Hospital-acquired infections are a major global public health issue. Multidrug-resistant (MDR) bacteria have provoked great concern due to the high risk of in-hospital death. To control nosocomial transmission of MDR bacteria, as well as antibiotic resistance, it is critical to specify the co-existence of antibiotic resistance genes (ARGs) and elucidate their mobility.

Metagenomic next-generation sequencing (mNGS) for patient specimens is a promising strategy for diagnosis of infection [1,2]. Quick and accurate identification of pathogens using mNGS prompts timely and appropriate antibiotic therapy. A disadvantage of mNGS is the false-positive rate of predictions [3]. The combination of whole-genome sequencing (WGS) and colony-independent mNGS would provide better medical specificity than optimisation of the workflow.

Klebsiella pneumoniae is notorious because of the antibiotic-resistance and worldwide nosocomial infections associated with this pathogen [4]. In addition, carbapenem-resistant K. pneumoniae (CR-Kp) is a concern in relation to hospital-acquired infections [5,6]. CR-Kp is one of the major nosocomial MDR opportunistic pathogens associated with high mortality in the intensive care unit (ICU) [7]. Early and accurate identification of CR-Kp isolates and their ARGs is critical for patient care and nosocomial infection control. Moreover, complex infection caused by KPC- and NDM-producing strains is a major therapeutic challenge [8,9].

Culture-based isolation and WGS enable rapid and effective identification of pathogens in the context of presumptive diagnosis by mNGS, as well as a better understanding of the occurrence and pattern of carbapenem resistance and high virulence in emerging CR-hvKp isolates [6,10]. Antibiotic resistance and virulence-encoding factors in K. pneumoniae are currently identified using machine learning of featural Raman [11] and Fourier transform infrared (FTIR) [12] spectroscopy. In addition to these innovative deep-learning approaches, culture-independent metagenomics surveillance is a rapid and concise method to identify underlying resistance and virulence determinants in hospital-derived specimens [13].

Here, an elderly male patient with simultaneous infection of KPC-2- and NDM-1-producing CR-Kp was identified. The emergence and dynamics of nosocomial CR-Kp infection were observed using mNGS during the patient's hospitalisation. CR-Kp producing serine β-lactamase (SBL, blaKPC-2) and metal β-lactamase (MBL, blaNDM-1) was isolated. CR-Kp producing KPC-2 (Ambler class A carbapenemase) was found first, and the abundance of NDM-1 CR-Kp significantly increased alongside antibiotic treatment. The genetic contexts of SBL- and MBL-producing CR-Kp were identified retrospectively, benefiting timely diagnosis. Interestingly, the genomic signature of high virulence was determined in one KPC-2-producing isolate. These findings indicate that these CR-Kp may be derived from different routes of nosocomial transmission, and reveal the ongoing spread of high virulence determinants and antimicrobial resistance.