Discussion

This report describes the clonal spread of NDM-producing E. coli, originating in a small cluster of 3 cases in a single healthcare institution and ultimately involving 9 facilities in different regions. First identified in Israel in 2010, NDM-producing Enterobacteriaceae remained rare in Israel for the first several years after their arrival. Until 2014, prior to the outbreak described here, only 11 cases of NDM-producing E. coli had been reported in Israel via the national surveillance system.

The outbreak described here highlights a number of characteristics of the epidemiology of CRE. First, it demonstrates that despite highly robust systems for the identification and isolation of patients harbouring CRE, new strains can still gain a foothold and spread within and between healthcare facilities. Second, this outbreak involved the spread of a single clone of NDM-producing E. coli through 9 healthcare facilities, leading to large-scale patient-to-patient transmission in 2 facilities, which was stopped only after the implementation of intensive infection control interventions. This outbreak illustrates the importance of interinstitutional patient transfer in the spread of multidrug-resistant (MDR) bacteria and other healthcare-associated pathogens of epidemiological significance. ^{Reference Won, Munoz-Price and Lolans9} In addition, our ability to identify the multi-institutional features of the outbreak and intervene in a timely manner highlights the necessity of MDR organism surveillance and containment efforts conducted at the regional and national levels. ^{Reference Schwaber, Lev and Israeli2,Reference Won, Munoz-Price and Lolans9} Third, the involvement of different types of healthcare facilities, and the pivotal role of one post-acute care hospital in propagating the outbreak strain, stress the need for infection control interventions across the continuum of healthcare. ^{Reference Ben-David, Masarwa and Fallach10}

During the outbreak period, we noted the occurrence of a new genotype of carbapenem-resistant E. coli containing both bla _NDM and bla _KPC in 2 of the affected facilities. Molecular analysis revealed that isolates with this genotype belonged to the same clone responsible for the first phase of the outbreak, having now acquired a second plasmid, containing bla _KPC. Whole-genome sequencing suggested 2 distinct acquisition events of bla _KPC in the same facility. This finding adds to the evidence that acquisition by carbapenemase-producing enterobacteria of additional carbapenemase genes may not be a rare phenomenon. ^{Reference Protonotariou, Poulou and Politi11} Indeed, subsequent to our study period, isolates of E. coli ST410 jointly harboring bla _NDM-5 and bla _OXA-181 were reported. ^{Reference Roer, Overballe-Petersen and Hansen12} In the current outbreak, it appears that what has since emerged as a globally distributed MDR bacterial sequence type, in this case a variant derived from the B3/H24Rx clade, took on a new resistance feature in the form of a plasmid harboring bla _KPC. ^{Reference Roer, Overballe-Petersen and Hansen12}

The formation and subsequent spread of a 2-carbapenemase-producing isolate, have potentially ominous implications, compounding the challenges of infection control and clinical treatment. The epidemiological ramifications of dual-carbapenemase–producing enterobacteria are significant; each transmission event results in the transfer of 2 distinct carbapenem resistance genes, increasing the reservoir of resistance. Moreover, horizontal gene transfer may not be limited to carbapenemase genes but may also entail concomitant transfer of elements coding resistance to additional antimicrobial classes, as our findings demonstrate. The clinical implications of this finding are less clear, as even single-carbapenemase-producing Enterobacteriaceae, such as our original isolate, can have high carbapenem MICs. Still, there is reason for concern on this account as well; new inhibitor combination antimicrobial agents may target a specific class of carbapenemase and may be rendered ineffective by pathogens acquiring a different class of carbapenemase. Scenarios such as these add merit to the Israeli guideline to separate patients in CPE isolation cohorts by carbapenemase type, thereby reducing the risk of horizontal gene transfer. ^{Reference Solter, Adler and Rubinovitch6}

Our report has a number of limitations. First, we did not uncover the origin of the outbreak strain; the NDM-producing E. coli isolates in Israel prior to this outbreak were unavailable for comparison. Second, we were unable to find an epidemiological link to explain the isolation of a dual-carbapenemase–producing E. coli at facility B in 2015, suggesting the existence of 1 or more carriers who were either not screened for CRE or had a CRE carriage load below the level of detection. However, we were able to construct clear epidemiological pathways for all other cases involved in the outbreak. Third, although we were able to glean some insights into the carbapenemase gene-carrying plasmids harbored by the isolates, the limitations of short-read sequencing did not allow us to confirm the exact composition and transmission dynamics of mobile genetic elements in each isolate. These details would require long-read sequencing technologies that are increasingly available.

In conclusion, we have described a multi-institutional CRE outbreak involving a previously rare pathogen in Israel, which evolved genetically more than once during the outbreak period. Our findings illustrate multiple facets of CRE spread, the utility of WGS in supporting outbreak investigation, and the importance of ongoing vigilance in detecting outbreaks and early intervention to contain spread.