Introduction

Bacteremia is the most important infectious complication of high-risk neutropenic patients, leading to an increase in mortality, length of hospital stay, and healthcare costs [1,2,3]. In recent decades, an epidemiological change has been observed in the etiology of bacteremia in this population, with a predominance of gram-negative bacilli (GNB) as the leading cause, and Enterobacterales as the etiological agents most frequently involved [4, 5]. Likewise, growing antibiotic resistance is a major concern worldwide, especially in Latin America, where the most frequent resistance mechanism is the production of extended-spectrum beta-lactamases (ESBL) [6]. In addition, carbapenemase-producing Enterobacterales are on the rise [7, 8]. Several factors could be involved, with misuse and overuse of antibiotic therapy (AT) being the most significant. In this regard, receiving AT ≥ 10 days proved to be a risk factor for carbapenemase-producing Enterobacterales bacteremia in cancer and hematopoietic stem cell transplant (HSCT) patients [9]. Therefore, shortening the duration of AT should be an important measure in antimicrobial stewardship programs in this population.

Several guidelines on AT in oncohematological and neutropenic patients have been published [10,11,12,13,14]. Some of them suggest that antibiotic therapy should be continued for at least 7 days, until the infection is microbiologically eradicated and all signs of infection are resolved, with the patient afebrile for at least 4 days [13, 14]. However, this remains a controversial issue given the limited available data on high-risk neutropenic patients.

The efficacy and safety of short course AT have been evaluated in three randomized controlled studies in patients with uncomplicated GNB bacteremia. The rates of clinical cure, bacteremia relapse, and 30-day mortality of all those patients treated for 7 days were similar to those who received 14 days of AT [15,16,17]. Recently, a meta-analysis including the previously mentioned studies evaluated the subgroup of immunocompromised patients and found no differences between those who received AT for 7 days vs. 14 days. However, bacteremia was mainly isolated from urinary source and was observed in hemodynamically stable patients. Moreover, high-risk neutropenic patients and those with carbapenemase-producing Enterobacterales bacteremia were excluded [18].

This study was designed to describe and compare the clinical, epidemiological, and outcome characteristics, such as bacteremia relapse, Clostridiodes difficile infection, 30-day mortality and length of hospital stay, of high-risk neutropenic patients with Enterobacterales bacteremia based on 7 or 14 days of AT.

Material and methods

Setting, patients and study design

A prospective, observational, multicenter study was performed in 8 referral teaching centers (5 private and 3 public) specialized in the management of oncohematological and HSCT patients in Argentina.

All the episodes of monomicrobial Enterobacterales bacteremia in adult patients (≥ 18 years of age) managed as inpatients from May 2014 to March 2023 were included, provided that the following criteria were met: (a) patients presenting with hematologic malignancies (HM) or autologous and allogeneic HSCT; (b) high-risk febrile neutropenia; (c) appropriate empirical AT; (d) clinical response within 7 days; and (f) a total of 7 or 14 days of AT.

Patients with polymicrobial or recurrent bacteremia and those receiving palliative care were excluded from the analysis, as well as those with a source of bacteremia involving prolonged treatment (endocarditis/endovascular infections, severe skin and soft-tissue infections, central nervous system infections and osteomyelitis), or with a clinical source that required surgery.

Patients who received 14 days of antibiotic therapy were included since the beginning of the study, as this was the usual duration of AT for GNB bacteremia in this population [10]. In comparison, those who received 7 days were recruited since 2018, when short AT for Enterobacterales bacteremia was first carried out in selected patients by participating centers, in accordance with the ECIL-4 guidelines [13].

Patients were identified as a result of a positive blood culture and were then followed up prospectively. Data were obtained from medical records and direct patient care, with a double check made with microbiological records from the laboratory. Clinical, microbiological, treatment, and outcome variables were evaluated. Empirical individualized AT was initiated based on the patient’s clinical and epidemiological features, pursuant to each center institutional guidelines and IDSA and ECIL recommendations [10, 12, 13]. The study investigator chose definitive therapy (DT) based on Enterobacterales isolates and their antibiotic resistance profile.

Patients were followed for 30 days after the episode (by direct patient care in hospitalized cases, or by phone calls to discharged patients), or until the patient’s death, provided that it happened before (assessed by direct patient care in those still hospitalized, or by a local healthcare database in each center).

Definitions

Neutropenia was defined as an absolute neutrophil count < 500 cells/mm3. High-risk febrile neutropenia was defined according to the Multinational Association for Supportive Care in Cancer (MASCC) as a score < 21 and one or more clinical criteria [10]. The clinical source of bacteremia was determined based on the isolation of Enterobacterales from the suspected source and/or the associated clinical signs and symptoms. It was classified pursuant to the US CDC criteria [19]. Neutropenic enterocolitis was defined as fever with abdominal symptoms and bowel wall thickening > 4 mm [20].

Complicated bacteremia was defined as an episode presenting with hypotension, septic shock, or pneumonia as the clinical source [10, 12].

High doses of corticosteroids were defined as having received prednisone (or equivalent) at doses ≥ 20 mg/day for a period ≥ 2 weeks prior to bacteremia, and the use of biological agents and/or anti-lymphocyte therapies, having received these drugs within six months prior to bacteremia.

Patients presenting one of the abovementioned factors were considered as severely immunosuppressed.

Carbapenemase-producing Enterobacterales colonization was defined as “previous” when it occurred within six months prior to hospital stay, and “recent” when detected within the week prior to the episode of bacteremia.

Bacteremia was classified as nosocomial, healthcare-associated, or community-acquired according to Friedman et al. [21]. Hypotension was defined as systolic blood pressure < 90 mm Hg on the day of the positive blood culture [22]. Septic shock was defined as the need for vasopressors to maintain mean arterial pressure ≥ 65 mmHg and serum lactate level > 18 mg/dl [23]. Infection severity and mortality probability were defined using Pitt and APACHE-II scores. Relapse of bacteremia was defined as a new episode of bacteremia within 30 days of AT completion, with the same isolated Enterobacteral. EAT was considered appropriate provided that it was initiated after blood cultures were drawn, and one or more antibiotics used were active in vitro against the isolated bacteria. In patients with ESBL-producing Enterobacterales, empirical AT with piperacillin/tazobactam or cefepime monotherapy was considered inappropriate [24]. In patients with isolation of any Enterobacterales species, tigecycline as monotherapy was also deemed unsuitable. Clinical response on day 7 of AT was defined as absence of fever for at least four days, source control of bacteremia, absence of hypotension, and clinical resolution of all signs and symptoms of infection. In catheter-related bacteremia, catheters were removed on the day of diagnosis. Mortality was related to infection provided that there was microbiological, histological, or clinical evidence of active infection.

Microbiology

Bacteremia was defined as the isolation of a pathogenic bacteria in at least one bottle of blood culture (BD BACTEC F Aerobic and Anaerobic, analyzed with BACTEC FX BD, BacTALERT 3D BioMérieux depending on the method available at each center) for a minimum incubation period of five days. MDR-GNB were defined as Enterobacterales resistant to three or more of the following categories of antibiotics: carbapenems, piperacillin/tazobactam, third and fourth generation cephalosporins, aztreonam, fluoroquinolones, or aminoglycosides [25, 26]. Microbiological identification and susceptibility testing were done with manual biochemical and microbiological methods, disk diffusion (according to the CLSI recommendations), and/or Etest, VITEK II Compact (BioMérieux), PHOENIX 100 BD (Becton Dickinson), VITEK MS (BioMérieux) and MALDI-TOF (Microflex from Bruker). ESBL production was determined by disk diffusion method using both ceftazidime and cefotaxime, alone and combined with clavulanic acid. Carbapenemase production was investigated in carbapenem-resistant bacteria using the modified Hodge method, disk synergy tests with a carbapenem disk placed close to the boronic acid disk test for KPC, and the EDTA disk for identification of metallo-β-lactamases. The presence of genes coding for blaKPC and blaOXA-48 was investigated by monoplex or multiplex polymerase chain reaction (PCR) using specific primers depending on the method available at each center. Multiplex PCR for blaVIM, blaNDM, blaIMP, blaKPC and blaOXA-48 was used to investigate carbapenemase-producing Enterobacterales isolates at the National Reference Laboratory of Microbiology (ANLIS-Malbrán) [27]. In order to detect colonization with KPC-producing Enterobacterales, rectal swabs were routinely collected (once a week and in every pre-transplant evaluation) using chromogenic methods and/or PCR.

Clostridioides difficile was investigated by immunochromatography in every patient with diarrhea in order to determine the presence of glutamate dehydrogenase (GDH) antigen and toxins A and B. Those samples with positive GDH and negative toxins were analyzed by real-time PCR available in 7 centers [28, 29].

Statistical analysis

The study population was characterized by descriptive statistics. For continuous variables, centrality (median) and dispersion (IQR) measures were used according to the distribution of variables. Categorical variables were analyzed using absolute frequency and percentage. Groups were compared using the U Mann–Whitney test for continuous variables and the Fisher exact test or the chi-square test for categorical variables. For all tests, a 95% level of statistical significance was used. The analyses were performed with the SPSS (Statistics for Windows, Version 22.0. Armonk, NY, USA) software packages.

Results

A total of 1,037 patients with HM and HSCT who presented Enterobacterales bacteremia were evaluated, and 837 were excluded because they did not meet the eligibility criteria (Fig. 1). Thus, the study population consisted of 200 patients, 100 in each treatment group. The baseline and epidemiological characteristics of bacteremia episodes that received AT for 7 and 14 days are described in Table 1. Most patients had acute leukemia as the underlying disease (127, 63.5%), followed by lymphoma (42, 21%). Sixty-five patients (32.5%) received HSCT (32, 46.1% allogeneic). HM was active in 144 (72%) patients; 171 (85.5%) had received chemotherapy one month prior to bacteremia episode, 49 (24.5%) and 40 (20%) had received high-dose corticosteroids or biological agents and/or anti-lymphocyte therapies, respectively. One hundred and nine (54%) patients had previously received antibiotics, with piperacillin-tazobactam being the most common (60, 30%). Recent and previous colonization by KPC-carbapenemase producing Enterobacterales were detected in 24 (12%) and 21 (11.5%), respectively. The Enterobacterales most frequently isolated were Escherichia coli (95, 47.5%) and Klebsiella sp. (81, 40.5%). Fifty-eight (29%) were MDR-GNB, with ESBL producers (44, 22%) and carbapenemase-producing Enterobacterales (12, 6%) being the most frequent, with similar distribution between both groups. Regarding resistance profile, more than 20% of the isolates were resistant to piperacillin-tazobactam and cefepime, and 6% to meropenem. Resistance to fluoroquinolones was 35%, being significantly higher in the group of patients treated for 14 days, and resistance to the other antibiotics was low in both groups. The microbiological characteristics and resistance profiles of bacteremia episodes are described in Figs. 2, 3, and 4.

Fig. 1
figure 1

Selection of the study population. Abbreviation: HM, hematologic malignancies; HSCT, hematopoietic stem cell transplants; EB, Enterobacterales; AT, antibiotic therapy

Full size image
Table 1 Baseline and epidemiological features
Full size table
Fig. 2
figure 2

Comparative etiology of gram-negative bacilli bacteremia between the 7- and 14-day antibiotic therapy groups. *p-value obtained by chi-square or Fisher exact tests

Full size image
Fig. 3
figure 3

Comparative frequency of multidrug-resistant gram-negative bacilli and resistance mechanisms detected between the 7- and 14-day antibiotic therapy groups. Abbreviations: MDR-GNB, multidrug-resistant gram-negative bacilli; ESBL, extended spectrum beta-lactamase; KPC, Klebsiella pneumoniae serin-carbapenemase; NDM, New Delhi metallo-beta-lactamase; AmpC; AmpC beta-lactamase. *p-value obtained by chi-square or Fisher exact tests

Full size image
Fig. 4
figure 4

Comparative antibiotic resistance profile of gram-negative bacilli bacteremia between the 7- and 14-day antibiotic therapy groups. Abbreviation: CIP, Ciprofloxacin; CRO, Ceftriaxone; FEP, Cefepime; PTZ, Piperacillin-tazobactam; MER, Meropenem; AMK, Amikacin; COL, Colistin; TIG, Tigecycline; FOS, Fosfomycin; CZA, Ceftazidime-avibactam. ¶: Colistin resistance available in 176 isolates. ¥: Tigecycline and Fosfomycin resistance available in 169 isolates. § Ceftazidime-avibactam resistance available in 85 isolates. *p-value obtained by chi-square or Fisher exact tests

Full size image

Table 2 shows the clinical characteristics, AT, and outcome of patients with Enterobacterales bacteremia according to the treatment group. Of 111 (55.5%) episodes of bacteremia with a clinical source, abdominal –evidenced as colitis (50, 25%), and catheter-related bacteremia (32, 16%) were the most frequent. EAT was combined in 88 (44%) cases, with carbapenems (57%) being the most commonly used antibiotics. DT as monotherapy was significantly higher in patients who received 7 days of AT (99% vs. 92%, p = 0.03), and carbapenems (33.5%) and piperacillin-tazobactam (28.2%) were the antibiotics most frequently prescribed. Forty-two patients (21%) received AT in prolonged infusion, which was largely used in patients with 7-day treatment (33% vs. 9%, p < 0.001). AT discontinuation before neutrophil recovery was mainly carried out in patients under 7-day treatment (46% vs. 26%, p = 0.003). No patients returned to antibiotic prophylaxis.

Table 2 Clinical presentation, treatment, and outcome
Full size table

Bacteremia episodes were complicated in 76 (38%) patients with no differences between groups. Septic shock at onset occurred in 25 (12.5%) patients, and was more frequently observed in those who received 14 days of treatment (18% vs. 7%, p = 0.01). Clostridioides difficile infection and bacteremia relapse occurred in 9 (4.5%) patients. Thirty-day mortality was 2%, in no case related to infection, with no differences between both groups. Length of hospital stay since bacteremia had a median of 13 days (IQR: 8–20), being significantly lower in the group that received 7 days of treatment. The differences observed in treatment and outcome between the two groups are highlighted in Table 2.

Discussion

This study evaluated a total of 7-day AT as a feasible strategy for high-risk neutropenic patients with Enterobacterales bacteremia who received adequate empirical AT with clinical response. The cohort included several patients with complicated bacteremia, an identifiable clinical source of infection, and even septic shock. Moreover, many of them presented bacteremia due to MDR-GNB, including carbapenemase-producing Enterobacterales. Both groups had similar baseline, microbiological, clinical, and epidemiological characteristics. A low rate of bacteremia relapse and mortality was observed, which was similar in both groups. However, length of hospital stay since bacteremia diagnosis was significantly reduced in patients who received 7 days of AT.

Short courses of AT in hematological and neutropenic patients with GNB bacteremia have been previously evaluated in three single-center studies.

The first one is a retrospective cohort study that compared the efficacy of short (median 6 days, IQR: 6–7) versus prolonged (median 11 days, IQR: 9.5–14) AT for bacteremia in patients with acute myeloid leukemia and febrile neutropenia [30]. A total of 71 patients presenting 104 episodes of bacteremia were included, and 46% received a short course of AT. In the total population, 65% of bacteremias were primary, and 51% were caused by GNB, mostly Enterobacterales. Of them, 23% were ESBL producers, but carbapenemase-producing Enterobacterales bacteremias were not included. Bacteremia relapse and mortality were similar between both groups. This study largely differs from ours. Although the population was homogeneous regarding the underlying disease, patients were not stratified according to the risk of neutropenia; bacteremias were mainly primary, and not all of them were caused by GNB. The implemented AT was not described, the variables usually associated with mortality were not considered, and more than one episode per patient were included. These issues may cause bias in the interpretation of the study results.

The second one is a retrospective study in which three groups of neutropenic patients with GNB bacteremia were compared based on AT duration: ≤ 10 days, 11–14 days, and ≥ 15 days. The primary objective was to evaluate mortality and bacteremia relapse at day 90 [31]. Although most patients had HM and HSCT, subjects with different immunosuppressive diseases were also included. As in our study, more than 50% of bacteremias had a clinical source, with abdominal and central venous catheters being the most frequent. However, their etiology was heterogeneous since in the short treatment group, of 67 bacteremias, 13.4% and 23.9% were caused by Pseudomonas aeruginosa and other GNB, respectively. Moreover, bacteremias caused by carbapenemase-producing Enterobacterales were excluded. Bacteremia relapse and mortality were similar between the three groups, adjusted by a propensity score analysis. Of note, the short treatment group included patients treated for up to 10 days. However, it might be controversial to assume that this is a short duration.

The last one is a prospective observational study that compared the duration of AT with a median of 7 days (IQR: 7–7) versus 14 days (IQR: 14–14) in 74 patients with cancer and HSCT presenting GNB bacteremia [32]. The inclusion criteria were the same as those in the present study. However, a small percentage of patients with bacteremia caused by Pseudomonas aeruginosa were included and only around 60% were neutropenic. The groups were comparable in terms of baseline characteristics, clinical presentation, severity, epidemiology, and microbiology. As in our study, 30-day mortality and bacteremia relapse were low and similar, with a significant reduction in length of hospital stay in those who received short AT.

These are the major findings of the present study: first, besides being neutropenic, a considerable number of patients was severely immunosuppressed, presented complicated bacteremia, and had a high APACHE II score. Notwithstanding that, mortality was low and in no case related to bacteremia, which suggests that early implementation of adequate empirical AT and source control of bacteremia is crucial. Second, in almost half of the patients that received 7-day AT, antibiotics were discontinued before neutrophil recovery, which shows that this approach could be safe. Third, in most of the patients who received AT for 7 days, DT was monotherapy, highlighting the importance of de-escalating antibiotics in high-risk neutropenic patients with a clinical response [33]. Fourth, one-third of patients in the short treatment group received antibiotics in prolonged infusion as the strategy implemented by some centers over the last years [34]. This variable could have had an impact on the result. However, there is still no conclusive data available to determine whether this strategy offers clinical benefits in high-risk neutropenic patients with GNB bacteremia [35]. Finally, patients who received short treatment had a 9-day median duration of hospital stay since bacteremia. Of note, hospital stay ≥ 10 days is known to be an independent risk factor for developing KPC-carbapenemase producing Enterobacterales bacteremia [9].

Our study has some limitations that should be considered. First, patients received different AT, which could induce a bias in the result. However, both cohorts were prescribed adequate AT, with a similar proportion of combination therapy and monotherapy. In addition, empirical AT was based on IDSA and ECIL recommendations. Second, some variables that could have influenced the outcomes, such as septic shock and higher Charlson comorbidity score, were more frequent in patients who received 14-day AT. Given that DT was a personalized decision, we cannot rule out that the study investigators might have decided to prolong the AT based on these factors. Third, although we did not observe statistical differences in relapse rates between the groups, the study might have been underpowered to detect this outcome. Fourth, although a significant rate of MDR-GNB bacteremias was included, the number of carbapenemase-producing Enterobacterales was low. Therefore, we cannot conclude that the findings obtained can be applied to these microorganisms. Finally, since the number of patients who presented with septic shock was low, we cannot extrapolate the final results to high-risk neutropenic patients with hemodynamic failure.

The strengths of our study rely on its prospective and multicenter design, carried out in healthcare facilities specialized in the treatment of high-risk neutropenic patients. Likewise, a homogeneous and comparable population was included in terms of clinical, microbiological, and outcome variables in cohorts allocated to the same duration of AT.

In conclusion, this study showed that AT for 7 days might be adequate for selected high-risk neutropenic patients with Enterobacterales bacteremia who received adequate empirical AT, with clinical response. Furthermore, this strategy was associated with a significant reduction in length of hospital stay, and this might have potential benefits, such as the reduction of healthcare costs and the emergence of MDR-GNB. Randomized studies are needed to support these findings and confirm the efficacy and safety of short courses of AT in this population.