- Email: [email protected]
Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection
Clinical Microbiology and Infection xxx (xxxx) xxx
Contents lists available at ScienceDirect
Clinical Microbiology and Infection journal homepage: www.clinicalmicrobiologyandinfection.com
Original article
Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection rez-Lledo 3, O. Rodríguez-Nún ~ ez 2, D. Viasus 1, y, P. Puerta-Alcalde 2, y, C. Cardozo 2, M. Sua r 2, F. Marco 4, 5, M. Chumbita 2, E. Moreno-García 2, L. Morata 2, C. Fehe rrez-Garcia 3, 6, J.A. Martínez 2, 6, J. Mensa 2, M. Rovira 3, 6, ndez-Avile s 3, G. Gutie F. Ferna 3, 6 2, 6 , A. Soriano , C. Garcia-Vidal 2, 6, * J. Esteve 1)
Health Sciences Division, Universidad del Norte, and Hospital Universidad del Norte, Barranquilla, Colombia Infectious Diseases Department, Hospital Clínic-IDIBAPS, Barcelona, Spain 3) Haematology Department, Hospital Clínic-IDIBAPS, Barcelona, Spain 4) ISGlobal, Hospital ClínicdUniversitat de Barcelona, Barcelona, Spain 5) stic Biom Microbiology Department, Centre Diagno edic, Hospital Clínic, Barcelona, Spain 6) University of Barcelona, Barcelona, Spain 2)
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2019 Received in revised form 7 June 2019 Accepted 1 July 2019 Available online xxx
Objectives: To assess risk factors for multidrug-resistant Pseudomonas aeruginosa (MDR-PA) infection in neutropenic patients. Methods: Single-centre retrospective analysis of consecutive bloodstream infection (BSI) episodes (2004 e2017, Barcelona). Two multivariate regression models were used at BSI diagnosis and P. aeruginosa detection. Significant predictors were used to establish rules for stratifying patients according to MDR-PA BSI risk. Results: Of 661 Gram-negative BSI episodes, 190 (28.7%) were caused by P. aeruginosa (70 MDR-PA). Independent factors associated with MDR-PA among Gram-negative organisms were haematological malignancy (OR 3.30; 95% CI 1.15e9.50), pulmonary source of infection (OR 7.85; 95% CI 3.32e18.56), nosocomial-acquired BSI (OR 3.52; 95% CI 1.74e7.09), previous antipseudomonal cephalosporin (OR 13.66; 95% CI 6.64e28.10) and piperacillin/tazobactam (OR 2.42; 95% CI 1.04e5.63), and BSI occurring during ceftriaxone (OR 4.27; 95% CI 1.15e15.83). Once P. aeruginosa was identified as the BSI aetiological pathogen, nosocomial acquisition (OR 7.13; 95% CI 2.87e17.67), haematological malignancy (OR 3.44; 95% CI 1.07e10.98), previous antipseudomonal cephalosporin (OR 3.82; 95% CI 1.42e10.22) and quinolones (OR 3.97; 95% CI 1.37e11.48), corticosteroids (OR 2.92; 95% CI 1.15e7.40), and BSI occurring during quinolone (OR 4.88; 95% CI 1.58e15.05) and b-lactam other than ertapenem (OR 4.51; 95% CI 1.45e14.04) were independently associated with MDR-PA. Per regression coefficients, 1 point was assigned to each parameter, except for nosocomial-acquired BSI (3 points). In the second analysis, a score >3 points identified 60 (86.3%) out of 70 individuals with MDR-PA BSI and discarded 100 (84.2%) out of 120 with non-MDR-PA BSI. Conclusions: A simple score based on demographic and clinical factors allows stratification of individuals with bacteraemia according to their risk of MDR-PA BSI, and may help facilitate the use of rapid MDRdetection tools and improve early antibiotic appropriateness. D. Viasus, Clin Microbiol Infect 2019;▪:1 © 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.
Editor: M. Paul Keywords: Bacteraemia Multidrug-resistant Pseudomonas aeruginosa Neutropenia Risk factors
Introduction * Corresponding author. C. Garcia-Vidal, Infectious Diseases Department, Hospital Clínic-IDIBAPS, Carrer de Villarroel 170, 08036, Barcelona, Spain. E-mail addresses: [email protected], [email protected] (C. Garcia-Vidal). y Diego Viasus and Pedro Puerta-Alcalde contributed equally to this manuscript.
Bloodstream infection (BSI) is an important cause of mortality in onco-haematological patients with febrile neutropenia [1]. Chemotherapies, new treatment modalities, indwelling intravascular
https://doi.org/10.1016/j.cmi.2019.07.002 1198-743X/© 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
2
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
catheters, prolonged hospitalizations and high consumption of antibiotics render febrile neutropenic patients highly vulnerable to multidrug-resistant (MDR) pathogen colonization or infections [2]. Pseudomonas aeruginosa continues to be a major pathogen causing infections with morbidity and mortality in neutropenic patients [3,4]. Febrile neutropenia is a strong predictor of P. aeruginosa infection; guidelines for the management of neutropenic fever recommend that empirical antibiotic treatment is started with antipseudomonal agents [5,6]. However, empirical coverage for P. aeruginosa has become a challenge due to the rising incidence of MDR P. aeruginosa (MDR-PA) worldwide. Cattaneo et al. [7] and Trecarichi et al. [8] reported rates of MDR among BSI P. aeruginosa isolates of 33% and 71%, respectively. Consequently, inadequate empirical therapy has become quite frequent, as well as being notably associated with an increase in mortality [3,7,9,10]. It is therefore essential that the detection of BSI episodes at risk of MDR-PA in neutropenic patients is improved. Rapid detection methods for bacterial pathogens in blood cultures have become widespread: as recently demonstrated, the great majority of blood cultures in febrile neutropenia are positive within the first 24 h [11]. Similarly, microbiological methods for rapid detection of antibiotic resistance in positive cultures have also been recently developed; however, such methods are expensive and not affordable for every patient [12e14]. Hence, two scenarios are important when modifying empirical antibiotic therapy in neutropenic patients with bacteraemia: the first one is when an individual is diagnosed with bacteraemia (positive blood culture) and the second one is when the causative agent of the bacteraemia has been documented. The availability of susceptibility tests follows pathogen identification by at least 18e24 hours. We hypothesized that some factors might be useful in predicting the risk of MDR-PA in neutropenic patients. Our aim was to assess these factors at BSI diagnosis to facilitate a physician's decisionmaking process concerning broad-spectrum antibiotic coverage. Second, once identification of P. aeruginosa in blood cultures had been completed, we assessed the risk factors for MDR-PA. These factors might help to rationalize the use of antibiotics and fast MDR-detection tools, and optimize clinical resources. Methods Setting, patients and study design This study was performed at a 700-bed university hospital in Barcelona (Spain). Our institution has been conducting a bacteraemia surveillance programme since 1991, with a senior infectious diseases specialist assessing every episode of bacteraemia to determine the source of infection and collecting clinical and microbiological data, and data on antibiotic therapy and outcomes (30 days). For this study, we performed a retrospective analysis of a prospective cohort of all consecutive episodes of BSI occurring in adults with neutropenia from January 2004 to December 2017. The Etico study was approved by the ethics committee board (Comite n Clínica del Hospital Clínic de Barcelona) of our de Investigacio institution. Definitions Neutropenia was defined as an absolute neutrophil count of <500 cells/mm3. Shock was defined as having a systolic pressure <90 mmHg, unresponsive to fluid treatment or requiring vasoactive drug therapy. Previous antibiotic therapy was defined as the use of any antimicrobial agent for 3 days during the month before the BSI occurrence. According to the protocols of our hospital, patients
with an expected neutropenia of >10 days received prophylaxis with fluoroquinolone. Definitions of co-morbidities, site of infection and catheter-related infections have been previously provided [3,15]. Stem cell transplants included both allogeneic and autologous transplants. Pseudomonas aeruginosa was defined as MDR-PA when it was non-susceptible to at least three classes of antibiotics: carbapenems, ureidopenicillins, cephalosporins (ceftazidime and cefepime), monobactams, aminoglycosides and fluoroquinolones. Appropriate empirical antibiotic therapy was reported when including at least one in vitro active antibiotic against the isolated microorganism. Overall mortality was defined as death by any cause within the first 30 days of onset; BSI-related mortality was defined as death in individuals with bacteraemia by unrelenting clinical sepsis. Microbiological methods Blood samples were processed for an incubation period of 5 days with the BACTEC 9240 system or Bactec FX system (Becton-Dickinson Microbiology Systems, Sparks, MD, USA). Isolates were identified by standard techniques and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Antimicrobial susceptibility testing was performed with either a microdilution system (Sensititre, Trek Diagnostic Systems Ltd, East Grinstead, UK or Phoenix System, Becton Dickinson, Franklin Lakes, NJ, USA) or the E-test method (AB Biodisk, Solna, Sweden, or biorieux, Marcy l’Etoile, France). The CLSI or EUCAST breakpoints Me were used to define susceptibility or resistance to antimicrobial agents; intermediate susceptibility was considered as resistance. All MDR strains were confirmed by the E-test methodology throughout the study period. Statistical analysis To identify significant differences among groups, we used the
c2 test or Fisher exact test for categorical variables and the Student's t test or ManneWhitney U test for continuous variables, when deemed appropriate. Gram-negative BSI other than MDRPA and susceptible P. aeruginosa BSI were compared with MDRPA BSI. We performed a multivariate analysis using a forward stepwise approach. Factors analysed as risk factors that were planned in advance were those related with demographics and co-morbid conditions, previous antibiotic therapy, site of acquisition, source of bacteraemia, clinical features, corticosteroid use, and BSI occurring within antibiotic therapy. Two multivariate regression models were used to identify independent risk factors for MDR-PA in individuals with BSI caused by Gram-negative bacilli (when Gram-negative bacilli were identified in blood cultures) as well as in patients with P. aeruginosa BSI (when P. aeruginosa is identified in blood cultures). The multivariate logistic regression analysis of factors potentially associated with MDR-PA included significant variables in the univariate analysis (p < 0.05) with clinical importance. The goodness of fit of the multivariate models was assessed by the HosmereLemeshow test. We also constructed two rules to stratify patients into risk groups for MDR-PA BSI. The resulting b-coefficients of significant predictors of MDR-PA BSI were used to assign a value (risk score) to each variable. A score chart method was followed. We then divided by the smallest coefficient, which by definition had a score of 1. The other predictors received rounded scores [16]. Accuracy of the scores was assessed by the area under the receiver operating characteristic (AUC) curve. With final models, internal validation was performed, using the Bootstrap re-sampling technique. All analyses were carried out with SPSS software (version 18.0; SPSS Inc., Chicago, IL, USA).
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
Results Out of the 661 episodes of Gram-negative BSI, 190 (28.7%) P. aeruginosa BSI were identified. Among P. aeruginosa isolates, 37.2% were resistant to ceftazidime, 38.2% to piperacillin/tazobactam, 42.5% to ciprofloxacin, 38.3% to carbapenems, 1.6% to amikacin and 0% to colistin. MDR-PA was found in 70 (36.8%) individuals, without significant changes over the study period (c2 for trend p 0.086; see Supplementary material, Table S1). Risk factors for BSI caused by MDR-PA when bacteraemia diagnosis is performed Characteristics of both patients with MDR-PA BSI and those with BSI caused by other Gram-negative, non-MDR pathogens are outlined in Table 1. Independent factors associated with MDR-PA among Gram-negative infections in our entire cohort of neutropenic patients were: haematological malignancy (OR
3
3.30; 95% CI 1.15e9.50), pulmonary source of infection (OR 7.85; 95% CI 3.32e18.56), nosocomial-acquired BSI (OR 3.52; 95% CI 1.74-7.09), previous use of antipseudomonal cephalosporin (OR 13.66; 95% CI 6.64e28.10) and piperacillin/tazobactam (OR 2.42; 95% CI 1.04e5.63), and BSI occurring during ceftriaxone treatment (OR 4.27; 95% CI 1.15e15.83). The goodness of fit of the multivariate model was assessed using the HosmereLemeshow test (0.625) (see Supplementary material, Tables S2A and S2B). Results from the multivariate analysis were used to establish the clinical prediction rule. Per regression coefficients, 1 point was assigned to each parameter, except for those of previous use of antipseudomonal cephalosporin (3 points), pulmonary source of infection, and BSI occurring during ceftriaxone treatment (2 points each). The discriminatory power of the clinical prediction ruledas evaluated by the AUCdwas 0.84 (95% CI 0.79e0.89), demonstrating a good ability to predict MDR-PA BSI when Gram-negative bacteraemia diagnosis is confirmed (Fig. 1).
Table 1 Demographic and clinical characteristics of bloodstream infection episodes in neutropenic patients Non-MDR GNB BSI other than PA n ¼ 407 Demographics Age, mean (SD), years 58.4 (16.8) Male sex 233 (57.2) Co-morbid conditions Chronic cardiac disease 32 (7.9) Chronic pulmonary disease 16 (3.9) Diabetes mellitus 46 (11.3) Haematological malignancies 307 (75.4)* Solid neoplasia 79 (19.4)* Transplantation 69 (17)* Previous hospital admission 155 (39.9)* Previous antibiotic therapy Quinolone 101 (24.8)* Aminoglycoside 16 (3.9)* Antipseudomonal cephalosporin 27 (6.6)* Ceftriaxone 21 (5.2) Piperacillin/tazobactam 39 (9.6)* Carbapenems not ertapenem 47 (11.5)* Ertapenem 5 (1.2) Site of acquisition Nosocomial 164 (40.3)* Health care 203 (49.9)* Community 40 (9.8)* Source of bacteraemia Pneumonia 28 (6.9)* Urinary tract 44 (10.8) Skin and soft-tissue infection 7 (1.7) Abdominal 21 (5.2) Catheter-related 55 (13.5) Unknown origin 231 (56.8)* Clinical features at onset Fever 378 (94.5) Septic shock 108 (26.7) BSI occurring during treatment with antibiotic Quinolone 122 (30) Aminoglycoside 5 (1.2) Antipseudomonal cephalosporin 0 (0) Ceftriaxone 9 (2.2)* Piperacillin/tazobactam 13 (3.2) Carbapenems nor ertapenem 19 (4.7)* Ertapenem 0 (0) b-lactams nor ertapenem 52 (12.8)* Corticosteroids use 159 (42.3)* Inappropriate empiric antibiotic therapy 51 (12.5)* Overall-mortality 94 (23.3)* BSI-related mortality 69 (17.1)*
Non-PA MDR GNB BSI n ¼ 64
Non-MDR PA n ¼ 120
MDR-PA BSI n ¼ 70
52.7 (13.9) 34 (53.1)
62.3 (17.3)*** 70 (58.3)
56.9 (13.2) 42 (60)
1 (1.6) 3 (4.7) 4 (6.3) 51 (79.7) 7 (10.9) 10 (15.6)** 32 (50)
10 14 18 76 32 25 49
(8.3) (11.7) (15) (63.3)*** (26.7)*** (20.8)*** (42.6)
2 (2.9) 6 (8.6) 10 (14.3) 62 (88.6) 6 (8.6) 27 (38.6) 35 (53.8)
30 (46.9) 4 (6.3) 6 (9.4)* 10 (15.6) 10 (15.6) 16 (25) 2 (3.1)
14 (11.7)*** 6 (5)*** 15 (12.5)*** 4 (3.3)*** 12 (10)*** 13 (10.8)*** 6 (5)
28 (40) 11 (15.7) 37 (52.9) 8 (11.4) 15 (21.4) 18 (25.7) 0 (0)
44 (68.8) 19 (29.7) 1 (1.6)
30 (25)*** 72 (60)*** 18 (15)***
51 (72.9) 18 (25.7) 1 (1.4)
3 (4.7)** 6 (9.4) 2 (3.1) 2 (3.1) 11 (17.2) 35 (54.7)**
31 (25.8) 3 (2.5) 9 (7.5) 5 (4.2) 18 (15) 48 (40)
23 (32.9) 4 (5.7) 4 (5.7) 1 (1.4) 13 (18.6) 23 (32.9)
60 (93.8) 15 (23.4)
114 (95) 40 (33.3)
65 (95.6) 26 (37.7)
25 (39.1) 0 (0) 1 (1.6) 9 (14.1) 9 (14.1) 9 (14.1) 0 (0) 28 (43.8) 28 (43.8)** 15 (23.4)** 8 (12.5)** 5 (7.8)**
9 (7.5)*** 0 (0)*** 1 (0.8) 1 (0.8)*** 0 (0)*** 3 (2.5)*** 3 (2.5) 9 (7.5)*** 52 (46)*** 9 (7.6)*** 34 (28.6) 30 (25.1)
25 (35.7) 3 (4.3) 1 (1.4) 7 (10) 3 (4.3) 18 (25.7) 0 (0) 30 (42.9) 45 (68.2) 31 (45.6) 29 (42.6) 26 (37.1)
BSI, bloodstream infection; GNB, Gram-negative bacilli; MDR, multidrug-resistant; PA, Pseudomonas aeruginosa; SD, standard deviation. *p < 0.5 comparing non-MDR Gram-negative BSI other than P. aeruginosa versus MDR-PA BSI. **p < 0.5 comparing Non-P. aeruginosa MDR GNB BSI versus MDR-PA BSI. ***p < 0.5 comparing P. aeruginosa BSI other than MDR-PA versus MDR-PA BSI.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
4
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
Fig. 1. Prediction rule and receiver operating characteristics curve for predicting multidrug-resistant Pseudomonas aeruginosa bloodstream infection when Gram-negative bacteria are identified in blood cultures.
Risk factors for BSI caused by MDR-PA when P. aeruginosa bacteraemia is identified Independent factors associated with MDR-PA BSI among patients with P. aeruginosa BSI were nosocomial acquisition of BSI (OR 7.13; 95% CI 2.87e17.67), haematological malignancy (OR 3.44; 95% CI 1.07e10.98), previous use of antipseudomonal cephalosporin (OR 3.82; 95% CI 1.42e10.22) and quinolones (OR 3.97; 95% CI 1.37e11.48), corticosteroid use (OR 2.92; 95% CI 1.15e7.40), and BSI occurring during quinolone (OR 4.88; 95% CI 1.58e15.05) and blactam other than ertapenem antibiotic therapy (OR 4.51; 95% CI 1.45e14.04). The goodness of fit of the multivariate model was assessed using the HosmereLemeshow test (0.69) (see Supplementary material, Tables S3A and S3B). Results from the multivariate analysis were also used to establish the clinical prediction rule. Per regression coefficients, 1 point was assigned to each parameter, except for those of nosocomialacquired BSI and BSI occurring during quinolone treatment (2 points each). The discriminatory power of the clinical prediction ruledas evaluated by the AUCdwas 0.90 (95% CI 0.85e0.95), demonstrating an excellent ability to predict the risk of MDR-PA BSI among patients with P. aeruginosa BSI (Fig. 2). Among high-risk patients (>3 points), 75.3% were resistant to ceftazidime, 76.1% to piperacillin/tazobactam, 79.5% to ciprofloxacin, 78.4% to meropenem and 2.7% to amikacin. When evaluating possible combinations, association of ciprofloxacin barely improved b-lactam susceptibility (73.2%, 73.9% and 76.4% resistance, for combinations with ceftazidime, piperacillin/tazobactam and meropenem, respectively). Association between the different b-lactams also did not result in any improvement. The strains resistant to amikacin were also resistant to all other antibiotics, except for colistin. Table 2 details various cut-off points with different degrees of sensitivity and specificity for prediction optimization per clinical setting. With final models, internal validation was performed to control the apparent optimism of the model, using the Bootstrap resampling technique by extracting 1000 samples. We calculated
the possible deviation of the b-coefficients (bias), with their respective confidence intervals, and the results of each model were compared both in calibration and discrimination, finding a very high similarity between them. Discussion This retrospective analysis of a prospective cohort describes both demographic and clinical features of P. aeruginosa BSI, as well as risk factors for MDR-PA BSI in a current cohort of individuals with neutropenia. We found that MDR-PA BSI is frequent among patients with P. aeruginosa BSI. We identified independent factors related with MDR-PA BSI in two scenarios, that is, when Gramnegative bacteraemia diagnosis is performed and when aetiological identification of P. aeruginosa BSI is completed. Simple, practical clinical scores were generated to stratify the risk of MDRPA BSI, possessing good ability to predict this pathogen among neutropenic patients with BSI. According to our findings, ecologically adverse effects of antibiotic therapy play an important role in the development of MDR-PA BSI. We found a high prevalence of MDR-PA. This observation is concordant with other current studies in haematological patients [7,9,17]; although, specific information about neutropenic patients with bacteraemia is scarce. This high and increasing prevalence of MDR-PA makes P. aeruginosa coverage a major problem worldwide. Interestingly, we found that previous antibiotic use or antibiotic therapy during a BSI episode could have a critical role in the development of MDR-PA BSI. We documented that previous use of some b-lactams and quinolones were independent risk factors for MDR-PA BSI. Similarly, BSI occurring during b-lactams other than ertapenem was a factor independently associated with this pathogen. In this regard, broad-spectrum antibiotics can encourage the selection and undesirable development of MDR organism colonization or infection [18,19]. Similarly, studies have documented that previous quinolone use is a risk factor for subsequent infection with quinolone-resistant and extended-spectrum b-lactamase-producing microorganisms, including P. aeruginosa [19,20]. Consumption of cephalosporins is also associated with an increase in
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
5
Fig. 2. Prediction rule and receiver operating characteristics curve for predicting multidrug-resistant Pseudomonas aeruginosa bloodstream infection when P. aeruginosa is identified in blood cultures.
P. aeruginosa isolation rate; for example, ceftazidime use leads to a significant reduction in P. aeruginosa susceptibility to this antibiotic [21]. Therefore, initiatives that promote the responsible use of antibiotics and limit pressure set by selection for antimicrobialresistant strains should be encouraged in these patients. Clinical decisions regarding empirical treatment for P. aeruginosa or MDR-PA is one of the most notable challenges for physicians attending neutropenic patients. Recently, inappropriate treatment for MDR-PA in individuals with acute leukaemia and BSI has been identified as the only modifiable risk factor independently associated with mortality [3]. New antipseudomonal antibiotics (ceftolozane/tazobactam and ceftazidime/avibactam) have been recently approved. A Spanish nationwide study found that the most active antipseudomonal agents among a total of 1445 P. aeruginosa isolates were ceftolozane/tazobactam and colistin (both 94.6% susceptible), followed by ceftazidime/avibactam (94.2% susceptible) [22]. Similar data were found in the present study, among 35 P. aeruginosa tested for ceftolozane/tazobactam and ceftazidime/ avibactam since 2016, resistance was documented in one (2.85%) case. However, it is challenging to establish the role of these new antibiotics in clinical practice, especially as it concerns empirical treatment for febrile neutropenic patients. Offering broadspectrum coverage or even one of these new antibiotics empirically to every patient with febrile neutropenia to cover the possibility of MDR-PA does not seem a logical option, considering that
most patients will not need such coverage, and this strategy would further increase selection of antimicrobial resistance and be associated with a significant increase in health-care costs. This easy to use clinical tool could be employed to guide the use of antipseudomonal therapy in patients with a recent diagnosis of BSI, and rationalize the use of new microbiological molecular techniques for rapid detection of resistance in high-risk patients. The strengths of this study are the large number of patients included, the prospective collection of the data, and the comprehensive clinical and microbiological data gathered. This information allowed us to identify specific antibiotic therapies associated with an increased risk of developing MDR-PA BSI. In addition, by means of scoring tools, this study aimed to identify and stratify patients with increased risk for MDR-PA BSI within this context. The accuracy of the tool was good, as evaluated by the AUC. Lastly, we performed multivariate analysis and controlled for confounding variables, including optimization of control group selection and comorbid illnesses. However, there are some limitations that should be acknowledged. Our study was conducted at a single centre performing quinolone prophylaxis in patients with an expected neutropenia over 10 days; some factors found might differ from other epidemiological contexts. Additionally, the present scoring tool should be validated in further investigations. Moreover, the first score could have limited utility in institutions that routinely perform rapid diagnosis tests for bloodstream isolates given that
Table 2 Sensitivity and specificity of different operating cut off points Sensitivity
Specificity
PLR
NLR
Predicting multidrug-resistant Pseudomonas aeruginosa BSI when the diagnosis of bacteraemia is performed >1 91.4% 51.1% 1.8 0.1 >3 64.2% 89.3% 6.0 0.4 >5 24.2% 98.4% 15.9 0.7 Predicting multidrug-resistant Pseudomonas aeruginosa BSI when the aetiological identification of P. aeruginosa BSI is performed >1 96.9% 39.8% 1.6 0.07 >3 86.3% 84.0% 5.4 0.1 >5 39.3% 98.2% 22.2 0.8
PPV
NPV
18.3% 41.6% 65.3%
98.0% 94.4% 91.6%
48.5% 75.9% 93.3%
96.0% 91.7% 73.7%
BSI, bloodstream infection; NLR, negative likelihood ratio; NPV, negative predictive value; PLR, positive likelihood ratio; PPV, positive predictive value.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
6
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
there are a few hours between Gram stain results and microbial identification. However, such methods are expensive and not affordable for every patient everywhere. Likewise, other factors, such as previous colonization by MDR-PA, cross-transmission, and previous duration of antibiotic treatment using cumulative dose were not recorded in the present study. Finally, the low number of patients who received certain antibiotics, such as ertapenem, limited their evaluation as risk factors associated with MDR-PA. In conclusion, MDR-PA BSI is frequent in neutropenic patients with P. aeruginosa BSI and is associated with high morbidity and poor prognosis. The risk of MDR-PA BSI can be estimated by evaluating simple demographic and clinical factors. The prediction rule allows for patients to be stratified according to their risk of MDR-PA BSI and possibly new antimicrobial strategies worthy of consideration. Finally, ecologically adverse effects of antibiotic therapy play a pivotal role in the development of MDR-PA BSI.
[8]
Transparency declarations
[9]
CG-V has received honoraria for talks on behalf of Gilead Science, Merck Sharp and Dohme, Pfizer, Jannsen, Novartis, Lilly and a grant support from Gilead Science and Merck Sharp and Dohme. AS has received honoraria for talks on behalf of Merck Sharp and Dohme, Pfizer, Novartis, Angellini, and a grant support from Pfizer. Funding This work was co-funded by a research grant from the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III (FIS PI18/01061), a European Regional Development Fund and a research grant from MSD (2018-00020723). Dr Carolina Garcia-Vidal has received the Grantda grant supported by the Catalan Health INTENSIFICACIO gic de recerca i innovacio en salutd Agency (PERIS (Pla estrate “Strategic Plan for Research and Innovation in HealthCare”)). Our group is recognized by the AGAUR (Project 2017SGR1432) of the Catalan Health Agency. Dr Pedro Puerta-Alcalde has received a predoctoral grant supported by the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III (RH041828). The funders had neither a specific role in study design or collection of data, nor in writing of the paper or decision to submit.
[3]
[4]
[5]
[6]
[7]
[10]
[11]
[12]
[13]
[14]
[15]
[16] [17]
Acknowledgements
[18]
We are grateful for the contributions made by Anthony Armenta in his corrections of the English language/syntax of the publication at hand.
[19]
Appendix A. Supplementary data [20]
Supplementary data to this article can be found online at https://doi.org/10.1016/j.cmi.2019.07.002. [21]
References [1] Nørgaard M, Larsson H, Pedersen G, Schønheyder HC, Sørensen HT. Risk of bacteraemia and mortality in patients with haematological malignancies. Clin Microbiol Infect 2006;12:217e23. [2] Gudiol C, Tubau F, Calatayud L, Garcia-Vidal C, Cisnal M, Sanchez-Ortega I, et al. Bacteraemia due to multidrug-resistant Gram-negative bacilli in cancer
[22]
patients: risk factors, antibiotic therapy and outcomes. J Antimicrob Chemother 2011;66:657e63. Garcia-Vidal C, Cardozo-Espinola C, Puerta-Alcalde P, Marco F, Tellez A, Agü Ero D, et al. Risk factors for mortality in patients with acute leukemia and bloodstream infections in the era of multiresistance. PLoS One 2018;13:1e12. Klastersky J, Ameye L, Maertens J, Georgala A, Muanza F, Aoun M, et al. Bacteraemia in febrile neutropenic cancer patients. Int J Antimicrob Agents 2007;30:51e9. Averbuch D, Orasch C, Cordonnier C, Livermore DM, Mikulska M, Viscoli C, et al. European guidelines for empirical antibacterial therapy for febrile neutropenic patients in the era of growing resistance: summary of the 2011 4th European Conference on Infections in Leukemia. Haematologica 2013;98: 1826e35. Freifeld AG, Bow EJ, Sepkowitz KA, Boeckh MJ, Ito JI, Mullen CA, et al. Highlights from clinical practice guideline for the use of antimicrobial agents in neutropenic patients with cancer: an update by the infectious diseases society of America. Infect Dis Clin Pract 2011;19:352e3. Cattaneo C, Antoniazzi F, Casari S, Ravizzola G, Gelmi M, Pagani C, et al. P. aeruginosa bloodstream infections among hematological patients: an old or new question? Ann Hematol 2012;91:1299e304. Trecarichi EM, Tumbarello M, Caira M, Candoni A, Cattaneo C, Pastore D, et al. Multidrug resistant Pseudomonas aeruginosa bloodstream infection in adult patients with hematologic malignancies. Haematologica 2011;96:e1e3. Tofas P, Samarkos M, Piperaki E-T, Kosmidis C, Triantafyllopoulou I-D, Kotsopoulou M, et al. Pseudomonas aeruginosa bacteraemia in patients with hematologic malignancies: risk factors, treatment and outcome. Diagn Microbiol Infect Dis 2017;88:335e41. Tumbarello M, Spanu T, Caira M, Trecarichi EM, Laurenti L, Montuori E, et al. Factors associated with mortality in bacteremic patients with hematologic malignancies. Diagn Microbiol Infect Dis 2009;64:320e6. rez-Lledo M, Rodríguez-Nún ~ ez O, Morata L, Puerta-Alcalde P, Cardozo C, Sua r C, et al. Current time-to-positivity of blood cultures in the febrile Fehe neutropenia: a tool to be used in stewardship de-escalation strategies. Clin Microbiol Infect 2018;46:1356e64. Boutal H, Vogel A, Bernabeu S, Devilliers K, Creton E, Cotellon G, et al. A multiplex lateral flow immunoassay for the rapid identification of NDM-, KPC-, IMP- and VIM-type and OXA-48-like carbapenemase-producing Enterobacteriaceae. J Antimicrob Chemother 2018;73:909e15. n D, Vergara A, Vila J, et al. García-Fern andez S, Morosini M-I, Marco F, Gijo Evaluation of the eazyplex® SuperBug CRE system for rapid detection of carbapenemases and ESBLs in clinical Enterobacteriaceae isolates recovered at two Spanish hospitals. J Antimicrob Chemother 2015;70:1047e50. Zboromyrska Y, Vergara A, Cosgaya C, Verger G, Mosqueda N, Almela M, et al. Rapid detection of b-lactamases directly from positive blood cultures using a loop-mediated isothermal amplification (LAMP)-based assay. Int J Antimicrob Agents 2015;46:355e6. De la Calle C, Morata L, Cobos-Trigueros N, Martinez JA, Cardozo C, Mensa J, et al. Staphylococcus aureus bacteremic pneumonia. Eur J Clin Microbiol Infect Dis 2016;35:497e502. Steyerberg EW. Clinical prediction models: a practical approach to development, validation, and updating. New York: Springer; 2008. Kim HS, Park BK, koo Kim S, Han SB, Lee JW, Lee D-G, et al. Clinical characteristics and outcomes of Pseudomonas aeruginosa bacteremia in febrile neutropenic children and adolescents with the impact of antibiotic resistance: a retrospective study. BMC Infect Dis 2017;17:500. Trecarichi EM, Pagano L, Candoni A, Pastore D, Cattaneo C, Fanci R, et al. Current epidemiology and antimicrobial resistance data for bacterial bloodstream infections in patients with hematologic malignancies: an Italian multicentre prospective survey. Clin Microbiol Infect 2015;21: 337e43. pez-Dupla M, Martínez J-A, Vidal F, Almela M, Soriano A, Marco F, et al. Lo Previous ciprofloxacin exposure is associated with resistance to b-lactam antibiotics in subsequent Pseudomonas aeruginosa bacteremic isolates. Am J Infect Control 2009;37:753e8. Zervos MJ, Hershberger E, Nicolau DP, Ritchie DJ, Blackner LK, Coyle EA, et al. Relationship between fluoroquinolone use and changes in susceptibility to fluoroquinolones of selected pathogens in 10 United States teaching hospitals, 1991e2000. Clin Infect Dis 2003;37:1643e8. M, Fa brega A, Cobos-Trigueros N, Zamorano L, Ferrer-Navarro M, Sole -Delpierre C, et al. In vivo evolution of resistance of Pseudomonas Balleste aeruginosa strains isolated from patients admitted to an intensive care unit: mechanisms of resistance and antimicrobial exposure. J Antimicrob Chemother 2015;70:3004e13. ~ o E, Zamorano L, Cortes-Lara S, Lo pez-Causape C, S del Barrio-Tofin anchezDiener I, Cabot G, et al. Spanish nationwide survey on Pseudomonas aeruginosa antimicrobial resistance mechanisms and epidemiology. J Antimicrob Chemother 2019.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
Contents lists available at ScienceDirect
Clinical Microbiology and Infection journal homepage: www.clinicalmicrobiologyandinfection.com
Original article
Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection rez-Lledo 3, O. Rodríguez-Nún ~ ez 2, D. Viasus 1, y, P. Puerta-Alcalde 2, y, C. Cardozo 2, M. Sua r 2, F. Marco 4, 5, M. Chumbita 2, E. Moreno-García 2, L. Morata 2, C. Fehe rrez-Garcia 3, 6, J.A. Martínez 2, 6, J. Mensa 2, M. Rovira 3, 6, ndez-Avile s 3, G. Gutie F. Ferna 3, 6 2, 6 , A. Soriano , C. Garcia-Vidal 2, 6, * J. Esteve 1)
Health Sciences Division, Universidad del Norte, and Hospital Universidad del Norte, Barranquilla, Colombia Infectious Diseases Department, Hospital Clínic-IDIBAPS, Barcelona, Spain 3) Haematology Department, Hospital Clínic-IDIBAPS, Barcelona, Spain 4) ISGlobal, Hospital ClínicdUniversitat de Barcelona, Barcelona, Spain 5) stic Biom Microbiology Department, Centre Diagno edic, Hospital Clínic, Barcelona, Spain 6) University of Barcelona, Barcelona, Spain 2)
a r t i c l e i n f o
a b s t r a c t
Article history: Received 23 February 2019 Received in revised form 7 June 2019 Accepted 1 July 2019 Available online xxx
Objectives: To assess risk factors for multidrug-resistant Pseudomonas aeruginosa (MDR-PA) infection in neutropenic patients. Methods: Single-centre retrospective analysis of consecutive bloodstream infection (BSI) episodes (2004 e2017, Barcelona). Two multivariate regression models were used at BSI diagnosis and P. aeruginosa detection. Significant predictors were used to establish rules for stratifying patients according to MDR-PA BSI risk. Results: Of 661 Gram-negative BSI episodes, 190 (28.7%) were caused by P. aeruginosa (70 MDR-PA). Independent factors associated with MDR-PA among Gram-negative organisms were haematological malignancy (OR 3.30; 95% CI 1.15e9.50), pulmonary source of infection (OR 7.85; 95% CI 3.32e18.56), nosocomial-acquired BSI (OR 3.52; 95% CI 1.74e7.09), previous antipseudomonal cephalosporin (OR 13.66; 95% CI 6.64e28.10) and piperacillin/tazobactam (OR 2.42; 95% CI 1.04e5.63), and BSI occurring during ceftriaxone (OR 4.27; 95% CI 1.15e15.83). Once P. aeruginosa was identified as the BSI aetiological pathogen, nosocomial acquisition (OR 7.13; 95% CI 2.87e17.67), haematological malignancy (OR 3.44; 95% CI 1.07e10.98), previous antipseudomonal cephalosporin (OR 3.82; 95% CI 1.42e10.22) and quinolones (OR 3.97; 95% CI 1.37e11.48), corticosteroids (OR 2.92; 95% CI 1.15e7.40), and BSI occurring during quinolone (OR 4.88; 95% CI 1.58e15.05) and b-lactam other than ertapenem (OR 4.51; 95% CI 1.45e14.04) were independently associated with MDR-PA. Per regression coefficients, 1 point was assigned to each parameter, except for nosocomial-acquired BSI (3 points). In the second analysis, a score >3 points identified 60 (86.3%) out of 70 individuals with MDR-PA BSI and discarded 100 (84.2%) out of 120 with non-MDR-PA BSI. Conclusions: A simple score based on demographic and clinical factors allows stratification of individuals with bacteraemia according to their risk of MDR-PA BSI, and may help facilitate the use of rapid MDRdetection tools and improve early antibiotic appropriateness. D. Viasus, Clin Microbiol Infect 2019;▪:1 © 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.
Editor: M. Paul Keywords: Bacteraemia Multidrug-resistant Pseudomonas aeruginosa Neutropenia Risk factors
Introduction * Corresponding author. C. Garcia-Vidal, Infectious Diseases Department, Hospital Clínic-IDIBAPS, Carrer de Villarroel 170, 08036, Barcelona, Spain. E-mail addresses: [email protected], [email protected] (C. Garcia-Vidal). y Diego Viasus and Pedro Puerta-Alcalde contributed equally to this manuscript.
Bloodstream infection (BSI) is an important cause of mortality in onco-haematological patients with febrile neutropenia [1]. Chemotherapies, new treatment modalities, indwelling intravascular
https://doi.org/10.1016/j.cmi.2019.07.002 1198-743X/© 2019 European Society of Clinical Microbiology and Infectious Diseases. Published by Elsevier Ltd. All rights reserved.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
2
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
catheters, prolonged hospitalizations and high consumption of antibiotics render febrile neutropenic patients highly vulnerable to multidrug-resistant (MDR) pathogen colonization or infections [2]. Pseudomonas aeruginosa continues to be a major pathogen causing infections with morbidity and mortality in neutropenic patients [3,4]. Febrile neutropenia is a strong predictor of P. aeruginosa infection; guidelines for the management of neutropenic fever recommend that empirical antibiotic treatment is started with antipseudomonal agents [5,6]. However, empirical coverage for P. aeruginosa has become a challenge due to the rising incidence of MDR P. aeruginosa (MDR-PA) worldwide. Cattaneo et al. [7] and Trecarichi et al. [8] reported rates of MDR among BSI P. aeruginosa isolates of 33% and 71%, respectively. Consequently, inadequate empirical therapy has become quite frequent, as well as being notably associated with an increase in mortality [3,7,9,10]. It is therefore essential that the detection of BSI episodes at risk of MDR-PA in neutropenic patients is improved. Rapid detection methods for bacterial pathogens in blood cultures have become widespread: as recently demonstrated, the great majority of blood cultures in febrile neutropenia are positive within the first 24 h [11]. Similarly, microbiological methods for rapid detection of antibiotic resistance in positive cultures have also been recently developed; however, such methods are expensive and not affordable for every patient [12e14]. Hence, two scenarios are important when modifying empirical antibiotic therapy in neutropenic patients with bacteraemia: the first one is when an individual is diagnosed with bacteraemia (positive blood culture) and the second one is when the causative agent of the bacteraemia has been documented. The availability of susceptibility tests follows pathogen identification by at least 18e24 hours. We hypothesized that some factors might be useful in predicting the risk of MDR-PA in neutropenic patients. Our aim was to assess these factors at BSI diagnosis to facilitate a physician's decisionmaking process concerning broad-spectrum antibiotic coverage. Second, once identification of P. aeruginosa in blood cultures had been completed, we assessed the risk factors for MDR-PA. These factors might help to rationalize the use of antibiotics and fast MDR-detection tools, and optimize clinical resources. Methods Setting, patients and study design This study was performed at a 700-bed university hospital in Barcelona (Spain). Our institution has been conducting a bacteraemia surveillance programme since 1991, with a senior infectious diseases specialist assessing every episode of bacteraemia to determine the source of infection and collecting clinical and microbiological data, and data on antibiotic therapy and outcomes (30 days). For this study, we performed a retrospective analysis of a prospective cohort of all consecutive episodes of BSI occurring in adults with neutropenia from January 2004 to December 2017. The Etico study was approved by the ethics committee board (Comite n Clínica del Hospital Clínic de Barcelona) of our de Investigacio institution. Definitions Neutropenia was defined as an absolute neutrophil count of <500 cells/mm3. Shock was defined as having a systolic pressure <90 mmHg, unresponsive to fluid treatment or requiring vasoactive drug therapy. Previous antibiotic therapy was defined as the use of any antimicrobial agent for 3 days during the month before the BSI occurrence. According to the protocols of our hospital, patients
with an expected neutropenia of >10 days received prophylaxis with fluoroquinolone. Definitions of co-morbidities, site of infection and catheter-related infections have been previously provided [3,15]. Stem cell transplants included both allogeneic and autologous transplants. Pseudomonas aeruginosa was defined as MDR-PA when it was non-susceptible to at least three classes of antibiotics: carbapenems, ureidopenicillins, cephalosporins (ceftazidime and cefepime), monobactams, aminoglycosides and fluoroquinolones. Appropriate empirical antibiotic therapy was reported when including at least one in vitro active antibiotic against the isolated microorganism. Overall mortality was defined as death by any cause within the first 30 days of onset; BSI-related mortality was defined as death in individuals with bacteraemia by unrelenting clinical sepsis. Microbiological methods Blood samples were processed for an incubation period of 5 days with the BACTEC 9240 system or Bactec FX system (Becton-Dickinson Microbiology Systems, Sparks, MD, USA). Isolates were identified by standard techniques and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Antimicrobial susceptibility testing was performed with either a microdilution system (Sensititre, Trek Diagnostic Systems Ltd, East Grinstead, UK or Phoenix System, Becton Dickinson, Franklin Lakes, NJ, USA) or the E-test method (AB Biodisk, Solna, Sweden, or biorieux, Marcy l’Etoile, France). The CLSI or EUCAST breakpoints Me were used to define susceptibility or resistance to antimicrobial agents; intermediate susceptibility was considered as resistance. All MDR strains were confirmed by the E-test methodology throughout the study period. Statistical analysis To identify significant differences among groups, we used the
c2 test or Fisher exact test for categorical variables and the Student's t test or ManneWhitney U test for continuous variables, when deemed appropriate. Gram-negative BSI other than MDRPA and susceptible P. aeruginosa BSI were compared with MDRPA BSI. We performed a multivariate analysis using a forward stepwise approach. Factors analysed as risk factors that were planned in advance were those related with demographics and co-morbid conditions, previous antibiotic therapy, site of acquisition, source of bacteraemia, clinical features, corticosteroid use, and BSI occurring within antibiotic therapy. Two multivariate regression models were used to identify independent risk factors for MDR-PA in individuals with BSI caused by Gram-negative bacilli (when Gram-negative bacilli were identified in blood cultures) as well as in patients with P. aeruginosa BSI (when P. aeruginosa is identified in blood cultures). The multivariate logistic regression analysis of factors potentially associated with MDR-PA included significant variables in the univariate analysis (p < 0.05) with clinical importance. The goodness of fit of the multivariate models was assessed by the HosmereLemeshow test. We also constructed two rules to stratify patients into risk groups for MDR-PA BSI. The resulting b-coefficients of significant predictors of MDR-PA BSI were used to assign a value (risk score) to each variable. A score chart method was followed. We then divided by the smallest coefficient, which by definition had a score of 1. The other predictors received rounded scores [16]. Accuracy of the scores was assessed by the area under the receiver operating characteristic (AUC) curve. With final models, internal validation was performed, using the Bootstrap re-sampling technique. All analyses were carried out with SPSS software (version 18.0; SPSS Inc., Chicago, IL, USA).
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
Results Out of the 661 episodes of Gram-negative BSI, 190 (28.7%) P. aeruginosa BSI were identified. Among P. aeruginosa isolates, 37.2% were resistant to ceftazidime, 38.2% to piperacillin/tazobactam, 42.5% to ciprofloxacin, 38.3% to carbapenems, 1.6% to amikacin and 0% to colistin. MDR-PA was found in 70 (36.8%) individuals, without significant changes over the study period (c2 for trend p 0.086; see Supplementary material, Table S1). Risk factors for BSI caused by MDR-PA when bacteraemia diagnosis is performed Characteristics of both patients with MDR-PA BSI and those with BSI caused by other Gram-negative, non-MDR pathogens are outlined in Table 1. Independent factors associated with MDR-PA among Gram-negative infections in our entire cohort of neutropenic patients were: haematological malignancy (OR
3
3.30; 95% CI 1.15e9.50), pulmonary source of infection (OR 7.85; 95% CI 3.32e18.56), nosocomial-acquired BSI (OR 3.52; 95% CI 1.74-7.09), previous use of antipseudomonal cephalosporin (OR 13.66; 95% CI 6.64e28.10) and piperacillin/tazobactam (OR 2.42; 95% CI 1.04e5.63), and BSI occurring during ceftriaxone treatment (OR 4.27; 95% CI 1.15e15.83). The goodness of fit of the multivariate model was assessed using the HosmereLemeshow test (0.625) (see Supplementary material, Tables S2A and S2B). Results from the multivariate analysis were used to establish the clinical prediction rule. Per regression coefficients, 1 point was assigned to each parameter, except for those of previous use of antipseudomonal cephalosporin (3 points), pulmonary source of infection, and BSI occurring during ceftriaxone treatment (2 points each). The discriminatory power of the clinical prediction ruledas evaluated by the AUCdwas 0.84 (95% CI 0.79e0.89), demonstrating a good ability to predict MDR-PA BSI when Gram-negative bacteraemia diagnosis is confirmed (Fig. 1).
Table 1 Demographic and clinical characteristics of bloodstream infection episodes in neutropenic patients Non-MDR GNB BSI other than PA n ¼ 407 Demographics Age, mean (SD), years 58.4 (16.8) Male sex 233 (57.2) Co-morbid conditions Chronic cardiac disease 32 (7.9) Chronic pulmonary disease 16 (3.9) Diabetes mellitus 46 (11.3) Haematological malignancies 307 (75.4)* Solid neoplasia 79 (19.4)* Transplantation 69 (17)* Previous hospital admission 155 (39.9)* Previous antibiotic therapy Quinolone 101 (24.8)* Aminoglycoside 16 (3.9)* Antipseudomonal cephalosporin 27 (6.6)* Ceftriaxone 21 (5.2) Piperacillin/tazobactam 39 (9.6)* Carbapenems not ertapenem 47 (11.5)* Ertapenem 5 (1.2) Site of acquisition Nosocomial 164 (40.3)* Health care 203 (49.9)* Community 40 (9.8)* Source of bacteraemia Pneumonia 28 (6.9)* Urinary tract 44 (10.8) Skin and soft-tissue infection 7 (1.7) Abdominal 21 (5.2) Catheter-related 55 (13.5) Unknown origin 231 (56.8)* Clinical features at onset Fever 378 (94.5) Septic shock 108 (26.7) BSI occurring during treatment with antibiotic Quinolone 122 (30) Aminoglycoside 5 (1.2) Antipseudomonal cephalosporin 0 (0) Ceftriaxone 9 (2.2)* Piperacillin/tazobactam 13 (3.2) Carbapenems nor ertapenem 19 (4.7)* Ertapenem 0 (0) b-lactams nor ertapenem 52 (12.8)* Corticosteroids use 159 (42.3)* Inappropriate empiric antibiotic therapy 51 (12.5)* Overall-mortality 94 (23.3)* BSI-related mortality 69 (17.1)*
Non-PA MDR GNB BSI n ¼ 64
Non-MDR PA n ¼ 120
MDR-PA BSI n ¼ 70
52.7 (13.9) 34 (53.1)
62.3 (17.3)*** 70 (58.3)
56.9 (13.2) 42 (60)
1 (1.6) 3 (4.7) 4 (6.3) 51 (79.7) 7 (10.9) 10 (15.6)** 32 (50)
10 14 18 76 32 25 49
(8.3) (11.7) (15) (63.3)*** (26.7)*** (20.8)*** (42.6)
2 (2.9) 6 (8.6) 10 (14.3) 62 (88.6) 6 (8.6) 27 (38.6) 35 (53.8)
30 (46.9) 4 (6.3) 6 (9.4)* 10 (15.6) 10 (15.6) 16 (25) 2 (3.1)
14 (11.7)*** 6 (5)*** 15 (12.5)*** 4 (3.3)*** 12 (10)*** 13 (10.8)*** 6 (5)
28 (40) 11 (15.7) 37 (52.9) 8 (11.4) 15 (21.4) 18 (25.7) 0 (0)
44 (68.8) 19 (29.7) 1 (1.6)
30 (25)*** 72 (60)*** 18 (15)***
51 (72.9) 18 (25.7) 1 (1.4)
3 (4.7)** 6 (9.4) 2 (3.1) 2 (3.1) 11 (17.2) 35 (54.7)**
31 (25.8) 3 (2.5) 9 (7.5) 5 (4.2) 18 (15) 48 (40)
23 (32.9) 4 (5.7) 4 (5.7) 1 (1.4) 13 (18.6) 23 (32.9)
60 (93.8) 15 (23.4)
114 (95) 40 (33.3)
65 (95.6) 26 (37.7)
25 (39.1) 0 (0) 1 (1.6) 9 (14.1) 9 (14.1) 9 (14.1) 0 (0) 28 (43.8) 28 (43.8)** 15 (23.4)** 8 (12.5)** 5 (7.8)**
9 (7.5)*** 0 (0)*** 1 (0.8) 1 (0.8)*** 0 (0)*** 3 (2.5)*** 3 (2.5) 9 (7.5)*** 52 (46)*** 9 (7.6)*** 34 (28.6) 30 (25.1)
25 (35.7) 3 (4.3) 1 (1.4) 7 (10) 3 (4.3) 18 (25.7) 0 (0) 30 (42.9) 45 (68.2) 31 (45.6) 29 (42.6) 26 (37.1)
BSI, bloodstream infection; GNB, Gram-negative bacilli; MDR, multidrug-resistant; PA, Pseudomonas aeruginosa; SD, standard deviation. *p < 0.5 comparing non-MDR Gram-negative BSI other than P. aeruginosa versus MDR-PA BSI. **p < 0.5 comparing Non-P. aeruginosa MDR GNB BSI versus MDR-PA BSI. ***p < 0.5 comparing P. aeruginosa BSI other than MDR-PA versus MDR-PA BSI.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
4
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
Fig. 1. Prediction rule and receiver operating characteristics curve for predicting multidrug-resistant Pseudomonas aeruginosa bloodstream infection when Gram-negative bacteria are identified in blood cultures.
Risk factors for BSI caused by MDR-PA when P. aeruginosa bacteraemia is identified Independent factors associated with MDR-PA BSI among patients with P. aeruginosa BSI were nosocomial acquisition of BSI (OR 7.13; 95% CI 2.87e17.67), haematological malignancy (OR 3.44; 95% CI 1.07e10.98), previous use of antipseudomonal cephalosporin (OR 3.82; 95% CI 1.42e10.22) and quinolones (OR 3.97; 95% CI 1.37e11.48), corticosteroid use (OR 2.92; 95% CI 1.15e7.40), and BSI occurring during quinolone (OR 4.88; 95% CI 1.58e15.05) and blactam other than ertapenem antibiotic therapy (OR 4.51; 95% CI 1.45e14.04). The goodness of fit of the multivariate model was assessed using the HosmereLemeshow test (0.69) (see Supplementary material, Tables S3A and S3B). Results from the multivariate analysis were also used to establish the clinical prediction rule. Per regression coefficients, 1 point was assigned to each parameter, except for those of nosocomialacquired BSI and BSI occurring during quinolone treatment (2 points each). The discriminatory power of the clinical prediction ruledas evaluated by the AUCdwas 0.90 (95% CI 0.85e0.95), demonstrating an excellent ability to predict the risk of MDR-PA BSI among patients with P. aeruginosa BSI (Fig. 2). Among high-risk patients (>3 points), 75.3% were resistant to ceftazidime, 76.1% to piperacillin/tazobactam, 79.5% to ciprofloxacin, 78.4% to meropenem and 2.7% to amikacin. When evaluating possible combinations, association of ciprofloxacin barely improved b-lactam susceptibility (73.2%, 73.9% and 76.4% resistance, for combinations with ceftazidime, piperacillin/tazobactam and meropenem, respectively). Association between the different b-lactams also did not result in any improvement. The strains resistant to amikacin were also resistant to all other antibiotics, except for colistin. Table 2 details various cut-off points with different degrees of sensitivity and specificity for prediction optimization per clinical setting. With final models, internal validation was performed to control the apparent optimism of the model, using the Bootstrap resampling technique by extracting 1000 samples. We calculated
the possible deviation of the b-coefficients (bias), with their respective confidence intervals, and the results of each model were compared both in calibration and discrimination, finding a very high similarity between them. Discussion This retrospective analysis of a prospective cohort describes both demographic and clinical features of P. aeruginosa BSI, as well as risk factors for MDR-PA BSI in a current cohort of individuals with neutropenia. We found that MDR-PA BSI is frequent among patients with P. aeruginosa BSI. We identified independent factors related with MDR-PA BSI in two scenarios, that is, when Gramnegative bacteraemia diagnosis is performed and when aetiological identification of P. aeruginosa BSI is completed. Simple, practical clinical scores were generated to stratify the risk of MDRPA BSI, possessing good ability to predict this pathogen among neutropenic patients with BSI. According to our findings, ecologically adverse effects of antibiotic therapy play an important role in the development of MDR-PA BSI. We found a high prevalence of MDR-PA. This observation is concordant with other current studies in haematological patients [7,9,17]; although, specific information about neutropenic patients with bacteraemia is scarce. This high and increasing prevalence of MDR-PA makes P. aeruginosa coverage a major problem worldwide. Interestingly, we found that previous antibiotic use or antibiotic therapy during a BSI episode could have a critical role in the development of MDR-PA BSI. We documented that previous use of some b-lactams and quinolones were independent risk factors for MDR-PA BSI. Similarly, BSI occurring during b-lactams other than ertapenem was a factor independently associated with this pathogen. In this regard, broad-spectrum antibiotics can encourage the selection and undesirable development of MDR organism colonization or infection [18,19]. Similarly, studies have documented that previous quinolone use is a risk factor for subsequent infection with quinolone-resistant and extended-spectrum b-lactamase-producing microorganisms, including P. aeruginosa [19,20]. Consumption of cephalosporins is also associated with an increase in
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
5
Fig. 2. Prediction rule and receiver operating characteristics curve for predicting multidrug-resistant Pseudomonas aeruginosa bloodstream infection when P. aeruginosa is identified in blood cultures.
P. aeruginosa isolation rate; for example, ceftazidime use leads to a significant reduction in P. aeruginosa susceptibility to this antibiotic [21]. Therefore, initiatives that promote the responsible use of antibiotics and limit pressure set by selection for antimicrobialresistant strains should be encouraged in these patients. Clinical decisions regarding empirical treatment for P. aeruginosa or MDR-PA is one of the most notable challenges for physicians attending neutropenic patients. Recently, inappropriate treatment for MDR-PA in individuals with acute leukaemia and BSI has been identified as the only modifiable risk factor independently associated with mortality [3]. New antipseudomonal antibiotics (ceftolozane/tazobactam and ceftazidime/avibactam) have been recently approved. A Spanish nationwide study found that the most active antipseudomonal agents among a total of 1445 P. aeruginosa isolates were ceftolozane/tazobactam and colistin (both 94.6% susceptible), followed by ceftazidime/avibactam (94.2% susceptible) [22]. Similar data were found in the present study, among 35 P. aeruginosa tested for ceftolozane/tazobactam and ceftazidime/ avibactam since 2016, resistance was documented in one (2.85%) case. However, it is challenging to establish the role of these new antibiotics in clinical practice, especially as it concerns empirical treatment for febrile neutropenic patients. Offering broadspectrum coverage or even one of these new antibiotics empirically to every patient with febrile neutropenia to cover the possibility of MDR-PA does not seem a logical option, considering that
most patients will not need such coverage, and this strategy would further increase selection of antimicrobial resistance and be associated with a significant increase in health-care costs. This easy to use clinical tool could be employed to guide the use of antipseudomonal therapy in patients with a recent diagnosis of BSI, and rationalize the use of new microbiological molecular techniques for rapid detection of resistance in high-risk patients. The strengths of this study are the large number of patients included, the prospective collection of the data, and the comprehensive clinical and microbiological data gathered. This information allowed us to identify specific antibiotic therapies associated with an increased risk of developing MDR-PA BSI. In addition, by means of scoring tools, this study aimed to identify and stratify patients with increased risk for MDR-PA BSI within this context. The accuracy of the tool was good, as evaluated by the AUC. Lastly, we performed multivariate analysis and controlled for confounding variables, including optimization of control group selection and comorbid illnesses. However, there are some limitations that should be acknowledged. Our study was conducted at a single centre performing quinolone prophylaxis in patients with an expected neutropenia over 10 days; some factors found might differ from other epidemiological contexts. Additionally, the present scoring tool should be validated in further investigations. Moreover, the first score could have limited utility in institutions that routinely perform rapid diagnosis tests for bloodstream isolates given that
Table 2 Sensitivity and specificity of different operating cut off points Sensitivity
Specificity
PLR
NLR
Predicting multidrug-resistant Pseudomonas aeruginosa BSI when the diagnosis of bacteraemia is performed >1 91.4% 51.1% 1.8 0.1 >3 64.2% 89.3% 6.0 0.4 >5 24.2% 98.4% 15.9 0.7 Predicting multidrug-resistant Pseudomonas aeruginosa BSI when the aetiological identification of P. aeruginosa BSI is performed >1 96.9% 39.8% 1.6 0.07 >3 86.3% 84.0% 5.4 0.1 >5 39.3% 98.2% 22.2 0.8
PPV
NPV
18.3% 41.6% 65.3%
98.0% 94.4% 91.6%
48.5% 75.9% 93.3%
96.0% 91.7% 73.7%
BSI, bloodstream infection; NLR, negative likelihood ratio; NPV, negative predictive value; PLR, positive likelihood ratio; PPV, positive predictive value.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002
6
D. Viasus et al. / Clinical Microbiology and Infection xxx (xxxx) xxx
there are a few hours between Gram stain results and microbial identification. However, such methods are expensive and not affordable for every patient everywhere. Likewise, other factors, such as previous colonization by MDR-PA, cross-transmission, and previous duration of antibiotic treatment using cumulative dose were not recorded in the present study. Finally, the low number of patients who received certain antibiotics, such as ertapenem, limited their evaluation as risk factors associated with MDR-PA. In conclusion, MDR-PA BSI is frequent in neutropenic patients with P. aeruginosa BSI and is associated with high morbidity and poor prognosis. The risk of MDR-PA BSI can be estimated by evaluating simple demographic and clinical factors. The prediction rule allows for patients to be stratified according to their risk of MDR-PA BSI and possibly new antimicrobial strategies worthy of consideration. Finally, ecologically adverse effects of antibiotic therapy play a pivotal role in the development of MDR-PA BSI.
[8]
Transparency declarations
[9]
CG-V has received honoraria for talks on behalf of Gilead Science, Merck Sharp and Dohme, Pfizer, Jannsen, Novartis, Lilly and a grant support from Gilead Science and Merck Sharp and Dohme. AS has received honoraria for talks on behalf of Merck Sharp and Dohme, Pfizer, Novartis, Angellini, and a grant support from Pfizer. Funding This work was co-funded by a research grant from the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III (FIS PI18/01061), a European Regional Development Fund and a research grant from MSD (2018-00020723). Dr Carolina Garcia-Vidal has received the Grantda grant supported by the Catalan Health INTENSIFICACIO gic de recerca i innovacio en salutd Agency (PERIS (Pla estrate “Strategic Plan for Research and Innovation in HealthCare”)). Our group is recognized by the AGAUR (Project 2017SGR1432) of the Catalan Health Agency. Dr Pedro Puerta-Alcalde has received a predoctoral grant supported by the Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III (RH041828). The funders had neither a specific role in study design or collection of data, nor in writing of the paper or decision to submit.
[3]
[4]
[5]
[6]
[7]
[10]
[11]
[12]
[13]
[14]
[15]
[16] [17]
Acknowledgements
[18]
We are grateful for the contributions made by Anthony Armenta in his corrections of the English language/syntax of the publication at hand.
[19]
Appendix A. Supplementary data [20]
Supplementary data to this article can be found online at https://doi.org/10.1016/j.cmi.2019.07.002. [21]
References [1] Nørgaard M, Larsson H, Pedersen G, Schønheyder HC, Sørensen HT. Risk of bacteraemia and mortality in patients with haematological malignancies. Clin Microbiol Infect 2006;12:217e23. [2] Gudiol C, Tubau F, Calatayud L, Garcia-Vidal C, Cisnal M, Sanchez-Ortega I, et al. Bacteraemia due to multidrug-resistant Gram-negative bacilli in cancer
[22]
patients: risk factors, antibiotic therapy and outcomes. J Antimicrob Chemother 2011;66:657e63. Garcia-Vidal C, Cardozo-Espinola C, Puerta-Alcalde P, Marco F, Tellez A, Agü Ero D, et al. Risk factors for mortality in patients with acute leukemia and bloodstream infections in the era of multiresistance. PLoS One 2018;13:1e12. Klastersky J, Ameye L, Maertens J, Georgala A, Muanza F, Aoun M, et al. Bacteraemia in febrile neutropenic cancer patients. Int J Antimicrob Agents 2007;30:51e9. Averbuch D, Orasch C, Cordonnier C, Livermore DM, Mikulska M, Viscoli C, et al. European guidelines for empirical antibacterial therapy for febrile neutropenic patients in the era of growing resistance: summary of the 2011 4th European Conference on Infections in Leukemia. Haematologica 2013;98: 1826e35. Freifeld AG, Bow EJ, Sepkowitz KA, Boeckh MJ, Ito JI, Mullen CA, et al. Highlights from clinical practice guideline for the use of antimicrobial agents in neutropenic patients with cancer: an update by the infectious diseases society of America. Infect Dis Clin Pract 2011;19:352e3. Cattaneo C, Antoniazzi F, Casari S, Ravizzola G, Gelmi M, Pagani C, et al. P. aeruginosa bloodstream infections among hematological patients: an old or new question? Ann Hematol 2012;91:1299e304. Trecarichi EM, Tumbarello M, Caira M, Candoni A, Cattaneo C, Pastore D, et al. Multidrug resistant Pseudomonas aeruginosa bloodstream infection in adult patients with hematologic malignancies. Haematologica 2011;96:e1e3. Tofas P, Samarkos M, Piperaki E-T, Kosmidis C, Triantafyllopoulou I-D, Kotsopoulou M, et al. Pseudomonas aeruginosa bacteraemia in patients with hematologic malignancies: risk factors, treatment and outcome. Diagn Microbiol Infect Dis 2017;88:335e41. Tumbarello M, Spanu T, Caira M, Trecarichi EM, Laurenti L, Montuori E, et al. Factors associated with mortality in bacteremic patients with hematologic malignancies. Diagn Microbiol Infect Dis 2009;64:320e6. rez-Lledo M, Rodríguez-Nún ~ ez O, Morata L, Puerta-Alcalde P, Cardozo C, Sua r C, et al. Current time-to-positivity of blood cultures in the febrile Fehe neutropenia: a tool to be used in stewardship de-escalation strategies. Clin Microbiol Infect 2018;46:1356e64. Boutal H, Vogel A, Bernabeu S, Devilliers K, Creton E, Cotellon G, et al. A multiplex lateral flow immunoassay for the rapid identification of NDM-, KPC-, IMP- and VIM-type and OXA-48-like carbapenemase-producing Enterobacteriaceae. J Antimicrob Chemother 2018;73:909e15. n D, Vergara A, Vila J, et al. García-Fern andez S, Morosini M-I, Marco F, Gijo Evaluation of the eazyplex® SuperBug CRE system for rapid detection of carbapenemases and ESBLs in clinical Enterobacteriaceae isolates recovered at two Spanish hospitals. J Antimicrob Chemother 2015;70:1047e50. Zboromyrska Y, Vergara A, Cosgaya C, Verger G, Mosqueda N, Almela M, et al. Rapid detection of b-lactamases directly from positive blood cultures using a loop-mediated isothermal amplification (LAMP)-based assay. Int J Antimicrob Agents 2015;46:355e6. De la Calle C, Morata L, Cobos-Trigueros N, Martinez JA, Cardozo C, Mensa J, et al. Staphylococcus aureus bacteremic pneumonia. Eur J Clin Microbiol Infect Dis 2016;35:497e502. Steyerberg EW. Clinical prediction models: a practical approach to development, validation, and updating. New York: Springer; 2008. Kim HS, Park BK, koo Kim S, Han SB, Lee JW, Lee D-G, et al. Clinical characteristics and outcomes of Pseudomonas aeruginosa bacteremia in febrile neutropenic children and adolescents with the impact of antibiotic resistance: a retrospective study. BMC Infect Dis 2017;17:500. Trecarichi EM, Pagano L, Candoni A, Pastore D, Cattaneo C, Fanci R, et al. Current epidemiology and antimicrobial resistance data for bacterial bloodstream infections in patients with hematologic malignancies: an Italian multicentre prospective survey. Clin Microbiol Infect 2015;21: 337e43. pez-Dupla M, Martínez J-A, Vidal F, Almela M, Soriano A, Marco F, et al. Lo Previous ciprofloxacin exposure is associated with resistance to b-lactam antibiotics in subsequent Pseudomonas aeruginosa bacteremic isolates. Am J Infect Control 2009;37:753e8. Zervos MJ, Hershberger E, Nicolau DP, Ritchie DJ, Blackner LK, Coyle EA, et al. Relationship between fluoroquinolone use and changes in susceptibility to fluoroquinolones of selected pathogens in 10 United States teaching hospitals, 1991e2000. Clin Infect Dis 2003;37:1643e8. M, Fa brega A, Cobos-Trigueros N, Zamorano L, Ferrer-Navarro M, Sole -Delpierre C, et al. In vivo evolution of resistance of Pseudomonas Balleste aeruginosa strains isolated from patients admitted to an intensive care unit: mechanisms of resistance and antimicrobial exposure. J Antimicrob Chemother 2015;70:3004e13. ~ o E, Zamorano L, Cortes-Lara S, Lo pez-Causape C, S del Barrio-Tofin anchezDiener I, Cabot G, et al. Spanish nationwide survey on Pseudomonas aeruginosa antimicrobial resistance mechanisms and epidemiology. J Antimicrob Chemother 2019.
Please cite this article as: Viasus D et al., Predictors of multidrug-resistant Pseudomonas aeruginosa in neutropenic patients with bloodstream infection, Clinical Microbiology and Infection, https://doi.org/10.1016/j.cmi.2019.07.002