The present study investigated the clinical characteristics and outcomes of P. aeruginosa bacteremia in FN children and adolescents. Mortality due to P. aeruginosa bacteremia remained high in the 2010s, and more than one-third of the isolated P. aeruginosa strains were MDR.
The mortality among immunocompromised patients with P. aeruginosa bacteremia was approximately 70% in the 1960s and 1970s [1, 2, 18], which decreased to 20–25% in the 1990s with the use of anti-pseudomonal antibiotics [18, 19]. However, the mortality in the 2000s was 20–39%, similar to that in the 1990s [5–7, 11], and 38.9% of FN children and adolescents with P. aeruginosa bacteremia died in the present study. This recent slowdown in improving outcomes in P. aeruginosa bacteremia patients might be associated with increasing prevalence of antibiotic-resistant strains. MDR P. aeruginosa comprised 1.6–8.2% of the identified P. aeruginosa strains until the early 2000s [20, 21]; however, the proportion of MDR strains increased to 30.7–71.1% in the late 2010s [5, 6, 11]. In Korea, 11.3% of P. aeruginosa bacteremia cases diagnosed in hospitalized children, including immune-competent and -compromised children, were caused by MDR strains in the 2000s ; however, 36.1% of P. aeruginosa bacteremia were caused by MDR strains in the present study.
The appropriateness of empirical antibiotic therapy as well as infection due to MDR strains was associated with mortality in patients with P. aeruginosa bacteremia, a relationship that has been previously reported [5–7, 18, 22–25]. In sum, antibiotics to which MDR P. aeruginosa strains are susceptible should be administered empirically in order to improve the outcomes of immunocompromised patients with P. aeruginosa bacteremia. The P. aeruginosa antibiotic susceptibility rates in the present study were 100% to aminoglycosides and colistin and 97.2% to ciprofloxacin, which were higher than those to anti-pseudomonal β-lactam agents such as piperacillin/tazobactam and cefepime. Previous studies on P. aeruginosa bacteremia in children also reported higher antibiotic susceptibility rates to amikacin and fluoroquinolones compared with those of β-lactam agents [10, 11]. However, the use of fluoroquinolones has been restricted in children due to concerns of skeletal adverse effects, and empirical use of colistin may not be appropriate considering its nephrotoxicity and neurotoxicity . Aminoglycosides are not effective as a single agent against Gram-negative bacterial infections including pseudomonal infections [22, 24, 27, 28]. As a result, anti-pseudomonal β-lactam agent and aminoglycoside combination therapy may be helpful to broaden the antibiotic coverage for MDR strains and consequently improve the outcomes of patients with P. aeruginosa bacteremia. In the present study, although the appropriateness of the empirical β-lactam agents did not differ significantly between the survived and deceased groups, the combination with aminoglycosides significantly increased the appropriateness of the empirical antibiotics in the survived group. However, the contribution of the β-lactam agent and aminoglycoside combination to antibiotic synergism, improved clinical outcomes, and suppressed the emergence of antibiotic resistance has not been confirmed [29–31]. Therefore, this antibiotic combination can be maintained for early (3 to 5 days) bacteremia, followed by targeted antibiotic therapy based on the antibiotic susceptibility results .
The relationship between infections due to MDR strains and mortality of patients with P. aeruginosa bacteremia in the present study underscore the need to decrease the prevalence of MDR strains. Infections due to MDR P. aeruginosa were associated with recent use of carbapenems, ventilator care, and P. aeruginosa infection or colonization within the previous year [32, 33]. In Korean children, the primary risk factor for MDR P. aeruginosa bacteremia was admission to the intensive care unit within 1 month ; however, no patient in the present study had been admitted to the intensive care unit within 2 months before developing P. aeruginosa bacteremia. Almost all patients in the present study had received repeated anti-pseudomonal antibiotic therapy due to their underlying hematologic/oncologic disorders; therefore, recent use of anti-pseudomonal antibiotics was not significantly associated with MDR strain infections. However, the effect of recent antibiotic use on the MDR strain infections cannot be ignored, considering the relationship between breakthrough and MDR strain infections. Previous studies have reported that various β-lactam agents and fluoroquinolones were related to MDR P. aeruginosa infections [20, 30, 34]. In addition, the induction rate of antibiotic resistance in P. aeruginosa was affected by the type of previously administered antibiotics, and imipenem showed a higher rate of resistance induction compared with those of other antibiotic agents . Accordingly, restriction of carbapenem use may reduce the emergence of MDR P. aertuginosa strains. Carbapenems have an additional antibiotic effect against extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae beyond other anti-pseudomonal β-lactam agents commonly used for FN patients. Our previous study, however, showed that a combination of empirical β-lactam agent and aminoglycoside instead of carbapenems did not cause unfavorable outcomes in FN patients with ESBL-producing E. coli and K. pneumoniae infections . As a result, empirical anti-pseudomonal β-lactam and aminoglycoside combination therapy in FN patients may reduce carbapenem use and subsequently prevent the emergence of antibiotic resistance without worsening prognosis due to Gram-negative bacterial infections. In our hospital, anti-pseudomonal β-lactam and aminoglycoside combination therapy has been used as first-line empirical therapy for FN patients. However, carbapenems have been administered as a second-line empirical antibiotic agent for patients with persistent fever despite the first-line empirical antibiotic therapy until the recovery of neutropenia. Eventually, many patients might have received carbapenems for longer days than anti-pseudomonal penicillins or anti-pseudomonal cephalosporins during their hospitalization. Such prolonged use of carbapenems might cause the emergence of MDR P. aeruginosa strains in our hospital, and therefore, further efforts to shorten the duration of empirical carbapenem use should be performed.
The present study had several limitations. First, P. aeruginosa comprise about 10% of the pathogens identified in FN patients; thus, the number of FN patients with P. aeruginosa bacteremia was small. The increase in the number of enrolled patients may reveal additional factors related to mortality and MDR strain infections. A multicenter study is necessary to overcome this limitation; however, each hospital may have their own strategies for chemotherapy, transplantation, and antibiotic therapy in FN patients. In addition, the antibiotic resistance patterns of each hospital reflect the resistance patterns of individual communities and countries . Therefore, interpretation of the results of a multicenter or multinational study on the antibiotic susceptibilities may be difficult. Also, this study was a retrospective observational study, and therefore, a well-designed prospective cohort study or case-control study is necessary to overcome such limitations. Second, the appropriateness of targeted antibiotic therapy rather than empirical therapy can affect the outcomes of patients with bacteremia. In the present study, we could not evaluate the effect of targeted therapy on the outcomes because 14 types of targeted antibiotic therapy were performed in 36 episodes.