Increasing Rates of Fluoroquinolone Resistance in Escherichia Coli Isolated From the Blood and Urine of Patients with Hematologic Malignancies and Stem Cell Transplant Recipients

Fluoroquinolone (FQ) antibiotics have been shown to reduce mortality and the number of febrile episodes when used as prophylaxis during neutropenia. Prior studies suggest that prophylaxis may result in increasing rates of FQ resistance. Fluoroquinolone non-susceptibility trends in Escherichia coli isolated from blood and urine cultures were evaluated over a 16-year period during which prophylaxis was initiated in patients with hematologic malignancies and stem cell transplants. Non-susceptibility rates increased after the introduction of prophylaxis, with yearly non-susceptibility rates rising from 30%–33% to 40%–88% in blood isolates. The high rates of non-susceptibility now observed raise concerns about the continued efficacy of FQ prophylaxis. This concern exists particularly in those patients undergoing stem cell transplants where the total FQ non-susceptibility rates over the study period were 82.3%. Further evaluation of the effect of FQ prophylaxis on antibiotic resistance and its efficacy in the setting of increased rates of resistance is warranted.


INTRODUCTION
Neutropenia and the associated risk of infection is a consistent problem in the treatment of leukemia and lymphoma. Studies have shown that 12%-43% of patients undergoing treatment for leukemia have an episode of bacteremia during the course of their care [1][2][3][4][5][6][7]. Guidelines for the management of neutropenia in both hematologic malignancy (HM) patients and hematopoietic stem cell transplant (HSCT) recipients recommend fluoroquinolone (FQ) prophylaxis based on studies showing a significant reduction in the number of documented infections and episodes of bacteremia [7][8][9][10]. Importantly, the Infectious Disease Society of America (IDSA) neutropenia in cancer guidelines also recommend having a strategy to monitor rates of FQ resistance among gram-negative bacilli.
In this retrospective study, the primary goal was to evaluate the trend of FQ non-susceptibility rates in E. coli isolates from patients' blood and urine cultures at our institution from 2000 to 2015. FQ non-susceptibility rates in isolates obtained from an initial blood or urine culture from patients with HM/HSCT were compared to non-HM/HSCT patients.

METHODS
Between 2005 and 2015, HSCT recipients and high risk patients with HM who were neutropenic (absolute neutrophil count (ANC) < 500 cells/mm 3 ) received universal FQ prophylaxis (levofloxacin 500mg PO daily) per institutional protocol. Prior to 2005, use of FQ prophylaxis was at the discretion of the treating physician. Patients with HM are considered high risk if anticipated duration of neutropenia > 7 days and/or anticipated ANC < 100 cells/mm 3 . FQ prophylaxis was discontinued when the ANC was > 500 cells/mm 3 . The study was performed under a waiver of informed consent. The study was approved by the IRB of the University of North Carolina (IRB 14-0842).
All E. coli isolates were tested for susceptibility to ciprofloxacin, levofloxacin, or both. Isolates were defined as non-susceptible if testing determined them to be either intermediate or resistant to fluoroquinolones using standardized testing and interpretive criteria as published by the Clinical and Laboratory Standards Institute. Susceptibility testing from 2000 to mid-2009 was performed using Kirby-Bauer disk diffusion. After June of 2009 susceptibility testing was performed using VITEK2 (bioMerieux, Durham, NC).
The percentage of cultures showing FQ non-susceptibility per year was calculated. Each patient was included in the evaluation only once based on their first culture positive for E. coli (blood or urine). Comparisons between HM/HSCT and non-HM/HSCT patients were performed with Fisher's exact or Pearson's chi-square test.
For the blood culture results, an interrupted time series analysis was performed with the objective of assessing differences in the trend in rates of FQ resistance before and after instituting a protocol of antibiotic prophylaxis for neutropenia. The HM/HSCT and non-HM/HSCT cohorts were analyzed separately using the linear regression function 'lm' in the software package R, which assumes independent errors. The interrupted time series model was performed using time and preprotocolized vs postprotocolized antibiotic prophylaxis as the covariates. No significant time effects were seen in the analyses, possibly due to small sample size.  Table 1.

RESULTS
The FQ non-susceptible rates in E. coli isolates in blood cultures were 81 out of 149 (54%) in the HM/HSCT group and 462/1445 (32%) in the non-HM/HSCT group (P < 0.001). Within the HM/ HSCT group, HSCT patients were more likely to have bacteremia caused by E. coli non-susceptible to FQ, compared to HM patients; 42 of 51 (82.3%) vs 39 of 98 (39.8%), P < 0.001.
In urine isolates the FQ non-susceptibility rates for E. coli isolates were 76 of 267, (28%) in the HM/HSCT group and 3408 of 23 382 (14.6%) in the non-HM/HSCT group (P < 0.001). Similar to the results from E. coli blood isolates, E. coli from HSCT patients displayed a higher rate of resistance (27 of 57, 47%), as compared to HM patients (49 of 210, 23%, P < 0.001).
The annual rate of FQ non-susceptibility in E. coli increased across the study period of 2000-2015 for blood isolates, and the annual rate of FQ non-susceptibility in E. coli increased across the study for both blood and urine isolates (Figures 1 and 2). Starting in 2004 the numbers of E. coli that were FQ non-susceptible began to increase at a greater rate in the HM/HSCT group compared to the control group in both blood and urine culture groups. The total rate of non-suscepti-

DISCUSSION
This trend of increasing FQ non-susceptibility in the HM/HSCT group presents questions about the efficacy of FQ antibiotics for neutropenic prophylaxis. The marked increase in E. coli non-susceptible to FQ from 2005 to 2007 in the HM/HSCT group is of particular concern because prophylaxis with levofloxacin became part of the standard neutropenia protocol in 2005. In prior studies looking at rates of FQ non-susceptibility after the introduction of prophylaxis, similar increases in non-susceptibility were seen in some [11][12][13][14], while others found no change in rates of non-susceptibility after the onset of prophylaxis [15][16][17]. The rise in resistance seen in this study is most likely an effect of increased FQ exposure, with the more significant rise seen in the HM/HSCT group likely reflecting the initiation of FQ prophylaxis. The observed highest rates of resistance in HSCT patients probably reflects increased FQ exposure, as these patients generally have had multiple neutropenic episodes.
Another concern pertains to the efficacy of prophylaxis in the setting of high baseline rates of FQ resistance. It is unknown whether the evidence for the benefit of prophylaxis continues to hold true in the setting where non-susceptibility rates are 40%-60%. This uncertainty exists particularly in the HSCT population where 82.3% of E. coli isolates were non-susceptible to FQ. However, these findings may reflect successful suppression of FQ-susceptible isolates, such that only FQ non-susceptible organisms cause breakthrough infections.
In conclusion, FQ non-susceptibility rates in E. coli isolated from blood and urine cultures increased over the study period, especially in patients with hematologic malignancies and HSCT patients. This increase in rates of FQ non-susceptibility coincided with the implementation of levofloxacin prophylaxis for patients with neutropenia at our institution. Whether this is a reversible trend remains to be determined.