Antibiotic Prophylaxis Revisited: The Impact of Fluoroquinolones on Epidemiology and on the Emergence of Resistant Strains in Acute Leukemia Patients. Results of a Prospective Study By the Rete Ematologia Lombarda (REL)

Introduction. Multiresistant (multiR) bacteria are emerging pathogens in hematologic cancer patients (pts) and may negatively impact on the outcome of bloodstream infections (BSI). The antibiotic pressure, including fluoroquinolones (Fq) prophylaxis, may be one of the factors responsible for this phenomenon. In order to better define the very recent epidemiology and outcome of BSI in a real-life setting, we planned a prospective study collecting all consecutive febrile/infectious episodes occurring in acute leukemia (AL) pts admitted to 9 hematological institutions participating to REL, representing about 75% of the entire AL population treated in Lombardy.

Patients and Methods. From Dec-2012 to Jul-2014, all febrile/infectious episodes were recorded. The following data concerning BSI were analysed: age, gender, type and phase of leukemia, neutropenia, Fq prophylaxis, presence of central venous catheter (CVC), concomitant pulmonary infiltrates, antibiotic resistance.

Results. In 218 AL pts (M/F 131/87; median age 54y, range 17-80; AML/ALL 181/37), 314 BSI were diagnosed. In 180 (57.3%) BSI occurred in pts on Fq prophylaxis. In 71 (22.6%) pneumonia was also present. Gram-positive cocci (GPC) were isolated in 145/314 BSI (46.2%); Gram-negative rods (GNR) in 117 (37.3%), polymicrobial (PMB) aetiology in 48 (15.3%) and fungi (F) in 4 (1.3%). Coagulase-negative staphylococci (CoNS) were the most frequent GPC (105/314, 33.4%); S. aureus, S. viridans and enterococci were observed in 7 (2.2%), 17 (5.4%) and 37 (11.8%) cases, respectively. Methicillin-resistant strains accounted for 75.9% of all staphylococci and vancomycin-resistant strains for 2.7% of enterococci. CVC-related BSI, which accounted for 31.4% of BSI occurring in pts with CVC, was independently correlated with GPC aetiology (OR 1.9, CI 1.1-3.1, p=0.01). Considering GNR, Enterobacteriaceae were isolated in 113/314 BSI (36%) and P. aeruginosa in 31 (9.9%). GNR occurred more frequently during consolidation cycles (45.8% vs 31.1%) and in pts not on Fq prophylaxis (44.2% vs 32.2%). Both conditions were independent risk factors at multivariate analysis (p=0.002 and p=0.02, respectively). Frequency of Enterobacteriaceae BSI was also higher during consolidation cycles in comparison with other AL phases (48.1% vs 27.3%, p=0.0005), but Fq prophylaxis was no longer a protective factor (35% vs, 37.3%). Fq resistance was observed in 63/113 (55.7%) of Enterobacteriaceae; in 84.1% of cases were detected during Fq prophylaxis. ESBL+ strains, which accounted for 24.8% of Enterobacteriaceae, were also associated with Fq prophylaxis (OR 2.8, CI 1.1-7.1, p=0.02) at multivariate analysis. Carbapenemase producing (CP) strains occurred in 8.8% of Enterobacteriaceae. Among P. aeruginosa strains, 19.4% were multiR. Thirty-day mortality was 7.6% (24/314); it was lower for GPC (5.5%) in comparison with GNR (9.4%) and PMB BSI (11.6 %). CP Enterobacteriaceae or multiR P. aeruginosa BSI (30d mortality: 33.3%; OR 21.3, CI 4.4-102.4, p=0.0001) but not ESBL+ strains were independent predictors of death. Furthermore, having relapsed/resistant AL (18.2%; OR 9, CI 2.7-30.7, p=0.0004), and the presence of concomitant pulmonary infiltrates (26.8%; OR 15, CI 4.8-46.8, p<0.001) significantly correlated with the risk of death at multivariate analysis.

Conclusions. The proportion of multiresistant GNR is becoming a major problem in our real-life prospective study. Fq prophylaxis reduced the proportion of GNR BSI as a whole, but not those incurred by Enterobacteriaceae, mainly observed during consolidation phases. Moreover, Fq exposure was associated with ESBL+ aetiology, which also occurred during consolidation cycles. While ESBL+ did not impact on survival, both multiR- and CP-bacteria correlated with a higher risk of death. Interestingly, also a concomitant diagnosis of pneumonia during BSI was a strong predictor of a poorer outcome. The usefulness of Fq prophylaxis has to be reanalysed in this new epidemiologic scenario.