We whole-genome sequenced 29 carbapenemase-producing K. pneumoniae isolates: 18 from neonatal carriage (isolates from swabs of neonates) (4), 3 from the NICU environment (cots and trolley handles, incubator doors, tables), and 8 from neonatal bloodstream infections. These samples were isolated from the NICU of Korle-Bu Teaching Hospital (Accra, Ghana) from September 2017 through February 2019 (5) (Table; Appendix).

Figure 1

Figure 1. Phylogenetic tree of 28 carbapenemase-producing Klebsiella pneumoniae isolates and their acquired resistance genes from the neonatal intensive care unit at Korle-Bu Teaching Hospital, Accra, Ghana, 2017–2019. The tree was produced by...

Twenty-eight of the 29 isolates were sequence type (ST) 17 and capsular type KL25. We excluded 1 isolate from further analysis that was ST48 and KL64. Core-genome phylogeny showed a close genetic relationship of all ST17 isolates (0–32 single-nucleotide polymorphism [SNP] differences; median 5 SNP differences), suggesting a localized outbreak (Figure 1). We estimated that the most recent common ancestor of the outbreak emerged in April 2017 (year 2017.3; 95% highest posterior density interval 2017.0–2017.6) with an estimated mean substitution rate of 2.1 × 10^–6 SNPs/site/year (9.9 SNPs/year) (Appendix Figure).

All isolates were resistant to amoxicillin/clavulanic acid, gentamicin, amikacin, cefuroxime, ceftriaxone, ceftazidime, tazobactam/piperacillin, and ciprofloxacin. The isolates were susceptible to colistin and had MICs of <1 μg/mL. All outbreak isolates harbored the carbapenemase bla_OXA-181 and extended-spectrum β-lactamase bla_CTX-M-15 in addition to other β-lactamases (bla_TEM-1B, bla_SHV-94). We also found several genes encoding resistance to other antimicrobial drugs: aminoglycosides (rmtB, aph(3¢')-Ib, aph(3¢)-Ia, aph(6)-Id, aac(3)-IId, aadA2, aadA2b); fluoroquinolones (qnrS, oqxA, oqxB); fosfomycin (fosA); macrolide (mph (A)); phenicols (catA2); sulphonamides (sul2, sul1); tetracyclines (tetA, tetG); and trimethoprim (dfrA12, dfrA14) (Figure 1).

Figure 2

Figure 2. Binary rational tree illustrating genetic diversity (presence–absence of genes) of the accessory genome of carbapenemase-producing Klebsiella pneumoniae isolates from the neonatal intensive care unit at Korle-Bu Teaching Hospital, Accra, Ghana, 2017–2019....

All isolates contained 4 common plasmid incompatibility (Inc) groups (IncX3, IncF1B (Mar), IncQ1, IncColKP3). Eighteen isolates also contained incompatibility groups IncFIB (K) and IncFII, and 3 contained additional IncColRNA (Figures 1, 2). Further analysis revealed that recently recovered isolates had more plasmid Inc groups than did older isolates (Figure 1). The accessory genome of the isolates showed large variation in gene content (Figure 2). These data illustrate that this variation existed at the time of the first sampling in September 2017, when the isolates formed 3 distinct clusters (Figure 2). The clustering is associated with differences in plasmid content of the isolates and represents the uptake or loss of 205 genes. On the basis of the phylogeny and metadata, we hypothesize that 4 major evolutionary events caused changes in Inc groups and the ancestor of the cluster of isolates with Inc groups IncX3, IncFIB, IncQ1 and ColKP3 (Figure 2).

A study in South Africa identified a fully closed plasmid carrying bla_OXA-181 (6). Using the short-read sequencing applied in this study, we cannot determine whether bla_OXA-181 is carried on a plasmid or located in the chromosome. Mapping of raw reads toward the fully closed plasmid reveals complete coverage across the whole plasmid for 24 of the 28 isolates; the remaining 4 most recent isolates had reads covering the whole plasmid (except for 4 genes). This finding might indicate these isolates have a similar plasmid containing bla_OXA-181, although we cannot rule out that these reads might belong to other related plasmids and not the previously reported plasmid (6).

We identified an outbreak of ST17 K. pneumoniae carrying bla_OXA-181 in a NICU in Ghana. Outbreak isolates were resistant to all antimicrobial drugs commonly used to treat neonatal infections (although it was susceptible to colistin). Similar outbreaks of ST17 OXA-181–producing K. pneumoniae have been documented in South Africa (7), further confirming the spread of this type of resistance into nonendemic regions (2). Time-based phylogenetic analysis showed the outbreak isolates share a recent ancestor (approximately April 2017). This finding suggests that the outbreak strain had been introduced recently into the NICU or that the outbreak strain had limited genetic diversity because of a recent bottleneck or selective sweep in the outbreak strain population.

K. pneumoniae is an entry point of antimicrobial resistance into the family Enterobacteriaceae (8). Thus, carbapenemase-producing K. pneumoniae in the NICU might transmit resistance to other Enterobacteriaceae species. Other studies have associated bla_OXA-181 with the insertion sequence element ISEcp1, which can spread cephalosporinases and extended-spectrum β-lactamases (9). In our study, all isolates possessed the IncX3 plasmid. This plasmid is self-transmissible and associated with worldwide dissemination of New Delhi metallo-β-lactamases 1 and 5 (10,11). Recent studies from countries in Africa have found bla_OXA-181 carried on the IncX3 plasmid in Enterobacteriaceae species, including K. pneumoniae (2,6,7).

In Europe, the spread of carbapenem-resistant K. pneumoniae has been driven by 4 carbapenemase-positive clonal lineages that are often transmitted in hospitals (8). The isolates from the NICU were genetically diverse, especially in the plasmid content of the accessory genome. This diversity indicates the genome evolved rapidly, similar to isolates from an outbreak of K. pneumoniae in Beijing, China. In the outbreak in China, the isolates underwent rapid genotypic evolution mainly through rearrangement (including the gain and loss of genes) in the accessory genome (12). Antimicrobial pressure in hospitals might lead to adaptation and resistance transmission of K. pneumoniae in the hospital environment (8).

From our data, we infer the background transmission of carbapenemase-producing K. pneumoniae in the NICU before its detection. Neonatal carriage or environmental contamination by carbapenemase-producing K. pneumoniae might have started or maintained the outbreak. Improved surveillance of multidrug-resistant organisms, buttressed with improved infection prevention and control activities, are required to detect and control outbreaks in low-resource settings.