Nationwide epidemiology of carbapenem resistant Klebsiella pneumoniae isolates from Greek hospitals, with regards to plazomicin and aminoglycoside resistance

Background To evaluate the in vitro activities of plazomicin and comparator aminoglycosides and elucidate the underlying aminoglycoside resistance mechanisms among carbapenemase-producing K. pneumoniae isolates collected during a nationwide surveillance study in Greek hospitals. Methods Three hundred single-patient carbapenemase-producing K. pneumoniae isolates were studied, including 200 KPC-, 50 NDM-, 21 VIM-, 14 KPC & VIM-, 12 OXA-48-, two NDM & OXA- and one KPC & OXA-producing isolates. Susceptibility testing was performed by broth microdilution, and minimum inhibitory concentrations (MICs) interpreted per EUCAST breakpoints. Carbapenemase-, aminoglycoside modifying enzyme- and 16S rRNA methylase- encoding genes were detected by PCR. Results Of 300 isolates tested, 5.7% were pandrug resistant and 29.3% extensively drug resistant. Plazomicin inhibited 87.0% of the isolates at ≤2 mg/L, with MIC50/MIC90 of 0.5/4 mg/L. Apramycin (a veterinary aminoglycoside) inhibited 86.7% of the isolates at ≤8 mg/L and was the second most active drug after plazomicin, followed by gentamicin (S, 43%; MIC50/MIC90, 4/> 256) and amikacin (S, 18.0%; MIC50/MIC90, 32/128). Twenty-three (7.7%) isolates (16 KPC-, 6 VIM- and one KPC & OXA-48-producers) exhibited MICs ≥64 mg/L for plazomicin, and harbored rmtB (n = 22) or armA (n = 1). AAC(6′)-Іb was the most common aminoglycoside modifying enzyme (84.7%), followed by AAC(3΄)-IIa (25.3%), while those two enzymes were co-produced by 21.4% of the isolates. Conclusions Plazomicin retains activity against most carbapenemase-producing K. pneumoniae isolated from Greek hospitals, with MICs consistently lower than those of the other aminoglycosides, even in the presence of aminoglycoside modifying enzymes. Dissemination of 16S- rRNA methylases in 8% of the isolates is an unwelcome event that needs strict infection control measures and rigorous stewardship interventions. Electronic supplementary material The online version of this article (10.1186/s12879-019-3801-1) contains supplementary material, which is available to authorized users.


Background
Hospital infections caused by carbapenem-resistant Klebsiella pneumoniae constitute a worldwide problem associated with high morbidity, mortality, and prolongation of hospitalization and associated costs [1]. The spread of carbapenemases in K. pneumoniae has created therapeutic dilemmas for clinicians as those isolates often demonstrate resistance to many other classes of antibiotics, thus limiting our therapeutic options. Furthermore, few new antibiotics are in line to replace carbapenems [2].
Aminoglycosides are broad-spectrum antibiotics which have been used for the treatment of life-threatening infections. Many mechanisms of acquired resistance to aminoglycosides have emerged, with the aminoglycoside-modifying enzymes (AMEs) being the most prevalent. These enzymes include N-acetyltransferases, O-nucleotidyltransferases and O-phosphotransferases, which inactivate aminoglycosides by covalently modifying specific amino or hydroxyl moieties on the drugs [6]. Another less common mechanism of resistance is the up-regulation of efflux pumps and reduction in membrane permeability developed by bacteria to affect the transport of hydrophilic aminoglycosides across cell membranes. Additionally, 16S rRNA methyltransferases (RMTs), which occur at a low incidence in clinical isolates, modify bacterial 16S rRNA, the molecular target of aminoglycosides and confer high-level resistance to all widely used aminoglycosides [7].
Plazomicin is a next-generation aminoglycoside that was developed to overcome common aminoglycoside-resistance mechanisms for the treatment of patients with serious infections caused by multidrug-resistant Enterobacteriaceae, including ESBL producing and carbapenem-resistant Enterobacteriaceae [8]. Plazomicin is a semi-synthetic derivative of sisomicin, not affected by any known aminoglycoside-modifying enzymes (AMEs), except N-acetyltransferases (AACs) AAC(2′)-Ia, −Ib and -Ic (found only in Providencia spp.) [8]. Like sisomicin, it lacks the 3′-and 4′-OH groups, thus is protected from the O-phosphotransferase (APH) APH (3′) and O adenyltransferase (ANT) ANT (4′) enzymes that generate resistance to amikacin. The hydroxy-aminobutyric acid substitute introduced at the N1 position of sisomicin provides protection from the AAC(3), ANT(2″) and APH(2″) AMEs, while the hydroxyethyl substitute at the 6′ position blocks the multitude of AAC(6′) AMEs, without reducing potency, as occurred in previous efforts to shield this position [8]. Plazomicin (ZEMDRI™) was approved in June 2018 by the U.S. Food and Drug Administration, for adults with complicated urinary tract infections (cUTI), including pyelonephritis, caused by certain Enterobacteriaceae in patients who have limited or no alternative treatment options. ZEMDRI is an intravenous infusion, administered once daily.
In this study, we evaluated the in vitro activities of plazomicin and comparator aminoglycosides (amikacin, gentamicin and tobramycin) and elucidated the underlying aminoglycoside resistance mechanisms among 300 carbapenemase-producing K. pneumoniae isolates collected during a nationwide surveillance study in Greek hospitals.

Clinical isolates
A total of 300 single-patient carbapenem-non-susceptible Κ. pneumoniae isolates, collected between November 2014 and April 2016, from 14 Greek hospitals in six cities, were included. The isolates were collected prior to this study, during a prospective Greek multicenter study, organized by the Hellenic Society of Chemotherapy, and were archived and anonymised.

Typing
Genetic relatedness among carbapenemase producing K. pneumoniae isolates was evaluated by pulsed-field gel electrophoresis (PFGE) analysis of chromosomal restriction fragments obtained following cleavage with SpeI (New England BioLabs Inc., GmbH Frankfurt am Main, Germany). A dendrogram was generated from the homology matrix with a coefficient of 1.5% using the unweighted pair-group method using arithmetic averages ("UPGMA") to describe the relationships among PFGE profiles. Isolates were considered to belong to the same PFGE group if their Dice similarity index was ≥80%.
Only twenty-four Κ. pneumoniae isolates (8.0%) were aminoglycoside-susceptible, whereas the remaining 276 isolates (92.0%) were resistant to at least one of the indicated aminoglycosides tested, i.e. amikacin, gentamicin, tobramycin or netilmicin, using the clinical and epidemiological breakpoints defined by EUCAST. Gentamicin was the most active in vitro aminoglycoside in clinical use, with 43.0% being susceptible, followed by amikacin (S, 18.0%). The highest resistance rates were observed for tobramycin, with 89 and 83.3% resistant according to EUCAST and CLSI breakpoints, and netilmicin, with 87.3 and 84.3% respectively. Finally, 154 (51.3%) and 80 (26.7%) isolates were non-susceptible and resistant, respectively, to the four clinically available aminoglycosides (amikacin, gentamicin, tobramycin and netilmicin) per EUCAST breakpoints. 79 (26.3%) and 43 (14.3%) were non-susceptible and resistant, respectively, when CLSI breakpoints were applied. The in vitro activity of tested aminoglycosides against the studied collection of 300 clinical isolates is presented in Table 1. Of note, 87.0% of the isolates were inhibited by plazomicin at ≤2 mg/L, which is the breakpoint approved by the FDA, and 91.3% at ≤4 mg/L. Plazomicin was the most active aminoglycoside tested with an MIC 90 value ≥32 times lower than that of all aminoglycosides in clinical use tested, > 64 times lower than neomycin and 4 times lower than apramycin (Table 1). Among isolates that were non-susceptible or resistant to the four aminoglycosides in clinical use (according to the EUCAST breakpoints), plazomicin exhibited an MIC 50 of 1 and 2 mg/L, respectively. The activities of the aminoglycosides against all isolates as well as isolates categorized according to the specific carbapenemase produced are summarized in Table 1. Further, the plazomicin MIC distribution and the cumulative percentage inhibited, in relation to the carbapenemase produced and in relation to amikacin and/or gentamicin susceptibility are presented in Tables 2 and 3, respectively. Twenty-three strains (7.7%), isolated in seven of the 14 hospitals, were highly resistant to all indicated aminoglycosides (MICs ≥256 mg/L), had highly elevated plazomicin MICs (≥64 mg/L) and harbored a RMT gene (Tables 4 and 5). Fifteen KPC-, 6 VIM-and one KPC & OXA-48 -producing K. pneumoniae isolates harbored rmtB, and one KPC-producing K. pneumoniae isolate harbored armA. It is of note that none of the NDMproducing K. pneumoniae isolates produced a RMT although bla NDM -carrying plasmids frequently are associated with armA, rmtB, rmtC, and rmtF [7].
For strains carrying no RMT gene (n = 277), the MIC 90 of plazomicin was 2 mg/L, with 94.2% of isolates being susceptible and the highest MIC observed at 8 mg/L. Among those 277 isolates, 148 (53.4%) and 225 (80.5%) were non-susceptible to gentamicin and amikacin, respectively. One hundred thirty-one (47.3%) isolates were non-susceptible to both gentamicin and amikacin and against 120 (91.6%) of these, the MIC of plazomicin was ≤2 mg/L (Table 3). Isolates with plazomicin MIC 8 mg/L (n = 3) were non-susceptible to all other aminoglycosides tested including apramycin.
Associations of AMEs, and AME combinations, with aminoglycoside MICs are shown in Table 5. Additionally all isolates non-susceptible to neomycin (n = 169) harboured the aph(3′)-I (n = 167) or/and the aph(3′)-VI  (MIC 1-2 mg/L), although those two genes confer resistance to amikacin and gentamicin, respectively. PFGE genotyping revealed a multiclonal population of KPC-producing K. pneumoniae, with a prevalent PFGE profile (42.0%), consisting of nine variants, all detected in more than one center. Additionally, there were two more PFGE profiles identified, consisting of three and two variants each, and another 24 PFGE profiles that included a few isolates each.
The 21 VIM-, the 14 KPC-and VIM-and the 12 OXA-48-like -producing K. pneumoniae isolates were multiclonal, with no particular clone prevailing as different clones between hospitals and within hospitals were observed.
However, PFGE genotyping of NDM-producing isolates demonstrated great genetic similarity in the 48 (96.0%) isolates (dominant clone A), consisting of three main variants (A1 to A3), two of which were detected in more than one center. Additionally, two isolates, each with a distinct PFGE profile, were also detected (4.0%). Moreover, two isolates carrying both bla NDM and bla OXA-48-like belonged to dominant clone A.
The 23 K. pneumoniae isolates that harboured an RMT (rmtB or armA) and had highly elevated plazomicin MICs, belonged to six clonal types, suggesting that these isolates were not clonal, although clonal dissemination of rmtB positive VIM or KPC producing isolates was observed in three hospitals.

Discussion
Among contemporary carbapenem-resistant K. pneumoniae isolates from Greece, where KPC-producing pathogens remain predominant, followed by NDM-producing isolates, plazomicin was more potent than that of the comparator aminoglycosides gentamicin and amikacin. These results are similar to those reported in previous studies with carbapenem-resistant enterobacteriaceae from diverse geographic regions [14][15][16][17][18][19].
There are several noteworthy findings in our study. We observed that in Greece, aminoglycoside resistance in carbapenemase-producing K. pneumoniae clinical isolates is predominantly caused by the production of AMEs (85.3% of isolates), while the occurrence of RMTs was observed in 7.7%. A remarkable AME diversity was observed. Overall, 23 different AME patterns (maximum of five genes/isolate) correlating with different levels of aminoglycoside resistance were identified.
The aac(6′)-Ib enzyme was the most common gene detected and found in ≥71.4% of carbapenemase producing strains, regardless of the carbapenemase present, whilst the aac(3′)-IIa gene was mainly associated with NDM-and OXA-48-producing isolates. The ant(3′)-I gene was always associated with VIM-, while aph(3΄)-Ia was mainly associated with KPC and VIM-producing K. pneumoniae.
Notably, the aminoglycoside resistance phenotype was not always a reliable predictor of the AME genotype. For instance, in 49.7% of the gentamicin non-susceptible isolates the aac(3′)-IIa, ant(2′)-Ia, aac(3′)-Ia or aac(3)-IV genes, were not detected, while aac(3′)-IIa was detected in 4% of the gentamicin susceptible isolates (MICs 1-2 mg/L). All amikacin non-susceptible isolates harbored the aac(6′)-Ib or an RMT gene, while 20 (6.7%) amikacin  susceptible isolates harbored also the aac(6′)-Ib gene. This is consistent with previous studies where, despite the presence of aac(6′)-Ib, low amikacin MICs (2-8 mg/L) have been reported in K. pneumoniae and E. coli strains [16,20,21]. The contribution of multiple concurrent resistance mechanisms and differentiations in catalytic activity of AME genes is probably the explanation for this. There are 45 non-identical AAC(6′)-Ib-related entries in the NCBI database, with 1 to 8 amino acid differences and a total of 24 positions showing amino acid variations. Among them, 32 have identical name but a non-identical amino acid sequence (97-99.5% similarity). Some of these variants have conserved specificity, while others have not, i.e. AAC(6′)-Ib 11 has an extended resistance spectrum that includes gentamicin or AAC(6′)-Ib' confers resistance to gentamicin but not to amikacin [22]. On the contrary, the presence of high resistance (MIC ≥256 mg/L) to both amikacin and gentamicin correctly predicted (95.5%) the presence of an RMT gene, which also displayed similar highly elevated plazomicin MICs, which is consistent with the limitations of plazomicin and the aminoglycoside class. There were 23 isolates that encoded both a carbapenemase and an RMT. Sixteen KPC-(8.0%), six VIM-(28.6%) and one KPC & OXA-48 -producing K. pneumoniae isolates harbored either rmtB or armA.
Acquired aminoglycoside resistance mediated by 16S-RMTases is a relatively new mechanism described in the early 2000s. Co-association of 16S-RMTases with carbapenemases leads to XDR and, in some instances, to PDR phenotypes [7].
In previous literature reports, plazomicin MICs were predominantly ≤4 mg/L, except for CRE isolates that produced the NDM-1 metallo-β-lactamase [15]. Interestingly, our findings showed that all 52 NDM-producing K. pneumoniae had plazomicin MICs ≤2 mg/L. This was similar to results found against NDM-producing Enterobacteriaceae from Brazil, which exhibited plazomicin MICs ≤4 mg/L [19]. In both countries, bla NDM gene has been reported to be located on an IncFII-type plasmid [23][24][25][26], while aminoglycoside susceptibility was variable, suggesting that the mechanism of resistance was      According the recently approved FDA breakpoint due to the presence of AMEs rather than 16S rRNA methyltransferase. Plazomicin MICs in RMT-negative isolates were consistently lower than those of the other aminoglycosides, and further, the activity of plazomicin was not affected by the number or type of AMEs produced or by the presence of any carbapenemase. As plazomicin was designed to evade modifications conferred by most AMEs [14], these findings are not surprising.
Another noteworthy finding in our study was the apramycin susceptibility. Apramycin is a structurally unique aminoglycoside, a veterinary agent that has not been approved for clinical use, which is likely due to its narrow therapeutic index [15]. It is not inactivated by most of the known AMEs [27], and it is active against producers of the most common N7-G1405 RMTs [15]. Apramycin inhibited 86.7% of the K. pneumoniae isolates at ≤8 mg/L and it was the second most active drug after plazomicin. This is in accordance with previous evidence that apramycin has broad-spectrum activity against carbapenem-susceptible and carbapenem-resistant Enterobacteriaceae strains from the US, the UK and China [15,28,29], suggesting that apramycin may be a candidate for modification to potentially generate new potent aminoglycosides.

Conclusions
In conclusion, plazomicin was active against most of the contemporary carbapenemase-producing K. pneumoniae isolates collected from 14 Greek hospitals, with 87.0% of the isolates inhibited by an MIC≤2 mg/L, while 94.2% of the isolates that did not carry a RMT gene were inhibited by an MIC≤2 mg/L. Plazomicin demonstrated the most potent in vitro inhibitory activity of all aminoglycosides (regardless of the AMEs produced) and of all other drugs typically used today to treat infections caused by such strains, suggesting that this agent may play an important role for the treatment of MDR K. pneumoniae infections. Dissemination of 16S-RMTases among already MDR organisms is an unwelcome event. Strict infection control measures have to be elaborated to prevent the spread of MDR organisms such as those described here that co-produced carbapenemases and RMTS. Fig. 1 Aminoglycoside-modifying enzymes (AMEs) and 16S rRNA methylases (RMTs) in carbapenemase-producing K. pneumoniae isolates. a Distribution of AMEs and RMTs. AMEs and RMTs that were present alone and those that were present in combination with other AMEs/RMTs are indicated by gray and black bars, respectively. b Number of AMEs/RMTs possessed by the isolates