Precise Species Identification by Whole-Genome Sequencing of Enterobacter Bloodstream Infection, China Wenjing Wu,1 Li Wei,1 Yu Feng, Yi Xie, Zhiyong Zong
Total Page:16
File Type:pdf, Size:1020Kb
Precise Species Identification by Whole-Genome Sequencing of Enterobacter Bloodstream Infection, China Wenjing Wu,1 Li Wei,1 Yu Feng, Yi Xie, Zhiyong Zong (2), E. agglomerans to genus Pantoea (3), and E. sakazakii The clinical importance of Enterobacter spp. remains unclear because phenotype-based Enterobacter spe- to genus Cronobacter (4). Currently, 14 Enterobacter spp. cies identification is unreliable. We performed a genomic with validly published names exist, and 3 additional En- study on 48 cases of Enterobacter-caused bloodstream terobacter spp. have tentative species designations await- infection by using in silico DNA–DNA hybridization to ing validation under the rules of the International Code identify precise species. Strains belonged to 12 species; of Nomenclature of Bacteria (Bacteriological Code) Enterobacter xiangfangensis (n = 21) and an unnamed (Appendix 1 Table 1, https://wwwnc.cdc.gov/EID/ species (taxon 1, n = 8) were dominant. Most (63.5%) article/27/1/19-0154-App1.pdf). Enterobacter strains (n = 349) with genomes in Gen- Several Enterobacter spp., such as E. asburiae, Bank from human blood are E. xiangfangensis; taxon 1 E. cloacae, and E. hormaechei, cause infections in hu- (19.8%) was next most common. E. xiangfangensis and mans (1). Enterobacter strains extracted from clinical taxon 1 were associated with increased deaths (20.7% samples are usually reported as E. cloacae, and some- vs. 15.8%), lengthier hospitalizations (median 31 d vs. 19.5 d), and higher resistance to aztreonam, cefepime, times E. asburiae, E. hormaechei, or E. kobei, by auto- ceftriaxone, piperacillin-tazobactam, and tobramycin. mated microbial identification systems such as Vi- Strains belonged to 37 sequence types (STs); ST171 (E. tek II (bioMérieux, https://www.biomerieux.com). xiangfangensis) was most common (n = 6). Four ST171 However, such phenotype-based tests are unreliable strains belonged to a defined clone. Precise species for species identification ofEnterobacter and can result identification has greater implications for epidemiology in misidentification (1). For instance, all Enterobacter and infection control than treatment. spp. have a positive reaction for β-galactosidase, ar- ginine dihydrolase, citrate utilization, sucrose, amyg- nterobacter spp. belongs to the family Enterobacte- dalin, arabinose, and D-glucose but are negative for lysine decarboxylase, H S production, urease activity, Eriaceae and is a common pathogen in a variety of 2 infections, such as bloodstream and intraabdominal indole production, deaminase, and gelatinase (5–7). infections, most of which are healthcare associated (1). Differentiating Enterobacter spp. by biochemical reac- Enterobacter spp. is the third most common human tions commonly used in clinical microbiology labora- pathogen, after Escherichia coli and Klebsiella pneumoniae, tories is therefore difficult. The differences in clinical and is therefore of clinical importance (1). Enterobacter importance of each Enterobacter species remain large- consists of several closely related species (1) that cannot ly unknown because they are regularly misidentified typically be identified precisely by common phenotypic in clinical microbiology laboratories. tests. The taxonomy of Enterobacter is complicated by Because of the substantial reduction in cost of the reassignment to other genera of some species that whole-genome sequencing for bacterial strains, we formerly belonged to the Enterobacter genus. For ex- are entering the era of genomic microbiology (8). ample, E. aerogenes has been moved to genus Klebsiella Newly created methods can determine the overall nucleotide identities between genome sequences and therefore enable more precise species identifica- Author affiliations: West China Hospital, Sichuan University, tion (9). Calculation of average nucleotide identity Chengdu, China (W. Wu, L. Wei, Y. Feng, Y. Zie, Z. Zong); State (ANI) between genomes is widely used for species Key Laboratory of Biotherapy, Chengdu (W. Wu, Y. Feng, Z. Zong) DOI: https://doi.org/10.3201/eid2701.190154 1These authors contributed equally to this article. Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 27, No. 1, January 2021 161 RESEARCH identification. It has been proposed that ANI >96% treatment was not working and patients felt death would guarantee species assignation, whereas ANI was imminent. We categorized patients who chose of <93% can be considered species differentiation to be discharged but were likely to die in the hos- (10). However, ANI values in the range of 93%–96% pital (judged by the consensus of 2 physicians re- represent a vague zone in which the boundary of viewing blind data) as patients with predicted a species might fall (10). DNA–DNA hybridization death. BSI, the type of BSI (primary or secondary (DDH) remains the standard for species identifica- to infection of other sites), central line–associated tion, with a >70% cutoff recommended to define a BSI (CLABSI), and healthcare-associated infection species. However, DDH is cumbersome, prone to were determined by using criteria established by fluctuation, and requires the availability of type the Centers for Disease Control and Prevention’s strains. To overcome the shortcomings of DDH, in National Healthcare Safety Network (13,14). We silico DDH (isDDH) mimics DDH by comparing conducted the study in accordance with the amend- genome sequences and can be a reliable and conve- ed Declaration of Helsinki. The Ethics Committee nient tool for species assignation. To provide insight of West China Hospital approved the study and into the potential clinical importance of different En- waived informed consent. terobacter spp., we performed a genomic study using isDDH to identify bloodstream infection (BSI)–caus- Short-Read Genome Sequencing, Analysis, ing Enterobacter strains to the species level. and Precise Species Identification All strains underwent whole-genome sequenc- Materials and Methods ing by using the HiSeq X10 platform (Illumina, https://www.illumina.com). Genomic DNA was Strain and Susceptibility Tests prepared by using the QIAamp DNA mini kit We collected nonduplicate Enterobacter strains re- (QIAGEN, https://www.qiagen.com). We used covered from blood cultures during January 2016– Unicycler version 0.4.3 (15), in the conservative June 2018 at West China Hospital of Sichuan Uni- mode for increased accuracy, to perform a de novo versity (Appendix 2 Table 1, https://wwwnc.cdc. hybrid assembly. Precise species identification was gov/EID/article/27/1/19-0154-App2.xlsx). West established by determining the pairwise isDDH be- China Hospital is a 5,000-bed major referral hospital tween the genome sequence of the query strain and in western China. Initial species identification and those of type strains of Enterobacter spp., including in vitro susceptibility testing were performed by us- the validly published species and the species await- ing Vitek II. We determined MICs of colistin by us- ing validation (Appendix 1 Table 1). This process ing the broth microdilution method of the Clinical was performed by using the Genome-to-Genome and Laboratory Standards Institute (CLSI) and inter- Distance Calculator, formula 2 (16). A >70% cut- preted susceptibility following CLSI guidelines (11). off was applied to define a species. In addition, we For colistin and tigecycline, no CLSI breakpoints are determined the pairwise ANI of the genome se- available, so we used breakpoint standards defined quence of the query strain and those of type strains by the European Committee on Antimicrobial Sus- of Enterobacter spp. (Appendix 1 Table 1) by using ceptibility Testing (https://www.eucast.org). Mul- JSpecies software (https://imedea.uib-csic.es/ tidrug resistance was defined based on the criteria jspecies) with a >96% ANI cutoff to define a spe- for Enterobacteriaceae (12). cies (10). Sequence types (STs) were determined by using the genomic sequences to query the multilo- Patient Data cus sequence typing database of E. cloacae (https:// West China Hospital has a comprehensive hospital pubmlst.org/ecloacae). Antimicrobial resistance information system, which allowed us to retrieve genes were identified from genome sequences by patient data including age, sex, length of hospital- using the ABRicate program (https://github.com/ ization, and clinical outcomes (death or discharge) tseemann/abricate) to query the ResFinder data- from electronic medical records. One patient (with base (https://genomicepidemiology.org). strain 090040) had an unusually long hospital stay The genome sequences of all Enterobacter strains (578 d) because of a medical dispute and was re- recovered from human blood (n = 349, Appendix 2 moved from our analysis of length of stay. Accord- Table 2) were retrieved from GenBank (accessed 2018 ing to social customs in China, dying at home is Nov 1). These Enterobacter genomes were subjected to preferred over the hospital; it is likely many pa- precise species identification by using the Genome- tients chose to stop treatment and return home if to-Genome Distance Calculator as described. 162 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 27, No. 1, January 2021 Enterobacter Bloodstream Infection, China Clonal Relatedness on the Basis of for categorical variables. Pearson χ2 test was used Single-Nucleotide Polymorphisms when sample size (n) was >40 and theoretical fre- We performed single-nucleotide polymorphism quency (T) >5, Yates correction for continuity when (SNP) calling for genome sequences to untangle the n>40 and