Simultaneous Infection with Enterobacteriaceae and Pseudomonas aeruginosa Harboring Multiple Carbapenemases in a Returning Traveler Colonized with Candida auris

We report our clinical experience treating a critically ill patient with polymicrobial infections due to multidrug-resistant Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa in a 56-year-old woman who received health care in India and was also colonized by Candida auris. A precision medicine approach using whole-genome sequencing revealed a multiplicity of mobile elements associated with NDM-1, NDM-5, and OXA-181 and, supplemented with susceptibility testing, guided the selection of rational antimicrobial therapy.

ABSTRACT

We report our clinical experience treating a critically ill patient with polymicrobial infections due to multidrug-resistant Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa in a 56-year-old woman who received health care in India and was also colonized by Candida auris. A precision medicine approach using whole-genome sequencing revealed a multiplicity of mobile elements associated with NDM-1, NDM-5, and OXA-181 and, supplemented with susceptibility testing, guided the selection of rational antimicrobial therapy.

INTRODUCTION

Carbapenem-resistant Enterobacteriaceae (CRE) and Pseudomonas aeruginosa are major causes of health care-associated infections and have been declared urgent public health threats (1). Infections with these multidrug-resistant organisms (MDROs) are associated with high mortality rates, while their ability to acquire and disseminate antimicrobial resistance (AMR) determinants makes them particularly problematic (2, 3). The presence of multiple AMR determinants complicates the selection of antimicrobial therapy, and transcontinental spread of MDROs has the potential to alter local susceptibility patterns. Phenotypic tests for detection of carbapenem resistance (CR) are highly variable and unreliable (4), and therefore, providing a genomic context to aid in surveillance and epidemiological analysis is key in controlling the global dissemination of CR organisms (5). The advent of economical sequencing using both short- and long-read technology allows the microbiology laboratory to empower physicians with knowledge of the types of mobile genetic elements (MGE) and AMR determinants present in a patient and the potential for horizontal gene transfer and can inform treatment decisions at the bedside. (Part of this work was presented at IDWeek 2018, San Francisco, CA.)

CASE PRESENTATION

A 56-year-old woman presented to our institutional intensive care unit as a transfer from a rehabilitation hospital with concerns about sepsis. She had a history of intra-abdominal surgery in New Delhi, India, complicated by bowel perforation, secondary peritonitis, and bacteremia due to carbapenem-resistant Escherichia coli and Klebsiella pneumoniae treated with intravenous (i.v.) colistin. On initial presentation, her temperature was 36.8°C with a blood pressure of 87/55 mm Hg, heart rate of 118 beats/minute, and respiratory rate of 33 breaths/minute. The patient had a tracheostomy from a prior hospital stay, and at admission required ventilator support with a new oxygen requirement and an increase in tracheal secretions. Chest radiographs showed new right-sided pleural effusion and bilateral infiltrates in the lung bases. Laboratory results were significant with a white blood cell (WBC) count of 16,100 cells/μl and a serum creatinine of 1.5 mg/dl (baseline 0.7 mg/dl), and urinalysis after urinary catheter exchange showed >100 WBC/high power field. The patient was started on an empirical regimen of i.v. colistin (2.75 mg/kg/day divided every 12 hours), ceftaroline, and tobramycin. Subsequently, urine cultures grew E. coli and K. pneumoniae, respiratory tracheal aspirate cultures grew P. aeruginosa, and body site surveillance cultures grew Candida auris. Antimicrobial susceptibilities from the microbiology laboratory identified each isolate to be extensively drug resistant (XDR) (6) (Table 1).

TABLE 1.

Antimicrobial susceptibility results of Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Candida auris isolated from patient cultures^a

Drug	MICs (μg/ml) for:
	E. coli	K. pneumoniae	P. aeruginosa	C. auris
Antibiotic
Amikacin	16 (S)	16 (S)	>32 (R)
Ampicillin-sulbactam	>32/16 (R)	>32/16 (R)
Cefepime	>256 (R)	>16 (R)	>16 (R)
Ceftazidime			>16 (R)
Ceftriaxone	>32 (R)	>32 (R)
Ceftazidime-avibactam (CZA)*	>256/4 (R)	>256/4 (R)	>256/4 (R)
Ceftolozane-tazobactam*	>256/4 (R)	>256/4 (R)	>256/4 (R)
Ciprofloxacin	>2 (R)	>2 (R)	>2 (R)
Colistin*	0.38 (S)	0.38 (S)	0.38 (S)
Gentamicin	<4 (S)	<4 (S)	>8 (R)
Meropenem	>8 (R)	4* (R)	>8 (R)
Meropenem-vaborbactam*	24/8 (R)	4/8 (S)	>256/8 (R)
Piperacillin-tazobactam	>64/4 (R)	>64/4 (R)	>64/4 (R)
Plazomicin*	0.094 (S)	0.38 (S)	>256
Tobramycin	<4 (S)	>8 (R)	>8 (R)
Aztreonam*	>256 (R)	>256 (R)	4 (S)
Aztreonam + CZA*	0.5	0.094	4
Antifungal
Fluconazole				>256
Micafungin				0.125
Amphotericin B				2

^aAll susceptibilities were performed on a Vitek 2 instrument with the exception of those with an asterisk (*), which were performed using Etest. Susceptibility interpretations provided in parenthesis (S, susceptible; I, intermediate; R, resistant) according to Clinical Laboratories Standards Institute 2019 M100 (20) where applicable, except for plazomicin, for which FDA breakpoints are given (https://www.fda.gov/drugs/development-resources/plazomicin-injection). In vitro synergy testing for ceftazidime-avibactam (CZA) and aztreonam was performed by applying aztreonam Etest on Mueller-Hinton II agar containing CZA at a final concentration of 2.2 μg/ml of the avibactam component (see Supplementary Methods for details).

CHALLENGE QUESTION

Based on the antimicrobial susceptibility profile, how would you alter the empirical antimicrobial regimen for this patient?

• A.No change, continue i.v. colistin, ceftaroline, and tobramycin.
• B.Change tobramycin to plazomicin.
• C.Change ceftaroline to ceftolozane-tazobactam.
• D.Change therapy to ceftazidime-avibactam plus aztreonam.
• E.Change ceftaroline to meropenem-vaborbactam.

TREATMENT AND OUTCOME

In the setting of coinfection with multiple MDR organisms, it is a challenge for the clinician to select an effective treatment regimen. Given the patient’s history of medical care in India and the MDR phenotype of the infecting isolates, PCR was used to screen for potential carbapenemase genes (see Table S1 in the supplemental material). The E. coli, K. pneumoniae, and P. aeruginosa all harbored bla_NDM, the Enterobacteriaceae also harbored bla_CTX-M, and the E. coli contained bla_OXA-48; no bla_KPC was detected. Detection of genes encoding the NDM metallo-β-lactamase, which does not hydrolyze aztreonam, in conjunction with genes encoding CTX-M-15 and OXA-48-like enzymes that are inhibited by avibactam led us to consider the combination of ceftazidime-avibactam and aztreonam. In vitro synergy testing confirmed that the combination was active against the E. coli and K. pneumoniae; fortunately, P. aeruginosa was susceptible to aztreonam (MIC, 4 μg/ml) (Table 1). As the patient remained hypotensive with leukocytosis on colistin, the treatment regimen was switched to ceftazidime-avibactam (2.5 g i.v. every 8 hours) plus aztreonam (2 g i.v. every 8 hours) on day 3, and over the next 24 hours, the patient’s WBC count improved from 16.1 to 11.9 cells/μl, and she no longer required vasopressor support. Respiratory cultures were repeated on day 9 of illness and grew K. pneumoniae, P. aeruginosa, and Stenotrophomonas maltophilia. Trimethoprim-sulfamethoxazole was added to the antimicrobial regimen. On the 11th day of illness, the pleural fluid was drained and cultures from the fluid resulted in no growth. On the 20th day of illness, the patient experienced an episode of hypotension requiring vasopressor support. Blood, urine, and bronchoalveolar lavage cultures were negative for growth; however, computed tomography of the abdomen demonstrated a 5-cm abscess in the retroperitoneal space. Linezolid and metronidazole were added to the antimicrobial regimen, and percutaneous drainage of the abscess was performed on day 22. Cultures from the abscess fluid resulted in no growth, and after drainage, the WBC count normalized. All antibiotics were stopped on day 26, and the patient did not have any further signs of infection. Unfortunately, the clinical course was complicated by a large-territory cerebrovascular infarction on day 28, and the patient ultimately died after the decision was reached to withdraw care.

We performed whole-genome sequencing using both short-read (Illumina MiSeq) and long-read (Oxford Nanopore MinION) platforms to close the genome of each organism, identify plasmids, and resolve the genomic context of each MGE (see the Supplemental Methods for a detailed description of sequence analysis). This approach utilizes the long reads as a scaffold to bridge repeat regions associated with MGEs that cannot be assembled using short reads alone. The E. coli isolate was sequence type 167 (ST167) (BioSample accession no. SAMN12307470), a clone associated with dissemination of bla_NDM-5 and bla_NDM-1 (7–9). This isolate harbored five identifiable plasmids, three of which carried clinically relevant beta-lactamase genes (Fig. 1A), as well as a variety of other resistance determinants (Table S2). The K. pneumoniae isolate was ST16 (BioSample accession no. SAMN12307651), a clone also associated with the international spread of bla_NDM-1 (10), and harbored 8 identifiable plasmids. The P. aeruginosa isolate was ST773 (BioSample accession no. SAMN12307670) without identifiable plasmids (Fig. 1B). A maximum-likelihood phylogenetic analysis indicated that the colonizing C. auris isolate (BioSample accession no. SAMN12307774) clustered with other strains from South Asia, a notable finding given the patient’s travel history.

FIG 1.

Genomic context of clinically relevant AMR determinants. (A) Plasmids carried by E. coli ST167 and K. pneumoniae ST16 isolates with beta-lactamases annotated. Additional AMR determinants are labeled in red. (B) P. aeruginosa ST773 chromosome (GenBank accession no. CP041945.1) with the beta-lactamases, class I integron, and putative integrative conjugative element (ICE) annotated. Note the close physical proximity of rmtB with bla_NDM-1, suggesting cocarriage. (C) Intrapatient IncFII plasmid comparisons, including NDM variant contexts for E. coli, K. pneumoniae, and P. aeruginosa. The top DNA strand represents the E. coli ST167 circular 83,648-bp IncFII conjugative plasmid (GenBank accession no. CP041957.1), and the bottom linear DNA strand represents the K. pneumoniae ST16 circular 75,307-bp IncFII conjugative plasmid (GenBank accession no. CP041948). By BLAST identity, 99.99% of the K. pneumoniae IncFII aligns with the E. coli IncFII. The third strand is the 110,299-bp putative ICE showing a completely different bla_NDM-1 context for P. aeruginosa compared to the other IncFII bla_NDM-5 genes. AMR coding DNA sequences (CDSs) are labeled in red, IS26 transposases in white, ΔISAba125 in orange, the IS91 transposase in purple, class 1-like integrases in green, other integrases in light green, Tn3-like transposon elements in yellow, and other CDSs in gray. (D) Intrapatient IncX3 plasmid comparisons. The top DNA strand represents the E. coli ST167 circular 74,945-bp IncX3 conjugative plasmid (GenBank accession no. CP041956), and the bottom linear DNA strand represents the K. pneumoniae ST16 circular 47,983-bp IncX3 conjugative plasmid (GenBank accession no. CP041947). By BLAST identity, 99.99% of the K. pneumoniae IncX3 aligns with the E. coli IncX3. Predicted CDSs are labeled as in panel C with the addition that other insertion sequence (IS) transposases are labeled in peach.

Interestingly, the E. coli and K. pneumoniae isolates both harbored bla_NDM-5 on an IncFII conjugative plasmid (83 kbp and 75 kbp, respectively), where 99.9% of the K. pneumoniae IncFII aligned with the E. coli IncFII by BLAST identity, suggesting a potential horizontal gene transfer (HGT) event (Fig. 1C). Notably, these plasmids share significant homology with IncFII plasmids harboring bla_NDM variants in K. pneumoniae ST147 isolates from patients in the United Arab Emirates (UAE) (pABC143C-NDM) (11) and South Korea (pCC1410-2) (12) (Fig. S1). The E. coli and K. pneumoniae also harbored IncX3-like plasmids (75 kbp and 48 kbp, respectively; Fig. 1D) carrying bla_CTX-M-15 and bla_TEM-1. Additional transposable elements containing the bla_OXA-181 (similar to bla_OXA-48) and bla_CMY-4 genes were also present in E. coli. The P. aeruginosa isolate harbored a chromosomally encoded bla_NDM-1 in a different genetic context from the Enterbacteriaceae species, suggesting an independent acquisition event of this AMR determinant (Fig. 1C). In addition, the Pseudomonas isolate harbored a 16S-rRNA methyltransferase (rmtB) which was previously reported to colocalize with genes encoding MBLs (13). Interestingly, bla_NDM-1 and rmtB were located on an 110,299-bp integrative conjugative element (ICE) that is flanked by two direct-repeat 19-bp attL site-specific recombination sites.

The only single agent with in vitro activity across all isolates in this case was colistin. The patient, however, did not improve after 3 days of colistin. The plasma kinetics of colistin are highly variable (14), and while urinary excretion of the drug may favor its use for a urinary source of infection, the patient’s concomitant pneumonia and acute kidney injury made this a less attractive choice. The utility of newer agents introduced to combat carbapenem-resistant organisms, such as meropenem-vaborbactam, ceftolozane-tazobactam, and plazomicin, were compromised due to the nature of the resistance mechanisms present. Vaborbactam, a boronic acid inhibitor, possesses activity against Ambler class A and C serine β-lactamases but is not active against the class B MBLs or class D OXA-type enzymes carried by the organisms isolated in this case (15). Ceftolozane-tazobactam, which retains activity against some CR P. aeruginosa due to increased stability in the presence of AmpC and by overcoming porin- or efflux-mediated mechanisms of resistance, is not active in the setting of acquired carbapenemases (16). Plazomicin, a novel semisynthetic aminoglycoside that is stable in the presence of aminoglycoside-modifying enzymes, retained activity for both the Enterbacteriaceae. However, the P. aeruginosa carried a 16S ribosomal methyltransferase encoded by rmtB, which confers resistance to plazomicin (17). Although clinical data on the use of ceftazidime-avibactam plus aztreonam are limited to case reports (18, 19), the combination was used in this case, as it was active against the E. coli and K. pneumoniae; as noted, P. aeruginosa was susceptible to aztreonam. Thus, the answer to the challenge question is choice D. Notably, the complexity of standardizing high-throughput in vitro synergy testing along with the lack of clinical data and validation studies makes it difficult for a clinical microbiology laboratory to execute such testing.

In conclusion, an understanding of the molecular mechanisms of resistance in conjunction with whole-genome sequencing analysis and in vitro susceptibility testing to demonstrate synergistic activity of rational antibiotic combinations allowed precise therapy of a complex MDR-polymicrobial infection. Genomic studies of clinical patterns of AMR determinant dissemination can aid the clinician in choosing effective therapy for the treatment of MDROs. The global success of this strategy, including in countries most burdened by MDROs, depends on the increasing affordability of genomic diagnostic platforms and the continued development of active antibiotics.

This Journal section presents a real, challenging case involving a multidrug-resistant organism. The case authors present the rationale for their therapeutic strategy and discuss the impact of mechanisms of resistance on clinical outcome. Two expert clinicians then provide a commentary on the case.