Genomic Investigation of a Putative Endoscope-Associated Carbapenem-Resistant Enterobacter cloacae Outbreak Reveals a Wide Diversity of Circulating Strains and Resistance Mutations

Abstract

A genomic epidemiologic investigation of a putative carbapenem-resistant Enterobacter cloacae outbreak revealed few plausible instances of nosocomial transmission, highlighting instead the frequent importation of E. cloacae into our hospital. Searching for genetic determinants of carbapenem resistance demonstrated that most resistance is due to convergent mutations in phylogenetically diverse E. cloacae.

Carbapenem-resistant Enterobacteriaceae (CRE) are a major cause of antibiotic-resistant, healthcare-associated infections. Outbreaks of CRE are most often associated with globally distributed epidemic lineages for which resistance to carbapenems is conferred by plasmid-associated carbapenemases [1]. In the United States, the majority of CRE infections are due to Klebsiella pneumoniae that harbor a bla-KPC carbapenemase, with 70% of these infections being associated with the globally distributed ST258 lineage [2]. However, in recent years there have been increasing reports of CRE infections caused by other members of the Enterobacteriaceae family, with one of the most common being carbapenem-resistant Enterobacter cloacae (CREC) [3, 4]. In addition to sporadic CREC cases due to local spread of mobile resistance determinants, recent reports in the midwestern United States and elsewhere hint at the emergence of a CREC lineage, ST171, with epidemic potential [5, 6].

In 2015, the infection-prevention department at our institution identified CREC infections in 4 patients who had previous procedures with duodenoscopes. There have been several recent reports of CRE outbreaks linked to contaminated duodenoscopes, with subsequent outbreak investigations indicating that a small number of device-associated infections can be a sentinel of larger transmission chains [7]. We therefore undertook a genomic epidemiologic investigation to evaluate the possibility of an outbreak and the potential for duodenoscopes to be a source of CREC transmission. In addition, we sought to identify genetic determinants of carbapenem resistance and virulence in circulating CREC in order to better understand the evolution and epidemiology of these clinically important phenotypes.

METHODS

Outbreak Investigation

The University of Michigan Health System is a 990-bed tertiary academic hospital. All available CREC isolates that were collected as part of the clinical diagnostic workflow from September 2012 through December 2015 were included in the investigation (Supplementary Table S1). Isolates were considered carbapenem resistant if they were resistant to ertapenem, meropenem, doripenem, or imipenem or were intermediate and displayed a positive modified hodge test (1 isolate, UM-CRE-9). Minimum inhibitory concentration interpretations were made per Clinical and Laboratory Standards Institute guidelines. Clinical metadata were collected from the hospital’s laboratory information system. The University of Michigan Medical School Institutional Review Board approved this protocol.

Genomic Analysis

For the outbreak investigation, whole genome sequencing was performed on 42 isolates comprising putative CRECs and epidemiologically related cases. Pairwise genetic distance was used to rule out epidemiologically implausible transmission linkages. To detect variants associated with carbapenem resistance and virulence, we sequenced an additional set of 38 contemporaneous carbapenem-sensitive bloodstream isolates. Also, we searched for genes/pathways enriched in variation among carbapenem-resistant and bloodstream isolates, respectively (see Supplementary Materials for detailed methods). Sequence data are available under Bioproject PRJNA401340.

RESULTS

Transmission of Carbapenem-Resistant Enterobacter cloacae Was Infrequent and Duodenoscopes Were Not a Prominent Mode of Transmission

Investigation into a putative endoscope-mediated CREC outbreak led to the identification of 37 patients with 42 CREC infection isolates; 6 of these patients had a duodenoscopy prior to their infection. Overall, isolates displayed highly similar antibiotic resistance profiles (Supplementary Table S1), and there was extensive spatiotemporal overlap in the hospital among case patients (Supplementary Figure S1), supporting the possible clonal spread of CREC.

To determine whether case patients were linked by nosocomial transmission, we performed whole genome sequencing on the 42 CREC isolates. Pairwise comparison of sequenced isolates revealed few plausible transmissions, with a conservative threshold of 100 core-genome variants, yielding only 8 patients with a putative transmission linkage (Supplementary Figure S2). Focus on isolates from the 6 endoscope-exposed patients revealed only 1 pair that differed by <100 core-genome variants (UM-CRE-2, UM-CRE-3; Figure 1A), with the remaining isolates differing by >17000 variants. Further investigation into this putative transmission link revealed that these patients occupied the same ward prior to their infections (Figure 1B), thus leaving no genomically plausible cases of nosocomial transmission where endoscopes were the only epidemiologic link.

Figure 1.

Genomic epidemiology investigation of carbapenem-resistant Enterobacter cloacae (CREC). A, The number of core-genome variants between CREC isolates from patients who were exposed to duodenoscopes was calculated and log₁₀ transformed for visualization purposes. Four of the 6 isolates from patients who were exposed to duodenoscopes prior to CREC infection are >17000 core-genome variants from the other isolates, providing strong evidence against direct transmission. In contrast, isolates UM-CRE-2 and UM-CRE-3 are only distinguished from each other by 62 core-genome variants, indicating a plausible transmission relationship. B, Patient bed traces for the patients from panel A are shown, with the order of patients (y-axis) corresponding to the phylogenetic relationship of their CREC isolates. Each patient is represented by 2 rows, which correspond to the wards where patients resided while in the hospital (colors represent unique wards), and duodenoscopy procedure/E. cloacae infection dates (black/red). Note that patients 2 and 3, whose isolate’s pairwise genetic distance is consistent with recent transmission, overlapped on the same ward. C, A maximum-likelihood phylogeny was constructed for all 80 isolates sequenced as part of this study and a representative set of E. cloacae available in Genbank using variants identified in the 1.4 Mb core genome (scale in mutations per site in core genome). The presence of the KPC gene is shown for all isolates (black triangle). Duodenoscope exposure (green circle) is indicated for the 42 isolates that were included in the outbreak investigation. Carbapenem susceptibility (blue circle), infection type (blue-green circle), and the presence of an AmpD variant (yellow circle) are shown only for the 80 genomes that were sequenced for the current study (42 outbreak isolates and 38 contemporaneous carbapenem-sensitive bloodstream isolates).

Most bla-KPC-Carrying CREC Are ST171 While CREC Strains Lacking Carbapenemase Are Genetically Diverse

With genomic analysis indicating that most patients’ CREC isolates were not clonal, we the determined whether the incidence of CREC in our institution was due to a circulating resistance element [1]. A search of the 42 CREC genomes for carbapenemase-encoding genes revealed only 7 patients with isolates that harbored a carbapenemase, all of which were of the KPC type. Five of these KPC+ isolates were clustered on the phylogeny and within 100 variants of one another (Figure 1C, lower right). The other 2 isolates were distantly related and harbored distinct resistance elements. Closer inspection of this KPC+ cluster in the context of previously sequenced CREC revealed that they belonged to the emerging ST171 lineage, which appears to be stably associated with KPC-3 and has been associated with outbreaks in midwestern healthcare systems [6].

Evidence of Convergent Evolution of Carbapenem Resistance Among Highly Diverse E. cloacae

The observation that carbapenemase acquisition could only explain resistance in 7 of 42 isolates indicated that core-genome variation drove the majority of resistance. To facilitate the identification of core genomic loci associated with carbapenem resistance, we sequenced an additional set of 38 contemporaneous carbapenem-sensitive clinical E. cloacae isolates from our institution to serve as references, relative to which resistance-conferring variants could be identified. Next, we searched for carbapenem resistance loci by identifying genes that contained an excess of indels and nonconservative amino acid changes in resistant isolates, as compared to susceptible controls. This analysis yielded an association between carbapenem resistance and variants in ampD (Fisher exact test, P = .04), which is a negative regulator of the endogenous beta-lactamase ampC and has been previously implicated in carbapenem resistance in E. cloacae [8]. Of note, all 9 observed ampD mutations were distinct, providing strong evidence of convergent resistance evolution. Except for ampD, no other significant genes were identified, leaving the genetic basis for carbapenem resistance unexplained in 29 of the 42 isolates.

Evidence of Convergent Evolution of Bloodstream Infection Among Highly Diverse E. cloacae

While in an increasing number of studies genomics have been successfully applied to identify genetic signatures associated with antibiotic resistance, there are few examples of genetic association studies yielding predictors of other clinically important phenotypes [9]. In the hope of prioritizing patients who are at the highest risk of invasive E. cloacae infections, we searched for genetic signatures associated with isolates from bloodstream infections (BSIs). Examination of the distribution of BSI isolates on the phylogeny did not show evidence of invasive lineages, with BSI isolates spanning the full diversity of observed E. cloacae (Figure 1C). Therefore, if there are microbial factors associated with risk of BSI, these must have arisen repeatedly, either during the evolution of different lineages or within colonized patients. To test this hypothesis, we searched for evidence of recurrent mutational signatures associated with BSI isolates as compared to isolates from other types of infections. While no associations reached statistical significance, several of the most strongly associated pathways were involved in short-chain fatty acid (SCFA) metabolic pathways (Supplementary Figure S3).

DISCUSSION

In evaluating infection control measures, it is vital to know what drives the prevalence of healthcare-associated infections and the extent to which nosocomial transmission or importation is responsible. Here, we performed a genomic epidemiology investigation that allowed us to rule out a putative duodenoscope-mediated outbreak of CREC. Moreover, the observation that CREC at our hospital span the full diversity of previously sequenced isolates, with few cases of patients sharing closely related strains, strongly suggests that importation, rather than transmission, has driven the incidence of CREC infection at our institution.

While a great deal is known about the epidemiology of CRE-carrying carbapenemases, far less is known about the epidemiologic importance of mutational carbapenem resistance [10]. One hypothesis is that common mutations associated with carbapenem resistance carry a significant fitness cost, making these resistant strains less transmissible [11]. In agreement with this, we found few instances of plausible transmission of CREC that lacked a carbapenemase and no cases where an identified resistance mutation was shared among multiple patient isolates. In contrast to the apparent dead-end associated with mutational carbapenem resistance, we observed multiple patients with bla-KPC CREC from the ST171 clonal lineage that was stably associated with a bla-KPC-containing plasmid and appeared to be a lineage with significant epidemic potential [6].

In addition to understanding the genetic basis for antibiotic resistance, we also searched for genetic signatures associated with invasive infection. While the diversity of our isolate collection limited our power, we observed a signature of recurrent mutation in SCFA metabolic pathways among BSI isolates. This observation is consistent with a model wherein successful colonization, and potentially modification, of the gut environment precedes invasive infection [12]. In particular, the SCFA butyrate is known to play a role in the reinforcement of the colonic epithelial barrier, raising the possibility that altered SCFA metabolism could compromise the intestinal barrier and facilitate introduction of CRE into the bloodstream. In the future, it will be important to assess the robustness of these findings and search for additional genetic signatures associated with virulence.

Our genomic epidemiologic investigation provides a glimpse into the epidemiology of CREC infection and adds to a growing body of literature that demonstrates the potential for genomics to help prioritize patient risk of clinical outcomes and guide infection control practices.

Supplementary Data

Supplementary materials are available at Clinical 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.

Notes

Financial support. This work was supported by the University of Michigan Medical School Host Microbiome Initiative. S. E. H. was supported by the National Institutes of Health (T32-GM113900).

Acknowledgments. We thank Ali Pirani for bioinformatics support, Mike Taveirne and Carol Young for technical support, and Christy Scipione and Carolyn Dombecki for assistance with the initial outbreak investigation.

Potential conflicts of interests. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.