Multidrug-resistant Klebsiella pneumoniae in a pediatric ICU: Emergence of newly identified ST7120 and the first report of ST792 in Taiwan

Abstract

Background

Multidrug-resistant (MDR) Klebsiella pneumoniae outbreaks in pediatric intensive care units (PICUs) pose serious risks to vulnerable patients. We report a one-year cluster of MDR K. pneumoniae in a PICU, along with the genetic relatedness, antimicrobial resistance profiles, and resistance mechanisms of the isolates.

Methods

We investigated a one-year cluster of MDR K. pneumoniae in a 14-bed PICU from September 2021 to August 2022. Selected clinical isolates underwent pulsed-field gel electrophoresis, multilocus sequence typing, and resistance gene detection. Patient location and transfer records were reviewed to infer possible transmission pathways.

Results

Twenty K. pneumoniae isolates from 12 patients, including 10 preterm infants, were obtained and classified into six sequence types. Two predominant sequence types (STs) were identified: ST792 (CTX-M-14, SHV-12), not previously reported in Taiwan, and the newly identified ST7120 (CTX-M-15, SHV-11, TEM-1). The cluster exhibited two temporal peaks, driven by different STs. During the second peak, up to three STs circulated simultaneously. One patient carried both ST792 and ST7120, suggesting coinfection or sequential acquisition. Three invasive infections occurred, two with carbapenem-resistant ST7120 (including one fatal case) and one with extended-spectrum β-lactamase-producing ST792, which had severe neurologic sequelae.

Conclusion

This MDR K. pneumoniae cluster involved a newly identified ST, ST7120, and the first reported case of ST792 in Taiwan, both were associated with prolonged colonization and invasive disease. Molecular typing revealed complex transmission routes not detectable through phenotypic methods alone. Integrating molecular epidemiology into routine MDRO surveillance could improve early outbreak detection and containment.

Highlights

• •One-year MDR K. pneumoniae cluster in a pediatric ICU.
• •First report of ST792 in Taiwan and a newly assigned ST7120.
• •Molecular typing revealed complex transmission routes not detected phenotypically.
• •ST792 and ST7120 were repeatedly detected; invasive infections occurred.
• •Findings support integrating molecular epidemiology into routine MDRO surveillance.

1. Introduction

Nosocomial infections pose a significant burden on patient outcomes and healthcare costs. In pediatric or neonatal intensive care units (ICUs), outbreaks of multidrug-resistant organisms (MDROs) are particularly associated with increased mortality and morbidity [[1], [2], [3], [4], [5]]. Among common MDROs, Klebsiella pneumoniae is a major pathogen that causes both healthcare-associated and community-acquired infections and contributes significantly to mortality in children worldwide [1,[6], [7], [8]].

In Taiwan, the prevalence of carbapenem-nonsusceptible Enterobacterales remains relatively low, with most cases attributed to K. pneumoniae. According to a national surveillance study conducted from 2012 to 2015, the rate of carbapenem nonsusceptibility in K. pneumoniae varied regionally, ranging from 0.5% to 3% [9]. However, in a more recent multicenter study from 2018 to 2021 involving 360 K. pneumoniae isolates, carbapenem nonsusceptibility increased to 3.3% in community-acquired bloodstream infections and 21.2% in hospital-acquired bloodstream infections (BSIs) [10]. With respect to carbapenem-resistant K. pneumoniae (CRKP), the largest study by the National Health Research Institutes from 2012 to 2015, which included 1457 carbapenem-nonsusceptible strains, reported a carbapenemase prevalence of 31.4% [9]. In a national multicenter study of K. pneumoniae isolates nonsusceptible to ertapenem, 65.7% were found to produce extended-spectrum beta-lactamases (ESBLs), and 55.6% harbored AmpC enzymes [11]. These findings suggest that carbapenemase production is not the primary resistance mechanism in Taiwan, with many nonsusceptible isolates instead producing ESBLs and/or AmpC enzymes [12].

The resistance mechanisms and sequence type (ST) distribution of CRKP differ between strains isolated from pediatric and adult populations. Previous studies have identified ST11 as the predominant ST among adults, with this clone emerging as the primary strain responsible for CRKP infections across many regions [13]. In contrast, pediatric populations demonstrate greater diversity in terms of dominant sequence types (STs), with significant geographic variation. STs such as ST36, ST37, ST45, ST629, ST2407, and ST2735 have been reported among pediatric patients in different regions [[13], [14], [15], [16]]. Given this diversity, pediatric-focused regional surveillance is essential for identifying emerging clones and guiding targeted infection control strategies.

Contact tracing is a critical step in preventing the spread of MDROs, as it can clarify strain transmission dynamics. Traditional contact tracing methods, which often rely on phenotypic characteristics such as carbapenem resistance, have notable limitations [17,18]. To address these gaps, molecular typing methods such as multilocus sequence typing (MLST) and pulsed-field gel electrophoresis (PFGE) provide higher resolution for clarifying outbreak dynamics.

In this study, we investigated a one-year cluster of multidrug-resistant K. pneumoniae in a pediatric ICU. We hypothesized that integrating molecular typing methods with spatiotemporal mapping could reveal hidden transmission routes that might be missed by conventional phenotypic surveillance. By applying these molecular tools, we aimed to characterize the clonal relationships, resistance mechanisms, and transmission dynamics of the predominant strains, thereby informing future surveillance strategies in high-risk pediatric ICU settings.

2. Methods

2.1. Medical setting

This study was approved by the Institutional Review Board (IRB No. 11311-003). The hospital in this study was a tertiary referral center in the city. The pediatric ICU comprises 14 beds divided into two sections. Section A primarily admitted low-birth-weight premature infants and included an adjacent room dedicated to isolation or hospice care, allowing family members to accompany the patient. Section B was more spacious and designed to accommodate term babies or older children. The layout in Section B facilitated easier movement and accommodated additional equipment necessary for the care of these patients.

2.2. Bacterial isolates and antimicrobial susceptibility testing

Clinical K. pneumoniae isolates were collected during a one-year clustering period. The collection period spanned one year, from September 1, 2021, to August 31, 2022, at a medical center. Specimens were collected on the basis of clinical judgment and the presence of symptoms or for deisolation purposes. The samples were processed, and the isolates were identified using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS; Bruker Daltonik, Bremen, Germany) at our institution. The antibiotic susceptibility of all available isolates was determined using the routine disk diffusion method and interpreted according to the Clinical and Laboratory Standards Institute (CLSI) M100 (34th edition, 2024).

2.3. Phenotypic screening criteria and isolate selection for molecular typing

Phenotypes were detected according to the following definitions:

• 1.ESBL phenotype: Isolates resistant to one or more third-generation cephalosporins, including cefotaxime, ceftriaxone, and ceftazidime.
• 2.Carbapenem resistance phenotype: Isolates resistant to at least one of the following carbapenems: ertapenem, meropenem, imipenem, or doripenem.

Criteria for serial isolate collection from the same patient were as follows:

• 1.Isolated from distinct or sterile sites.
• 2.Isolates exhibited a different phenotype than previous isolates from the same patient.
• 3.Isolation occurred at least one month after the previous isolation from the same patient.

2.4. PFGE

In selected K. pneumoniae isolates, PFGE was performed with a CHEF DR II apparatus (Bio-Rad Laboratories, Hercules, CA, USA) as previously described. In brief, DNA was digested by XbaI. Electrophoresis was then performed via a 1% agarose gel. The electrophoresis conditions used were as follows: Initial switch time 1.0 s, final switch time 40.0 s, run time 21 h, gradient 6 V/cm, angle 120°, and temperature 14 °C. Bacteriophage lambda ladder pulsed-field grade (PFG) and low-range PFG molecular weight markers were loaded onto all the gels. The similarities of the PFGE profiles of each strain were compared using a Dice coefficient at 1.0% tolerance and 1.0% optimization.

2.5. MLST and eBURST analysis

In selected K. pneumoniae isolates, MLST was performed using internal fragments of the seven housekeeping genes (gapA, infB, mdh, pgi, phoE, rpoB, and tonB) as previously described [19]. To verify the diverse STs of the K. pneumoniae isolates, we used the MLST scheme available from the Pasteur Institute K. pneumoniae MLST website (https://bigsdb.pasteur.fr/klebsiella/). Additionally, eBURST analysis (https://www.phyloviz.net/goeburst/) was performed to investigate the evolutionary relationships and clonal complexes among the isolates. Triple-locus variants are connected with a line.

2.6. Molecular identification of carbapenemase, ESBL, and outer membrane proteins (OMPs)

Plasmid DNA and genomic DNA were extracted using the Geneaid Plasmid DNA Mini Kit and Geneaid Genomic DNA Mini Kit (Geneaid Biotech, New Taipei City, Taiwan), respectively. Polymerase chain reaction (PCR) was used to characterize the ESBL genes (bla_TEM, bla_SHV, and bla_CTX-M), ampC genes (bla_DHA and bla_CMY-2), and carbapenemase genes (bla_KPC, bla_NDM, bla_VIM, bla_IMP, and bla_OXA-48 like). Specific primers were used as previously described [20,21]. In addition to screening for ESBL and carbapenemase genes, the OMP genes ompK35 and ompK36, including their upstream promoter regions, were amplified by PCR. The primers used were designed on the basis of the K. pneumoniae reference sequence (GenBank accession no. KN046818.1).

2.7. Functional analysis of OmpK35 and OmpK36

Outer membrane proteins were prepared using a modified previously described method. Briefly, the isolates were grown overnight in nutrient broth at 37 °C, washed twice, and resuspended in cold 10 mM HEPES. The cells were sonicated on ice, and the cell debris was removed by centrifugation at 3000×g for 10 min at 4 °C.

The supernatant was subsequently centrifuged at 14,000×g for 30 min at 4 °C. The resulting pellet was treated with 2% sodium lauroyl sarcosinate in 10 mM HEPES for 30 min at room temperature to solubilize inner membrane proteins. Outer membrane proteins were recovered by centrifugation, separated by 10% SDS‒PAGE, and visualized by Coomassie Brilliant Blue staining. K. pneumoniae ATCC 13883 was used as the reference strain.

3. Results

During the study period, a total of 12 patients with 20 K. pneumoniae isolates were identified (Table 1). Most of the patients (10 out of 12) were preterm infants whose gestational ages ranged from 26 to 33 weeks. The hospitalization duration for these patients ranged from 31 to 216 days. The days on which K. pneumoniae isolates were identified ranged from 3 to 110 days after admission.

Table 1.

Characteristics of included children with identified Klebsiella pneumoniae isolates.

	Age	Gender	Underlying disease	Hospitalization days	Isolation day (after admission)	Isolate ID	Site
P1	0d	Male	GA: 26 weeks	118	36	P1-1	Blood
						P1-2	Blood
P2	15y	Male	Acute necrotizing pancreatitis	63	25	P2-1	Blood
P3	0d	Male	GA: 28 weeks, NEC stage IIIb post operation	216	110	P3-1	Ascites
						P3-2	Rectal swab
						P3-3	Blood
P4	0d	Female	GA: 27 weeks, bilateral IVH	142	51	P4-1	Sputum
P5	0d	Male	GA: 28 weeks	129	21	P5-1	Sputum
						P5-2	CSF
						P5-3	CSF
P6	0d	Male	GA: 30 weeks, twin B	43	7	P6a-1	Blood
						P6a-2	Rectal swab
						P6a-3	Rectal swab
						P6b-1	Rectal swab
P7	0d	Male	GA: 33 weeks	31	3	P7-1	Sputum
P8	1y4m	Male	Infantile spasm with WASF1 mutation	50	41	P8-1	Urine
P9	0d	Female	GA: 26 weeks	99	39	P9-1	Urine
P10	1d	Male	GA: 35 weeks, pulmonary hemorrhage	45	19	P10-1	Sputum
P11	0d	Female	GA: 30 weeks	39	23	P11-1	Urine
P12	0d	Female	GA: 30 weeks	58	21	P12-1	Blood

Abbreviations: GA: gestational age; IVH: intraventricular hemorrhage; NEC: necrotizing enterocolitis; CSF: cerebrospinal fluid.

To investigate the genetic relatedness of these isolates, PFGE was performed. The genetic characterization and clustering of the K. pneumoniae isolates identified in our study are shown in Fig. 1. The isolates were categorized into six distinct clusters, with two major groups (A and D). Group A was associated with ST792, and all the isolates in this group carried the SHV-12 and CTX-M-14 genes. One isolate in this group (KP6b-1) was identified as carbapenem resistant, but no carbapenemase gene was detected. However, an insertion in the ompK36 gene likely caused loss of function. Group D represented a newly identified sequence type, ST7120, and consisted of isolates that carried SHV-11, CTX-M-15, and TEM-1, distinguishing them from Group A. This group had a greater number of carbapenem-resistant isolates, with five identified, although no carbapenemase genes were detected. Truncations in ompK36 were observed in these strains. To further evaluate whether the identified ompK36 alterations were associated with loss of porin expression, outer membrane protein profiles of representative isolates were analyzed by SDS‒PAGE (Fig. S1). Compared with the reference strain, the carbapenem-resistant isolates lacked the OmpK36 band, whereas the OmpK35 band was preserved, which is consistent with porin loss. All the isolates lacking detectable OmpK36 expression exhibited a carbapenem-resistant phenotype. MLST allelic analysis revealed that ST7120 differed from ST43 at a single locus (pgi: allele 1 vs. 5), indicating that ST7120 is a single-locus variant (SLV) of ST43. eBURST analysis further supported the close evolutionary relationship between ST7120 and ST43 (Fig. S2).

Fig. 1.

Pulsed-field gel electrophoresis (PFGE) patterns, multilocus sequence typing (MLST), and antimicrobial resistance gene profiles of clinical Klebsiella pneumoniae isolates. The red dashed line indicates the 80% similarity threshold used to define closely related strains.

On the basis of the PFGE and MLST results, we elucidated the distribution of strains during the cluster period. The numbers of patients who carried ESBL-producing K. pneumoniae (ESBL KP) and CRKP during the study period are shown in Fig. 2. The cluster had two peaks. The first peak included ST6968 and ST1810, whereas the second peak included four different STs: ST792, ST1962, ST2150, and ST7120. During the second wave of the cluster, up to three STs were in circulation simultaneously. We further performed eBURST to examine the relationships between different STs, as shown in Fig. S1. Apart from the close relationship between ST7120 and ST43, the remaining STs were evolutionarily distant, indicating substantial genetic diversity among the isolates.

Fig. 2.

Temporal distribution of multidrug-resistant (MDR) Klebsiella pneumoniae in the pediatric intensive care unit. The upper panel shows the number of patients with MDR K. pneumoniae over time. The y-axis represents the number of patients (colonized or infected). The solid black areas indicate patients with carbapenem-resistant K. pneumoniae (CRKP), and the dashed lines indicate patients with ESBL-producing K. pneumoniae. The lower panel shows the temporal distribution of major sequence types (STs), with horizontal bars indicating the time span from first to last detection of each ST. ESBL-producing and carbapenem-resistant isolates are displayed together to highlight that identical STs included different resistance phenotypes, indicating the limitation of phenotype-based surveillance in delineating true transmission clusters.

3.1. Hypothesized transmission links during the second peak of the cluster

Since genetic characterization was performed retrospectively, environmental samples and hand cultures of healthcare workers were not collected. In the absence of whole-genome sequencing, fine-scale transmission events could not be resolved. Therefore, we reviewed the bed positions of these patients and their transfer records. Possible transmission routes during the peak of the cluster are illustrated in Fig. 3.

Fig. 3.

Hypothesized transmission links of multidrug-resistant Klebsiella pneumoniae in this study. Bed map overview from April 13 to May 31 shows the patient locations and STs of K. pneumoniae isolates. The arrows indicate patient transfer. Possible transmission links involving ST792 and ST7120 are inferred from spatiotemporal overlap, patient movement, and repeated detection in adjacent beds.

On April 13, P5 was transferred from Section A to Section B, which could have been the initial point of cross-section transmission. By April 17, new detections were observed in Section A, with patient P7 acquiring ST792 and P6 acquiring ST7120, indicating possible environmental contamination from P5 or an asymptomatic carrier within Section A.

On April 24, Section A continued to have patients P3, P6, and P7 carrying ST7120 and ST792. In Section B, P5 continued to carry ST792, and a new patient, P9, was admitted. By April 30, patient P6 in Section A was infected with both ST7120 and ST792, suggesting possible cross-contamination or coinfection within Section A. Additionally, P8, who resided in a separate room, acquired ST7120.

On May 24, patient P3 in Section A continued to carry ST7120, while Section B saw new admissions, including P10, and the persistence of ST792 in P5. By May 31, Section B had patients P5, P9, and P10 all carried ST792, indicating the stable persistence of this strain in Section B.

3.2. Clinical characteristics and outcomes of invasive infections

Three patients developed invasive infections during the outbreak (Table 2). Two cases were caused by CRKP ST7120, including one fatal bloodstream infection and one case of neonatal sepsis with recovery. The remaining case, caused by ESBL-producing ST792, involved recurrent ventriculitis and resulted in severe hydrocephalus despite appropriate antibiotic therapy.

Table 2.

Clinical and microbiological characteristics of invasive K. pneumoniae strains in patients.

MLST	Patient	Disease spectrum	Site	Outcome	Resistance profile	Treatment
ST792	P5	Recurrent ventriculitis	CSF	Severe hydrocephalus	GEN, CAZ, CIP, PIP-TAZ	Meropenem
ST7120	P3	Intestinal failure following necrotizing enterocolitis with recurrent bacteremia	Blood	Expired	GEN, CAZ, CIP, PIP-TAZ, ETP	Meropenem, ceftazidime/avibactam
ST7120	P6	Early onset neonatal sepsis	Blood	Alive	GEN, CAZ, CIP, PIP-TAZ, ETP	Meropenem

Abbreviations: CAZ, ceftazidime; CSF, cerebrospinal fluid; CIP, ciprofloxacin; ETP, ertapenem, GEN, gentamicin, PIP-TAZ, piperacillin–tazobactam.

4. Discussion

In this study, we characterized a one-year cluster of multidrug-resistant K. pneumoniae in a pediatric ICU that involved six genetically distinct STs, suggesting the complexity of transmission dynamics in this setting. The most important finding of this study was the emergence of a previously undescribed ST, ST7120, and the first report of ST792 in Taiwan. These two STs predominated during the cluster in the ICU. ST7120 represents a newly identified ST not previously reported in the literature, whereas ST792, although rare, has been described in southern China as carrying KPC-2 [22,23]. However, the resistance gene profile of our ST792 strains differed from that in those reports, suggesting a distinct origin. The emergence of these two multidrug-resistant strains is a significant public health concern, as both strains were repeatedly detected over time and were observed in patients who developed invasive infections in this high-risk PICU population. Given the small number of invasive cases, these observations should not be interpreted as evidence of increased virulence of any specific ST. Because multiple STs co-circulated and several STs were represented by only one or two patients, the observed pattern may also be compatible with endemic circulation with sporadic introductions rather than a single continuous outbreak.

With respect to the newly identified ST7120 strains, MLST analysis demonstrated that ST7120 differs from ST43 at only a single locus, indicating a close evolutionary relationship. ST43 is distinct from dominant adult high-risk clones such as ST11 and ST45 and has been reported mainly in adult respiratory infections, often with multidrug resistance but uncommon hypervirulence [[24], [25], [26], [27]]. Data on ST43 in pediatric populations remain limited. However, endemic dissemination of ST43 in a neonatal ICU has been reported [26]. Together with our identification of the closely related ST7120 in a pediatric ICU, these findings suggest that pediatric ICUs may serve as settings for the emergence and local expansion of non-dominant or previously unrecognized clones. Therefore, continuous molecular surveillance tailored to pediatric settings is essential for the early detection of emerging strains and timely infection control interventions.

The cluster period lasted nearly one year, but resistance phenotype-based classification suggested two distinct peaks. MLST and PFGE confirmed that each peak was driven by different strains, highlighting the limitation of relying solely on phenotypic data for outbreak tracking. Contact tracing remains a critical infection control tool, enabling targeted isolation, patient cohorting, and integration into multifaceted control bundles [28,29]. However, our findings reveal that without molecular typing, coinfection or sequential infection with multiple STs, as observed in patient P6, the coinfection or sequential infection with multiple STs may go undetected, leading to inappropriate cohorting and potential cross-transmission.

This study had several limitations. It was conducted in a single pediatric ICU and involved a small number of patients and isolates, which may limit the generalizability of the findings. The retrospective design also prevented timely environmental sampling and healthcare worker screening, raising the possibility that important reservoirs or transmission vectors were missing. Therefore, the proposed transmission links should be regarded as hypothesis-generating rather than confirmatory. In addition, although ST7120 was identified as a newly assigned sequence type, our molecular characterization was limited to PFGE, MLST, and targeted resistance and outer-membrane protein-encoding gene analyses. In the absence of whole-genome sequencing, we were unable to determine whether ST7120 carried specific virulence islands, detailed capsular and plasmid features, or whether it belonged to an internationally recognized high-risk clone. In addition, the small number of invasive infections limits any assessment of clinical virulence by sequence type. These aspects warrant further investigation in future studies.

5. Conclusion

We investigated a one-year cluster of multidrug-resistant K. pneumoniae in a pediatric ICU. Molecular typing revealed complex transmission routes that were not detectable through phenotypic methods alone, highlighting its value for strengthening outbreak detection and containment in high-risk settings. In addition, we identified a novel ST, ST7120, and the first report of ST792 in Taiwan as two emerging STs. Their detection emphasizes the importance of molecular surveillance in recognizing new and locally unreported clones with epidemic potential.

CRediT authorship contribution statement

Yu-Chin Chen: Writing – original draft, Validation, Investigation. Cheng-Yen Kao: Resources, Project administration, Methodology. Chi-Chung Chen: Project administration, Methodology, Data curation. Hung-Jen Tang: Validation, Resources, Funding acquisition. Tu-Hsuan Chang: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Figure S1.

Outer membrane protein profiles of selected Klebsiella pneumoniae isolates. Outer membrane proteins were extracted and analyzed by SDS‒PAGE. K. pneumoniae ATCC 13883 was used as a reference strain. Arrowheads indicate the position of OmpK36, and arrows indicate the position of OmpK35. Isolates P6b-1, P3-2, P3-3, and P6a-3 lacked a detectable OmpK36 band, whereas OmpK35 expression was preserved. All the isolates lacking OmpK36 exhibited a carbapenem-resistant phenotype. Molecular weight markers (kDa) are shown on the left.

Figure S2.

eBURST diagram of Klebsiella pneumoniae sequence types (STs) based on MLST analysis. Highlighted STs (e.g., ST792, ST7120, and ST2150) represent the isolates identified in this study.