During the past 10 years, Chinese clinicians have witnessed a dramatic increase in the rates of carbapenem resistance among clinical isolates of Klebsiella pneumoniae. According to Mohnarin and CHINET, two leading antimicrobial resistance surveillance networks in China that cover all regions except Tibet, carbapenem resistance rates among K. pneumoniae escalated from 0.7% in 2000 to 2.9% in 2009 [1]. Moreover, in a university hospital in Shanghai, the incidence of carbapenem-resistant K. pneumoniae in 2009 reached 12.9% [2]. In China, carbapenem resistance in K. pneumoniae has mainly resulted from the rapid dissemination of Klebsiella pneumoniae carbapenemase (KPC) [2–4], and ST11 was demonstrated to be the dominant clone of KPC-producing K. pneumoniae (KPC-KP) [4].

KPC-KP isolates were usually multidrug resistant, but might remain susceptible to one or more aminoglycoside agents as well as to colistin and/or tigecycline [4–6]. Nevertheless, 16S rRNA methylases (ArmA and RmtB) were reported in a few KPC-producing Enterobacteriaceae recently, conferring high level resistance to almost all clinically important aminoglycosides [7–9]. Co-production of 16S rRNA methylases in KPC-producing pathogens and their potential spread could leave few choices for antimicrobial treatment.

Since its first isolation in 2007 [10], KPC-KP has been increasingly emerging at our hospital. Most of these isolates were still susceptible to one or more aminoglycoside antibiotics until 2009; nevertheless, high-level aminoglycoside resistance has emerged and gradually become prevalent in the last two years (data not shown). In May 2010, an Enterobacter amnigenus as well as a K. pneumoniae isolate, both co-producing bla _KPC and rmtB on a single plasmid, were simultaneously identified from a patient in our neurosurgery ward. And a small outbreak of KPC-RmtB-producing K. pneumoniae was revealed in the neurosurgery department [7]. This promoted us to perform a hospital-wide screening of rmtB gene in KPC-KP strains. Our study focused on the comparisons of microbiological, molecular and clinical characteristics between rmtB-positive KPC-KP (RPKP) and rmtB-negative KPC-KP (RNKP) isolates.

The epidemic curve revealed three phases (Figure 1): period 1 (January to April), during which sporadic cases of KPC-KP were identified and RPKP initiated to emerge; period 2 (May to July), during which a hospital-wide outbreak of KPC-KP was observed, with dramatic increase in both RNKP and RPKP isolates, and RPKP became predominant gradually; period 3 (August to December), during which high-level epidemic of KPC-KP persisted, and RPKP overtook RNKP to become the dominant KPC-KP.

Molecular typing results and resistance determinant profiles are shown in Table 3. PFGE analysis revealed 10 different pulsetypes (PTs), designated as A-J (Figure 2). The most widely disseminated PT was A, consisting of 63 clinical isolates (75%). By MLST, 10 distinct sequence types (STs) were identified, including 3 novel STs (ST689, ST690, and ST691). ST11 was the predominant ST, with 68 isolates (80.9%). Most of MLST data were consistent with results generated by PFGE. ST11-PTA was the dominant molecular type in both RPKP and RNKP. However, a greater diversity of STs and more complicated relatedness between the results of PFGE and MLST were observed in RNKP, indicating more diverse genetic backgrounds in RNKP.

Groups	MLST	PFGE	Isolate	Resistance determinants
RPKP (n = 48)	11	A	39	bla CTX-M-14 -bla KPC-rmtB-bla SHV-bla TEM
	11	B1	4	bla CTX-M-14 -bla KPC-rmtB-bla SHV-bla TEM
	11	E	1	bla CTX-M-14 -bla KPC-rmtB-bla SHV-bla TEM
	11	I	1	bla CTX-M-14 -bla KPC-rmtB-bla SHV-bla TEM
	542	D	1	bla CTX-M-3 -bla KPC-rmtB-bla TEM
	655	C	2	bla KPC-rmtB-bla SHV-bla TEM (one strain encoded qnrS)
RNKP (n = 36)	11	A	22	bla CTX-M-14 -bla KPC-bla SHV-bla TEM (two strains encoded qnrS)
	11	H1	1	bla CTX-M-14 -bla KPC-bla SHV-bla TEM
	214	A	1	bla CTX-M-14 -bla KPC-bla SHV
	412	A	1	bla CTX-M-14 -bla KPC-bla SHV-bla TEM-bla OXA-1-bla OXA-10-aac(6)-Ib
	412	J	1	bla CTX-M-14 -bla KPC-bla SHV-bla OXA-1-bla OXA-10-aac(6)-Ib
	412	H2	1	bla CTX-M-14 -bla KPC-bla SHV-bla OXA-1-bla OXA-10-aac(6)-Ib
	494	F1	4	bla CTX-M-14 -bla KPC-bla SHV-bla TEM
	494	B2	1	bla KPC-bla SHV-bla TEM
	689	E	1	bla CTX-M-3 -bla KPC-bla SHV-bla TEM -bla OXA-1
	690	F1	1	bla KPC-bla SHV-bla TEM
	691	F2	1	bla KPC-bla SHV-bla TEM
	161	G	1	bla KPC-bla SHV-bla TEM-bla OXA-1-bla OXA-10-qnrS- aac(6)-Ib

Footnotes. RPKP, rmtB-positive KPC-producing K. pneumoniae; RNKP, rmtB-negative KPC-producing K. pneumoniae; MLST, multilocus sequence typing; PFGE, pulse field gel electrophoresis; STs, sequence types.

Apart from bla _KPC-2 in all isolates and rmtB in RPKP, bla _CTX-M-14, bla _SHV-12 and bla _TEM-1 were revealed in most isolates. In addition, four RNKP belonging to ST412 or ST692 also encoded bla _OXA-1–type and bla _OXA-10–type β-lactamases and aac(6’)-Ib-cr, and a single ST689 RNKP encoded bla _OXA-1. Only one RPKP and three RNKP strains carried qnrS. No other 16S rRNA methylase-encoding genes were identified.

In total, 210 environmental samples were obtained from the general ICU department. Among the 210 samples, only one was RPKP positive, which was identified from a folder of patient chart from an empty bed with a RPKP patient discharged. The environmental RPKP was assigned to ST11-PTA clone, and carried bla _CTX-M-14, bla _SHV-12 and bla _TEM-1 in addition to bla _KPC and rmtB genes. It also showed similar antibiogram as its clinical counterparts (data not shown).

Nowadays, KPC-KP isolates have become major hospital pathogens, and their worldwide spread makes them a great threat to currently available antibiotic-based treatments. Tigecycline, colistin, and aminoglycosides if active, are agents recommended for severe infections caused by carbapenemase-producing Enterobacteriaceae[17]. However, despite of good in vitro activity of colistin and tigecycline against most of these strains, clinical data on their in vivo efficacy are quite limited [17]. Emergence of tigecycline resistance and colistin resistance during therapy with these agents is another issue to be considered [18–20]. A colistin-resistant RNKP isolate was also identified from a patient in our study, even without history of colistin therapy. Fortunately, a high proportion of KPC-KP isolates still showed in vitro susceptibility to aminoglycosides [6, 21, 22], including most of recently emerging colistin-resistant or tigecycline-resistant KPC-KP strains [18–20]. The colistin-resistant isolate found in our study was susceptible to aminoglycosides as well. In addition, a more recent study on KPC-KP bacteriuria demonstrated that treatments including aminoglycosides achieved a significantly higher rate of microbiologic clearance than did treatments with polymyxin B or tigecycline [23]. These findings together indicated that aminoglycosides may serve as effective antimicrobials in treating KPC-KP infections. Emergence of aminoglycoside-resistance in KPC-KP isolates therefore threatens the viability of this therapeutic option and further limits choices of available active agents.

In the present study, we documented the emergence and rapid spread of rmtB-positive KPC-KP (RPKP) in a Chinese university hospital. Due to the presence of rmtB gene, RPKP isolates showed high level resistance to almost all clinical available aminoglycosides. In addition, RPKP were usually resistant to many more clinically useful antimicrobials compared with RNKP, including minocycline and fosfomycin, which could serve as salvage remedies alone or in combination for infections due to carbapenem-resistant K. pneumoniae[21, 24]. The fact that RPKP only exhibited sufficient susceptibility to colistin and tigecycline has led to the dead-end of antimicrobial therapy in our hospital, since both colistin and tigecycline were unavailable in China during the study period.

Our results are concordant with the observation that ST11 was the dominant clone of KPC-KP in Mainland China [4]. ST11 is a single-locus variant of the international hyper-epidemic lineage ST258, and has also been identified in Singapore and more recently in Taiwan [25, 26]. It is increasingly recognized that ST258 is the predominant clone of KPC-KP across the whole world and has caused outbreaks in many countries [5, 6, 22]. Furthermore, ST258 has extended to the recently reported colistin-resistant KPC-KP [18, 27]. In the present study, ST11 KPC-KP also seemed to show dissemination advantage over other clones and be good colonizers to capture and accumulate resistance determinants. Of particular concern is that increasing prevalence of RPKP coincides with decreasing prevalence of RNKP in our hospital, which might suggest a bigger dissemination advantage of ST11 RPKP over ST11 RNKP, even in the absence of selective pressure. Another particular concern is the fact that 3 RPKP isolates (all belonged to ST11-PTA) were imported from other local hospitals, suggesting the presence of inter-hospital dissemination of this extremely multi-drug resistant pathogen. This indicates a dangerous possibility that ST11 RPKP clone might spread widely outside our hospital in the future.

The fact that most RPKP isolates belonged to ST11-PTA and were acquired in our hospital indicates the possibility that monoclonal cross transmission is the main mode of spreading. However, the failure to find a common environmental reservoir indicated patient-to-patient transmission as the main mechanism of RPKP spread in our study. Frequent bed transfers of case patients, particularly after isolation of RPKP, combined with lack of adequate preventive measures might have facilitated this process. It is also worth mentioning that the only one environmental RPKP isolate was identified from a decontaminated empty bed, which obliged us to enhance the effectiveness of decontamination procedures in our hospital.

Infections with pathogens resistant to more antibiotics have been reported to be associated with poorer functional status and worse outcomes [28, 29]. However, in our study, despite of diverse clinical characteristics, no significant differences were observed between the two cohorts. The similar clinical characteristics between RPKP and RNKP patients were likely multifactorial. Firstly, patients in both cohorts were at poor functional status, with severe underlying diseases and highly vulnerable to the colonization or infection of multidrug resistant pathogens. Secondly, RPKP and RNKP were both extremely drug-resistant pathogens. They showed sufficient susceptibility only to colistin and tigecycline. However, neither of them was readily available in China during our study period. Finally, although most RNKP (88.9%) were susceptible to at least one aminoglycoside in vitro, treatment containing aminoglycoside agents was very limited in our hospital, due to limited data on their in-vivo efficacy on KPC-KP infections and their nephrotoxicity [30]. Only three patients with RNKP were treated with antimicrobial regimens containing aminoglycosides in our study, but all of them had good clinical outcomes (data not shown).

In summary, we documented the emergence and wide dissemination of rmtB-positive KPC-KP in a Chinese university hospital. These isolates showed much severer resistance phenotypes compared with their rmtB-negative counterparts and seemed to disseminate more easily, representing a major threat for hospitalized patients. Moreover, patients suffered from RPKP infections were usually severely ill and ended up with rather high attributable mortality. These results, together with the fact that discovery of new antibiotics is reduced in recent years, suggest that prudent and conservative use of available active agents combined with good control practices are mandatory.