of patients with bacteraemia caused by different genospecies of ACB complex are compared in Table 2. Appropriate initial and definitive antimicrobial therapy was administered more frequently to patients with Acinetobacter gen. sp. 13TU bacteraemia than to patients with A. baumannii bacteraemia (P < 0.001 and P = 0.004, respectively). Time to appropriate antimicrobial therapy was longer among patients with A. baumannii bacteraemia than among those with Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3 bacteraemia (P = 0.009 and P = 0.007, respectively). In addition, A. baumannii bacteraemia was associated with higher 14 day, 30 day and in-hospital mortality than Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3 bacteraemia [30 day mortality: 42.7% versus 15.6% (P < 0.001) and 42.7% versus 14.3% (P = 0.015), respectively].

Table 4 shows the results of the univariate and multivariate analyses of prognostic factors for 30 day mortality of patients with ACB complex bacteraemia. In the univariate analysis, A. baumannii, Charlson co-morbidity index, Pitt bacteraemia score, length of stay before bacteraemia, ICU onset, primary bacteraemia, multidrug resistance and administration of appropriate initial antimicrobials were associated with 30 day mortality. The independent prognostic factors for 30 day mortality identified in the multivariate logistic regression analysis included the Charlson co-morbidity index [adjusted odds ratio (aOR) 1.385, 95% confidence interval (CI) 1.176–1.631; P < 0.001)] and Pitt bacteraemia score (aOR 1.554, 95% CI 1.296–1.864; P < 0.001). Multidrug resistance tended to be associated with a poor prognosis, although there was no significant association (aOR 2.830, 95% CI 0.923–8.679; P = 0.069).

Table 4.

Multivariate logistic regression analysis of prognostic factors associated with 30 day mortality in patients with A. calcoaceticus–A. baumannii complex bacteraemia

	Survivors	Non-survivors	Univariate analysis			Multivariate analysis
	n = 150, n (%)	n = 65, n (%)	aOR (95% CI)		P value	aOR (95% CI)		P value
Age, years (mean ± SD)	57.37 ± 19.795	61.85 ± 17.040	1.013	(0.997–1.030)	0.116	0.996	(0.974–1.018)	0.718
Charlson co-morbidity index	3.87 ± 2.638	5.05 ± 2.452	1.187	(1.059–1.331)	0.003	1.385	(1.176–1.631)	<0.001
Pitt bacteraemia score	2.14 ± 2.194	5.30 ± 2.980	1.566	(1.367–1.795)	<0.001	1.554	(1.296–1.864)	<0.001
Intensive care unit onset	45 (30.0)	38 (58.5)	3.284	(1.794–6.010)	<0.001	1.363	(0.513–3.621)	0.534
Length of stay before bacteraemia (days)	18.47 ± 23.17	26.48 ± 29.154	1.012	(1.000–1.023)	0.043	1.001	(0.987–1.015)	0.941
Primary bacteraemia	31 (20.7)	5 (7.7)	0.320	(0.118–0.865)	0.025	0.831	(0.293–2.888)	0.771
Catheter-related bloodstream infection	86 (57.3)	42 (64.6)	1.359	(0.744–2.483)	0.319
Pneumonia	24 (16.0)	27 (41.5)	3.730	(1.930–7.208)	<0.001	0.970	(0.377–2.499)	0.950
Polymicrobial bacteraemia	51 (34.0)	27 (41.5)	1.379	(0.759–2.508)	0.292
Multidrug resistancea	40 (26.7)	43 (66.2)	5.375	(2.867–10.076)	<0.001	2.830	(0.923–8.679)	0.069
Appropriate initial antimicrobials	84 (56.4)	26 (40.0)	0.516	(0.285–0.516)	0.029	0.668	(0.270–1.665)	0.384
Appropriate definitive antimicrobials	111/147 (75.5)	24/30 (80)	1.297	(0.492–3.423)	0.599
Time to appropriate antibiotic therapy (days)	0.90 ± 1.441	1.08 ± 1.730	1.080	(0.856–1.363)	0.518
Acinetobacter baumannii	67 (44.7)	50 (76.9)	4.129	(2.133–7.995)	<0.001	1.804	(0.697–4.673)	0.224

P values marked in bold if <0.05.

^aMultidrug resistance was defined as resistance to three or more of the following classes of antimicrobials: antipseudomonal cephalosporins, antipseudomonal carbapenems, ampicillin/sulbactam, fluoroquinolones and aminoglycosides.

Among the 77 Acinetobacter gen. sp. 13TU isolates, PFGE patterns of 50 groups (clusters) had ≥87% similarity. Twenty-seven of these groups contained more than one isolate and no clustering was found. Among the 21 Acinetobacter gen. sp. 3 isolates, all exhibited different PFGE patterns, indicating the absence of clustering of the isolates.

Discussion

In this study, bacteraemia due to MDR strains but not A. baumannii per se appears to be associated with a poor prognosis for survival. The results of two epidemiological studies conducted in Taiwan suggested that A. baumannii was a poor prognostic factor for sepsis-related mortality.^20,21 Our patient population had higher heterogeneity (hospitalized patients versus ICU patients) than the population studied by Chuang et al.,²¹ which may influence the diversity of underlying diseases and the severity of ACB complex bacteraemia. In addition, polymicrobial bacteraemia was more common in our study than in the study by Lee et al.²⁰ Furthermore, we applied a more stringent inclusion criterion in our study; only patients with two sets of documented bacteraemia were recruited. However, Lee et al.,²² who did not identify the genospecies of Acinetobacter, found that the mortality rate among patients with MDR ACB complex bacteraemia was higher than that among patients with non-MDR ACB complex bacteraemia. Therefore, it is reasonable to assume that multidrug resistance influenced the clinical outcome in the current study.

Antimicrobial resistance is common among A. baumannii isolates, thereby limiting treatment options.^3,4,23,24 In the current study, A. baumannii, multidrug resistance and appropriate antimicrobial therapy were associated with 30 day mortality in the univariate analysis, but were shown not to be independent predictors of 30 day mortality in the multivariate analysis. This result could be due to a contributory or confounding effect among these three factors. The analysis, however, revealed that the severity of underlying co-morbidities (Charlson co-morbidity index) and bacteraemia (Pitt bacteraemia score) were independent predictors of poor outcome.

Bacteraemia caused by A. baumannii was associated with higher 30 day mortality than bloodstream infections due to Acinetobacter gen. sp. 13TU or Acinetobacter gen. sp. 3. Previous retrospective matched cohort and case–control studies^25–28 found that mortality attributable to nosocomial ACB complex bacteraemia ranged from 7.8% to 36.5%. This wide variability in clinical prognosis might be related to a heterogeneous study population without genospecies identification. In the present study, A. baumannii bacteraemia was associated with higher 30 day mortality and in-hospital mortality than bacteraemia due to the other two species of bacteria in the ACB complex (42.7% versus 15.6%, 49.6% versus 23.4%, respectively). Therefore, after genospecies identification, the crude mortality of patients with A. baumannii bacteraemia may be higher than previously considered.

McDonald et al.²⁹ reported the persistence and clonal spread of a single strain of Acinetobacter gen. sp. 13TU in the intensive therapy unit of a large Scottish teaching hospital. Lee et al.³⁰ identified three major clones (A, B and C) of imipenem-resistant Acinetobacter gen. sp. 13TU by PFGE analysis, and found that the PFGE patterns of the A and B clones were similar and accounted for the clonal spread in a hospital; although, the number of isolates (14) was small. In two of our previous studies, we provided evidence of the clonal spread of A. baumannii isolates, particularly MDR isolates, in our hospital.^31,32 The clonal spread of A. baumannii isolates has also been demonstrated by recent molecular epidemiological surveillance of A. baumannii isolates (data not shown). However, in the current study, PFGE analysis revealed no clustering of Acinetobacter gen. sp. 13TU isolates and no strain dissemination of Acinetobacter gen. sp. 3 isolates.

In the present study, Acinetobacter gen. sp. 13TU and gen. sp. 3 isolates were less susceptible to colistin than A. baumannii. In total, 5 (6.5%) isolates of Acinetobacter gen. sp. 13TU and 1 (4.8%) isolate of Acinetobacter gen. sp. 3 were resistant to colistin (MIC ≥ 2 mg/L). The emergence of colistin-resistant Acinetobacter spp. isolates has been noted in South Korea as well.³³ Park et al.³³ found that Acinetobacter gen. sp. 13TU isolates accounted for 34 of 54 Acinetobacter spp. isolates with colistin resistance. Adams et al.³⁴ reported that resistance to colistin in A. baumannii was associated with mutations in the pmrA and pmrB two-component system. However, resistance mechanisms in different genospecies have yet to be investigated.

This study had several limitations. First, erroneous records and missing data in medical records can occur in any retrospective study, despite detailed review. Second, the current study could only evaluate the crude mortality of patients with ACB complex bacteraemia. Attributable mortality needs to be addressed in a prospective study.

In conclusion, there were significant variations in the antimicrobial susceptibility patterns, clinical features and outcomes among patients with bacteraemia due to A. baumannii, Acinetobacter gen. sp. 13TU and Acinetobacter gen. sp. 3. We found that bacteraemia due to MDR strains but not A. baumannii per se appears to be associated with poor outcome. The association between bacteraemia due to these three different genospecies of the ACB complex and patient outcome needs further investigation.

Funding

This study was supported by internal funding.

Transparency declarations

None to declare.