Introduction

Bloodstream infection (BSI) refers to the isolation of at least one pathogen (bacteria and fungi) from blood samples. Patients with hematologic malignancies often develop BSI, with incidences ranging from 11 to 50%, and mortality rates ranging from 10 to 35.3% [1,2,3,4]. Acute leukemia (AL) is a common hematologic malignancy. Due to factors such as immunosuppression, bone marrow suppression, high-dose combination chemotherapy, and neutropenia, patients with AL often develop BSI during treatment, which can be life-threatening [5].

Recent studies showed that BSI caused by gram-negative bacteria (GNB) were more common than those caused by gram-positive bacteria [6, 7]. Our previous study also manifested that GNB BSI accounted for 70.4% in hematological malignancies [8]. According to epidemiological surveys in the United States, there were 279,000 cases of GNB BSI annually, with 41,900 deaths [9]. Moreover, the isolation of antibiotic-resistant strains in GNB BSI is increasing, leading to serious concerns about antimicrobial resistance. It is estimated that antimicrobial resistance causes 700,000 deaths globally each year, and this number was predicted to rise to 10 million by 2025 [10]. Several studies showed that the mortality rates of BSI in patients with hematologic malignancies caused by multidrug-resistant (MDR) bacteria and carbapenem-resistant (CR) bacteria were 20.2-33.3% and 20-74.1%, respectively [11,12,13,14,15]. Therefore, early diagnosis and timely effective treatment of BSI caused by antibiotic-resistant bacteria in AL patients are crucial. On the one hand, understanding the local prevalence patterns and antibiotic susceptibilities of pathogens helps clinicians choose appropriate antibiotic treatment. Studies revealed that inappropriate empiric antibiotic therapy was associated with poor prognosis in patients with hematologic malignancies who develop BSI [16, 17]. On the other hand, identifying risk factors for antibiotic-resistant bacterial BSI helps physicians recognize infections with antibiotic-resistant bacteria and intervene promptly to improve patient outcomes.

However, data on antibiotic-resistant GNB BSI in AL patients are limited, especially in the southwestern region of China. Therefore, our study aims to analyze local patterns of antibiotic-resistant GNB and antibiotic susceptibilities, evaluate risk factors associated with the 30-days mortality of GNB BSI, and identify risk factors associated with antibiotic-resistant bacterial BSI, in order to provide meaningful guidance for the prevention and treatment of antibiotic-resistant bacterial BSI in AL patients.

Methods

Setting and study design

This retrospective study was conducted at West China Hospital of Sichuan University, which has over 4300 beds. The subjects of this study were patients who were diagnosed with AL and blood cultures positive for GNB at the Department of Hematology, West China Hospital, from June 2016 to June 2021. The following conditions were excluded: (1) patients diagnosed with non-AL; (2) contaminated blood culture specimens; (3) incomplete clinical and laboratory data. The study was approved by the Ethics Committee of West China Hospital of Sichuan University.

Data collection

The data were collected by the Hospital Information System (HIS) of West China Hospital of Sichuan University. The detail information was as follows: age, gender, underlying disease, treatment, hospital stays, complications, septic shock, laboratory results (neutrophil count, platelet count, pathogens of BSI, drug sensitivity test), invasive procedure (urinary catheter, central venous catheter (CVC), or peripherally inserted central catheter (PICC)), use of antibiotics, past 30 days of antibiotic exposure, clinical outcome (30 days after BSI). Data were extracted using a standardized data collection form and build an electronic database for analysis.

Definitions

GNB BSI was defined as the presence of clinical manifestations of infection in a patient (such as fever, chills, tachycardia, or hypotension), along with the identification of GNB in blood cultures. MDR bacteria referred to bacteria resistant to three or more classes of antimicrobial agents. The definition of CR bacteria referred to isolates that exhibit intermediate or resistant to one or more carbapenems. An absolute neutrophil count of less than 0.5 × 109 /L was defined as neutropenia, while less than 0.1 × 109 /L was defined as profound neutropenia. Empirical antibiotic therapy referred to the administration of one or more antibiotics to suspected BSI patients within 48 h of blood culture collection and before obtaining drug sensitivity test. Appropriate empiric antibiotic therapy was defined as at least one of the empirically administered antibiotics demonstrates sensitivity in drug sensitive test; otherwise, it was deemed inappropriate empiric antibiotic therapy. Antibiotic exposure referred to patients receiving any intravenous or oral antibiotics for more than 48 h within the 30 days prior to the onset of BSI.

Statistical analysis

Categorical variables were presented as frequencies and percentages, and between-group differences were compared using the Chi-square test or Fisher’s exact test. The median and interquartile range (IQR) were used to represent continuous data, while the t-test or Mann-Whitney U test was applied to compare differences between groups. Variables with p < 0.05 in univariate analysis were included in the multivariable logistic regression analysis to determine the risk factors for developing MDR GNB BSI, CR GNB BSI in patients with AL, as well as the risk factors for 30-day mortality in patients with GNB BSI. The Kaplan-Meier method was applied to plot survival curves, and the log-rank test was used to compare differences. p < 0.05 was regarded as statistically significant. All analyses were conducted in IBM SPSS 26.0.

Results

Patient characteristics

This study included a total of 206 patients with GNB BSI in AL, among which acute myeloid leukemia (AML) accounted for 68.4% (141/206), and acute lymphoblastic leukemia (ALL) accounted for 31.6% (65/206). Among the 206 patients with AL, 74 (35.9%) had a disease status of relapsed or refractory. The median age of all patients was 41 (IQR, 27.8–54.0) years, with 84 (40.8%) being female. The median length of stay before BSI was 20 (IQR, 13.0-37.3) days, and hospital-acquired infection was observed in 194 (94.2%) patients. The majority of patients number (85.4%) developed pulmonary infections, 31 (15.0%) developed perianal infections, and 15 (7.3%) developed gastroenteritis. Additionally, 43 (20.9%) patients experienced septic shock. Before the onset of BSI, 177/206 (85.9%) patients had neutropenia, with 152/206 (73.8%) being profound neutropenia. Patients with urinary catheters and PICC/CVC were 26 (12.6%) and 107 (51.9%), respectively. Of the 206 patients, 58 (28.2%) received corticosteroid therapy. In all patients, 48 (23.3%) received inappropriate empiric antibiotic therapy at the onset of BSI, while 158 (76.7%) received appropriate empiric antibiotic therapy, with 41(19.9%) receiving monotherapy and 117 (56.8%) receiving combination therapy. General characteristics of the above patients are detailed in Table 1.

Table 1 Characteristics of GNB BSI patients stratified by outcome
Full size table

Microbiology

Table 2 shows the common GNB and their antibiotic resistance characteristics. A total of 206 GNB were isolated, with Escherichia coli (E. coli) (81/206, 39.3%), Klebsiella pneumoniae (K. pneumoniae) (46/206, 22.3%), and Pseudomonas aeruginosa (P. aeruginosa) (34/206, 16.5%) being the most common. Among all GNB, MDR GNB accounted for 62.1% (128/206), most commonly observed in E. coli (46.9%), followed by P. aeruginosa (22.7%) and K. pneumoniae (16.4%). Forty-four strains (21.4%) were identified as CR bacteria, mainly found in E. coli (22.7%), P. aeruginosa (18.2%), and K. pneumoniae (18.2%). Five strains of A. baumannii were isolated in the study, all of which were MDR and CR. Overall, as shown in Fig. 1, there was a trend of initial increase followed by decrease in MDR GNB and CR GNB from 2016 to 2021. MDR K. pneumoniae also showed a similar trend of increase followed by decrease over the five years, while MDR E. coli and CR K. pneumoniae displayed an increasing trend from 2019 to 2021.

Table 2 Antimicrobial resistance profiles of all GNB and isolated MDR and CR strains
Full size table
Fig. 1
figure 1

Trend in MDR (A) and CR (B) strains of BSI in AL patients. EC: Escherichia coli; KP: Klebsiella pneumoniae

Full size image

As shown in Table 2, all GNB showed relatively low resistance to amikacin (7.3%), meropenem (16.1%), and imipenem (20.9%). In common GNB, E. coli and K. pneumoniae exhibited low resistance to amikacin (4.9%, 2.2%), meropenem (7.4%, 17.4%), and imipenem (8.6%, 17.4%). While P. aeruginosa had low resistance to amikacin (2.9%), ciprofloxacin (2.9%), and gentamicin (5.9%). All A. baumannii isolates were completely resistant to most antibiotics (piperacillin/tazobactam, cefoperazone subactam, ceftriaxone, cefepime, ciprofloxacin, and imipenem). Overall, MDR and CR bacteria exhibited relatively high resistance to most antibiotics, except for amikacin (11.7%, 25.0%).

Risk factors for MDR GNB BSI and CR GNB BSI

Univariate analysis showed that female sex, placement of PICC/CVC, and exposure to carbapenems/aminoglycosides/quinolones within the preceding 30 days were potential risk factors for MDR GNB BSI in AL patients (Table 3). Further multivariate analysis revealed that exposure to quinolones within the preceding 30 days (Odds ratio (OR) = 3.111, 95% confidence interval (95%CI): 1.623–5.964, p = 0.001) was an independent risk factor for MDR GNB BSI in AL patients (Table 4). Similarly, differences were found between CR GNB BSI and Non-CR GNB BSI in univariate analysis for factors such as placement of urinary catheter and exposure to cephalosporins and carbapenems within the preceding 30 days (Table 3). Multivariable logistic regression analysis indicated that placement of urinary catheter (OR = 6.311, 95%CI: 2.478–16.073, p < 0.001), exposure to cephalosporins (OR = 2.340, 95%CI: 1.090–5.025, p = 0.029), and carbapenems (OR = 2.558, 95%CI: 1.190–5.497, p = 0.016) within the preceding 30 days were independently associated with CR GNB BSI in AL patients (Table 4). Furthermore, this study also showed that the 30-day mortality rate was significantly higher in AL patients with CR GNB BSI (59.1%) compared to those with Non-CR GNB BSI (17.3%) (p < 0.001), while there was no difference between MDR GNB BSI (25.8%) and Non-MDR GNB BSI (26.9%) (p = 0.857) (Table 3).

Table 3 Univariate analysis of factors associated with BSI caused by MDR and CR GNB
Full size table
Table 4 Multivariate analysis of risk factors for MDR and CR GNB BSI, as well as 30-day mortality risk factors in GNB BSI patients
Full size table

Outcome of GNB BSI

This study showed that the 30-day mortality rate among AL patients with GNB BSI was 26.2%. We analyzed the risk factors for 30-day mortality in AL patients with GNB BSI, and the results revealed differences between survivors and non-survivors in the following factors: female sex, length of stay before BSI, CR GNB BSI, relapsed/refractory AL, pulmonary infection, septic shock, duration of neutropenia before BSI, platelets < 30 × 109/L before BSI, and inappropriate empiric antibiotic therapy (Table 1). The above-mentioned variables were included in the multivariate analysis, the results indicated that CR GNB BSI (OR = 2.960, 95% CI: 1.016–8.624, p = 0.047), relapsed/refractory AL (OR = 3.035, 95% CI: 1.265–7.354, p = 0.013), septic shock (OR = 5.108, 95% CI: 1.794–14.547, p = 0.002), platelets < 30 × 109/L before BSI (OR = 7.785, 95% CI: 2.055–29.492, p = 0.003), and inappropriate empiric antibiotic therapy (OR = 3.140, 95% CI: 1.171–8.417, p = 0.023) were independent risk factors for 30-day mortality in AL patients with GNB BSI (Table 4). Survival analysis was conducted on AL patients with GNB BSI, revealing that CR GNB BSI (p < 0.001), relapsed/ refractory AL (p < 0.001), septic shock (p < 0.001), platelets < 30 × 109/L before BSI (p < 0.001), and inappropriate empiric antibiotic therapy (p < 0.001) were associated with poorer 30-day survival rates (Fig. 2).

Fig. 2
figure 2

Kaplan-Meier curves of the estimated 30-day probability of survival for GNB BSI caused by CR strain and non–CR strain (A); patients with and without relapsed/refractory AL (B); patients with and without septic shock (C); patients with and without platelets < 30 × 109/L before BSI (D); and patients with and without inappropriate empiric treatment (E)

Full size image

Discussion

With the widespread use of antibiotics, bacteria have developed various mechanisms, such as producing inactivating enzymes (carbapenemases), mutations in target sites, and reduced bacterial outer membrane permeability [18]. Antimicrobial resistance has become a global concern. However, data on the epidemiology and prognosis of BSI caused by drug-resistant bacteria in patients with hematologic malignancies are limited [11,12,13], especially in AL patients. Our study revealed that among AL patients with GNB BSI, the proportion of antibiotic-resistant bacterial BSI was considerable, with MDR GNB BSI accounting for 62.1% and CR GNB BSI accounting for 21.4%. Moreover, our hospital data showed that previous antibiotic exposure within 30 days was a significant factor associated with the occurrence of MDR GNB BSI and CR GNB BSI. This underscores the importance of judicious antibiotic use to avoid overuse. The 30-day mortality rate of AL patients with GNB BSI in our hospital was 26.2%, consistent with other studies (21.6-28.6%) [13, 19, 20]. However, the mortality rate of antibiotic-resistant bacteria BSI, especially CR GNB BSI, was high, with a 30-day mortality rate of 59.1%. Previous studies have also shown mortality rates for CR bacterial BSI ranging from 42.5 to 74.1% [13, 15, 21]. This highlights the significant threat of antibiotic-resistant bacterial BSI to patient life, necessitating timely identification of pathogens, appropriate antibiotic treatment, and regular hospital surveillance to improve patient management and prognosis.

In this study, GNB causing BSI was most commonly E. coli, followed by K. pneumoniae and P. aeruginosa, consistent with reports from China, Italy, and Brazil [6, 13, 22]. However, a retrospective study in Saudi Arabia showed that GNB BSI in AL patients was more commonly caused by K. pneumoniae (33.3%) [23]. This indicates regional differences in the prevalence patterns of pathogens, highlighting the importance of ongoing local pathogen surveillance. Our study found that isolated GNB exhibited varying degrees of resistance to commonly used antibiotics. E. coli and K. pneumoniae showed low resistance to amikacin and carbapenems (imipenem and meropenem), while P. aeruginosa exhibited low resistance to amikacin and quinolones. Therefore, the above appropriate antibiotics can be selected for clinical infections caused by these bacteria.

Antibiotic-resistant bacteria were a major focus of our study. Antibiotic-resistant bacterial infections not only lead to poor prognosis but also increase patient hospitalization time and economic burden. Wang et al. found that MDR bacterial BSI accounted for 57% in AML patients [24], while Schonardie et al. showed that CR bacterial BSI accounted for 17.3% of BSI in patients with hematologic malignancies [13]. Our study revealed that MDR GNB BSI accounted for 62.1% and CR GNB BSI accounted for 21.4% of all BSI in AL patient. The monitoring system of the China Antimicrobial Surveillance Network (CHINET) in 2021 showed a continuous decline in the detection rates of various important CR bacteria in recent years, despite their increasing trend over several years [25]. Our study also found a trend of initially increasing and then decreasing MDR GNB and CR GNB from 2016 to 2021, possibly due to the attention given by the hematologists in our hospital to antibiotic-resistant bacteria and the proper use of antibiotics. However, because of antibiotic overuse, resistant bacterial infections have become a significant threat to patient health, necessitating the development of new antibiotics targeting resistant bacteria [26].

CR GNB BSI was associated with poor prognosis, with a previous study showing an increased risk of death in patients with CR bacterial BSI [27]. Our study similarly found CR GNB BSI to be one of the independent risk factors for 30-day mortality in AL patients. Although MDR GNB BSI did not increase the risk of death in our study, another study showed it to be a risk factor for mortality [12]. This difference may be due to sample size and regional variations.

As antibiotic-resistant bacterial BSI is closely related to prognosis, our study analyzed the risk factors for the occurrence of antibiotic-resistant bacterial BSI. Our study demonstrated that prior use of quinolone antibiotics within the past 30 days was a risk factor for developing MDR GNB BSI. Studies by Tacconelli et al. and Zhao et al. also showed that exposure to quinolones was associated with MDR bacterial BSI [12, 28]. As shown in previous studies, exposure to carbapenems in the past was independently associated with CR bacterial BSI [12, 29]. Our study also confirmed this result. Therefore, clinicians need to use antibiotics cautiously to prevent misuse. Clinicians can avoid overuse of these antibiotics by using a few available antibiotics more judiciously to prevent the emergence of more antibiotic-resistant bacteria. In addition, our study found that placement of urinary catheter and exposure to cephalosporins within 30 days were independent risk factors for CR GNB BSI. A multicenter retrospective study verified a significantly higher proportion of placement of urinary catheter developing CR bacterial infections [30].

Antibiotics are indispensable in the treatment of BSI, and timely and effective empiric antibiotic therapy is key to controlling BSI. Ricard et al. found that delayed antibiotic use was associated with increased in-hospital mortality in septic patients [31]. A series of studies emphasized that inappropriate empiric antibiotic therapy was a risk factor for death in patients with hematologic malignancies [16, 32, 33]. In our study, 23.3% of patients received inappropriate empiric antibiotic therapy, and we found that inappropriate empiric antibiotic therapy was one of the independent risk factors for 30-day mortality in AL patients with GNB BSI. Therefore, antibiotic therapy is crucial for AL patients with BSI. Understanding the local distribution of pathogens and antibiotic susceptibility is helpful for clinicians to choose appropriate antibiotics.

In addition to CR GNB BSI and inappropriate empiric antibiotic therapy being associated with 30-day mortality in AL patients with GNB BSI, our study and other studies confirmed that relapsed/refractory AL and septic shock were risk factors for 30-day mortality in patients with GNB BSI [21, 22]. Therefore, patients with clinical manifestations such as hypotension, low oxygen saturation, and altered consciousness need to be identified early and provided with timely and effective interventions to restore circulation through fluid resuscitation, vasopressors, and other treatments. We also found that platelets < 30 × 109/L before BSI were associated with 30-day mortality. Besides its role in hemostasis, platelets also participate in preventing bacterial infections. Platelets can recognize microbial antigens and enhance immune responses, thereby preventing bacterial infections [34].

BSI is common in patient with AML or ALL. Previous studies have shown that the type of AL (AML or ALL) was not an independent risk factor for antibiotic-resistant bacterial BSI or mortality of BSI in patients with hematologic diseases [12, 13, 35]. Similarly, our study did not find that the subtype of AL was a risk factor for 30-day mortality in BSI or for the occurrence of MDR and CR GNB BSI. In contrast, Micozzi et al. showed that AML was an independent risk factor for the occurrence of BSI in hematologic malignancy patients colonized with CR K. pneumoniae [36]. These conflicting conclusions may stem from the fact that most of current studies are retrospective, their conclusions may be limited by small sample sizes, with possible bias. Therefore, whether the subtype of AL affects the occurrence and prognosis of BSI is unclear, which requires further evaluation through larger, multicenter, and prospective studies. In our study, we did not find that neutropenia was an independent risk factor for the occurrence of antibiotic-resistant bacterial BSI or 30-day mortality in BSI patients. Currently, there is no consensus on whether neutropenia affects the prognosis of BSI patients with hematological diseases. Some studies indicated that neutropenia was unrelated to mortality in patients with hematologic diseases who develop BSI [12, 37, 38], while others demonstrated that the neutropenia or the duration of neutropenia impacted the prognosis of these patients [39,40,41]. Additionally, previous research showed that neutropenia was not significantly associated with the occurrence of CR BSI [13, 35]. Therefore, larger prospective studies are needed to investigate whether neutropenia affects the occurrence and prognosis of BSI in patients with hematologic diseases.

This study has some limitations, primarily due to its single-center retrospective nature and the limited number of collected BSI patients. Since the epidemiology of BSI varies by region, the conclusions of this study may not apply to centers with significant regional differences, and these limitations may also result in discrepancies with findings from other studies.

In conclusion, our study confirmed that the incidence of antibiotic-resistant bacterial BSI was not low among AL patients, and the mortality rate was high for CR GNB BSI. Prior antibiotic exposure was one of the independent risk factors for developing MDR GNB BSI and CR GNB BSI. Furthermore, CR GNB BSI increased the risk of mortality in AL patients with GNB BSI. Prospective studies are warranted to better understand the characteristics of antibiotic-resistant bacterial BSI in AL.