Carbapenemase-Producing Enterobacteriaceae (CPE) Bloodstream Infections (BSI) in Neutropenic Patients (pts) with Acute Leukemia (AL). Comparison of 2 Strategies

Introduction. The global increase in CPE infections is probably the strongest threat to the success of AL therapy nowadays. CP-K.pneumoniae (CP-Kp) BSI in particular have a high mortality in neutropenic pts, tend to reappear in subsequent courses and lead to delays in chemotherapy (CT). To make things worse, no effective decolonization is available, isolation measures are controversial, and optimal treatment is largely unknown. The widespread use and abuse of carbapenems perpetuate the problem.

An outbreak of CP-Kp BSI began in our center in Nov/16, leading to profound changes in empirical and directed antibacterial protocols. This report analyzes the impact of these modifications on clinical outcomes.

Methods. Until Feb/18, our empirical regimen for a first neutropenic fever was monotherapy with meropenem; for documented CP-Kp BSI a double or triple combination, including colistin, tigecycline, high-dose/prolonged infusion meropenem or an aminoglycoside, was selected according to MICs. Since Mar/18 empirical carbapenems were banned, and a new empirical strategy was implemented: pts who were not colonized (rectal swab screening at started Sept/17) received piperacillin-tazobactam; known colonized pts received colistin + tigecycline. This regimen was modified 48h later on the basis of blood culture results, with de-escalation (if no growth of CP-Kp) versus immediate prescription of ceftazidime-avibactam (CF-A) which was given even for colistin-susceptible strains, with the dual intention of reducing nephrotoxicity and delaying resistance to colistin. Univariate analysis was performed to identify factors related to mortality, using the χ2 test and Fisher's exact test.

Results. From Nov/16 to Aug/18, 54 consecutive CP-Kp BSI were documented in 45 pts (9 pts surviving a first episode developed a new one in a later CT course). Of note, 3 of these 54 BSI occurred in 2016 (1% of all BSI and 4% of Kp BSI that year), 32 in 2017 (11% and 38%, respectively), and 19 in 2018 until August (13% and 32%). Median age of pts was 58 (17-85), 34 were male, and hospitalization was mainly intended for induction (74%) or consolidation (18%) cycles. Median duration of neutropenia < 500/μl at the onset of bacteremia was 7 days (0-74). A clinical source was identified in most pts: pneumonia in 26%, soft tissue infection in 22%, enterocolitis in 17%. Of the 30 episodes registered since screening began, 16 occurred in colonized patients (11 with KPC, 5 OXA-48). Septic shock developed in 41% of the episodes, and ICU admission was necessary in 30%. In 19 episodes (35%) empirical antibacterial treatment was (retrospectively) considered appropriate if it had included colistin and/or tigecycline. CF-A was prescribed as directed therapy in 28 cases (52%). Microbiological cure (negative blood cultures) was verified in 86% of 49 evaluable episodes. Attributable mortality (due to this infection) at day 14 of the onset of bacteremia was 22% (12 cases) and global mortality during hospitalization was 31% (17 cases). Fifteen of these 17 deaths occurred in the 40 episodes registered before protocol modification in Mar/18 and only 2 in the 14 episodes after that date (37% vs 14%). Adverse outcomes were significantly related to unresolved bacteremia, to occurrence of septic shock and to ICU admission (all p<0.001). Survival was more likely if microbiological cure was achieved (p<0.001), if appropriate antibiotics were empirically given at the onset of fever, and if directed treatment consisted of CF-A (p=0.02).

Conclusions. In this series of profoundly neutropenic pts under intensive CT, the mortality rate from CP-Kp BSI was very high. Unresolved neutropenia and severity of the infection itself (as measured by the presence of septic shock and/or ICU admission) had, as expected, a major impact on death. Two treatment-related factors (time to appropriate empirical coverage and directed use of CF-A) had a major influence on survival, both presumably through faster microbiological resolution. The strategy we implemented has two aims: allowing appropriate and timely empirical coverage in colonized patients while sparing carbapenems, and optimizing directed therapy while reducing colistin use and toxicity. Although the present data need confirmation in a larger series, we believe that the recent sharp decrease in mortality observed in our pts with CP-Kp bacteremia stems from these measures.