Risk factors for bloodstream infections due to extended-spectrum β-lactamase producing Enterobacteriaceae in cancer patients

Introduction: Bloodstream infection (BSI) caused by Enterobacteriaceae is associated with mortality in cancer patients receiving chemotherapy. The aim of this study is to identify the risk factors and outcomes related to BSIs caused by extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae in cancer patients. Methodology: Hematology/oncology patients, who were diagnosed with BSIs caused by Enterobacteriaceae by positive blood cultures were evaluated retrospectively. Patients were divided into two groups by ESBL-positive and ESBL-negative Enterobacteriaceae bacteremia. Patients' demographic features, underlying conditions, comorbidity, neutrophil count, duration of neutropenia, antibiotic use in the previous three months before infection, mechanical ventilation, steroid use, central venous catheter implementation, total parenteral nutrition (TPN), hospitalization in the past three months, stay in intensive care unit, quinolone prophylaxis, and history of infection with ESBL-producing Enterobactericeae were evaluated. Risk factors related to BSIs caused by ESBL-producing Enterobacteriaceae and mortality were assessed. Results: A total of 122 patients were evaluated retrospectively. Quinolone propyhlaxis, TPN, infection with Extended Spectrum BetaLactamase positive ESBL-P Enterobacteriaceae during the previous three months, treatment with piperasillin-tazobactam or carbapenems in the previous three months were found to be independent risk factors for ESBL-P BSIs. Longer duration of neutropenia before BSI and complication at the beginning of BSI were found to be independent risk factors for mortality related to infection. Conclusions: ESBL-producing Enterobacteriacea should be treated with an appropriate antibiotic that is associated with better outcomes in hematology/oncology patients with BSIs. History of broad-spectrum antibiotic use and stay in hospital in the previous three months should be taken into consideration upon commencing antibiotic therapy.


Introduction
Bloodstream infection (BSI) is common in cancer patients receiving chemotherapy and it is associated with high morbidity and mortality rates [1].These patients are immunocompromised due to chemotherapy, neutropenia, major surgery, malnutrition, transplantation or immunosuppressive treatment in addition to their underlying malignity [2].
An increase in the frequency of bacteremias caused by Gram-negative rods has been reported in many centers worldwide, with Gram-negative pathogens becoming either predominant or at least as frequent as Gram-positive pathogens [3].This trend has been confirmed in a recent literature review and European surveillance study performed in 2011 in 39 hematology centers from 18 countries for the Fourth European Conference of Infections in Leukemia (ECIL-4) [4].The high percentage of multidrug resistant microorganism is a world-wide problem.Extendedspectrum beta-lactamase (ESBL) production is the most common resistance type for Enterobacteriaceae [2,5,6].The knowledge of risk factors for ESBL bloodstream infection (BSI) will contribute to identify patients who are at higher risk, in order to start empirical therapy with proper coverage against these microorganisms.There are few reports comparing the clinical, epidemiological and microbiological factors of ESBL-producing (ESBL-P) Enterobacteriaceae BSI in adult cancer patients.This study was performed to evaluate the epidemiology of ESBL-P BSI in cancer patients and to assess the clinical impact of ESBLs on patients' outcome.

Methodology
The retrospective cohort study was conducted in a 550 beds teaching hospital.Hematology/oncology patients with culture positive BSIs with E. coli or Klebsiella spp, during a period of two years from January 2013 to December 2014, were included.Other Enterobacteriaceae strains were not included as they were very few in numbers.Patients' data and laboratory findings were collected by the electronic medical record database of the hospital retrospectively.Patients were divided into two groups, ESBL -positive (ESBL-P) and ESBL-negative (ESBL-N), according to ESBL production of the microorganism.If there was more than one positive blood culture, the first strain was included for each infection.
Demographic and infection-related data like age, gender, underlying disease, type of malignancy, chemotherapy regimen, comorbidities like diabetes mellitus (DM), Chronic Obstructive Pulmonary Disease (COPD), hematopoietic stem cell transplantation (HSCT), hospital unit in which the infection was detected, type of BSI (primary or secondary), neutrophil count and duration of neutropenia (< 500 neutrophils/mm 3 ) were recorded.Potential risk factors for infection with ESBL-P Enterobacteriaceae including history of hospitalisation and previous antibiotic use during the last three months, quinolone prophylaxis, duration of hospital stay before infection, stay in intensive care unit (ICU), history of mechanical ventilation support, total parenteral nutrition (TPN), steroid use, Granulocyte-colony stimulating factor (GCSF) use and central venous catheter use were recorded.
Outcomes of BSIs with ESBL-P Enterobactericeae were also evaluated.Appropriate initial antibiotic therapy was defined as a regimen that included at least one antibiotic that the isolated strain was susceptible to in vitro and started during the first 24 hours after the blood culture was obtained.If the first regimen was not appropriate, the day an adequate drug started was recorded.Presence of complications like septic shock, hypoxia and organ failure, length of stay in the hospital and mortality related with infection and overall mortality (infection and all other causes) in 30 days were also recorded.Mortality was defined as related to infection if it occurred in 14 days after positive blood culture and there was no other reason related to primary disease.
Identification of the bacteria and antimicrobial susceptibility testing were performed using conventional methods and a VITEK2 automated system (bioMérieux, Marcy l'Etoile, France) according to the recommendations of the Clinical and Laboratory Standard Institute (CLSI) [7].

Statistical Analysis
Statistical analyses were done using SPSS for Windows software, version 21.0.Chi-squared and Student's t tests were used for the univariate analysis of categorical and continuous variables of patient characteristics, respectively.Continuous variables with non-normal distribution were compared by using Mann-Whitney U test.Independent risk factors for ESBL-P bloodstream infections were evaluated by a multivariable model using logistic regression.

Results
122 patients with E. coli or Klebsiella spp.positive blood cultures were included in the study.The mean age ± standard deviation of the patients was 44.5 ± 15.55 and 89 (73%) of them were male.112 patients (91.8%) had hematologic malignancy and 10 (8.2%) of them had solid tumors.37 patients (30.3%) had hematopoietic stem cell transplantation (HSCT).84 (68.8%) of the isolated microorganisms were E. coli and 38 (31.2%) of them were Klebsiella spp.70 of the patients (57.4%) had BSI with ESBL-P microorganisms and 52 (42.61%) of them were ESBL-N.Demographic and clinical data of the patients are evaluated in Table 1.
We did not find a statistical significance between patients with ESBL-P and ESBL-N BSIs when we compared age, gender comorbidities, history of stay in ICU or mechanical ventilation support, central venous catheter use, neutrophil count and duration of neutropenia.There was no statistical significance in use of aminoglycosides and polymyxin E in the previous three months.
We assessed the outcomes of ESBL-P and ESBL-N BSIs, too.The length of hospital stay was higher in patients with ESBL-P BSI but it was not statistically significant (P: 0.105).Complications like septic shock and organ failure (P: 0.046 CI: 0.996-4.961OR: 2.22), all-cause 30 days mortality (P: 0.034 CI: 1.13-6.95OR: 2.69) and mortality related to infection (P: 0.025 CI: 1.41 -14.68 OR: 2.69) were significantly higher in patients with ESBL-P BSIs.Outcomes of patients with ESBL-P and ESBL-N BSIs are shown in Table 3.
The all-cause mortality was higher in patients that were treated with inappropriate antibiotics in the beginning.(P: 0.03, r: -267) Univariate analysis identified BSI with ESBL-P microorganism in the previous three months, neutropenia, longer duration of neutropenia before BSI, inappropriate treatment at the beginning, delay at starting appropriate treatment and complication at the beginning of BSI as risk factors for mortality related to infection.Multivariate logistic regression identified longer duration of neutropenia before BSI and complication at the beginning of BSI as independent risk factors for mortality related to infection.Risk factors for mortality related to infection are shown in Table 4.

Discussion
Infections with resistant Gram-negative microorganisms in cancer patients are relevant because of an inherent risk of treatment failure of the infection, which contributes to morbidity and mortality.In this study, patients infected with Enterobacteriaceae spp.were evaluated and ESBL positivity was found to correlate with higher mortality.
Early diagnosis and treatment of ESBL-positive patients were found to decreasemortality.In this study, we found that infection with ESBL-positive Gramnegative bacteria in the past three months, history of quinolone, piperasilin-tazobactam or carbapenem treatment in the past three months and total parenteral nutrition were independent risk factors for ESBL positivity in cancer patients.Prevention, early diagnosis and prompt treatment of ESBL-positive Gram-negative infections are vital in immunocompromised patients with underlying malignity.The prevalence of ESBL-positive Gramnegative infections is increasing among communityacquired as well as health care-associated infections [8].
The use of immunosuppressive agents and corticosteroids is associated with the development of ESBL-positive Gram-negative infections.A study in mice showed that the use of immunosuppressive drugs causes bacterial translocation from the intestinal tract and bacteremia.ESBL-positive microorganisms lead to more frequent infections in this population [9,10].ESBL-positive microorganisms have been shown to be located in the lower intestinal tract, and this region is usually the source for these infections [11].
In this study, 122 patients with hematologic and solid organ malignancies were evaluated.All of the patients were receiving immunosuppressive treatment.Fluoroquinolone prophylaxis is used to prevent infections in patients with profound prolonged neutropenia.Fluoroquinolone prophylaxis was used during the neutropenic period after allogeneic HSCT until engraftment for a short period in our hospital.Breakthrough infections with fluoroquinolone resistant Gram-negative bacteria are reported in recent studies [12,13].Fluoroquinolone prophylaxis has a potential role in infection with MRSA and Clostridium difficile and ESBL-producing Enterobacteriaceae as well as fluoroquinolone-resistant strains [6,14,15].However, there are some studies showing that fluoroquinolone prophylaxis does not increase the risk of infection with resistant microorganisms.Satlin et al. reported that levofloxacin prophylaxis did not increase MDR Gram-negative bacterial infection in 475 multiple myeloma patients undergoing autologous HSCT [16].We identified fluoroquinolone prophylaxis as a risk factor for ESBL-P bloodstream infection as well as for quinolone resistance.
Antibiotic exposure during the last three months has been reported as a risk factor for infections with resistant Gram-negative microorganisms [17].We found that using piperacillin-tazobactam, quinolones and carbapenems during the previous three months is a risk factor for ESBL-P infections.There was no statistical significance in use of aminoglycosides and polymxyn E, which may be due to the low number of patients treated with these drugs before BSI in our study.There was no usage of cephalosporins during the last three months so we could not demonstrate the impact of these drugs on ESBL-P bacteremia.
In a study involving 118 cancer patients, longer duration of hospitalization and cephalosporin, macrolide, quinolone and aminoglycoside treatment before infection were reported as independent risk factors for ESBL-positive BSIs.There was no significant difference in mortality in patients with ESBL-positive and negative BSIs in this study [18].
Risk factors for colonization / infection with ESBLproducing microorganisms were assessed in a metaanalysis including 14 studies (746 cases and 1257 controls).Parenteral nutrition, mechanical ventilation, use of central venous catheter, previous ampicillin, gentamicin and cephalosporin treatment were found to increase colonization / infection with ESBL-positive microorganisms [19].
Piperacillin/tazobactam is a good choice for the treatment of hospital-acquired infections, caused by Pseudomonas spp.and ESBL-positive microorganisms [20].It is thought that it can be used as an alternative to carbapenems to treat infections with resistant Gramnegative bacteria [21,22].Although piperacillin/tazobactam is considered to be a good option for treatinginfections with Gram-negative microorganisms, there are some studies showing that it promotes accumulation of ESBL-positive Enterobacteriaceae in the intestinal flora of patients [23,24].Piperasilin-tazobactam treatment in the previous three months was identified as an independent risk factor for BSI with ESBL-P in our study.
Prior colonization or infection by resistant organisms is one of the most important risk factors for developing an infection with ESBL-P microorganisms [18,25,26].We do not perform surveillance cultures for screening multidrug-resistant Gram-negative bacteria, therefore we could not assess the impact of colonization with these microorganisms in our patients.However, we found that infection with ESBL-P Enterobacteriaceae during previous three months is an individual risk factor for ESBL-P bacteremia.Infection and colonization with resistant microorganisms increase mortality and morbidity.To prevent this, it is necessary to avoid inappropriate antibiotic treatment [18].
Prolonged hospital stay and/or repeated hospitalizations have also been found to be risk factors for infection with resistant bacteria [14,27].Duration of hospitalization was reported as an independent risk factor for infection with multidrug-resistant ESBL-P E. coli and K. pneumoniae BSIs in one study from Turkey.[28].Although the number of patients staying in hospital during previous three months was higher and the duration of hospital stay prior to BSI was longer in the ESBL-P group than in the ESBL-N group, this was not reflected in a statistically significant difference in our study.
Total parenteral nutrition (TPN) is a risk factor for bacteremia and candidemia [16,29,30].We identified TPN as an independent risk factor for ESBL-P BSIs.Univariate analysis identified receiving TPN in the previous 30 days as a potential risk factor for acquiring ESBL-P Enterobacteriaceae among the paediatric population in one study [31].There is a need for more studies about the role of TPN in infections with resistant Gram-negative bacteria.
Infection with multidrug-resistant Gram-negative bacteria was reported to be associated with a high mortality in cancer patients in the literature [32][33][34][35][36]. Matsuma et al. found that the 30-day mortality was higher in patients with ESBL-positive BSIs than the ones with ESBL-negative BSIs [37].Conversely, ESBL positivity did not have a negative effect on mortality in some studies.Namikawa et al. reported that there was no significant difference in mortality according to age, underlying diseases, CRP level, white blood cell numbers of the patients and ESBL positivity of the microorganism [10].Although univariate analysis showed that the mortality rate was higher in ESBL-P group, we did not identifiy ESBL-P BSI as an independent risk factor for mortality.
Early initiation of treatment with appropriate antibiotics in bacteremic patients prevents sepsis and reduces mortality.
Metan et al. assessed the relationship between multidrug-resistant Gram-negative bacteremia and mortality and found that inappropriate antibiotic treatment was higher in patients who did not survive seven days after bacteremia onset.However, they could not demonstrate a significance of this finding by multivariate analysis [38].In our study, mortality related to infection was higher in the patients that were treated with inappropriate antibiotics in the beginning, but inappropriate therapy was not an independent risk factor for mortality related to infection when analysed by multivariate logistic regression.In this study we defined infection-related mortality as mortality occurring within 14 days after the first positive blood culture.30 (30.3 %) out of 99 survivors and 14 (60.9%) of the patients who did not survive had been neutropenic when they had BSI.Duration of neutropenia was higher in non-survivors.Higher duration of neutropenia was found to be an independent risk factor for mortality related to infection in our study.Complications like hypoxy, septic shock etc. were more common in patients who did not survive, which is in accordance with the literature.
Our study has some limitations.We only evaluated patients from one center.Patients'profiles may vary in other hospitals.The design of the study is retrospective, therefore we could not take all possible risk factors like travel history or colonisation with ESBL-P bacteria before the infection and outcomes related to ESBL-P BSIs into consideration.

Conclusions
ESBL-P Enterobacteriaceae should be kept in mind in BSIs of hematology/oncology patients with previous broad spectrum antibiotic use and stay in hospital in previous 3 months, as starting appropriate antibiotic treatment is life-saving in these patients.Broadspectrum antibiotics should promptly be given to patients with prolonged neutropenia and unstable clinical conditions like hypoxy, septic shock or organ failure.

Table 1 .
Demographic data and clinical features of patients with ESBL-positive or ESBL-negative bloodstream infection.

Table 2 .
Risk factors for EBL-P and ESBL-N bloodstream infections.

Table 3 .
Outcomes of ESBL-P and ESBL-N bloodstream infections.

Table 4 .
Risk factors for mortality related to infection.