Pseudomonas aeruginosa bacteremia in patients with liver cirrhosis: a comparison with bacteremia caused by Enterobacteriaceae

Background This study was performed to detect risk factors for Pseudomonas aeruginosa bacteremia in patients with liver cirrhosis. Methods A retrospective case–control study was designed to identify risk factors for P. aeruginosa bacteremia in cirrhotic patients. The cases were cirrhotic patients with P. aeruginosa bacteremia and the controls were cirrhotic patients with Enterobacteriaceae bacteremia. Results Sixty-one cases and the same number of controls were enrolled. In a multivariate analysis, younger age {adjusted odds ratio (aOR) per one year: 0.96, 95% confidence interval: 0.93 - 0.99}, nosocomial acquisition (aOR 3.87, 95% confidence interval: 1.50 - 9.94), preexisting biliary disease (aOR 4.79, 95% confidence interval: 1.92 - 10.47), and recent exposure to immunosuppressive agent (aOR 3.10, 95% confidence interval: 1.23 - 7.82) were associated with P. aeruginosa bacteremia. In the case group the frequency of appropriate initial antibiotic regimens was considerably lower than in the control group: 29.5% vs. 65.6% (P <0.01). However, thirty day mortality did not differ significantly between cases and controls (19.7% vs. 24.6%). Conclusions Nosocomial acquisition, preexisting biliary disease, and recent use of immunosuppressive agents are strong predictive factors for P. aeruginosa bacteremia in cirrhotic patients.


Background
Blood stream infection by Gram negative bacteria is a common complication in cirrhotic patients, and the prognosis of the bacteremia in this group is poor [1,2]. Third generation cephalosporins are most commonly used to treat infectious complications in patients with liver cirrhosis, but their antimicrobial spectra against Pseudomonas aeruginosa are not identical to each other [3][4][5]. It has been reported that inadequate or delayed antibiotic therapy for pseudomonas bacteremia is associated with poor prognosis [6][7][8][9]. Hence, empirical antimicrobial therapy that lacks anti-pseudomonal activity could be harmful in case of pseudomonas bacteremia.
Thus, it is important for clinicians to be aware of the risk factors for pseudomonas bacteremia. Neutropenia following chemotherapy is a well-known risk factor [10], and some clinical predictors of pseudomonas bacteremia have been reported in patients without neutropenia [11][12][13]. Age over 90 years, healthcare-associated infection, indwelling central venous catheter, indwelling urinary device, and antimicrobial therapy within the previous 30 days, have been reported to be clinical predictors of pseudomonas infection.
However, no studies have focused on the clinical predictors or outcomes of pseudomonas bacteremia in patients with liver cirrhosis. The aim of this study was therefore to identify predictive factors and clinical outcomes of pseudomonas bacteremia in patients with liver cirrhosis.

Study design and subjects
We performed a case-control study to identify risk factors associated with pseudomonas bacteremia in cirrhotic patients. Cirrhosis was diagnosed by gastroenterologists based on patient history and physical, laboratory, and radiologic features. Cases were defined as patients with liver cirrhosis and bacteremia due to P. aeruginosa admitted to Seoul National University Hospital (SNUH) or Seoul National University Bundang Hospital (SNUBH) during the period January 2005 through December 2011. SNUH is a university-affiliated tertiary referral hospital with 1,600 beds, located in Seoul, Korea. SNUBH is also a university affiliated tertiary referral hospital and it is located in Seong-Nam, Korea. To select controls, we recruited at random cirrhotic patients with bacteremias caused by Enterobacteriaceae, who were admitted to SNUH or SNUBH in the same period. The numbers of case and control patients were the same. When patients experienced two or more episodes of bacteremia, we selected the first.

Data collection
We reviewed electronic medical records and following data have been collected: age, sex, etiology of liver cirrhosis, liver function measured by Child-Pugh score, severity of bacteremia measured by Pitt bacteremia score, Charlson weighted index of comorbidity, associated diseases or conditions (such as diabetes, renal failure, malignancies, chronic obstructive pulmonary disease, bronchiectasis, etc.), primary focus of bacteremia (e.g. primary bacteremia, urinary tract, biliary tract, vascular catheters, and etc.), potential risk factors of pseudomonas infection (such as nosocomial or healthcare-associated infection, antibiotic exposure in recent 30 days, use of corticosteroids or antineoplastic agents, central venous catheter, urinary catheter, nasogastric tube, neutropenia), laboratory data at the time of diagnosis of bacteremia (such as serum creatinine, prothrombin time, bilirubin, albumin), adequacy of initial antibiotics, and 30-day mortality.

Definitions
Community acquired bacteremia: bacteremia developed before or in the initial 48 hours of hospitalization and not due to healthcare-associated bacteremia.
Nosocomial acquisition: nosocomial bacteremia or healthcare-associated bacteremia. Nosocomial bacteremia was diagnosed when the bacteremia developed more than 48 hours after admission. We considered it healthcareassociated if any of the following conditions were present: 1) hemodialysis in the previous 30 days, 2) antibiotic treatment in the previous 30 days, 3) a history of > 48 hours hospital admission (including nursing home and long term care facility) in the previous 90 days.
Recent exposure to immunosuppressive drugs: exposure to anti-neoplastic drugs, cytotoxic agents, corticosteroids, or monoclonal antibodies used to suppress the immune system within 30 days of the onset of bacteremia.
Appropriate antibiotic treatment: at least one antibiotic shown to be effective in microbiologic tests administrated within 24 hours.

Microbiologic test
The VITEK II automated system (bioMérieux Inc., Marcy L'Etoile, France) was used to identify bacterial organisms. In the VITEK II, system bacterial organisms are identified with standard identification cards.

Statistics
Student's t-test was used for normally distributed continuous variables, and the Wilcoxon rank sum test for non-normally distributed continuous variables. The Chisquare or Fisher's exact test was used to analyze categorical variables. A logistic regression model was adopted to adjust for confounding variables and to identify risk factors. To this end, we performed a stepwise multivariate logistic regression analysis. All variables with P <0.1 in the univariate analysis were included in the multivariate analysis, and variables with P <0.05 were selected in the final model. SAS v9.3 (SAS Institute Inc., Cary, NC, USA) was used for all statistical analyses.

Study approval
This study was approved by the institutional review board of the Seoul National University Hospital Medical Center (Protocol No; H-1201-002-391) according to the Helsinki Declaration. Informed consent was waived by the board.

Predictive factors for P. aeruginosa bacteremia
The mean age (± standard deviation) of the case group was slightly lower than that of the control group: 56.8 ± 13.1 and 60.7 ± 8.9, respectively (Table 1). Males predominated in both groups. Biliary origin was more common in the case group (52.5% vs. 26.3%, P <0.01), as was nosocomial acquisition (83.6% vs. 54.1%, P < 0.01). Neither the severity of cirrhosis measured by Child-Pugh class nor the severity of comorbid conditions measured by Charlson weighted index or the severity of bacteremia measured by Pitt bacteremia score differed between the two groups. Liver transplantation was performed more frequently in the case group (32.8% vs. 16.4%, P = 0.04), and preexisting biliary disease was also more common (63.9% vs. 29.5%, P <0.01), as was the use of immunosuppressive drugs (39.3% vs. 18.0%, P <0.01).

Clinical outcomes and the impact of inadequate antibiotics
In the case group the frequency of appropriate initial antibiotic regimens was considerably lower than in the control group: 29.5% vs. 65.6% (P <0.01). However, 30day mortality did not differ: 12/61 (19.7%) in the case group, 15/61 (24.6%) in the control group (P = 0.51). In the case group, 18/61 (29.5%) of the patients received adequate initial antibiotic therapy. Of these, eleven received beta-lactam antibiotics, and seven, quinolones. The clinical characteristics and outcomes in the case group according to adequacy of empirical antibiotics are shown in Table 3. 30-day mortality was lower in the patients who were prescribed adequate initial antibiotic agent(s) than in those who were not, (11.1% vs. 23.3%), but the difference was not significant (P = 0.48).

Discussion
In this study, nosocomial acquisition was an important predictor of pseudomonas bacteremia with aOR 3.87 (95% confidence interval: 1.50 -9.94, P <0.01). P. aeruginosa is a significant cause of infection in patients exposed to hospital environments; the second most common cause of nosocomial pneumonia, and the third most common cause of nosocomial urinary tract infection [14]. Thus this finding is plausible, and consistent with previous reports [11,13,15,16].
It is notable that pseudomonas bacteremia was significantly associated with the presence of biliary disease in cirrhotic patients, with aOR 4.79 (95% confidence interval: 1.92 -10.47, P <0.01). P. aeruginosa is one of the most frequent etiologic organisms of bacteremic biliary tract infection [17][18][19]. In our study, 33 of the 39 case patients with biliary tract disease had preexisting biliary catheters, including percutaneous transhepatic biliary drains, endoscopic retrograde biliary drains, and endoscopic nasobiliary drains. P. aeruginosa can form firm biofilms on foreign material and cause foreign materialassociated infections [14,20,21]. Thus biliary catheters may play a role in the origin of biliary sepsis involving P. aeruginosa in patients with biliary diseases, and the results of previous studies support this notion [22][23][24]. Twenty of the 61 (32.8%) patients with pseudomonas bacteremia had received liver transplants. During such operations, biliary-to-biliary anastomosis is made, and biliary catheters are frequently used. This suggests that clinicians should consider prescribing empirical antibiotics with anti-pseudomonas activity when gram negative bacteremia develops in cirrhotic patients with biliary diseases or preexisting biliary catheters.
It is well known that immune deficiency conditions, such as HIV infection, neutropenia, and use of immunosuppressive agents, are risk factors for pseudomonas infection [14]. Recent use of immunosuppressive agents was also associated with pseudomonas bacteremia in this study. We found that the mean age of patients with pseudomonas bacteremia was lower than that of patients with Enterobacteriaceae bacteremia. However the difference was too small to have clinical significance (aOR: 0.96, 95% confidence interval: 0.93 -0.99, P = 0.03).
Inappropriate antibiotic therapy is known to result in poor prognoses in P. aeruginosa bacteremia [6][7][8][9]12,25,26]. In our study, inappropriate initial antibiotics were prescribed more frequently in the case group, but mortality was similar in the two groups. We think that the difference between our findings and previous ones may be due to differences in the study populations: in the earlier studies, pneumonia, non-hepatobiliary tract origin, presence of neutropenia, and initial presentation with septic shock, were prognostic factors for poor outcomes in patients with pseudomonas bacteremia [7,11]. In our study, the most common source of bacteremia was the biliary tract, and only a small proportion of the  [11]. In that study the pseudomonas bacteremia originated from biliary tract infections in 50% of the patients. It is well known that interventions to relieve biliary tree obstruction are the mainstay of treatments for biliary tree infection [14]. As far as we know, this is the first study examining the clinical predictors of pseudomonas bacteremia in cirrhotic patients. It has several limitations. First, it is a retrospective study, and there may be unidentified confounding factors. Second, although we enrolled patients in two tertiary hospitals over seven years, the study population was relatively small, and we could not assess the significance of small differences. Third, control group was more heterogeneous than the case group, because the control group included patients infected with a variety of species of the family Enterobacteriaceae.

Conclusions
Nosocomial acquisition, biliary disease, and exposure to immunosuppressive agents are significantly related to P. aeruginosa bacteremia in cirrhotic patients, compared to cirrhotic patients with bacteremia caused by Entrobacteriaceae.