Surveillance and Genomic Analysis of Third-Generation Cephalosporin-Resistant and Carbapenem-Resistant Klebsiella pneumoniae Complex in Germany

To analyse the epidemiology and population structure of third-generation cephalosporin-resistant (3GCR) and carbapenem-resistant (CR) Klebsiella pneumoniae complex isolates, patients were screened for rectal colonisation with 3GCR/CR K. pneumoniae complex on admission to six German university hospitals (2016–2019). Also collected were 3GCR/CR and susceptible K. pneumoniae isolates from patients with bloodstream infections (2016–2018). Whole-genome sequencing was performed followed by multilocus sequencing typing (MLST), core-genome MLST, and resistome and virulome analysis. The admission prevalence of 3GCR K. pneumoniae complex isolates during the 4-year study period was 0.8%, and 1.0 bloodstream infection per 1000 patient admissions was caused by K. pneumoniae complex (3GCR prevalence, 15.1%). A total of seven K. pneumoniae complex bloodstream isolates were CR (0.8%). The majority of colonising and bloodstream 3GCR isolates were identified as K. pneumoniae, 96.7% and 98.8%, respectively; the remainder were K. variicola and K. quasipneumoniae. cgMLST showed a polyclonal population of colonising and bloodstream isolates, which was also reflected by MLST and virulome analysis. CTX-M-15 was the most prevalent extended-spectrum beta-lactamase, and 29.7% of the colonising and 48.8% of the bloodstream isolates were high-risk clones. The present study provides an insight into the polyclonal 3GCR K. pneumoniae population in German hospitals.

Among the 3GCR/CR K. pneumoniae complex bloodstream isolates, 39 different STs were identified ( Table 2, Supplementary Tables S1 and S2), the most common being HiR clone ST307 (n = 13, 16.2%), HiR clone ST15 (n = 7, 8.8%), HiR clone ST48 (n = 7, 8.8%), ST219 (n = 5, 6.3%), and HiR clone ST147 (n = 4, 5%) ( Table 2). Other HiR clones identified were ST101 (n = 3), ST17 (n = 1), ST37 (n = 1), ST258 (n = 1), ST383 (n = 1), and ST395 (n = 1). Using cgMLST analysis, only a few clusters of closely related isolates could be identified, including a cluster of five K. pneumoniae bloodstream isolates assigned as ST219 that were recovered from three study centres, A, B, and C; a cluster of four HiR clone ST48 isolates recovered in 2017 from study centre D; and a cluster of four HiR clone ST307 isolates recovered from study centre A in 2017 and 2018. Finally, six small clusters of two K. pneumoniae isolates each, namely, ST13 (centre A and B), HiR clone ST15 (centre B), HiR clone ST101 (centre C), HiR clone ST147 (centre C), ST607 (centre F), and ST1825 (centres A and C), were also detected in the present study ( Figure 3 and Supplementary Figure S3). A total of 77 STs were identified among the 3GCS and CS bloodstream isolates ( Table 2, Supplementary Tables S1 and S2). HiR clone ST37 (n = 7, 7.4%) was the most common ST, followed by HiR clones ST14 and ST17 (each; n = 3, 3.2%). The HiR clones ST20 and ST101 were singletons. cgMLST analysis revealed the presence of only a few clusters of closely related 3GCS/CS K. pneumoniae isolates. Two ST160 isolates from study centre B and C and recovered in 2017 and 2018, respectively, clustered together, as well as two HiR clone ST14 isolates from centres C and F. Furthermore, two ST3640 isolates recovered from two different patients in centre C in 2018 were identical by cgMLST. Finally, two ST6069 K. pneumoniae isolates from centre D recovered in 2017 and 2018 were identical ( Figure 4 and Supplementary Figure S4).
isolates from centres C and F. Furthermore, two ST3640 isolates recovered fro different patients in centre C in 2018 were identical by cgMLST. Finally, two ST6 pneumoniae isolates from centre D recovered in 2017 and 2018 were identical ( Figure  Supplementary Figure S4).

Virulence Genes
Since various virulence properties may contribute to infectivity and persistence of K. pneumoniae, several virulence-associated factors were analysed in the present study. Forty-nine 3GCR/CR colonising isolates were grouped into 28 different KL types, and 42 isolates were classified as unknowns (Table S2). KL102 (n = 9) was the most prevalent K locus and was identified in 6 of 10 HiR clone ST307 and three of three ST1653 colonising isolates. Furthermore, KL2 (n = 3) associated with biosynthesis of the K2 capsule serotype was identified in three of four HiR clone ST14. Finally, one of five ST45, one of one HiR clone ST48, and one of one ST791 were typed as KL62 (n = 3). The siderophore yersiniabactin ybt locus was identified in 41 3GCR/CR K. pneumoniae complex isolates. The most prevalent yersiniabactin locus was ybt 10 harboured by the integrative and conjugative element ICEKp4 (n = 10), followed by ICEKp3 harbouring ybt 9 (n = 6), ICEKp5 harbouring ybt 14 (n = 6), and plasmid-encoded ybt 4 (n = 4). Two colonising isolates encoded the yersiniabactin ybt 17 co-located in ICEKp10 with colibactin clb 3, while another isolate harboured the aerobactin locus icu 3.
The virulence gene content of the 3GCR/CR K. pneumoniae bloodstream isolates was also investigated. A total of 20 bloodstream isolates were grouped into 11 KL types, while 60 isolates were classified as unknown (Table S2). KL62 (n = 6) was the most common locus associated with the K62 capsular serotype and was identified only in HiR clone ST48 isolates. In addition, KL102 (n = 3) was carried by HiR clone ST307, KL24 (n = 2) by HiR clone ST15, and KL114 (n = 2) by ST219. A total of 34 3GCR/CR isolates carried a siderophore yersiniabactin ybt lineage. The most common locus was ybt 10 harboured by ICEKp4 (n = 18), while ICEKp3 encoding ybt 9 and ICEKp12 carrying ybt 16 were found in five and three bloodstream isolates, respectively. In addition, an ST25 isolate with unknown KL type carried the aerobactin icu 1 and salmochelin iro 1 loci. In the present study, a total of 14 KL loci were identified among the susceptible bloodstream isolates, while 77 were classified as unknown (Table S2). KL2 (n = 4) was the most common capsule type identified in two of four HiR clone ST14, one of two ST39, and one of one ST380 isolates followed by KL14 (n = 2) in one of seven HiR clone ST37 and one of one ST76 isolates. A total of 32 3GCS/CS K. pneumoniae isolates harboured the virulence locus yersiniabactin, and the most common was ybt 9 harboured by ICEKp3 (n = 6), followed by ybt 10 encoded by ICEKp4 (n = 5) and ybt 14 carried by ICEKp5 (n = 4), while five 3GCS/CS bloodstream isolates harboured a colibactin locus. Furthermore, five K. pneumoniae isolates harboured aerobactin loci and six isolates salmochelin, while a total of three isolates harboured the hypermucoidy locus rmpADC.

Discussion
Increasing rates of 3GCR K. pneumoniae causing bloodstream infections have been reported by several countries according to EARS-Net. In detail, in 2020, 15% of the reporting countries 3GCR rates were <10%, while in 44% of the countries reporting data to EARS-Net, resistance rates were ≥50%, including mainly countries in the southern and eastern parts of Europe, e.g., Bulgaria (79.1%), Greece (74.5%), and Romania (67.9%) [9]. In the present multicentre surveillance study addressing the prevalence of 3GCR K. pneumoniae complex colonising patients on hospital admission, the carriage rate over the four-year study period remained stable, ranging between 0.5 and 1%. This result is in agreement with a previous survey conducted in Germany in 2014 reporting an admission prevalence of 0.8% for 3GCR K. pneumoniae [10]. In the present three-year surveillance study of patients with bloodstream infections, the prevalence of 3GCR isolates of K. pneumoniae complex was found to be 15.1% (2016, 16%; 2017, 19.2%; 2018, 11.5%), showing a slight decrease over the study period. Resistance to third-generation cephalosporins was mainly mediated by bla CTX-M-like betalactamases, with no difference between colonising and bloodstream isolates. EARS-Net reported for Germany slightly lower 3GCR rates in K. pneumoniae isolates causing invasive infections, i.e., 13.6% in 2016, 14.6% in 2017, and 12.9% in 2018 [9]. Finally, no colonising CR K. pneumoniae complex isolates were detected in the present study, while the prevalence of carbapenem resistance among bloodstream isolates was rather low (0.8%). Similar results were also reported by EARS-Net and other studies, suggesting that carbapenem resistance in K. pneumoniae complex is currently not a threat in Germany [9,10].
In the present study, the vast majority of 3GCR/CR bloodstream isolates (98.8%) were K. pneumoniae, confirming that this species is the major cause of invasive K. pneumoniae complex infections in healthcare settings in Germany [4,31]. Similarly, 96.7% of the colonising isolates were identified as K. pneumoniae. However, 25.4% 3GCS/CS bloodstream isolates presumptively identified as K. pneumoniae by the participating study centres were re-identified as non-K. pneumoniae by sequencing, highlighting that susceptible K. variicola and K. quasipneumoniae isolates are often causing bloodstream infections in Germany and are underreported. Similar frequencies of non-K. pneumoniae bloodstream infections, i.e., 24.4% K. variicola and 6.5% K. quasipneumoniae, were reported in the Stockholm area [16].
Overall, both colonising and bloodstream K. pneumoniae complex isolates were classified into a large number of STs (3GCR/CR colonising, n = 58; 3GCR/CR, n = 39; and 3GCS/CS, n = 77 bloodstream isolates), indicating that the population structure of K. pneumoniae in German hospitals is heterogenous and consists of multiple genotypes with even more diversity among 3GCS/CS vs. 3GCR/CR bloodstream isolates. K. pneumoniae HiR ST307 was the most frequent clone among the 3GCR/CR colonising and bloodstream isolates in the present study, accounting for 10 (11%) and 13 (16.2%) of isolates, respectively. HiR ST307 has a global distribution and has caused numerous outbreaks in healthcare settings worldwide [32]. Furthermore, an increasing prevalence has been observed for ST307 carbapenemase-producing K. pneumoniae isolates, even replacing other successful HiR clones such as ST258 in Italy and Colombia [21,33]. Outbreaks of XDR ST307 carrying resistance plasmids encoding either NDM-1, CTX-M-15, or OXA-48, have been reported previously in four medical facilities in north-eastern Germany [34,35]. Among the 3GCR colonising K. pneumoniae isolates, 29.7% were identified as HiR, as were 48.8% of the 3GCR/CR bloodstream isolates. When only the ST is used as inclusion criterion and the susceptible phenotype is not taken into account, HiR clones were also identified among 3GCS/CS bloodstream isolates, including ST14, ST17, ST20, ST37, and ST101, accounting for 15.8% of these isolates. These data suggest that although various K. pneumoniae HiR clones are circulating in the six German university hospitals participating in the study, no particular clone was predominant.
The cgMLST analysis of both colonising and bloodstream K. pneumoniae complex isolates revealed only a few small clusters of closely related K. pneumoniae isolates, indicating potential transmission events, whereas the non-K. pneumoniae isolates were all singletons. Representatives of HiR clones such as ST13, ST15, ST48, ST101, ST147, and ST307 also formed small clusters comprising between two and six isolates that were mainly centre-specific and likely reflect the local epidemiology of each study centre, while isolates representing other well-known HiR clones such as ST258 were singletons. Numerous previous studies have reported the presence of HiR clones such as ST101, ST147, or ST258 in different hospitals in Germany, highlighting that these strains were already circulating in the country [36][37][38].
Resistance  [10,39,40]. The high prevalence of ESBLs, such as CTX-M-15, in the polyclonal population of colonising and bloodstream isolates of the present study may suggest that horizontal plasmid transmission or dissemination of other mobile genetic elements may be responsible for the spread of acquired beta-lactamases, which needs to be further investigated. Of note, three of four carbapenemases detected in bloodstream isolates at study sites B and C were found in HiR clones, i.e., OXA-48, ST101; KPC-2, ST258; and VIM-19, ST383, which also co-harboured ESBLs and other antimicrobial resistance genes. In Enterobacterales, OXA-48, VIM-1, and KPC-2 are the most frequently detected carbapenemases in Germany, as reported by the German National Center for Multidrug-Resistant Gram-Negative Bacteria and have been frequently involved in healthcare-associated infections and outbreaks [37,38,41,42].
Numerous studies have reported that hypervirulent K. pneumoniae, a pathotype that is more virulent than classical K. pneumoniae, carries large virulence plasmids and causes infections in healthy individuals in the community. Furthermore, increased virulence has been associated with certain capsular serotypes, e.g., K1 and K2 [4,24,26]. In the present study, KL102 was the most frequent capsule locus among the colonising isolates, mainly detected in HiR ST307 (67%), concurring with previous findings from a K. pneumoniae ST307 outbreak in north-eastern Germany [35]. Conversely, KL62 was the most frequent capsular serotype among 3GCR/CR bloodstream isolates and was only detected in HiR ST48 isolates as previously reported in China [43]. Different acquired virulence factors such as capsule and aerobactin production have been linked to hypervirulent strains [24,26]. Siderophores such as yersiniabactin were identified in the present study in 45%, 42.5%, and 33.7% of the 3GCR/CR colonising, 3GCR/CR, and 3GCS/CS bloodstream isolates, respectively. Furthermore, aerobactin, colibactin, salmochelin, and the hypermucoidy determinant rmpA were also detected in a small number of isolates. In the present study, the diverse population structure of Klebsiella spp. isolates was also confirmed by the heterogeneity of acquired virulence factors.
To the best of our knowledge, the present work represents the largest study on the molecular epidemiology of 3GCR/CR colonising and bloodstream K. pneumoniae complex isolates in Germany; however, our study has several limitations. The data obtained from the admission prevalence study are not representative of the general population because many patients admitted to tertiary care hospitals represent a patient population with frequent hospital contacts and colonisation may have occurred during previous healthcare contacts. In addition, the present study did not investigate whether 3GCR/CR bloodstream infections were caused by the same isolates found at hospital admission.
In summary, the results of the present study describe the polyclonal population of 3GCR K. pneumoniae complex isolates that colonise patients on hospital admission and cause bloodstream infections. Although certain K. pneumoniae genotypes were more frequent, no extensive clonality was detected among the isolates, suggesting that a diverse 3GCR K. pneumoniae population circulates in German hospitals. MLST and virulome analysis supported these findings. Finally, our data demonstrate the presence of HiR clones circulating in Germany and provide epidemiological data on 3GCR and CR K. pneumoniae complex isolates that may contribute to more effective infection control.

Study Participants and Design
Two independent epidemiologic surveys involving patients (i) at hospital admission and (ii) with nosocomial BSI were conducted at six German tertiary care university hospitals in eastern (centre A), western (centre B), south-western (centres C and F), central (centre D), and northern (centre E) Germany.
In the prospective epidemiological study on the prevalence of 3GCR and/or CR (3GCR/CR) K. pneumoniae complex on hospital admission, patients aged ≥18 years were included between June 2016 and December 2019. Centre F did not participate in the survey in 2016. Excluded were patients from the departments of ophthalmology, paediatrics, psychiatry, and from intensive care units (ICUs). 3GCR/CR K. pneumoniae complex bloodstream isolates were collected from patients aged ≥18 years between October 2016 and December 2018. Patients from the departments of ophthalmology, paediatrics, and psychiatry were excluded. Finally, a random sample of third-generation cephalosporin-susceptible (3GCS) and carbapenem-susceptible (CS) K. pneumoniae complex bloodstream isolates collected between 2016 and 2018 at the same study centres were used as controls.

Species Identification and Antimicrobial Susceptibility Testing
Species identification was performed using MALDI-TOF MS (Bruker Daltonics GmbH, Bremen, Germany). Antimicrobial susceptibility testing (AST) for cefotaxime, ceftazidime, imipenem, and meropenem was carried out using VITEK ® 2 (bioMérieux, Nürtingen, Germany). VITEK ® 2 was routinely validated using characterised quality control strains. Isolates that were non-susceptible to cefotaxime and/or ceftazidime (MIC ≥ 2 mL/L) as well as isolates non-susceptible to imipenem and/or meropenem (MIC ≥ 4 mL/L) were included in the study and further characterised on the basis of the EUCAST resistance breakpoints for Enterobacterales (Version 6.0, January 2016, http://www.eucast.org/, accessed on 1 April 2022).

Whole-Genome Sequencing (WGS)
Total DNA was extracted using the MagAttract HMW DNA Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Sequencing libraries were prepared using the Nextera XT library prep kit (Illumina GmbH, Munich, Germany) for a 250 bp paired-end sequencing run on a MiSeq (Illumina GmbH) platform. The genomes were assembled de novo using Velvet [44].
Supplementary Materials: The following supporting information can be downloaded at https://www. mdpi.com/article/10.3390/antibiotics11101286/s1. Table S1: Overview of the STs of the K. pneumoniae complex colonising and bloodstream isolates including singletons. Table S2: Overview of the species, STs, and virulence factors of the K. pneumoniae complex 3GCR/CR colonising (a), 3GCR/CR (b), and 3GCS/CS (c) bloodstream isolates. Figure S1. (A) Prevalence of 3GCR K. pneumoniae complex carriage of patients on hospital admission per year and centre. (B) Prevalence of 3GCR among 880 K. pneumoniae complex bloodstream isolates per year and centre. Figure S2. Dendrogram generated using Ridom SeqSphere+ for the 3GCR/CR colonising K. pneumoniae complex isolates including HiR clones (n = 91) coloured by study centre and identified by ST type, ignoring missing values. Each box represents one isolate from an individual patient according to sequence analysis of 2358 cgMLST target genes. Figure S3. Dendrogram generated using Ridom SeqSphere+ for the 3GCR/CR K. pneumoniae complex bloodstream isolates including HiR clones (n = 80) coloured by study centre and identified by ST type, ignoring missing values. Each box represents one isolate from an individual patient according to sequence analysis of 2358 cgMLST target genes. Figure S4. Dendrogram generated using Ridom SeqSphere+ for the 3GCS/CS K. pneumoniae complex bloodstream isolates including HiR clones (n = 95) coloured by study centre and identified by ST type, ignoring missing values. Each box represents one isolate from an individual patient according to sequence analysis of 2358 cgMLST target genes.