Is Polymicrobial Bacteremia an Independent Risk Factor for Mortality in Acinetobacter baumannii Bacteremia?

This retrospective observational study assessed the differences between monomicrobial and polymicrobial A. baumannii bacteremia and identified possible independent risk factors for 14-day mortality. There were 379 patients with A. baumannii bacteremia admitted to a tertiary care center in northern Taiwan between August 2008 and July 2015 enrolled for data analysis. Among them, 89 patients (23.5%) had polymicrobial bacteremia and 290 patients (76.5%) had monomicrobial bacteremia. No significant difference in 14-day mortality was observed between patients with monomicrobial and polymicrobial A. baumannii bacteremia (26.9% vs. 29.2%, p = 0.77). Logistic regression controlled for confounders demonstrated that polymicrobial bacteremia was not an independent predictor of mortality, whereas appropriate antimicrobial therapy was independently associated with reduced mortality. Higher 14-day mortality rates were observed in the polymicrobial bacteremic patients with concomitant isolation of Escherichia coli, Pseudomonas aeruginosa, and Enterobacter spp. from the bloodstream. Compared with patients with monomicrobial multidrug-resistant A. baumannii (MDRAb) bacteremia, those with MDRAb concomitant with Gram-negative bacilli bacteremia had a worse outcome. Polymicrobial A. baumannii bacteremia was not associated with a higher 14-day mortality rate than that of monomicrobial A. baumannii bacteremia, although more deaths were observed when certain Gram-negative bacteria were concomitantly isolated. Appropriate antimicrobial therapy remains an important life-saving measure for A. baumannii bacteremic patients.

Acinetobacter baumannii is an emergent nosocomial pathogen with high morbidity and mortality rates [8][9][10]. Bacteremia caused by A. baumannii occurs as polymicrobial infections in 19%-35% of cases [6,9,11,12]. Our previous study found that carbapenem-resistant A. baumannii (CRAb) provided a sheltering effect for carbapenem-susceptible pathogens and enhanced the pathogenesis of polymicrobial infection [13]. However, no study has evaluated the impact of polymicrobial A. baumannii bacteremia on clinical outcomes. Therefore, this retrospective observational study assessed whether polymicrobial infection was an independent risk factor for 14-day mortality in patients with A. baumannii bacteremia.

Study Population
This retrospective study was conducted at Taipei Veterans General Hospital (T-VGH), a 2900-bed tertiary-care teaching hospital located in Taipei, Taiwan, during a seven-year period from August 2008 to July 2015. Charts were reviewed for all patients with at least one blood culture positive for Acinetobacter calcoaceticus-Acinetobacter baumannii (Acb) complex accompanied by two or more of the following symptoms and signs of infection: fever, hypothermia, tachypnea, tachycardia, leukocytosis, or leukopenia. We included only the first blood culture for patients with two or more positive blood cultures. Patients under 20 years of age and those with incomplete medical records were excluded. The protocol was approved by the institutional review boards (IRBs) of T-VGH (IRB No. 2016-05-019CC).

Data Collection
We reviewed the medical records to retrieve clinical information, including demographic characteristics, comorbidities, duration of intensive care unit (ICU) and hospital stays, time of receipt, dose and route of therapy with individual antimicrobial drugs, receipt of invasive procedures, time of bacteremia onset, and infectious foci. The onset of bacteremia was defined as the day when the blood culture that eventually yielded A. baumannii was drawn. Polymicrobial bacteremia was defined as the simultaneous isolation of A. baumannii and one or more microorganisms from the blood during the same bacteremic episode. However, Corynebacterium spp., Bacillus spp., Propionibacterium acnes, and skin flora (e.g., Micrococcus spp., Streptococcus viridans, and coagulase-negative staphylococci) were considered contaminants and not included in the analyses unless they were related to device infection and grown in two or more blood cultures [6,14]. Episodes of bloodstream infection were defined as ICU-acquired bacteremia if they appeared 48 h after ICU admission. Chronic kidney disease was defined as an estimated glomerular filtration rate <60 mL/min/1.73 m 2 . Neutropenia was defined as an absolute neutrophil count <0.5 × 10 9 neutrophils/L. Immunosuppressive therapy was defined as the receipt of cytotoxic agents within six weeks [15], corticosteroids at a dosage equivalent to or higher than 15 mg of prednisolone daily for at least seven days within four weeks, or other immunosuppressive agents within two weeks before bacteremia onset [15,16]. A history of recent surgery was defined as receipt of operations within four weeks before bacteremia onset. The severity of infection was evaluated using the Acute Physiology and Chronic Health Evaluation (APACHE) II score [17] within 24 h before bacteremia onset. The primary infection source of bacteremia was determined according to the definitions from the Centers for Disease Control and Prevention [18]. Appropriate antimicrobial therapy was defined as the administration of at least one antimicrobial agent to which the causative pathogen was susceptible in vitro within 48 h after the onset of bacteremia, with an approved route and dosage appropriate for end organ(s) function. In patients with polymicrobial bacteremia, a combination of antimicrobial agents covering all isolated microbes was deemed appropriate. Monotherapy with an aminoglycoside was not considered appropriate therapy for A. baumannii bacteremia. The primary outcome measure was all-cause 14-day mortality following bacteremia onset. Fourteen-day mortality was chosen as the endpoint to allow adequate ascertainment of treatment response. We reasoned that 30 days is too long for critically ill patients, as there are many competing causes of death, and seven days is too short a time to witness a response to treatment.

Microbiological Studies
Bacteria were phenotypically identified to Acb complex using a Vitek 2 system (bioMérieux, Marcy l'Etoile, France). Other bacterial species were also identified using this system. All isolates were regrown from -80 • C storage, identified to the species level, and tested for their susceptibility to various antimicrobials. A multiplex polymerase chain reaction (PCR) method was used to identify A. baumannii to the genomic species level [19]. Antimicrobial susceptibilities of A. baumannii and other concomitant isolated bacterial pathogens were determined by the agar dilution method [20] and interpreted according to the Clinical Laboratory Standards Institute (CLSI) criteria [21].
Multidrug resistance was defined as resistance to three or more of the following classes of antimicrobials: anti-pseudomonal cephalosporins, anti-pseudomonal carbapenems, ampicillin/sulbactam, fluoroquinolones, and aminoglycosides. Carbapenem resistance was defined as resistance to imipenem, meropenem, or doripenem.

Statistical Analysis
Chi-square tests with Yates correction or Fisher's exact tests were used to compare categorical variables; Student's t or Mann-Whitney rank-sum tests were used to analyze continuous variables, as appropriate. Logistic regression models were used to explore independent risk factors of 14-day mortality. We performed univariate analyses separately for each risk factor to ascertain the odds ratios (ORs) and 95% confidence intervals (CIs). All biologically plausible variables with a p <0.20 in the univariate analysis were considered for inclusion in the logistic regression model in the multivariable analysis. A backward selection process was utilized. Interactions between APACHE II score and the covariates were also examined in the logistic regression model. Time to mortality, defined as the interval between bacteremia onset and death, was analyzed by Kaplan-Meier survival analysis with log-rank tests. p < 0.05 was considered statistically significant. SPSS Statistics for Windows, version 22.0 (IBM Corp., Armonk, NY, USA) was used for all data analyses.

Results
During the seven-year study period, we enrolled 732 patients with at least one set of blood cultures positive for Acb complex. A total of 379 patients met the inclusion criteria, of which 89 (23.5%) patients had polymicrobial A. baumannii bacteremia. The demographic and clinical characteristics of patients with polymicrobial and monomicrobial A. baumannii bacteremia are summarized in Table 1. Patients with polymicrobial A. baumannii bacteremia were older, more likely to have coronary artery disease, less frequently experienced shock within three days of bacteremia onset, and more frequently received inappropriate antimicrobial therapy (13.5% vs. 38.6%, p < 0.001), compared with those in patients with monomicrobial bacteremia. No significant difference in 14-day mortality (29.2% vs. 26.9%, p = 0.770) and 30-day mortality (38.2% vs. 34.8%, p = 0.561) was observed between these two groups. To analyze the duration of in-hospital stay after the onset of bacteremia, we separated the population into two groups, survivors and non-survivors, at hospital discharge. In the patients who survived to discharge, those with polymicrobial A. baumanii bacteremia and monomicrobial A. baumanii bacteremia had a comparable hospital stay after the onset of bacteremia (21.9  vs. 21.9 [9-34] days, p = 0.997). In terms of the patients who died in hospital, those with polymicrobial A. baumanii bacteremia and monomicrobial A. baumanii bacteremia also had a comparable hospital stay after the onset of bacteremia (13.9 [0.75-26.5] vs. 13.7  days, p = 0.944). The 30-day re-admission rates of survived patients with polymicrobial and monomicrobial A. baumannii bacteremia were not significantly different (17.9% and 13.2%, respectively; p = 0.620). Logistic regression analysis was performed to determine the risk factors for 14-day mortality in the entire population with A. baumannii bacteremia, as shown in Table 2. Polymicrobial bacteremia was not independently associated with increased 14-day mortality. Malignancy, shock, and higher APACHE II score were independent predictors of 14-day mortality, whereas appropriate antimicrobial therapy was independently associated with reduced 14-day mortality (p = 0.001). The predictors for 30-day mortality among patients with A. baumannii bacteremia were analyzed in the same fashion as the above analyses. The results were similar to those for prediction of 14-day mortality (Table S1). We stratified the 89 patients with polymicrobial A. baumannii bacteremia according to the concomitantly isolated bacterial species (Table 3). Coagulase-negative staphylococci were the most common isolated microorganisms (20 cases). Among them, 17 patients (85.0%) were caused by a device-related infection. Nine patients (52.9%) had their infected device removed at a median duration of six days (interquartile range, 4-8) after the onset of the bacteremia. However, the removal of device did not affect their clinical outcome. There were 12 patients with polymicrobial A. baumannii bacteremia that received appropriate antimicrobial therapy. Half of them received combination therapy. Carbapenem plus vancomycin/teicoplanin was the most frequently used antimicrobial regimen for those with polymicrobial A. baumannii bacteremia (Table S2).
Although no significant difference in 14-day mortality rates was observed between patients with polymicrobial and monomicrobial A. baumannii bacteremia, the rate in polymicrobial A. baumannii bacteremia varied according to the concomitant pathogen. Concomitant Escherichia coli bacteremia had the highest 14-day mortality rate (71.4%), followed by Pseudomonas aeruginosa (50.0%) and Enterobacter spp. bacteremia (37.5%). The 14-day mortality rate was higher in patients who had polymicrobial A. baumannii bacteremia with concomitant Gram-negative bacilli (GNB) than in patients with concomitant Gram-positive cocci (GPC) ( Table 3; 38.9% vs. 19.1%, p = 0.053) (Figure 1; p = 0.038, log-rank test). The rates of methicillin-resistant Staphylococcus aureus (MRSA), extended-spectrum beta-lactamase (ESBL)-producing K. pneumoniae, ESBL-producing E. coli, and AmpC-producing E. coli were 80.0%, 16.7%, 14.3%, and 14.3%, respectively (Table S3). The presence of these resistant bacteria was not associated with significantly higher 14-day mortality.     To investigate the influence of A. baumannii antibiotic resistance on the mortality of patients with polymicrobial bacteremia with concomitant GNB infection, we extracted data of those with multidrug-resistant A. baumannii (MDRAb) or carbapenem-resistant A. baumannii (CRAb) infections for further analysis. Patients with MDRAb and concomitant GNB bacteremia had a higher 14-day mortality rate than patients with monomicrobial MDRAb bacteremia (Figure 2; p = 0.036, log-rank test). Patients with CRAb and concomitant GNB bacteremia had a higher 14-day mortality rate than patients with monomicrobial CRAb bacteremia (50.0% vs. 30.6%, p = 0.285). The difference was not statistically significant in this small cohort (n = 82). The survival analysis at 30 days was analyzed in the same fashion as the above analyses ( Figures S1  and S2). The results were similar to those for 14-day mortality.

Discussion
While polymicrobial A. baumannii bacteremia is not uncommon, its clinical significance has not previously been elucidated. This retrospective study assessed the differences between monomicrobial and polymicrobial A. baumannii bacteremia and identified possible independent risk factors for 14-day mortality for these infections. We found that polymicrobial bacteremia was not an independent predictor of mortality, whereas appropriate antimicrobial therapy was independently associated with reduced mortality.
This study, in agreement with previous studies [6,9,11,12], observed that approximately one-fourth of patients with A. baumannii bloodstream infections presented with polymicrobial bacteremia. Owing to its high prevalence, investigation of the clinical significance of polymicrobial A. baumannii bacteremia is warranted. These patients were older and more likely to have coronary artery disease and experienced shock at the time of bacteremia. Moreover, as few as one out of seven patients received appropriate antimicrobial therapy directed against every microbial species isolated from the blood. As the clinical characteristics differed between monomicrobial and polymicrobial A. baumannii bacteremia, studies on the clinical outcomes of patients with monomicrobial and polymicrobial A. baumannii bacteremia should evaluate the infections separately. Furthermore, when evaluating the efficacy of various regimens on the clinical outcome of A. baumannii bacteremia, polymicrobial A. baumannii bacteremia should be excluded, unless the antibiotic efficacy against the concomitant isolated microorganisms was also evaluated.
Our results showed that the 14-day mortality rate was not significantly higher in patients with polymicrobial than in patients with monomicrobial A. baumannii bacteremia. This finding is consistent with those of previous studies and suggests that polymicrobial infection was not an independent risk factor for mortality of A. baumannii bacteremia [10,11,22,23]. Although several studies have reported higher mortality rates in patients with polymicrobial bacteremia [3,5,6,24], the attributable mortality rate of polymicrobial bacteremia varied among causative pathogens [3,25,26]. Compared with monomicrobial bacteremia, polymicrobial bacteremia involving Staphylococcus aureus [27], Enterococcus spp. [28], and P. aeruginosa [29] have higher mortality rates. In contrast, polymicrobial Klebsiella pneumoniae bacteremia did not result in a worse prognosis [30]. These findings indicate that the influence of polymicrobial bacteremia on clinical prognosis should be investigated separately according to the causative pathogens.
The interbacterial interactions in polymicrobial infection, which vary among co-existing species, include metabolite exchange, co-aggregation, interspecies signaling, antibiotic resistance, and biofilm formation [31]. While most co-existing bacteria promote their pathogenesis through these interactions, some of them exhibit a competitive relationship [31]. Thus, these interbacterial interactions in polymicrobial infection may have two-sided effects on the disease severity. Therefore, the different co-existing bacteria in polymicrobial infections may have diverse effects on the clinical outcome. This in line with our findings that different concomitant bacterial pathogens had heterogeneous clinical outcomes. Compared with monomicrobial A. baumannii bacteremia, the higher mortality rate observed in patients with concomitant GNB bacteremia and lower mortality rate in those with concomitant GPC bacteremia suggested that the interactions between A. baumannii and GNB had negative effects, while those between A. baumannii and GPC positively influenced the disease severity.
On the basis of previous studies showing that beta-lactamase-producing pathogens can provide indirect pathogenesis by protecting the other pathogens in polymicrobial infection environments [13], we assessed whether the co-existence of drug-resistant A. baumannii would protect the concomitant GNB and result in a worse clinical outcome. Our findings were consistent with those of previous studies reporting higher 14-day mortality in patients with polymicrobial bacteremia with concomitant MDRAb and GNB than in those with monomicrobial MDRAb bacteremia. Although patients with polymicrobial bacteremia with concomitant CRAb and GNB had a higher 14-day mortality rate than those with monomicrobial CRAb bacteremia, the difference was not statistically significant in this small population.
Previous studies of monomicrobial A. baumannii bacteremia also reported inappropriate antimicrobial therapy to be an independent risk factor for mortality [9,11,12,22,32]. The results of the present study strengthened these findings by enrolling a large number of patients and including a number of cases with polymicrobial A. baumannii bacteremia. Ours and previous studies consistently found that inappropriate antimicrobial therapy was significantly associated with survival in patients with acquired polymicrobial bacteremia [2]. When studies evaluating the efficacy of appropriate antimicrobial therapy enroll both monomicrobial and polymicrobial A. baumannii bacteremia, appropriate therapy should be defined as a single or a combination of antimicrobial agents covering all isolated microbes.
The weakness of this study was the retrospective design required to enroll sufficient patients for detailed analysis at a single tertiary care medical center. The major strength of our study was its inclusion of a number of patients with genomically-defined A. baumannii bacteremia, consistency of patient care in a single center, stringent definition of the appropriateness of antimicrobial therapy, and a well-defined primary endpoint of 14-day mortality. We also conducted multivariate analysis to assess clinical variables that were significantly associated with mortality owing to A. baumannii bacteremia.

Conclusions
Polymicrobial A. baumannii bacteremia was not an independent risk factor for 14-day mortality. Regardless of monomicrobial or polymicrobial infection, appropriate antimicrobial therapy remains an important life-saving measure for A. baumannii bacteremia patients.
Supplementary Materials: The following are available online at http://www.mdpi.com/2077-0383/9/1/153/s1, Table S1: Logistic regression analysis of predictors for 30-day mortality among patients with Acinetobacter baumannii bacteremia. Table S2: Concomitantly isolated bacterial species and the appropriate antimicrobial regimens used for polymicrobial Acinetobacter baumannii bacteremia. Table S3: Resistant rates of the concomitant pathogens in polymicrobial bacteremia. Figure  The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflicts of Interest:
The authors declare no conflicts of interest.