Introduction
In both global and Malaysian contexts,
Klebsiella pneumoniae is the predominant species among carbapenem-resistant Enterobacterales (CRE). Carbapenem-resistant
K. pneumoniae (CRKP) represents a major global public health challenge, particularly in healthcare-associated settings [
1]. Surveillance data from Malaysia, as reported by the country’s National Health Institute, underscore the growing threat posed by CRKP [
2,
3]. The resistance rate of
K. pneumoniae to meropenem increased from 2.3% in 2018 to 5.0% in 2022, indicating a concerning upward trend. This increase is largely driven by the dissemination of carbapenemase-producing genes such as
bla_NDM and
bla_OXA-48, with
bla_NDM being the most frequently identified among Malaysian CRE isolates [
3].
CRKP poses a substantial clinical threat, driven primarily by the production of carbapenemase enzymes that render most β-lactam antibiotics ineffective and severely limit available treatment options. To elucidate the transmission dynamics of a prolonged CRKP outbreak at Hospital Canselor Tuanku Muhriz (HCTM), Universiti Kebangsaan Malaysia, a tertiary university hospital located in Cheras, Kuala Lumpur, we conducted a genomic epidemiology investigation integrating epidemiological data, enterobacterial repetitive intergenic consensus polymerase chain reaction (ERIC-PCR) genotyping, and whole-genome sequencing (WGS) of circulating CRKP strains isolated from the hospital.
Materials and Methods
Study Setting and Data Collection
This retrospective study was conducted at HCTM. The hospital has an infection prevention and control unit (IPCU) responsible for routine surveillance, infection prevention activities, staff training, and outbreak investigations. Routine IPCU surveillance data collected between January 1, 2022, and December 31, 2023, were reviewed. For each CRKP infection, information on patient demographics, ward or unit of admission, clinical diagnosis, and specimen type was retrieved.
Bacterial Identification and Antimicrobial Susceptibility Testing
CRKP identification was performed by HCTM’s Department of Diagnostic Laboratory Services (JPMD) using conventional biochemical tests and API 20E (bioMérieux) prior to the implementation of the automated VITEK 2 Compact system (bioMérieux) in July 2023, which utilized AST-N374 cards for Gram-negative organisms. Antimicrobial susceptibility testing was conducted using the Kirby–Bauer disc diffusion method on Mueller-Hinton agar for selected agents from the penicillin, cephalosporin, carbapenem, aminoglycoside, and fluoroquinolone classes and was interpreted according to Clinical and Laboratory Standards Institute (CLSI) M100 guidelines [
4]. Multidrug resistance was defined as non-susceptibility to at least 1 agent in 3 or more antimicrobial classes. For isolates with reduced susceptibility to carbapenems, minimum inhibitory concentrations for imipenem, meropenem, and ertapenem were determined using Etest gradient strips (bioMérieux).
Carbapenemase Detection and Classification
Carbapenemase production was assessed using the modified carbapenem inactivation method (mCIM) and, where applicable, the ethylenediaminetetraacetic acid-modified carbapenem inactivation method (eCIM), in accordance with CLSI guidelines [
4]. eCIM testing was performed for isolates that tested positive by mCIM.
ERIC-PCR and Cluster Identification
ERIC-PCR genotyping was performed for all 147 CRKP isolates collected during the study period using ERIC primers (forward: 5ʹ-ATG TAA GCT CCT GGG GAT TCAC-3ʹ and reverse: 5ʹ-AAG TAA GTG ACT GGG GTG AGC G-3ʹ) [
5]. The PCR protocol consisted of an initial denaturation at 95 °C for 5 minutes, followed by 30 cycles of denaturation at 94 °C for 1 minute, annealing at 48 °C for 1 minute, and extension at 72 °C for 1 minute, with a final extension at 72 °C for 10 minutes. PCR products were resolved on 1.5% agarose gels at 70 V for 1 hour and visualized under ultraviolet light. DNA fingerprint patterns generated by ERIC-PCR were analyzed using GelCompar II software to construct dendrograms based on unweighted pair group method with arithmetic mean clustering and dice similarity coefficients. Clinical and epidemiological data were cross-referenced to identify genetic clusters.
WGS, Genome Assembly, and Annotation
Selected strains representing ERIC-PCR clusters were subjected to further analysis using WGS. Genomic DNA extraction and purification were performed by bead homogenization using the ZymoBIOMICS DNA Extraction Kit (Zymo Research), according to the manufacturer’s instructions. For library preparation, 100 ng of genomic DNA was fragmented to approximately 350 bp using a Bioruptor and processed with the NEB Ultra II Library Preparation Kit (New England BioLabs) following the manufacturer’s protocol. Sequencing was conducted on an Illumina NovaSeq 6000 platform using a 2×150 bp paired-end configuration, generating approximately 1 Gb of sequencing data per isolate. Genome assembly was performed using SPAdes v4 [
6]. Genome annotation was carried out using Bacterial Annotation Toolkit (BAKTA) v1.9.3 with its standard database [
7], in which protein-coding genes were predicted using Prodigal v2.6.0 [
8] and annotated against multiple databases, including National Center for Biotechnology Information (NCBI) RefSeq, UniProt, Pfam, Rfam, KEGG, and Gene Ontology. Protein sequences were additionally submitted to the eggNOG-mapper and KofamKoala web servers [
9,
10].
High-Resolution Taxonomic Classification, Single-Nucleotide Polymorphism-Tree Construction, Multi-Locus Sequence Typing, and Antimicrobial Resistance Gene Prediction
For taxonomic classification, pairwise average nucleotide identity was calculated between genome assemblies and reference genomes in the GTDB r220 database using FastANI v1.33 [
11,
12]. Genome assemblies were further used to construct a single-nucleotide polymorphism (SNP)-based phylogenetic tree using maast v1.0.8 with default parameters [
13], in which SNPs were identified through pairwise comparisons to generate concatenated allele alignments. These nucleotide alignments were used as input for maximum-likelihood phylogenetic tree construction with FastTree 2 v2.1.11 [
14]. CRKP sequence types (STs) were assigned according to the PubMLST database using the standard 7-locus multi-locus sequence typing (MLST) scheme (gapA, infB, mdh, pgi, phoE, rpoB, and tonB), as implemented in the mlst tool [
15]. To identify antimicrobial resistance (AMR) genes, assembled genomes were searched against the NCBI AMRFinder, ResFinder, and CARD databases using Abricate v1.0.0 [
16–
18].
Integrated Epidemiological, Genotyping, and Genomic Analyses for CRKP Transmission Identification
Clinical and epidemiological data for WGS-selected isolates, including patient medical history, diagnosis, specimen type, ward or unit, and collection date, were reviewed in conjunction with ERIC-PCR genotyping and genomic findings (MLST, SNP, and AMR analyses) to infer potential transmission trajectories. Cases associated with strains exhibiting similar ERIC-PCR genotypes, STs, SNP profiles, and AMR gene content that also shared temporal overlap (ward admission periods) and spatial proximity (patient ward movement) were considered indicative of transmission events. In contrast, genetic or genomic dissimilarity among isolates with epidemiological overlap suggested independent introductions rather than direct transmission. This integrated analytical framework enabled higher-resolution mapping of CRKP transmission and provided critical insights into both clonal persistence and the coexistence of genetically diverse strains within HCTM.
Ethics Approval
This study was approved by the Medical Research Ethics Committee of Universiti Kebangsaan Malaysia (UKM PPI/111/8/JEP-2023-869).
Results
Most CRKP Isolates Were Multidrug-Resistant and Carbapenemase Producers
Thirty-eight and 109 cases of CRKP infections were recorded at HCTM in 2022 and 2023, respectively, resulting in the isolation of 147 CRKP isolates (
Table S1). The median age of patients from whom CRKP was isolated was 51.5 years (range, 14–89 years), and most patients were admitted to the general intensive care unit (GICU) (
n=46). Most CRKP isolates were recovered from blood specimens (
n=50) (
Table S1).
All study isolates were multidrug resistant, with susceptibility largely restricted to gentamicin and amikacin (81.5% and 82.1% susceptibility to gentamicin in 2022 and 2023, respectively; 78.4% and 92.2% susceptibility to amikacin in 2022 and 2023, respectively). A total of 124 isolates (84.4%) were carbapenemase producers. Ten isolates (6.8%) were mCIM-negative and classified as non-carbapenemase-producing CRKP. Twelve isolates (8.2%) were not tested because of the unavailability of mCIM reagents at the time of analysis. Among the carbapenemase-producing isolates, 87 (70.2%) were identified as metallo-β-lactamase producers (eCIM positive), whereas 5 (4.0%) were classified as serine carbapenemase producers (eCIM negative). eCIM testing was not performed for 32 isolates (25.8%), as testing was conducted for epidemiological surveillance rather than therapeutic guidance.
Diverse CRKP Genotypes Identified by ERIC-PCR with Inter-Ward Distribution
ERIC-PCR dendrogram analysis revealed 24 clusters at a 50% similarity threshold (
Figures S1,
S2). Strains within clusters 10, 11, 22, and 24 demonstrated closer similarity (>60%). No cluster was confined to a single ward, indicating the dissemination of multiple CRKP genotypes across hospital wards. Several isolates were further analyzed based on genotype similarity and epidemiological investigation (
Figure 1).
Improved Detection of CRKP Transmission Using Combined Epidemiological, Clinical, ERIC-PCR, and Genomic Approaches
Twelve CRKP isolates were selected for WGS based on ERIC-PCR-defined clustering and epidemiological linkage. Genome assembly statistics and genome completeness information are provided in
Table S2 and
Figure S3.
First, 4 isolates from ERIC-PCR cluster 22 (CR34/22, CR35/22, CR36/22, and CR37/22), collected from a single intensive care unit (ICU) patient in 2022 (
Table 1;
Figure 1A), were investigated. In addition to sharing similar ERIC-PCR genotypes, WGS revealed that these 4 isolates shared identical STs and highly similar AMR gene profiles (
Figure 2). This finding was consistent with persistent CRKP infection or colonization in a single patient during a prolonged ICU stay from late August to mid-September 2022, with multiple clinical specimens yielding isolates that were highly similar at the whole-genome level (
Figure 3).
Next, several ERIC-PCR genotypes (that of CR40/22, CR21/23, CR23/23, CR123/23) (
Table 1,
Figure 1B) corresponding to CRKP isolates recovered from the GICU over the study period (2022–2023) were analyzed. The ERIC-PCR profile of CR40/22 was initially unclear; however, WGS identified this isolate as ST4276 with a distinct AMR gene profile compared with other sequenced CRKP isolates (
Figure 2). In contrast, ERIC-PCR genotypes were indistinguishable for isolates CR21/23 and CR23/23 (
Figure 1B). These isolates were recovered from the same patient (patient D) in April 2023 and were later confirmed by WGS to belong to ST11 and to cluster on the same branch of the SNP-based phylogenomic tree (
Figure 3A), differing by only a small number of SNPs (
Figure 3B). Intriguingly, CR23/23, which was isolated 3 days later than CR21/3, did not carry
blaCTX-M-63,
blaTEM-1,
aac(3)-IId, or
bla-SHV-187, but instead harbored
blaSHV-1 (
Figure 2). Based on epidemiological investigation, the IPCU included the ERIC-PCR genotype of CR19/23, an isolate from patient C (a close contact of patient D in the ICU), for further ERIC-PCR and WGS analysis (
Figure 1C). CR19/23 was found to exhibit a different ERIC-PCR genotype from CR21/23 and CR23/23 and was confirmed by WGS to be phylogenomically distant from these isolates (
Figure 3). Further epidemiological investigation revealed that CR21/23, the first isolate recovered from patient D, originated from routine rectal swab screening and belonged to ERIC-PCR cluster 10 (ST11). This isolate likely originated from the community, was not hospital-associated, and did not appear to spread beyond its initial host (patient D). In contrast, CR19/23 and CR123/23 were identified as ST17 and shared a common ancestor with CRKP isolates recovered in 2022 (
Figure 3B). Overall, ST17 CRKP isolates recovered between August 2022 and October 2023 belonged to closely related ERIC-PCR clusters(clusters 10, 11, and 22)
Table 1;
Figure S2) and were predominantly isolated from the ICU, suggesting a prolonged outbreak of ST17 CRKP in this setting.
CR30/23, CR48/23, and CR58/23 from ERIC-PCR cluster 10 (
Table 1;
Figure 1D) were isolated from patients F, G, and H, respectively, during an outbreak in ward 6E in May and June 2023. Epidemiological contact tracing indicated that patient F (CR30/23) had close contact with patient D (CR21/23 and CR23/23) during their ICU stay, initially raising suspicion of CRKP transmission from the ICU to ward 6E. However, WGS demonstrated that CR30/23, CR48/23, and CR58/23 were ST17 and shared a more recent phylogenomic ancestor with each other than with CR21/23 and CR23/23 (
Figure 3), an indication that the hypothesis of transmission from patient D to patient F was incorrect. Indeed, all ST17 CRKP isolates subjected to WGS exhibited similar ERIC-PCR genotypes and were assigned to clusters 10, 11, or 22, which shared >60% similarity (
Figure 1E), highlighting the utility of ERIC-PCR as a screening tool for WGS strain selection in resource-limited hospital settings. Chronologically, patients F (CR30/23) and G (CR48/23) may have acquired ST17 CRKP in the ICU and subsequently transmitted the organism to patient H (CR58/23) during their stay in ward 6E. However, CR30/23 and CR48/23 were the only WGS isolates that carried
tet(A) and
blaTEM-235 (
Figure 2), suggesting independent acquisition of these AMR determinants. In addition, CR58/23 appeared to have diverged more recently from the ST17 ancestor than CR30/23 and CR48/23 (
Figure 3B). Sequencing additional isolates across the study period would allow a more comprehensive understanding of the evolutionary and transmission dynamics of ST17 CRKP.
Discussion
This study demonstrates the complementary yet critical roles of ERIC-PCR and WGS in the investigation of a CRKP outbreak in our university hospital. ERIC-PCR provided a rapid and cost-effective means of clustering isolates, making it a practical frontline screening tool in resource-limited settings [
19–
21]. However, poor gel image resolution, as observed for isolate CR40/22, may result from uneven PCR amplification of specific genomic regions and can complicate genotype assignment. In such situations, WGS offers substantially higher resolution and reveals genetic differences that ERIC-PCR cannot capture, consistent with previous reports highlighting the superiority of WGS for accurately delineating transmission events [
22,
23]. In our study, there was initial suspicion of CRKP transmission between patients C (CR19/23, ST17) and D (CR21/23 and CR23/23, both ST11), and subsequently between patients D (CR21/23 and CR23/23) and F (CR30/23) based on epidemiological investigation; notably, WGS demonstrated that these transmission events had not occurred. This finding further strengthened the confidence of our IPCU in applying WGS for future close-contact assessments and outbreak investigations.
A notable finding of this study was the persistence of CRKP ST17 across multiple time points during the study period, with possible horizontal acquisition of AMR mobile genetic elements in some isolates (CR30/23 and CR48/23). This persistence suggests that ST17 did not represent a short-lived outbreak but rather an endemic lineage entrenched within the ICU environment, with possible subsequent spread to ward 6E. Its prolonged survival points toward maintenance through environmental reservoirs and gaps in infection control measures [
24]. In contrast, lineages such as ST4276 and ST11 were detected only once, suggesting sporadic introductions rather than sustained persistence. The contrast between these 2 patterns illustrates the dual challenge faced by infection prevention teams: controlling both endemic high-risk clones and newly introduced strains.
The ICU setting plays a central role in facilitating both the persistence and spread of CRKP. Patients frequently undergo invasive procedures, receive broad-spectrum antibiotics, and rely on multiple medical devices, all of which increase susceptibility to colonization and infection. The coexistence of clonally related isolates such as ST17 with strains that appear to have independently acquired AMR determinants (e.g., CR30/23 and CR48/23 harboring
tet(A) and
bla_TEM-235) reflects the complex interplay between clonal expansion and repeated acquisition events, a pattern also observed in other high-risk settings [
25,
26]. These findings highlight the limitations of traditional epidemiological contact tracing. While epidemiological links provide valuable insights, they cannot fully capture the complexity of transmission dynamics, particularly in the presence of independent introductions. By integrating molecular tools such as ERIC-PCR and WGS into outbreak investigations, transmission pathways can be more accurately resolved, allowing infection prevention and control (IPC) teams to distinguish between clonal dissemination and multiple acquisition events [
27–
29].
The dissemination of ST17 from the ICU to ward 6E underscores the importance of hospital-level, rather than ward-specific, IPC strategies. Shared healthcare personnel, contaminated equipment, and patient transfers create opportunities for inter-ward spread, consistent with previous reports describing cross-unit dissemination of multidrug-resistant organisms [
30,
31]. Measures such as strict adherence to hand hygiene, environmental decontamination, and the use of dedicated medical devices for high-risk wards should therefore be prioritized [
32–
36]. In addition, inappropriate or excessive use of carbapenems may select for diverse resistant lineages, indicating that antimicrobial stewardship programs tailored to local resistance patterns are essential [
37,
38].
An important limitation in low-resource settings is the inability to perform WGS on all isolates because of financial and logistical constraints [
39]. In such contexts, ERIC-PCR serves as a valuable complementary method for rapidly and cost-effectively screening large numbers of isolates, thereby prioritizing representative samples for sequencing [
40]. This tiered approach enables a balance between resolution and feasibility, ensuring that outbreak investigations can still generate meaningful insights even when resources are constrained [
41–
43].
Overall, the persistence of ST17 in the ICU in our study illustrates how high-risk clones can establish long-term reservoirs in critical care environments, where they function as both amplifiers and sources of hospital-wide dissemination and potentially environmental contamination [
30,
31]. Sustained control of CRKP will require continuous genomic surveillance and bacterial characterization, ideally using WGS [
44,
45], integration of ERIC-PCR and WGS into routine workflows, and alignment of IPC and antimicrobial stewardship interventions [
46]. Future studies should focus on real-time sequencing and prospective epidemiological investigations to better elucidate the contributions of asymptomatic carriers and environmental reservoirs to the persistence of high-risk clones such as ST17.
Article information
Ethics Approval
This study was approved by the Medical Research Ethics Committee of Universiti Kebangsaan Malaysia (UKM PPI/111/8/JEP-2023-869) and conducted at Hospital Canselor Tuanku Muhriz in accordance with the principles of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study.
Conflicts of Interest
The authors have no conflicts of interest to declare.
Funding
This study was supported by the Faculty of Medicine Fundamental Grant (GFFP) (code: FF-2024-016), Universiti Kebangsaan Malaysia. No additional external funding was received.
Availability of Data
The whole-genome sequencing (WGS) project for Klebsiella pneumoniae isolates analyzed in this study has been deposited in the DDBJ/ENA/GenBank under the BioProject accession PRJNA1333618. Individual genome sequences are available under accession numbers JBROFO000000000–JBROFZ000000000 (BioSamples SAMN51830973–SAMN51830984). The version described in this paper is version JBROFO010000000–JBROFZ010000000.
Raw sequencing data and genome assemblies generated from this study have been submitted under the BioProject PRJNA1333618.
Authors’ Contributions
Conceptualization: UAZ, PP, SAS, HN; Data curation: UAZ, PP, XKC, NK, CLL, SAS, SS, NAAF, MFNY, GHM, HN, SRY; Formal analysis: UAZ, PP, HN, SAS, HMG; Funding acquisition: UAZ, PP, HN; Investigation: SAS, SS, UAZ, HN; Methodology: UAZ, HN; Project administration: all authors; Resources: all authors; Software: UAZ, HMG, SAS, HN; Supervision: PP, SAS, HN; Validation: PP, SAS, HMG, HN; Visualization: UAZ, HN; Writing–original draft: UAZ, HN; Writing–review & editing: all authors. All authors read and approved the final manuscript.
