Fluoroquinolone resistance in carbapenem-resistant Elizabethkingia anophelis: phenotypic and genotypic characteristics of clinical isolates with topoisomerase mutations and comparative genomic analysis Free

Abstract

Background

MDR Elizabethkingia anophelis strains are implicated in an increasing number of healthcare-associated infections worldwide, including a recent cluster of E. anophelis infections in the Midwestern USA associated with significant morbidity and mortality. However, there is minimal information on the antimicrobial susceptibilities of E. anophelis strains or their antimicrobial resistance to carbapenems and fluoroquinolones.

Objectives

Our aim was to examine the susceptibilities and genetic profiles of clinical isolates of E. anophelis from our hospital, characterize their carbapenemase genes and production of MBLs, and determine the mechanism of fluoroquinolone resistance.

Methods

A total of 115 non-duplicated isolates of E. anophelis were examined. MICs of antimicrobial agents were determined using the Sensititre 96-well broth microdilution panel method. QRDR mutations and MBL genes were identified using PCR. MBL production was screened for using a combined disc test.

Results

All E. anophelis isolates harboured the blaGOB and blaB genes with resistance to carbapenems. Antibiotic susceptibility testing indicated different resistance patterns to ciprofloxacin and levofloxacin in most isolates. Sequencing analysis confirmed that a concurrent GyrA amino acid substitution (Ser83Ile or Ser83Arg) in the hotspots of respective QRDRs was primarily responsible for high-level ciprofloxacin/levofloxacin resistance. Only one isolate had no mutation but a high fluoroquinolone MIC.

Conclusions

Our study identified a strong correlation between antibiotic susceptibility profiles and mechanisms of fluoroquinolone resistance among carbapenem-resistant E. anophelis isolates, providing an important foundation for continued surveillance and epidemiological analyses of emerging E. anophelis opportunistic infections. Minocycline or ciprofloxacin has the potential for treatment of severe E. anophelis infections.

Introduction

The genus Elizabethkingia, which was reclassified by Kim et al.¹ in 2005, comprises two former members of the genus Chryseobacterium (Chryseobacterium meningosepticum and Chryseobacterium miricola) that were identified by phylogenetic analysis with the 16S rRNA gene. Genome mapping has enabled further identification of species of this genus, including the Elizabethkingia meningoseptica, Elizabethkingia anophelis and Elizabethkingia miricola cluster.² Notably, accumulating evidence indicates that strains causing sporadic cases of meningitis and bacteraemia that were previously identified as E. meningoseptica were in fact E. anophelis, demonstrating the importance of modern taxonomic methods and molecular diagnostics (i.e. 16S rRNA sequencing and MALDI-TOF MS) in the identification of unusual and novel infectious agents.^3–5 By 2016, at least four E. anophelis infection outbreaks were identified, including a large multistate outbreak in the Midwestern USA causing significant morbidity and mortality.⁶ Like other members of this genus, E. anophelis is a Gram-negative, non-fermenting bacillus that is oxidase, indole, catalase and o-nitrophynyl, β-d-galactopyranoside (ONPG) positive.¹ Notably, the prevalence of MDR E. anophelis has been on the rise in recent decades, and thus treatment options for such infections are becoming increasingly limited.⁷

Carbapenems are well-known β-lactam antibiotics with broad-spectrum activity that have been commonly used for treating severe infections caused by Gram-negative bacteria.⁸ However, these agents are no longer effective owing to the development of different resistance mechanisms, including the production of MBLs, which are carbapenemases of worldwide concern as they possess activities against carbapenems and all β-lactam antibiotics.⁹ MBLs have been identified to be carried both on chromosomes and on plasmids, allowing for their rapid vertical and horizontal transmission within and among species. With the advent of next-generation sequencing (NGS) techniques, some antibiotic resistance genes were identified from the core and accessory genomes of E. anophelis,^10,11 including MBL genes such as blaGOB and blaB, which confer β-lactam resistance, and blaCME, encoding an ESBL.¹² The emergence of MBL-producing isolates is a further challenge to routine microbiology laboratories, which cannot always afford the time or expense of NGS to analyse the large amounts of data required for proper strain identification.

A variety of elements including efflux pumps, target-protecting proteins, plasmid-mediated resistance and even reduced porin expression have been shown to contribute to quinolone resistance.^13–15 In particular, mutations in genes that target the enzymes DNA gyrase and topoisomerase IV contribute to resistance in quinolone-resistant bacteria.^16,17 Rapid resistance to ciprofloxacin and levofloxacin is associated with chromosomal mutations in the QRDRs of the gyrA, gyrB, parC and parE genes in most species of bacteria. The prevalence of resistance to ciprofloxacin or levofloxacin in E. anophelis differed according to several studies; however, the mechanism has not been explored in detail to date.^5,7 Moreover, although instances of MDR are known to be common in E. anophelis, there are only a few reports on the specific occurrence of fluoroquinolone resistance in E. anophelis.¹⁸

Accordingly, the major aim of our study was to investigate the presence of the MBL gene in E. anophelis clinical isolates from our hospital that showed resistance to carbapenems, and to determine the genes conferring antimicrobial resistance to ciprofloxacin/levofloxacin. In addition, the AcrAB-TolC RND efflux pump has been shown to mediate the acquisition of fluoroquinolone resistance.¹⁹ Specifically, we used a rapid PCR-based method to detect carbapenemase genes carried by E. anophelis isolates and evaluated the phenotypic methods of MBL production based on the ability of metal chelators such as EDTA to inhibit the activity of MBL.

Methods

Ethics

This study was approved by the Tri-Service General Hospital Institutional Review Board (TSGH IRB 2-107-05-035), registered on 23 April 2018, and the requirement for patient consent was waived.

Bacterial isolates and identification of bacteria

A total of 115 non-duplicated isolates of E. anophelis were collected from patients at the Tri-Service General Hospital, a tertiary care centre in Taiwan, from 2017 to 2018. These isolates were collected from an abscess (n = 2), blood (n = 20), a central venous catheter tip (n = 1), sputum (n = 36), tracheal aspiration (n = 54), urine (n = 1) and wound discharge (n = 1) samples.

Antimicrobial susceptibility

The MICs of antimicrobial agents were determined using the Sensititre 96-well broth microdilution panel method following the manufacturer’s protocols. Antimicrobial susceptibility interpretation was based on the CLSI criteria²⁰ for other non-Enterobacteriaceae. The breakpoint used for colistin was that for Pseudomonas aeruginosa. The breakpoints used for doripenem and ertapenem were those reported for Enterobacteriaceae. The breakpoint used for tigecycline was determined according to US FDA criteria. Escherichia coli ATCC 25922 and P. aeruginosa ATCC 27853 were used as controls.

DNA extraction

Genomic DNA was isolated using a previously reported protocol.²¹ In brief, cellular lysis was achieved by a combination of EDTA/SDS detergent lysis and brief heat treatment. An additional phenol/chloroform step was further performed to deproteinate the preparation, yielding DNA of good quality. The concentrations of the purified genomic DNA were measured at 260 nm and the purity was estimated by measuring the ratio of the absorbance at 260 and 280 nm with a Picodrop spectrophotometer. DNA samples were stored at −20°C until used for PCR.

MALDI-TOF MS identification

All clinical E. anophelis isolates were identified using the VITEK MS Plus (bioMérieux) MALDI-TOF MS system as described previously.²² The spectra were acquired in linear positive-ion mode at a laser frequency of 50 Hz across m/z 2000 to 20 000 Da. The E. coli reference strain ATCC 8739 was used for instrument calibration on each section of the target slide, according to the manufacturer’s instructions. Under the spectral taxonomy tree, new folders of anophelis were established under the genus Elizabethkingia and the imported spectra were added to the respective folders. SuperSpectra of E. anophelis, E. miricola and E. meningoseptica were created with Saramis Premium software, and the frequency criterion of each peak was set to be >60%.

PCR for detection of bla genes responsible for carbapenem resistance in E. anophelis isolates

All isolates of E. anophelis were screened for class A carbapenemase (CIA and CME) and class B MBL (blaIND, blaGOB and blaB) genes using PCR with primers and conditions described previously,^23–26 with some modifications (Table 1).

The reaction mixture (50 μL) contained 10 mM Tris–HCl (pH 7.5), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM dNTPs, 10 pmol of each forward and reverse primer, 500 ng of genomic DNA and 0.8 U of Taq DNA polymerase. Amplification was carried out in a ProFlex PCR thermal cycler (Applied Biosystems, Foster City, CA, USA) with one initial denaturation step of 2 min at 95°C, 40 cycles of a denaturing step of 15 s at 94°C, an annealing step of 1 min at 45°C–58°C with corresponding genes and an extension step of 1 min at 72°C, and a final elongation step of 5 min at 72°C.

Combined disc diffusion method

Together with imipenem discs, imipenem/0.5 M EDTA combination discs were employed for detection in combined disc tests (CDTs) as described previously.²⁷ In brief, imipenem (10 μg) discs (BBL, France) were placed on a Mueller–Hinton agar plate for each tested isolate. A lawn culture of the test isolate (0.5 McFarland opacity standard) was conducted on Mueller–Hinton agar. Two 10 μg imipenem discs were placed on inoculated plates and 10 μL of 0.5 M EDTA solution was added to one of the discs. After overnight incubation, if the zone of inhibition of the imipenem+EDTA discs compared with that with imipenem alone was >7 mm, the test was considered to be positive.

Amplification and DNA sequencing of the QRDRs

All isolates of E. anophelis were screened for gyrA, gyrB, parC and parE genes by PCR amplification using specific primers, and the PCR products were sequenced for detection of amino acid polymorphisms.

The QRDRs of gyrase or topoisomerase IV genes were amplified with the primer pairs described in Table 1. The reaction mixture (50 μL) contained 10 mM Tris–HCl (pH 7.5), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM dNTPs, 10 pmol of each forward and reverse primer, 500 ng of genomic DNA and 0.8 U of Taq DNA polymerase. Amplification was carried out in a ProFlex PCR thermal cycler (Applied Biosystems) with one initial denaturation step of 2 min at 95°C, 40 cycles of a denaturing step of 15 s at 94°C, an annealing step of 1 min at 48°C–50°C with corresponding genes and an extension step of 1 min at 72°C, and a final elongation step of 5 min at 72°C. All PCR products were processed for DNA sequencing (Genomics, New Taipei City, Taiwan) with the same PCR primer sets. Raw sequences were reviewed using CLC Sequence Viewer 8 software (QIAGEN Bioinformatics). QRDR nucleotide sequences in gyrase or topoisomerase IV genes from each of the E. anophelis isolates were compared with the respective reference sequences in the GenBank database (NCBI Reference Sequence: Elizabethkingia anophelisNZ_CP007547.1).

Random amplification of polymorphic DNA (RAPD)-PCR and capillary gel electrophoresis

RAPD-PCR was performed using the primer 5′-GTCGATGTCG-3′ as described previously.²⁸ The reaction mixture (25 μL) contained 10 mM Tris–HCl (pH 7.5), 50 mM KCl, 2.5 mM MgCl₂, 0.2 mM dNTPs, 15 pmol of the RAPD primer, 50 ng of genomic DNA and 0.8 U of DyNAzyme II DNA polymerase (ABI, Thermo Fisher Scientific, Foster City, CA, USA). For every sample, each RAPD reaction was performed at least twice for each DNA extract. Amplification was carried out in a ProFlex PCR thermal cycler (Applied Biosystems) with one initial denaturation step of 5 min at 95°C, 40 cycles of a denaturing step of 1 min at 94°C, an annealing step of 1 min at 36°C and an extension step of 2 min at 72°C, and a final elongation step at 72°C for 8 min.

After PCR amplification, the products were analysed on a Qsep100 DNA Analyzer (BiOptic, Taiwan) according to the manufacturer’s instructions. PCR fragments were placed into a miniaturized single-channel capillary cartridge of the Qsep100 DNA-CE system with separation buffer. The run was performed using a high-resolution cartridge with a sample injection protocol of 8 kV for 10 s and separation at 5 kV for 300 s. The DNA alignment markers (20 bp, 1.442 ng/μL; and 5000 bp, 1.852 ng/μL) and the DNA size marker (50–3000 bp, 10.5 ng/μL) were obtained from BiOptic. Sample peaks were visualized using Q-Analyzer software (BiOptic).

Molecular pattern analysis

Isolates were categorized as identical, similar or unrelated according to their PCR banding patterns. The data were analysed using GelCompar II software (Applied Maths NV, Belgium). Dice similarity coefficients were calculated, and clustering was conducted by the unweighted pair group mean association method.

WGS of E. anophelis

To explore the mechanisms of fluoroquinolone resistance, WGS was performed for four isolates, comprising one isolate that was fluoroquinolone resistant without QRDR mutations, two isolates that were fluoroquinolone resistant with QRDR mutations and one isolate that was fluoroquinolone susceptible without QRDR mutations. The DNA of the isolate was prepared using a Qiagen genomic DNA purification kit according to the manufacturer’s instructions (Qiagen, Hilden, Germany) and the genome was sequenced using an Illumina MiSeq.2000 sequencing platform (Illumina, CA, USA). The short reads were assembled and optimized according to paired-end and overlap relationships via mapping reads to the contig using SOAP de novo. Subsystem technology prokaryotic genome annotations were based on Rapid Annotation using Subsystem Technology (RAST). The graphical map of the circular genome was generated using the CGView server. The virulence factors of the strain were analysed using the Virulence Factor Database (VFDB). Antimicrobial resistance genes were detected using HMMER3 (v.3.1b1) against the ResFams (Core v.1.2) with existing and putative new antibiotic resistance genes in bacterial genomes.¹¹ Bacterial Pan Genome Analysis (BPGA)²⁹ was used for comprehensive pan/core genome analysis, functional annotation of the core, accessory and unique genes to Cluster of Orthologous groups (COG) categories and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using default parameters.

Quantitative real-time PCR (qRT-PCR) of acrA, acrB and tolC in E. anophelis isolates

The mRNA expression levels of the efflux pump genes (acrB, acrA and tolC) were examined using qRT-PCR. Primers for each gene and optimal conditions for each primer pair are listed in Table S1 (available as Supplementary data at JAC Online). The experiments were performed using a housekeeping gene (16S) as an internal control, and the fold change of each gene was calculated by dividing the mRNA expression level of the above-mentioned genes by that of the 16S gene as described preciously.^19,25 In brief, cDNA (100 ng) was amplified by PCR with 40 cycles of denaturing (95°C, 15 s) and annealing (55°C, 30 s) using the Quantinova SYBR Green PCR kit (Qiagen). Quantitative analysis of the PCR products was carried out by a real-time PCR machine (Rotor-Gene Q, Qiagen) according to the manufacturer’s instructions. The experiments were performed in triplicate. The threshold cycle (Ct) value was defined as the cycle number at which the fluorescence generated within a reaction crossed the threshold value, and the relative Ct value of each target gene was compared with that of the levofloxacin-susceptible strain¹⁹ as a control (expression level = 1) to estimate the fold changes in relative mRNA expression among the samples.

Results

Characteristics of patients

All of the bacterial isolates tested were collected from different samples of hospitalized patients. Among the 115 isolates, 20 were obtained from blood cultures, which were cases of severe infections of particular concern. These 20 bacteraemic patients with E. anophelis infection were reported to have comorbidities such as malignancy, diabetes mellitus and COPD (Table 2). All of the patients included in our study had at least one underlying disease. The mortality rate of patients with E. anophelis infections was 25%.

RAPD results

RAPD analysis was employed to detect genotypic variation among 20 selected bacteraemic strains using one random primer. Amplified fragments in each strain ranged from three to seven bands, which varied in size from 50 to 1750 bp.

The RAPD dendrogram showed that the 20 bacteraemic isolates clustered into 11 groups with 85% similarity (Figure S1). There was no temporal or spatial overlap among the infected patients during hospitalization.

Antimicrobial susceptibility

The antimicrobial susceptibilities of the 115 E. anophelis isolates determined using the Sensititre 96-well broth microdilution panel method are summarized in Table 3. All isolates were resistant to amikacin, ceftazidime, colistin and imipenem, whereas all isolates were susceptible to minocycline. The susceptibilities of the isolates to doxycycline, piperacillin/tazobactam, ciprofloxacin and levofloxacin were 90%, 88%, 3% and 12%, respectively.

MBL gene detection and phenotypic detection

The PCR results showed that all 115 E. anophelis isolates harboured the blaB and blaGOB genes and these imipenem-resistant isolates were all positive for MBL production capability based on the CDT (Table 4) and MBL-producing test (Figure S2), with 100% consistency between the two methods.

QRDR detection

The associations of the amino acid substitutions in the QRDRs of the GyrA, GyrB, ParC and ParE regions of the 115 isolates with their respective ciprofloxacin/levofloxacin MICs are summarized in Table 4. Overall, 98 strains harboured a mutation in the QRDR of the gyrA gene, leading to an amino acid substitution of Ser83Ile or Ser83Arg. No other mutations were found in the gyrB, parC or parE genes. Clinical E. anophelis strains harbouring this mutation showed higher MIC values than those detected for WT strains. The Ser83Ile or Ser83Arg substitution in GyrA was associated with resistance to ciprofloxacin/levofloxacin. Among these isolates, only one did not have an amino acid substitution in GyrA, but nevertheless showed a high MIC of ciprofloxacin/levofloxacin.

The data summarized above suggested alternative mechanisms explaining the fluoroquinolone resistance in our E. anophelis isolates. Besides chromosomal mutations such as those in the genes encoding the protein targets (gyrA, gyrB, parC and parE), other mutations causing reduced drug accumulation and quinolone resistance genes associated with plasmids have been described.³⁰

WGS

The WGS analysis showed that four of the Elizabethkingia genomes contained antimicrobial resistance genes, including prominent genes encoding β-lactamases such as the MBL genes blaB and blaGOB; however, no plasmid-mediated or reduced porin expression for RND efflux systems was detected in any of our isolates regardless of their fluoroquinolone resistance status (Table S2). Moreover, these four isolates could be divided into three distinct strains based on core phylogenetic analysis (Figure S3). RAST annotation revealed components of the RND efflux system such as AcrA, AcrB and TolC. RND efflux pumps of the draft genomes of these four isolates are worthy of further attention, as they can be a major factor contributing to the resistance to fluoroquinolone in Gram-negative organisms.³¹

Gene expression of the AcrAB-TolC efflux pump

The expression levels of acrA, acrB and tolC were also examined using qRT-PCR. The expression level of the fluoroquinolone resistance-related efflux pump gene acrB showed a 12.7-fold increase in the isolate CMI02_16 compared with that of the control susceptible CMI02_67 isolate. Thus, the AcrAB efflux pump of isolate CMI02_16 plays a role in mediating fluoroquinolone resistance (Table S3).

Discussion

Among the 115 E. anophelis clinical isolates collected from a northern Taiwan hospital, the majority were obtained from adults with various types of underlying diseases, including malignancy, diabetes mellitus and COPD. This is in line with a review by Janda and Lopez,⁷ who indicated that primary infections associated with E. anophelis sepsis are typically found in very young patients (premature infants or neonates) or in adults with underlying medical conditions. A recent case report identified a healthy infant with E. anophelis infection with no previous medical or hospitalization history,³² suggesting the possibility that infection with this bacterium is not limited to adults; children might also be susceptible. The 20 bacteraemic patients included in our study had at least one underlying disease, with a fatality rate of 25%. Similarly, Figueroa Castro et al.³³ reported 11 patients with blood cultures that were positive for E. anophelis upon admission, all of whom had major comorbidities and had undergone recent treatments. The mortality rate was 18.2%, suggesting that E. anophelis is a true pathogen, especially in patients with multiple comorbidities.

E. anophelis is frequently misidentified as E. menigoseptica,^3,4 suggesting a requirement for the amendment of the discrimination rate and addition of non-claimed pathogens to databases of microbial identification systems. Although 16S rRNA sequencing enables the accurate identification of Elizabethkingia species by amplification using universal primer pairs, DNA sequencing of PCR products is laborious and time-consuming for routine clinical laboratories. Our results obtained using the less time- and labour-intensive MALDI-TOF MS method showed that all Elizabethkingia species isolates could be identified with excellent discrimination. Chew et al.³⁴ also pointed out that E. anophelis, rather than E. meningoseptica, is the predominant species found in blood cultures.

The information on susceptibilities of E. anophelis to antimicrobial agents is limited. Previous studies demonstrated that E. anophelis is typically resistant to most carbapenems, including β-lactams. Our study also showed a 100% rate of resistance to imipenem. In addition, there was a high rate (88%) of susceptibility to piperacillin/tazobactam (a β-lactam/β-lactamase inhibitor drug), which shows a wide range in the literature. All of the 115 E. anophelis isolates harboured the blaGOB and blaB genes, consistent with the high MICs of carbapenems and β-lactam drugs. This is in line with a previous report demonstrating that the genome of E. anophelis harbours multiple antimicrobial resistance genes.³⁵ This situation further exacerbates the difficulty in implementation of appropriate therapy. Figueroa Castro et al.³³ pointed out that combination antibiotic therapy would be recommended, pending the accumulation of more susceptibility data. Lin et al.¹⁸ also reported that all 67 E. anophelis isolates examined were susceptible to minocycline, consistent with our 115 E. anophelis isolates.

MBL production is an important mechanism of carbapenem resistance. Both the PCR method and the CDT method provided specific and accurate results on the MBL production capability of our isolates. However, it may not be practically possible for all laboratories to perform PCR owing to cost constraints and availability. Therefore, use of a simple screening test such as the CDT method could be an effective alternative tool to monitor these emerging MDR pathogens.

Quinolones are a widely used class of synthetic antimicrobials. Single amino acid changes in either gyrase or topoisomerase IV cause quinolone resistance,³⁰ and the resistance mutations are most commonly localized at the N-terminal domains of GyrA or ParC,³⁶ although mutations in specific domains of GyrB and ParE have also been shown to cause quinolone resistance in Gram-negative pathogens.^37,38 In the 2016 E. anophelis outbreak in Wisconsin, USA, 65 isolates were found to be susceptible to quinolones (ciprofloxacin and levofloxacin) and 1 isolate was resistant. Lin et al.³⁹ identified that E. anophelis strain EM361-9 isolated in Taiwan showed fluoroquinolone resistance (parC, parE, gyrA and gyrB) based on WGS, which was associated with a mutation of DNA gyrase subunit A (Ser83Ile). Lin et al.¹⁸ further reported susceptibility rates to ciprofloxacin and levofloxacin of 4.5% and 58.2%, respectively, in 67 E. anophelis isolates. Huang et al.⁴⁰ retrospectively evaluated cases of levofloxacin-resistant E. meningoseptica bacteraemia and found that they were associated with a high mortality rate; Huang et al.⁴¹ also described that inappropriate antibiotic use was associated with 14 day mortality in patients with E. meningoseptica bacteraemia. Seventeen of the E. anophelis isolates in our study were susceptible to ciprofloxacin and/or levofloxacin (MIC 1–4 mg/L), whereas others were resistant strains. The single amino acid substitution of Ser83Ile or Ser83Arg in GyrA was associated with high-level MICs of ciprofloxacin/levofloxacin against 98 E. anophelis isolates. However, no other mutations were found in the ParC, GyrB or ParE domains. Thus, early identification of ciprofloxacin/levofloxacin resistance in E. anophelis isolates appears to be important for tackling this MDR pathogen.

The development of NGS techniques has greatly contributed to our understanding of the drug resistance and infectivity of E. anophelis, allowing for the identification of multiple antibiotic resistance genes from the core and accessory genomes of E. anophelis.^42,43 Bulagonda et al.¹⁰ demonstrated that E. anophelis strains harbour several unique genes, and the ability for horizontal gene transfer indicated its distinct origin. Although one isolate in our study showed a high MIC against fluoroquinolone, there was no QRDR mutation identified, suggesting that another mechanism may drive this resistance. Recently, plasmid-mediated quinolone resistance genes were reported in Gram-negative bacteria, such as E. coli and Pseudomonas spp., and include the qnr, qep and oqx systems.⁴⁴ RamA and AcrR may be the major regulators of AcrAB-TolC-mediated fluoroquinolone resistance in Gram-negative bacteria.^19,45 Thus, further studies are needed to decipher the other fluoroquinolone resistance mechanisms in E. anophelis isolates.

The confirmed number of cases in the USA 2015–16 outbreak of E. anophelis infections was 65, which resulted in 20 deaths.⁶ Since E. anophelis was first isolated from mosquitoes, determining its relevance to human pathogenesis has been an ongoing pursuit.⁴⁶ A survey on the Wisconsin E. anophelis outbreak found bacterial strains that showed adhesion properties due to evolution.⁴⁷ This suggests the ability of E. anophelis to develop biofilms, which would further complicate treatment. Most bacteraemic patients in our study had used a ventilator system in the intensive care unit, suggesting a likely mode of transmission, considering that Li et al.⁴⁸ reported that Elizabethkingia spp. were the predominant bacteria identified in reused and disposable ventilator systems. Based on our results and these previous findings, empirical treatment of E. anophelis infection should include piperacillin/tazobactam plus quinolone, rifampicin or minocycline.

Overall, this study provides valuable information about carbapenem resistance genes and the quinolone resistance pattern, as well as the genetic diversity of clinical E. anophelis isolates. Moreover, we have demonstrated that the MALDI-TOF MS method with enriched databases could be a time- and cost-effective method for identifying E. anophelis clinical isolates. Nevertheless, more reports of E. anophelis infection are needed to determine whether cases such as those reported herein represent a situation of pathogen re-emergence or surveillance bias due to better case finding and organism identification. Regardless, these findings suggest the need for increased awareness of E. anophelis infections in clinical settings. Continuous monitoring and investigation of E. anophelis infections should be performed in collaboration with local health departments and healthcare providers. At the same time, further research should focus on elucidating the pathogenicity, modes of transmission and treatment regimens for E. anophelis.

Acknowledgements

We are grateful to Ching-Mei Yu for her assistance with the laboratory work.

Funding

This work was supported by Tri-Service General Hospital, Taipei, Taiwan, ROC (grant numbers TSGH-C104-203 and TSGH-C106-170).

Transparency declarations

None to declare.

References