Clinical Features and Genomic Epidemiology of Bloodstream Infections due to Enterococcal Species Other Than Enterococcus faecalis or E. faecium in Patients With Cancer

Abstract

Background

Non–Enterococcus faecium, non–E. faecalis (NFF) enterococci are a heterogeneous group of clinically pathogenic enterococci that include species with intrinsic low-level vancomycin resistance. Patients with cancer are at increased risk for bacteremia with NFF enterococci, but their clinical and molecular epidemiology have not been extensively described.

Methods

We conducted a retrospective review of all patients (n = 70) with NFF bacteremia from 2016 to 2022 at a major cancer center. The main outcomes assessed were 30-day mortality, microbiological failure (positive blood cultures for ≥4 days), and recurrence of bacteremia (positive blood culture <14 days after clearance). Whole-genome sequencing was performed on all available NFF (n = 65).

Results

Patients with hematological malignancies made up 56% of the cohort (77% had leukemia). The majority of solid malignancies (87%) were gastrointestinal in origin. The majority of infections (83%) originated from an intra-abdominal source. The most common NFF species were E. gallinarum (50%) and E. casseliflavus (30%). Most (61%) patients received combination therapy. Bacteremia recurred in 4.3% of patients, there was a 30-day mortality of 23%, and 4.3% had microbiological failure. E. gallinarum and E. casseliflavus isolates were genetically diverse with no spatiotemporal clustering to suggest a single strain. Frequencies of ampicillin resistance (4.3%) and daptomycin resistance (1.9%) were low. Patients with hematologic malignancy had infections with NFF enterococci that harbored more resistance genes than patients with solid malignancy (P = .005).

Conclusions

NFF bacteremia is caused by a heterogeneous population of isolates and is associated with significant mortality. Hematological malignancy is an important risk factor for infection with NFF resistant to multiple antibiotics.

Enterococci are commensal organisms of the gastrointestinal tract with the propensity to cause invasive disease in humans. The vast majority of infections in patients are caused by Enterococcus faecium and E. faecalis. A number of other enterococcal species also cause infection in humans, including E. gallinarum, E. casseliflavus, E. avium, E. raffinosus, E. durans, and E. hirae, among others, often referred to as non-faecalis, non-faecium (NFF) enterococci. Vancomycin resistance in enterococci (VRE) is a serious antimicrobial resistance public health threat [1]. While E. faecium (as well as E. faecalis, albeit less frequently) typically acquire vancomycin resistance through acquisition of the vanA/B gene clusters, some of the NFF enterococcal species (eg, E. gallinarum and E. casseliflavus) exhibit intrinsic chromosomally mediated vancomycin resistance via the vanC operon [2]. Of note, vancomycin resistance has been associated with worse clinical outcomes in patients with E. faecium infections [3, 4].

In the past, the epidemiology of NFF enterococcal infections has been difficult to estimate due to difficulties in accurate identification to the species level [5]. In the past 25 years, studies have found NFF enterococci to cause a range of 1%–17% of all enterococcal bloodstream infections (BSIs) [6–10]. The NFF enterococci are also prominent causes of hospital-acquired infections (HAIs), ranking 11th across all adult HAIs reported to the National Healthcare Safety Network [11]. Prior studies of NFF enterococcal bacteremia demonstrate a significant percentage of patients with a medical history of malignancy, ranging from 35% to 70% of studied patients [7–10, 12–14]. However, the clinical and molecular epidemiology of NFF enterococci has been incompletely explored within the population of immunocompromised patients with solid or hematologic malignancy. Here, we sought to describe the clinical characteristics, outcomes, and population structure and antimicrobial resistance gene content of NFF enterococci causing bacteremia in patients with solid organ or hematologic malignancy.

METHODS

Study Design

We conducted a retrospective chart review at a major cancer center in Houston, Texas, between January 2016 and May 2022. Cases were identified by searching the electronic medical record and clinical microbiology database for all blood cultures positive for enterococcal species that were NFF enterococci. Patients were included if they satisfied the inclusion criteria of age ≥18 years, were diagnosed with cancer, and had ≥1 blood culture positive for NFF enterococci.

Clinical Data and Definitions

Clinical data were managed using Research Electronic Data Capture (REDCap; Vanderbilt University). Data collected included patient demographics, medical comorbidities, transplant history, conditions and treatments causing immunocompromise, hospitalization history, risk factors for infection, antibiotic exposure 30 days preceding index bacteremia episode, bacterial species identification and antibiotic susceptibility tests, imaging results, and antibiotic treatment of infection. Immunocompromising medications included prednisone ≥10 mg per day or equivalent within 2 weeks of infection or chemotherapy, monoclonal antibody therapy, or chimeric antigen receptor T cell (CAR-T) therapy within 6 months of infection. Additionally, presence and duration of neutropenia, defined as absolute neutrophil count (ANC) <500 cells/µL, were recorded. Severity of illness was graded by the Pitt bacteremia score, and comorbidities were evaluated with the Charlson comorbidity index. The source of enterococcal BSI was assigned based on the infectious diseases consultant's or primary team physician's evaluation as recorded in the electronic medical record. If a source was not described in the medical record, it was determined upon independent review of the patient's clinical data by 2 infectious diseases physicians who were part of the study. Empiric antibiotic therapy was defined as antibiotics administered upon suspicion of infection after collection of blood cultures. Targeted antibiotic therapy was defined as the antibiotics that were continued or initiated after finalization of blood culture results. Antibiotic therapy was only recorded if the patient was treated for at least 24 hours.

Outcomes and Statistical Analysis

The primary outcome of the study was 30-day mortality, as all patients had close follow-up. Secondary outcomes included microbiological failure, defined as clearance of blood cultures ≥4 days after index blood culture while receiving at least 48 hours of antibiotics (as previously defined in our prospective study of enterococcal bacteremia [3] and other studies based in the same institution [15]). Recurrence of bacteremia was defined as a positive blood culture with the same organism within the 14 days after the documented eradication of bacteremia. Descriptive statistics were conducted for clinical and infection characteristics. Categorical variables were compared by the Pearson chi-square or Fisher exact test. Continuous variables were compared by Wilcoxon rank-sum test.

Clinical Microbiology and Genomic Sequencing

Enterococcal isolates from patients’ blood cultures were acquired from an institutional bank of bacterial isolates focused on patient bloodstream infections. Local identification of the isolates was performed previously by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) as part of standard of care in the clinical microbiology laboratory. Local antimicrobial susceptibility testing (AST) was performed via VITEK2 (bioMérieux) with AST for additional antibiotics (such as linezolid) performed on request via gradient strip testing (Etest, bioMérieux) according to Clinical and Laboratory Standards Institute (CLSI) protocols as part of standard of care. No rapid molecular diagnostic tests (ie, vanA/vanB detection) were performed on NFF enterococci. Genomic DNA was extracted using the DNeasy blood and tissue kit (QIAGEN, Germantown, MD, USA) as per the manufacturer's instructions, with the modification of predigestion of the cell wall with lysozyme and mutanolysin before extraction. DNA libraries were prepared using the DNA Prep kit (Illumina, San Diego, CA, USA) for short-read sequencing on either an Illumina MiSeq for 2 × 300 paired-end reads or an Illumina NextSeq2000 for 2 × 150 paired-end reads. The same input DNA was also used for long-read library preparation with SQK-RBK004 Rapid Barcoding Kit (Oxford Nanopore Technologies, Oxford, UK) sequencing on a GridION X5 with R9.4.1 flow cells. The long-read sequences were basecalled with Guppy, version 6.1.2 (Oxford Nanopore Technologies, Oxford, UK), using its high-accuracy model configuration.

De novo hybrid assembly utilizing both long- and short-reads was executed through a custom pipeline [16] employing flye [17] for initial Nanopore long-read draft assembly, along with subsequent polishing using Illumina short-reads. The assembly was then annotated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) [18], and additional antimicrobial resistance gene identification was carried out with NCBI Antimicrobial Resistance Gene Finder Plus (AMRFinderPlus) [19]. Taxonomic assignments, or species identification, of the genomic assemblies were confirmed via whole-genome Average Nucleotide Identity (ANI) [20, 21] as part of the PGAP pipeline. Unrooted and midpoint-rooted maximum likelihood phylogenetic trees were generated based on core genomes for both E. gallinarum and E. casseliflavus (the 2 species with the largest number of isolates). Trees were generated using RaxML [22] and visualized using iTOL [23].

In the research laboratory, linezolid susceptibility testing was performed by gradient strip method with Etest strips (bioMérieux, Inc., Durham, NC, USA). A .5 McFarland turbidity standard was prepared per strain and streaked on Mueller-Hinton agar. Antibiotic gradient strips (Etest, bioMérieux, Inc., Durham, NC, USA) were applied to the agar surface. Minimum inhibitory concentrations (MICs) were read after 24 hours of incubation. Interpretation of MICs was conducted based on the CLSI published breakpoints [24].

Patient Consent

The protocol for this study was approved by the institutional review board, which waived the requirement for individual consent from the patients based on the observational nature of the study.

RESULTS

Patient and Clinical Characteristics

Between January 2016 and May 2022, there were a total of 746 unique patients with enterococcal BSI, with 307 (41.1%) patients infected with E. faecium, 369 (49.5%) with E. faecalis, and 70 (9.4%) with NFF enterococci.

Patient demographics and clinical characteristics are displayed in Supplementary Table 1. The median age of patients (interquartile range [IQR]) was 63 (50–71) years, slightly more than half were male, and approximately three-quarters were Caucasian. Hematological cancer constituted 56% of the cancer diagnoses, with most categorized as leukemia (n = 30, 77% of hematologic cancers). Twenty-one patients (30% of patients overall) were stem cell transplant recipients, 80% of whom were allogeneic. Of the solid organ cancers, the majority (87%) were gastrointestinal in origin. In terms of other medical conditions, the most common were diabetes mellitus (31%) and ischemic heart disease (23%). Liver disease occurred in 17 (24%) patients; this was primarily mild liver disease (88%) related to malignancy. Similarly, biliary disease occurred in 26 (37%) patients, related to malignancy. The median Charlson comorbidity score (IQR) was 6 (4–8).

In terms of risk factors for infection, 52 patients (74%) had been hospitalized in the preceding 365 days. Many patients received immunosuppressing medications or had immunosuppressing conditions preceding their NFF BSI. Twenty-five (36%) patients had neutropenia with ANC <500 cells/μL at the time of their diagnosis with NFF BSI. Fifty-six patients (80%) had a central intravascular catheter at the time of diagnosis of NFF BSI. Two patients (2.9%) had a cardiac device in place at the time of infection. No patient had a history of endocarditis.

Thirty-six patients (51%) had had a screening vancomycin-resistant enterococci (VRE) rectal swab performed in the 3 months preceding infection, of which 2 (5.6%) were positive. Fifty-five patients (79%) had exposure to antibiotics in the 30 days preceding diagnosis with NFF enterococcal bacteremia. Thirty-nine patients (56%) had exposure specifically to anti-enterococcal antibiotics in the preceding 30 days. The largest number of patients had received treatment or prophylaxis with levofloxacin (28, 40%), cefepime (25, 36%), or linezolid (18, 26%).

Infection Characteristics

Of the 70 NFF enterococcal BSIs, half were caused by E. gallinarum, with the next most frequent being E. casseliflavus (30%), followed by E. avium (11%). Half of the NFF BSIs were polymicrobial. The most common additional bacteria present in the polymicrobial BSIs were Enterobacterales (46% of polymicrobial cultures) and coagulase-negative staphylococci (20%), in addition to several other bacterial species (Supplementary Table 2). Sixteen patients (23%) had a Pitt bacteremia score ≥2. Three patients had concurrent positive cultures from other sites positive for the same NFF within 48 hours of blood culture collection: 2 biliary fluid cultures (1 E. gallinarum, 1 E. casseliflavus) and 1 catheter tip culture from an intravascular line (E. gallinarum). Transthoracic echocardiogram (TTE) was performed in 36 patients (51%), 3 of whom also underwent transesophageal echocardiogram (TEE). The performance of TTE was determined by the treating team and included indications such as evaluation of shock, cardiac function, volume status, and infection. Of the 5 patients with unknown source of infection, 3 patients had TTE or TTE and TEE performed, which were negative for vegetations. None of the patients who underwent TEE had recurrence of bacteremia, microbiological failure, or 30-day mortality. No patient had confirmed endocarditis.

Infectious diseases consult occurred in 87% of cases, with a median (IQR) of 2 (1–3) days from initial blood culture collection. The source of BSI was determined to be gastrointestinal for the majority (83%) of NFF BSIs, followed by central line–associated (7.1%), primary bacteremia with unknown source (7.1%), and, least commonly, wound (sacral ulcer) (2.9%). No patients had evidence or diagnosis of genitourinary infection. Most patients (61%) received combination antibiotic therapy, the most frequent of which was daptomycin plus β-lactam(s) (31% of patients overall), followed by daptomycin plus β-lactam(s) plus tigecycline (10% of patients overall). The median dose of daptomycin used for combination therapy was 10 mg/kg, with 68% of patients receiving a daptomycin dose of ≥10 mg/kg (using actual body weight). Of the patients who received monotherapy, daptomycin was the most common (19% of patients overall), followed closely by β-lactams (17% of patients overall). The median dose of daptomycin used for monotherapy was 9.74 mg/kg, with 54% of patients receiving a daptomycin dose of ≥10 mg/kg. Thirty-six patients (51%) had antibiotic regimens containing some form of ampicillin or amoxicillin. Twelve patients (17%) underwent surgical or procedural source control measures: 9 patients had endoscopic retrograde cholangiopancreatography (ERCP) with biliary stent placement or exchange, and 3 patients underwent percutaneous biliary drainage by interventional radiology. Twenty-eight of 56 patients (50%) underwent central intravascular line removal after diagnosis of NFF BSI, with a median time from blood culture collection to central line removal of 3 (2–6) days. Of the 28 patients who had a central intravascular catheter removed, 21 patients had a diagnosis of hematologic malignancy and 14 patients had underlying neutropenia. Of the 28 patients who did not have removal of central venous catheter, 17 had a diagnosis of hematologic malignancy and 10 patients had underlying neutropenia.

Antibiotic Susceptibility Testing and Antibiotic Resistance Genes

Antibiotic susceptibility test (AST) categorizations and minimum inhibitory concentrations (MICs) were obtained from the clinical microbiology laboratory for all 70 patients. All E. gallinarum and E. casseliflavus isolates were reported as resistant to vancomycin in clinical AST reports. All NFF isolates underwent ampicillin AST. Three isolates had resistant ampicillin MICs: 1 E. raffinosus (MIC = 12 μg/mL), 1 E. gallinarum (MIC = 32 μg/mL), and 1 E. avium (MIC = 16 μg/mL). Of the 54 isolates that underwent daptomycin AST, 1 E. gallinarum was resistant to daptomycin (MIC = 8 μg/mL); isolates had a median daptomycin MIC (IQR) of 1.5 (1–2) μg/mL. A total of 28 isolates underwent linezolid AST, with 10 isolates reported with intermediate MICs and 3 reported with resistant MICs. The reported results for all available AST are listed in Table 1.

Table 1.

Antibiotic Susceptibility Testing Categorization for Non–E. faecalis, Non–E. faecium Enterococcal Bloodstream Isolates in Patients With Cancer; Antibiotic Susceptibility Categorization as Reported by Local Clinical Microbiology Laboratory at Time of Care

Antibiotic	Susceptible, No.	Intermediate, No.	Resistant, No.	Not Tested, No.	Total Tested, No.	% Susceptible Isolates
Ampicillin	67	−	3	0	70	95.7
Vancomycin	15	12	41	2	68	22.1
Linezolida	15	10	3	42	28	53.6
Daptomycin	53	1	0	16	54	98.1
Gentamicin (synergy)	11	−	0	59	11	100.0
Streptomycin (synergy)	8	−	3	59	11	72.7

^aRepeat linezolid gradient strip testing revealed that all available non-faecium, non-faecalis enterococcal isolates were susceptible to linezolid.

A total of 65 of NFF isolates from the 70 patients were able to be acquired from the BSI microbial bank for AST and whole-genome sequencing. Linezolid gradient strip testing performed on all 65 isolates acquired from the microbe bank resulted with susceptible linezolid MICs with a median MIC (IQR) of 1 (0.75–1.0) μg/mL. No isolate harbored the cfr, optrA, or poxtA genes associated with linezolid resistance.

Enterococcal isolates had antibiotic resistance genes identified for vancomycin, aminoglycosides, macrolide/lincosamide/streptogramin B phenotype, macrolides, chloramphenicol, trimethoprim, and tetracyclines. Sixty-three isolates (97%) had at least 1 antibiotic resistance gene detected. All E. gallinarum isolates acquired from the microbe bank (n = 35) harbored vanC (median vancomycin MIC [IQR], 4 [1–6]), and 1 E. gallinarum additionally harbored vanA (vancomycin MIC >256 μg/mL). All E. casseliflavus isolates (n = 21) also harbored vanC. The other NFF enterococcal isolates harbored no van genes.

Aminoglycoside resistance genes were detected in 17 (26%) NFF enterococcal isolates overall, with genes including aac(6′)-Ie-aph(2'’)-Ia (1 E. gallinarum, 1 E. avium), aac(6′)-Iid (2 E. hirae), ant(4′)-Ib (1 E. casseliflavus), ant(9)-Ia (3 E. avium, 2 E. raffinosus), aph(3′)-IIIa (6 E. gallinarum, 1 E. avium), and ant(6)-Ia/aad(6) (8 E. gallinarum, 1 E. avium). Other antibiotic resistance genes detected in the NFF enterococci are summarized in Supplementary Table 3.

Outcomes

Among the 16 patients (23%) who died at 30 days, 8 patients were infected with E. gallinarum, 5 patients with E. casseliflavus, and 1 each with E. avium, E. hirae, and E. raffinosus. Only Pitt bacteremia score ≥2 was associated with 30-day mortality in univariate analysis (P = .039). Patient demographics, cancer diagnoses, medical comorbidities, hospitalizations, measures of immunocompromise, NFF species, polymicrobial BSI status, infection source, antibiotic strategy, surgical management, and central line management were not associated with 30-day mortality (Supplementary Table 4). Three patients (4.3%) experienced microbiological failure. Three patients (4.3%) had recurrence of bacteremia; of note, 2 of these patients had 2 recurrent bacteremia episodes each. One patient had both microbiological failure and recurrence of bacteremia. All 5 patients with microbiological failure and recurrence of bacteremia had infection with E. gallinarum. Four out of the 5 patients had acute myeloid leukemia, and 4 out of 5 had neutropenia at the time of diagnosis of NFF BSI. All 5 patients had a central intravascular catheter at the time of NFF BSI, and 4 patients had removal of the catheter during the course of NFF BSI treatment. All patients with microbiological failure or recurrence of bacteremia received antibiotics to which the NFF enterococci were susceptible according to standard AST.

The 3 patients who experienced microbiological failure had the following respective treatment regimens before experiencing failure: (a) linezolid then daptomycin, (b) vancomycin then daptomycin, and (c) linezolid. The antibiotic regimens that resulted in negative blood cultures (ie, the eradication regimen) for these patients were (a) ampicillin, (b) linezolid plus daptomycin (12 mg/kg), and (c) linezolid plus ampicillin, respectively. In terms of recurrence of bacteremia, 1 patient was treated during initial bacteremia with linezolid and minocycline, and upon recurrence had eradication with daptomycin 10 mg/kg. Another patient was treated with linezolid plus daptomycin then a short course of ampicillin and was treated with ampicillin on recurrence; then, on second recurrence, the patient had eradiation with ampicillin plus daptomycin. The third patient was treated with linezolid plus vancomycin on initial bacteremia, then daptomycin on recurrence, then had eradication with ampicillin on second recurrence.

Phylogeny and Resistome of NFF Enterococci

We generated phylogenetic trees of each NFF species with our isolates and compared them with publicly available NFF genomes. E. gallinarum had the largest number of isolates. Clinical outcomes of 30-day mortality, microbiological failure, and recurrence of bacteremia were plotted on the phylogenetic tree for E. gallinarum and noted not to cluster. No specific patten of AMR genes was noted (Figure 1). There was no clustering of E. casseliflavus isolates with 30-day mortality outcome or specific pattern or clustering of AMR genes (Figure 2). There was no temporal relationship or spatial relationship of known hospital locations among the isolates in the phylogenetic trees of E. gallinarum or E. casseliflavus, indicating a low likelihood of clonal outbreaks.

Figure 1.

Phylogenetic tree of E. gallinarum bloodstream infection isolates in the setting of all publicly available E. gallinarum. E. gallinarum displayed in a midpoint-rooted maximum likelihood phylogenetic tree associated with cancer diagnosis of patients from whom the isolate was acquired, clinical outcomes, and antibiotic resistance mechanisms. Isolates from this study are indicated by a 2-digit integer. Additional E. gallinarum genomes obtained from the National Center for Biotechnology databank are indicated by a number prefaced by “GCA.” Cancer diagnoses and outcomes of 30-day mortality, microbiological failure, and recurrence of NFF bacteremia are indicated by dark blue squares. Resistance mechanisms are indicated by variable color squares. Abbreviations: GI, gastrointestinal; NFF, non–Enterococcus faecium, non–E. faecalis.

Figure 2.

Phylogenetic tree of E. casseliflavus bloodstream infection isolates in the setting of all publicly available E. casseliflavus. E. casseliflavus displayed in a midpoint-rooted maximum likelihood phylogenetic tree associated with cancer diagnosis of patients from whom the isolate was acquired, clinical outcomes, and antibiotic resistance mechanisms. Isolates from this study are indicated by a 2-digit integer. Additional E. casseliflavus genomes obtained from the National Center for Biotechnology databank are indicated by a number prefaced by “GCA.” Cancer diagnoses and outcomes of 30-day mortality, microbiological failure, and recurrence of NFF bacteremia are indicated by black squares. Resistance mechanisms are indicated by variable color squares. Abbreviations: GI, gastrointestinal; NFF, non–Enterococcus faecium, non–E. faecalis.

Patients with hematologic malignancy were noted to have infections with NFF enterococci harboring a higher number of AMR genes compared with the NFF enterococci infecting patients with solid malignancy (average 3.2 vs 1.6; P = .005).

DISCUSSION

Previous studies [6, 10, 12] have indicated that one of the main risk factors to develop an infection with NFF enterococci is the presence of malignancy. However, no studies have systematically investigated the burden of NFF enterococcal disease in patients with cancer. Moreover, prior studies did not evaluate the microbiological or genomic characteristics of infecting isolates in this patient population. Here, we assessed the largest cohort of cancer patients with NFF bacteremia to date. Furthermore, unlike previous studies [8, 9, 12–14, 25], our cohort included patients with solid tumors, patients with hematological malignancies, and recipients of bone marrow transplants.

During the study period, NFF enterococci caused 9.4% of enterococcal bacteremia, with a distribution of NFF enterococcal species causing BSI similar to other studies. Further, the proportion of polymicrobial BSI and the gastrointestinal source of NFF BSI were similar to previous reports (50% and 83%, respectively). However, we found several important differences as compared with previous findings. One of the main results of our study was that the 30-day mortality in our patient population was ∼2-fold higher (23%) than previously reported (8.8%–12.2%) [8, 13, 26]. This is likely attributable to the inherent patient complexity of cancer patients including those with hematologic malignancies and stem cell transplants. Also, our institution, as a national referral center, focuses on the care of patients with difficult-to-treat cancers, which could also have affected the mortality rate. It is important to note, however, that in our statistical analyses, 30-day mortality was not associated with cancer diagnosis, NFF species, infection source, antibiotic treatment, or different management strategies of central lines. The only direct factor associated with mortality was a higher Pitt bacteremia score (≥2), reflecting that the severity of the bacteremia episode is likely to be a major factor affecting the outcomes of cancer patients.

When comparing the mortality rate of NFF enterococci with in-hospital mortality caused by VRE bacteremia (mostly caused by E. faecium) in patients with cancer [3], the NFF bacteremia mortality was still lower (37.5% vs 23%, respectively) but higher than that of patients infected with vancomycin-susceptible enterococci (12.5%). Although it is difficult to compare these studies directly, these results likely indicate that bacteremia with enterococcal species that harbor many resistance determinants (see below) are likely to be indicators of poor prognosis in patients with cancer. Of note, previous studies have found the mortality rate in patients with NFF bacteremia to be equivalent to [14] or lower than mortality in patients with E. faecalis bacteremia [13].

A previous cohort study by our group [3] indicated that patients with enterococcal bacteremia who experienced microbiological failure (positive blood culture ≥4 days after the index culture) had a statistically significant increase in mortality. Indeed, microbiological failure was the strongest predictor of in-hospital mortality in patients with E. faecium bacteremia (hazard ratio, 5.03). Thus, we attempted to investigate the impact of microbiological failure or recurrence on mortality. Unfortunately, we were unable to obtain statistically significant analysis results for these outcomes due to small sample sizes. In our cohort with NFF enterococci in patients with cancer, the clinical features that were common to both microbiological failure and recurrence of bacteremia included patient diagnosis of acute myeloid leukemia, neutropenic status, and BSI caused by E. gallinarum.

In terms of clinical characteristics and risk factors for infection overall, we identified the following groups of patients: those with (i) solid malignancy and biliary disease (n = 27, 38.6%), (ii) profound neutropenia (n = 27, 38.6%), (iii) solid malignancy or lymphoma with invasive intestinal disease (n = 6, 8.6%), (iv) a central line–associated BSI (n = 5, 7.1%), and (v) non-malignancy-related infections of the gastrointestinal tract (n = 5, 7.1%). While solid malignancy causing biliary obstruction and subsequent infection has been described in patients with NFF BSI, we provide further evidence that neutropenia is an important risk factor for NFF BSI, particularly in patients with cancer.

Another interesting observation of our study relates to antimicrobial susceptibility testing. Although local standard AST indicated possible linezolid resistance in our NFF cohort, repeat AST performed on all isolates under strict conditions did not confirm the phenotype, and we were unable to find genetic determinants (mutations in 23S rRNA or ribosomal genes, presence of transferable genes previously associated with linezolid resistance) in the genomes of the infecting isolates. Discordance of linezolid AST results in enterococci is an established issue [27], reflecting differences due to AST methodologies as well as complications resulting from timing of MIC read and the potential presence of multiple zones of inhibition. Ampicillin nonsusceptibility occurred in 4.3% of NFF isolates, which is lower than previous reports of 8%–24% [14, 28–30]. Indeed, non-NFF VRE have a high frequency of ampicillin resistance, often reported to be up to 90%–95% [31, 32]. However, only half of patients received an antibiotic regimen containing ampicillin or amoxicillin. When evaluating patient treatment regimens that did not contain ampicillin/amoxicillin, we found that treating physicians seemed to be using broader-spectrum, nonenterococcal beta-lactams in combination with daptomycin or linezolid to treat a polymicrobial infection or intra-abdominal source. The frequency of daptomycin resistance in our isolates was low; indeed only 1 infecting isolate was daptomycin-resistant. Thus, most patients in this study were treated with daptomycin in either combination therapy or monotherapy.

Similar to other VRE, daptomycin constitutes an excellent empiric antibiotic for NFF enterococci, where linezolid use in the patient population in our study is limited by its myelosuppressive side effects. Furthermore, based on the resistance frequency in our study, it appears to be reasonable to use an aminopenicillin-based regimen to treat monomicrobial NFF infections when maximizing all efforts to achieve source control. We do note that in our study 2 of 3 patients with recurrent bacteremia ultimately had eradication with an ampicillin-containing regimen. The determination of the most efficacious antibiotic regimen for NFFs and the benefit of combination antimicrobial regimens over monotherapy require further study.

While E. gallinarum has been previously shown to cause clonal nosocomial outbreaks [33], there was no evidence of clonal transmission upon phylogenetic analysis. Indeed, the E. gallinarum and E. casseliflavus in this study represented diverse genetic lineages with no spatiotemporal clustering, and there was no clustering according to cancer diagnosis or antimicrobial resistance genes. The diversity of E. gallinarum and E. casseliflavus isolates, the largest group of NFF species in this study, indicates that the source of BSI is likely the gastrointestinal tract rather than a nosocomial or cryptic community clonal outbreak.

Limitations of the study include its single-center location, limited sample sizes due to the relative rarity of NFF BSI, and the lack of ability to compare the NFF enterococci with infections caused by similar organisms. Further studies are needed to accrue more patients with NFF BSI to explore trends in mortality and clinical outcomes. Future work could also concurrently comparatively evaluate E. faecalis and E. faecium BSI clinical outcomes to assess the impact of NFF enterococci among the more common enterococci.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.