Whole-Genome Sequencing Reveals Diversity of Carbapenem-Resistant Pseudomonas aeruginosa Collected through CDC's Emerging Infections Program, United States, 2016-2018

Abstract

The CDC's Emerging Infections Program (EIP) conducted population- and laboratory-based surveillance of US carbapenem-resistant Pseudomonas aeruginosa (CRPA) from 2016 through 2018. To characterize the pathotype, 1,019 isolates collected through this project underwent antimicrobial susceptibility testing and whole-genome sequencing. Sequenced genomes were classified using the seven-gene multilocus sequence typing (MLST) scheme and a core genome (cg)MLST scheme was used to determine phylogeny. Both chromosomal and horizontally transmitted mechanisms of carbapenem resistance were assessed. There were 336 sequence types (STs) among the 1,019 sequenced genomes, and the genomes varied by an average of 84.7% of the cgMLST alleles used. Mutations associated with dysfunction of the porin OprD were found in 888 (87.1%) of the genomes and were correlated with carbapenem resistance, and a machine learning model incorporating hundreds of genetic variations among the chromosomal mechanisms of resistance was able to classify resistant genomes. While only 7 (0.1%) isolates harbored carbapenemase genes, 66 (6.5%) had acquired non-carbapenemase β-lactamase genes, and these were more likely to have OprD dysfunction and be resistant to all carbapenems tested. The genetic diversity demonstrates that the pathotype includes a variety of strains, and clones previously identified as high-risk make up only a minority of CRPA strains in the United States. The increased carbapenem resistance in isolates with acquired non-carbapenemase β-lactamase genes suggests that horizontally transmitted mechanisms aside from carbapenemases themselves may be important drivers of the spread of carbapenem resistance in P. aeruginosa.

Keywords: Pseudomonas aeruginosa; antibiotic resistance; mechanisms of resistance; whole-genome sequencing.

FIG 1

Dendrogram of the 1,019 Emerging Infections Program (EIP) Pseudomonas aeruginosa genomes from isolates made from a core genome multilocus sequence typing (cgMLST) scheme of 4,440 core genes. Inner circle indicates sequence type (ST); outer circle indicates EIP site from which the specimen yielding the isolate was collected. Tree branches are scaled based on allele differences between nodes.

FIG 2

Frequency genetic markers of chromosomal mechanisms of carbapenem resistance found in the EIP P. aeruginosa isolates categorized by their susceptibility to three carbapenems (a to c; statistically significant differences between the resistant and susceptible/intermediate isolates as determined by Fisher’s exact test (P < 0.05) are shown with an asterisk [*]). Receiver operator characteristic (ROC) curves and area under the curve (AUC) values for the predictive power for carbapenem resistance of markers for the three chromosomal mechanisms of carbapenem resistance and combinations thereof, and a machine learning (ML) classifier made from the individual mutations on the genes associated with each mechanism (d to f). The ML line is the average of stratified, 5-fold cross-validations with shaded ±1 standard deviation.

FIG 3

Comparison of carbapenem susceptibility between 65 isolates with acquired bla genes (not including carbapenemases) and those without. (a) The bla-harboring genomes (bla+, in red) had different prevalence of markers of chromosomal carbapenem-resistance mechanisms. (b) Statistically significant differences as determined by Fisher’s exact test (P < 0.05) are shown with an asterisk (*).