Bloodstream infections in adult patients with malignancy, epidemiology, microbiology, and risk factors associated with mortality and multi-drug resistance

Background This study aimed to investigate the epidemiology, microbiology, and risk factors associated with mortality and multi-drug resistance bacterial bloodstream infections (BSIs) among adult cancer patients in Shiraz, Iran. We also report a four-year trend of antimicrobial resistance patterns of BSIs. Methods We conducted a retrospective study at a referral oncology hospital from July 2015 to August 2019, which included all adults with confirmed BSI. Results 2393 blood cultures tested during the four-year study period; 414 positive cultures were included. The mean age of our patients was 47.57 ± 17.46 years old. Central Line-Associated BSI (CLABSI) was more common in solid tumors than patients with hematological malignancies. Gram-negative (GN) bacteria were more detected (63.3%, 262) than gram-positive bacteria (36.7%, 152). Escherichia coli was the most common gram-negative organism (123/262, 47%), followed by Pseudomonas spp. (82/262, 31%) and Klebsiella pneumoniae (38/262, 14.5%). Coagulase-negative staphylococci (CoNS) was the most frequently isolated pathogen among gram-positive bacteria (83/152, 54.6%). Acinetobacter spp., Pseudomonas spp., E. coli, and K. pneumoniae were the most common Extended-Spectrum Beta-Lactamase (ESBL) producers (100, 96.2, 66.7%, and 60.7, respectively). Acinetobacter spp., Pseudomonas spp., Enterobacter spp., E. coli, and K. pneumoniae were the most common carbapenem-resistant (CR) isolates (77.8, 70.7, 33.3, 24.4, and 13.2%, respectively). Out of 257 Enterobacterales and non-fermenter gram-negative BSIs, 39.3% (101/257) were carbapenem-resistant. Although the incidence of multi-drug resistance (MDR) gram-negative BSI increased annually during 2015–2018, the mortality rate of gram-negative BSI remains unchanged at about 20% (p-value = 0.55); however, the mortality rate was significantly greater (35.4%) in those with resistant gram-positive BSI (p-value = 0.001). The overall mortality rate was 21.5%. Early (7-day mortality) and late mortality rate (30-day mortality) were 10 and 3.4%, respectively. Conclusions The emergence of MDR gram-negative BSI is a significant healthcare problem in oncology centers. The high proportion of the most frequently isolated pathogens were CR and ESBL-producing Enterobacterales and Pseudomonas spp. We have few effective choices against MDRGN BSI, especially in high-risk cancer patients, which necessitate newer treatment options. Supplementary Information The online version contains supplementary material available at 10.1186/s12879-021-06243-z.


Conclusions:
The emergence of MDR gram-negative BSI is a significant healthcare problem in oncology centers. The high proportion of the most frequently isolated pathogens were CR and ESBL-producing Enterobacterales and Pseudomonas spp. We have few effective choices against MDRGN BSI, especially in high-risk cancer patients, which necessitate newer treatment options.

Background
Bacterial bloodstream infection (BSI) is one of the most common complications of chemotherapy-induced neutropenia in patients with hematologic malignancies and solid organ tumors [1][2][3], which is associated with high mortality and morbidity [4][5][6][7][8]. Bacterial BSIs account for the etiologic cause of approximately 20 to 30% of all febrile neutropenic episodes in adult patients with malignancy [9,10]. While proper diagnosis and treatment are essential to decrease BSI-associated complications, inappropriate empiric antimicrobial therapy increases mortality [11]. BSI's reported crude mortality rates to reach as high as 34 to 50%, especially in MDR gramnegative BSI [9,12]. The antimicrobial stewardship program (ASP) helps to reduce the overuse of antibiotics and control the increased antimicrobial resistance. Surveillance of antimicrobial resistance is one of ASP's critical aspects and guides clinicians for appropriate empiric antimicrobial therapy [13,14].
This study aimed to determine the current epidemiology of bacterial BSI and its changes during the different study years in a large cohort of patients with solid organ and hematological malignancy. We also assessed BSI attributed mortality risk factors and MDR gramnegative BSI predictors.

Setting and data collection
This study performed at Amir oncology hospital, an educational 100-bed inpatient center. The adult units consist of four inpatient wards and an autologous hematopoietic stem cell transplantation ward. Since 2015 our institution has carried out a blood culture surveillance program using an automated blood culture system (BD BACTEC™). Hospital Information System (HIS) and microbiology department records used for data collection. Patients followed up 30 days after BSI by infectious disease specialist consultant (AA) and chief infection control unit staff (MS).

Study population and design
In this retrospective single-center study, we analyzed all consecutive episodes of BSI occurring in adult patients with hematological malignancies and solid organ tumors from July 2015 to August 2019. Each patient ≥18 years of age with positive blood culture was considered to BSI when he/she had clinical signs and symptoms of bacteremia. Patients with a fungal infection, contaminant result, or a mix of more than two organisms, excluded from this study. Finally, 414 patients recruited, and data analyzed for age, gender, underlying diseases, presence of central venous catheters, leucocyte count, neutropenia, etiologic microorganisms, susceptibility testing, and outcome.

Definitions
Patient with a recognized bacterial pathogen, which not included on the commensal list, identified from one or more blood specimens obtained by culture and at least one of the following signs or symptoms: fever (> 38°C), chills, or hypotension and not be related to an infection at another site considered as true BSI [15].
Febrile neutropenia defined as temperature > 38.5°C or two consecutive temperature > 38°C for 2 h and an absolute neutrophil count< 0.5 X 10 9 cell/L or expected to fall below < 0.5 X 10 9 cell/L. Imipenem, meropenem, cefepime, piperacillin-tazobactam, and colistin are available on our hospital formulary. We use piperacillintazobactam and carbapenem as the first-line agents for patients with febrile neutropenia.
Methicillin-resistant coagulase-negative staphylococci (MRCoNS) defined as cefoxitin-resistant strains. Enterobacteriaceae family, P. aeruginosa, and A. baumannii isolates resistant to ceftazidime or cefotaxime are considered extended-spectrum beta-lactamase (ESBL) producers [16]. Phenotypic confirmation of ESBL production carried out by using the double-disk synergy test [17]. Carbapenemresistant Enterobacterales (CREs), carbapenem-resistant Pseudomonas spp., and carbapenem-resistant Acinetobacter spp. isolates defined as Enterobacterales that test intermediate or resistant to one or more carbapenems using the CLSI current breakpoints. However, not all isolates tested against all carbapenems [16]. MDR defined as the strain non-susceptible to at least one agent in ≥3 classes of antibiotics, including carbapenems, combinations of betalactams plus beta-lactamase inhibitors, cephalosporins, aminoglycosides, and fluoroquinolones [18].

Microbiological methods
Blood cultures obtained by physicians based on clinical suspicion of bacteremia. BACTEC™ FX Automated Blood Culture Systems used for detection of BSI. Timeto-Detection (TTD) defined as the time between the placement of each blood culture bottle in the incubation cabinet and the detection of growth. Based on the result of the gram-stain, the subculture routinely was performed, and differentiation tests such as catalase, oxidase, coagulase, bacitracin, optochin, and CAMP test were applied. In our center, the susceptibility testing is done based on the disc diffusion method (Kirby-Bauer method) according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [17] using commercial antibiotic discs (MAST Group Ltd.; UK).

Statistical analysis
In this study, we used a univariate logistic regression model to examine the critical factors that may be influencing the patient's survival status and possible predictors of MDR gram-negative BSI. All variables in the univariate analysis (P ≤ 0.25) and variables with clinical significance entered into a multivariable model. The independent variables of sex, age, malignancy, ESR, CRP, leukopenia, neutropenia, febrile neutropenia, year, and Enterobacterales separately entered a logistic regression model. We reported odds ratio values and also the confidence interval of the odds ratio for each variable. A pvalue < 0.05 was considered statistically significant. The analysis was done by SPSS version 25 (IBM Corp., Armonk, NY, USA).

Clinical and laboratory features
The mean of white blood cell (WBC), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) were (6.94 ± 11.28) X 10 9 /L, (71.68 ± 37.26) mm/hour, (82.59 ± 41.44) mg/dl, respectively. ESR and CRP were not statistically different between two malignancy types (P = 0.6646; 95% CI: − 7.07 to 11.07 and P = 0.0663, 95% CI: − 19.64 to 0.64, respectively). Neutropenia was more prominent in those with hematologic malignancies (117, 69.6%) compare with patients with solid organ tumors (38,22.6%); P < 0.001. Febrile neutropenia also was significantly higher in those with hematologic malignancies (83, 74.8%; P < 0.001). Overall, 63.5% of positive culture was primary BSI, and 36.5% was CLABSI. CLABSI was more common in those with solid tumors; the difference was not significant p = 0.165. The most reported TTD was less than 12 h (188/414, 50%), while 30.3% detected during 12-24 h, 11.7% between 24 and 36 h, and only 8% diagnosed in more than 36 h by BACTEC™ Systems. The rate of admission, BSI, febrile neutropenia, and mortality during the study years are shown in Fig. 2. The summary characteristics of the patients with solid tumors and hematological malignancies could be found in Table S1 in the supplement file.
Antibiotic susceptibility results for the most common isolated bacteria are shown in Figs. 5 and 6. All E. coli were sensitive to polymyxin b. The sensitivity of E. coli for meropenem, colistin, amikacin, and imipenem were 92, 82.79, and 78%, respectively. All E. coli was resistant to piperacillin-tazobactam. More than 90% of Pseudomonas spp. were sensitive to ciprofloxacin, amikacin, gentamicin, and polymyxin-b, while 97% were chloramphenicol-resistant. All K. pneumoniae was sensitive to polymyxin-b, and 91% were susceptible to amikacin. More than 80% of K. pneumoniae was sensitive to meropenem, imipenem, and colistin. Trimethoprim/ sulfamethoxazole and ampicillin-sulbactam found less sensitive agents. Most coagulase-negative staphylococci (94%) were susceptible to linezolid, and more than 80% of them were susceptible to teicoplanin, rifampicin, chloramphenicol, and vancomycin.

Clinical influence of drug-resistance BSI and predictors of mortality
Based on the obtained results, third-generation cephalosporin-resistant E. Coli and P. aeruginosa, carbapenem-resistant P. aeruginosa, and polymyxinresistant K. pneumoniae BSIs associated with increased mortality; however, the difference was not statistically significant (p-value: 0.719, > 0.999, 0.521, and 0.467; respectively).
Based on results of univariate logistic regression analysis of variables investigated for survival in cancer patients with BSI, none of them associated with a greater risk of mortality (Table 2). Our results revealed that the odds of mortality decreased annually during 2015-2018 (Table 2).
Given the emergence of CRGN bacteria isolated in blood culture of patients with BSI during 2015-2018 (as we shown in Fig. 4) and the high proportion of CR isolates in non-fermenters (more than 70%) and also in Enterobacterales (about 25%), we investigated the odds of different possible risk factors for carbapenemresistant BSI by logistic regression model (Table 4). Accordingly, we found that age < 60-year-old, solid organ neoplasms, non-fermenter gram-negative BSI, thirdgeneration cephalosporine resistant gram-negative isolates, and polymyxin-resistance infections were significantly associated with carbapenem-resistant BSI based on univariate logistic regression analysis.
As illustrated in Fig. 7, based on both univariate and multivariate logistic regression analysis, Pseudomonasassociated BSI and polymyxin-resistance BSI were significantly associated with carbapenem-resistant BSI. Among tested antibiotics, only fluoroquinolones (including ciprofloxacin and levofloxacin) and aminoglycosides have acceptable sensitivity (resistance rate lesser than 10%) for Pseudomonas associated BSI. For E. coli-associated BSIs, polymyxins were the most active drug, while polymyxins and carbapenems were the best choices for K. pneumoniae-associated BSI (resistance rate < 15%).
Detail information regarding susceptibility profile of MDR, CR, and ESBL-producer E. coli, Pseudomonas spp., and K. pneumoniae isolates against different antimicrobial classes summarized in Table 5.

Discussion
Patients with malignancy are predisposed to developing BSI during their chemotherapy courses. Lots of evidence showed that the epidemiology of nosocomial infections in cancer patients changed over the past decades, with the reemergence of GNB as the predominant causative pathogens. The current study, therefore, conducted to describe the antibiotic-resistant patterns and outcomes of nosocomial infections caused by GNB in adult cancer patients.
Overall, despite the annual increase in the admission rate, no significant increase in the BSI incidence rate and attributed death occurred during the five-year study  (Fig. 2). Like some other reports, our study showed that CLABSI was more common in solid tumors than hematologic malignancies [19]; however, a higher prevalence of CLABSI detected in patients with hematological malignancies in other studies [20]. We found that gramnegative BSI was the most common etiology of BSI in cancer patients (63.3%), which agrees with other reports [21,22]. Like recent reports, we observed a gradual increase in the incidence of MDRGN and CRGN associated BSIs annually during our surveillance [23,24]. E. coli, Pseudomonas spp., and K. pneumoniae were the most common recovered gram-negative isolates in our study, which is in line with previous studies conducted in cancer patients [12,21,22,25]. We found a high proportion of ESBL producers and carbapenemresistant isolates, predominantly in non-fermenters (96.4 and 82.3%) and the Enterobacteriaceae family (64.1 and 28.8%). Although there are scarce reports on the incidence of carbapenem-resistant GNs in our region, obtained results are comparable to available reports [26]; however, compared to Europe, North America, Latin America global surveillance studies, and Asia-Pacific regional surveillance studies, our results showed significant higher-resistance rate [27]. We observed that the incidence of MDR BSI increased from 2015 to 2018.
We did not observe an association between investigated factors and survival; however, in a study by Chien-Yuan Chen et al. age ≥ 60 years, prior allogeneic transplantation and BSI due to VRE found as independent predictors for mortality [31]. CRGN BSI [26,[32][33][34], unresolved neutropenia, monotherapy, septic shock [32,35], and polymicrobial BSI [32] are other risk factors associated with increased mortality in cancer patients with GN BSI.
We identified MDR gram-negative infections more likely to occur in patients with WBC count< 4000 and non-fermenter gram-negative BSI (Table 3). Other studies found that male sex, age ≥ 60, previous antimicrobial use, liver disease, and bacteremia caused by K. pneumoniae are associated with increased risk of MDRGN BSI [36].
In agreement with current concerns regarding the efficacy of colistin against carbapenem-resistant pathogens, including CRE [37][38][39][40], we found that CR E. coli and K. pneumoniae isolates did not show acceptable sensitivity to our available antimicrobial choices, including colistin. However, ESBL producer K. pneumoniae and ESBL E. coli isolates had acceptable sensitivity to carbapenems and aminoglycosides (> 80%) and polymyxins (84.3%), respectively. Like the previous reports, our study reemphasized that carbapenems could still act as the drugs of choice in ESBL associated BSIs in cancer patients [41]. Besides, MDR E. coli and K. pneumoniae showed a high resistance rate to available antimicrobial agents. Our findings are consistent with previous studies concerning the global emergence of MDRGN pathogens as a significant healthcare burden that could be attributed to the overuse of antibiotics and necessitates the development of new antibiotics for treating CRE, CR P. aeruginosa, and CR Acinetobacter baumannii [42][43][44].
We also found that MDR, CR, and ESBL Pseudomonas spp. isolates still are sensitive to ciprofloxacin in our setting and could be considered a good treatment choice as recommended by current guidelines [45]. Remarkably, piperacillin-tazobactam, which is frequently preferred as one of the initial antibiotic therapies in our febrile neutropenic patients, may not be an excellent empiric choice against CR E. coli and K. pneumoniae. Our study showed that carbapenems might be less active against ESBL-producer Pseudomonas spp. and to some extent against ESBLproducer E. coli. In MDRGN BSI, none of our empiric treatments (carbapenems and piperacillintazobactam) could overcome GN bacteremia and need other choices such as colistin (Table 5). Although febrile neutropenia should be considered a medical emergency and a prompt administration of empirical antibiotic therapy is mandatory, increased mortality could be seen due to inappropriate empiric antibiotic therapy in setting with high rates of resistant pathogens [2,12,19,46]. Accordingly, regular epidemiological and microbiological surveillance of BSI should be encouraged strongly in oncology centers.

Limitation
Our study has some limitations. First, given the retrospective nature of this study, it was difficult to collect some variables (chemotherapeutic protocols, antibiotics treatment before admission, and some clinical and laboratory examination results) in this retrospective study.
So, there might be hidden biases in the analysis of the relationship. Second, carbapenemase type and enzyme were not investigated in this study. Finally, our research was conducted with data collected from a single center. Therefore, a prospective multicenter study is needed for further validation.