Outbreak of Infections with VIM-Positive Proteus mirabilis in a Hospital in Zagreb

1. Introduction

Proteus mirabilis is an opportunistic pathogen belonging to the family Enterobacterales, widely distributed in the environment and as a normal microbiota in the bowel of human and animals [1]. Along with other Enteroacterales, it is considered to be an important causative agent of urinary tract infections (UTI), particularly catheter associated urinary tract infections (CAUTI) due to its ability to form biofilm and wound infections [2]. P. mirabilis is intrinsically resistant to tigecycline, polymixins, tetracyclines and nitrofurantoin [3], but does not have any intrinsic β-lactamase.

Resistance to β-lactams in P. mirabilis is mediated by β-lactamases, enzymes which cleave β-lactam antibiotics, modification of PBR receptors, upregulation of efflux pumps and porin alteration or loss. The β-lactamases encountered in Proteus species are broad spectrum penicillinases (TEM-1, TEM-2, SHV-1, CARB, OXA-1-4), extended-spectrum β-lactamases (ESBLs), plasmid-mediated AmpC β-lactamase (p-AmpC) and recently carbapenemases [3,4,5,6]. ESBLs hydrolyze penicillins, all cephalosporins and monobactams, whereas p-Amp inactivate expanded-spectrum cephalosporins (ceftazidime, cefoaxime, ceftriaxone) and cephamycins, but spare carbapenems and cefepime, Carbapenems are potent β-lactam antibiotics and the last resort treatment for infections due ESBL and AmpC producing organisms [3]. Carbapenemases encountered in P. mirablis belong to class B or metallo-β-lactamases (MBLs) (VIM, IMP, NDM), class D or carbapenem hydrolyzing oxacillinases or CHDL (OXA-48, OXA-181, OXA-23, OXA-58) designated as CHDL and rarely A (KPC) [3,7]. The resistance to aminoglycosides in P. mirabilis is most frequently caused by enzymes which modify aminoglycosides and render them inactive. Adenyltransferases are encoded by aadA1 and aad2, acetyltransferases by aac(6'')-Ib and aacA4 genes and phosphorylases by aph(6)-Id, aph(6)-Ib and aph(6)-Ia) genes [3]. Panaminoglycoside resistance associated with 16S rRNA methylase was also reported in P. mirabilis [3]. P. mirabilis develops resistance to fluoroqinolones by mutation of the genes encoding DNA gyrase (gyrA) and topoisomerase (parC) or acquisition of plasmid-mediated qnr genes encoding qnr proteins which protect topoisomerase IV. Sul1 and sul2 genes usually confer resistance to sulphonamides while dfr genes encode dihydropholate reductase antagonizing the effect of trimethoprim [3].

Bibliographical data on β-lactam resistance mechanisms in P. mirabilis are mostly focused on ESBLs and p-AmpC. ESBLs belonged most frequently to TEM family in early 90-ties, but a shift to CTX-M was reported at the beginning of 2000-ties [3]. CMY is almost the only p-AmpC family reported in P. mirabilis [3]. In Croatia, an outbreak of infections with TEM-52 β-lactamase producing P. mirabilis was reported in Split in 2008 [8,9]. Later, an outbreak of infections with CMY-16 was described in a nursing home in Zagreb [10]. The same allelic variant was much later identified in a hospital in Split [11]. Intensive use of carbapenems in the treatment of infections with ESBL and p-AmpC producing organisms lead to proliferation of carbapenemases in P. mirabilis. The majority of studies detected carbapenemases belonging to VIM and NDM family and OXA-48 [3]. However, there are no reports addressing carbapenemases in this species from Croatia.

Here we report extensively drug-resistant carbapenemase producing P. mirabilis associated with an outbreak in a psychiatric hospital in Zagreb. We aimed to characterize β-lactamases and other resistance mechanisms carried by these isolates and their molecular epidemiology.

2. Materials and Methods

2.1. Patients and Bacterial Isolates

The bacterial isolates with reduced susceptibility to at least one carbapenem (imipenem, meropenem, ertapenem) were collected from 2023 to 2024 from a psychiatric hospital in Zagreb. The isolates were identified to a species level by MALDI-TOF MS (matrix-assisted laser desorption ionization–time of flight mass spectrometry) (Vitek MS, bioMerieux, France). The initial antibiotic susceptibility was tested according to EUCAST guidelines [12] in the Dr. Andrija Štampar Teaching Institute of Public Health which provides microbiological service for Psychiatric Hospital “Sveti Ivan”. Isolates which demonstrated reduced susceptibility to at least one carbapenem were sent to the Clinical Department for Clinical Microbiology and Infection Control and Prevention of the University Hospital Centre Zagreb for further analysis.

2.2. Antimicrobial Susceptibility Testing

Broth dilution method was applied to determine the minimum-inhibitory concentrations (MICs) of 13 antibiotic: amoxycillin alone and with clavulanic acid, piperacillin-tazobactam, cefuroxime, ceftazidime, cefotaxime, ceftriaxone, cefepime, imipenem, meropenem, gentamicin, amikacin and ciprofloxacin. The results were interpreted according to CLSI guidelines [13]. The susceptibility to ceftazidime-avibactam, sulphametoxazole-trimethoprim and chloramphenicol was determined only by disk-diffusion test. The antibiotic containing disks were provided by Oxoid (Basingstoke, UK). The isolates were classified as multidrug-resistant (MDR), extensively-drug resistant (XDR) or pandrug-resistant (PDR) as described previously by Magiorakos et al [14]. MDR is recognized as being non-susceptible to at least one antimicrobial agent in three antimicrobial classes. XDR is recognized as being resistant to at least one antimicrobial agent in all antimicrobial classes and susceptible to only two or less. PDR isolates are resistant to all antibiotics available for the specific species. Escherichia coli ATCC 25922 and Klebsiella pneumoniae 700603 served as quality control strains for MIC determination.

2.3. Phenotypic Detection of β-Lactamases

ESBLs were detected by a double disk- synergy test [15] and combined disk test with cephalosporins and clavulanic acid [13]. Plasmid-mediated AmpC β-lactamases were detected among cefoxitin resistant isolates by combined disk test using cephalosporin disks combined with cloxacllin [6]. Initial screening for carbapenemases was carried out for the purpose of routine microbiology testing by immunochromatographic test [16]. A modified Hodge test (MHT) with imipenem disk was used to confirm the production of carbapenemases [17]. Since identification of carbapenemases in P.mirabilis poses an identification challenge due to low level of resistance in this species, additional tests were done. Carbapenem inactivation method (CIM) was done to assess hydrolysis of carbapenems with disks of imipenem and meropenem [18]. A suspension of the test strain was adjusted to McFarland 0.5 (10⁸ CFU/mL) and a meropenem disk was placed in the suspension. The suspension was incubated for 2 h at 37 °C. E. coli ATCC 25922 was inoculated on Mueller–Hinton agar (MH) and the disk was placed in the middle of the plate. The plates were incubated overnight and the lack of inhibition zone, decreased inhibition zone (<15 mm) or colonies within the inhibition zone indicated carbapenem hydrolysis. The isolates positive in CIM test were subjected to eCIM for detection of MBLs. The eCIM test was performed in the same way, but with one tube containing 0.5 mM EDTA to inhibit MBLs. The test indicated MBL positivity if there was ≥5 mm increase in zone diameter in eCIM experiment compared to the control sample without EDTA [18]. The strains from own collection, known to be positive for KPC, VIM, NDM and OXA-48, were used as positive and negative controls [18]. Since there was a high rate of negativity in eCIM among isolates positive in immunochromatographic method, they were additionally tested by combined disk tests with imipenem and meropenem alone and combined with PBA, 0.1 M EDTA, or both to detect for KPC, MBLs, or simultaneous production of KPC and MBL, respectively [17].

2.4. Molecular Detection of Resistance Genes

DNA was extracted by thermal lysis at 100 C for 10 minutes. Cellular debris was removed by centrifugation and the supernatant was used as DNA template. Antimicrobial resistance genes comprising β-lactam (bla_TEM, bla_SHV, bla_CTX-M) [19,20,21] and fluoroquinolones (qnrA, qnrB, qnrS) [22], were determined by singleplex PCR as described previously. Carbapenemases coding genes belonging to class A (bla_KPC), B or MBLs (bla_IMP, bla_VIM, bla_NDM) and class D (bla_OXA-48), [23], bla_OXA-23, bla_OXA-24/40, bla_OXA-51, bla_OXA-58) [24] blaampC [25] and the genes of five clonal groups of CTX-M β-lactamases were amplified via multiplex PCR according to the protocols reported previously [26]. The randomly selected amplicons (11 CTX-M and 8 VIM) obtained by singleplex PCR were sequenced with forward and reverse using Eurofin Genomic service Eurofins Genomics (https://eurofingenomics.eu) and the sequences were compared with those from the GenBank nucleotide database at http://www.ncbi.nlm. nih.gov/blast (accessed on 1st February 2025. The flanking regions of bla_CTX-M genes were investigated by PCR mapping, using forward primer for ISEcp combined with universal reverse primer for CTX-M encoding genes (MA3) [27].

2.5. CarbaResist Inter-Array Genotyping Kit

Genotyping of six randomly selected isolates was conducted using the microarray-based CarbaResist Genotyping Kit, according to the manufacturer’s instructions, version 1012012100004 (INTER-ARRAY, fzmb GmbH, Bad Langensalza, Germany). In short, genomic DNA was isolated from monoclonal overnight cultures using the Qiagen DNeasy Blood and Tissue Kit, according to the manual. The unfragmented DNA was linear amplified using one primer for each target sequence (antisense) and internally labeled with biotin dUTP. The obtained ssDNA (single-stranded) products were transferred into the ArrayWells for hybridization. These wells contain 230 probes corresponding to distinct genes for the most relevant carbapenemases, ESBL and AmpC, as well as genes associated with β-lactam-, aminoglycoside-, fluoroquinolone-, sulphonamide-, trimethoprim- and colistin-resistance. After washing steps to remove any unbound DNA, horseradish peroxidase (HRP)-conjugated streptavidin was bounded to all hybridized sections, resulting in dark spots on the chip due to an enzymatic reaction. The detection of these spots and data acquisition was performed automatically using the INTER-VISION Reader.

2.6. Whole Genome Sequencing (WGS)

Four representative strains were subjected to WGS. First, the strains were cultivated in Tryptic Soy Broth (Merck Millipore, MA, USA) at 37 °C overnight. Then, the genomic DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Subsequently, the extracts were barcoded via the Rapid Barcoding Kit 96 V14 and sequenced using the minION (Oxford Nanopore Technologies). For the assembly of the single reads Flye (https://www.pnas.org/doi/full/10.1073/pnas.1604560113, accessed on 20 June 2025) was utilized. Subsequently, the data was analyzed using the webservers and services of the Center for Genomic Epidemiology (http://www.genomicepidemiology.org, accessed on 20 June 2025) [48]. The phylogenetic tree was generated with REALPHY online tool (https://doi.org/10.1093/molbev/msu088, accessed on 20 June, 2025) and visualized using phylo.io (https://doi.org/10.1093/molbev/msw080, accessed on 20 June 2025).

2.7. Characterization of Plasmids

PCR-based replicon typing (PBRT) was according to Carattoli et al. [29]. Since it was observed previously that PBRT can be inefficient in identifying L/M plasmid incompatibility type, an updated method designated to identify and distinguish between IncL and IncM plasmids was applied [30]. Positive control strains were kindly provided by Dr. Alessandra Carattoli, Instituto Superiore di Sanita, Rome, Italy.

2.8. Detection of Virulence Determinants

Urease activity was determined by inoculating the strains in the urea-containing medium. The change of the color to pink was recorded as a positive result. The haemolytic activity was tested by culturing the strains on a 10% sheep blood plate with the addition of trimethoprim to inhibit swarming. For the motility assay, one colony was stabbed with a one-microliter loop into a semisolid nutrient medium. After overnight incubation at 37 °C, motility was measured as turbidity of the medium [2].

2.9. Genotyping

Phylogenetic tree was constracted using WGS.

3. Results

3.1. Patients and Bacterial Isolates

In total 20 isolates with reduced susceptibility to carbapenems were collected from November 3rd 2023 until 24th July 2024. All isolates originated from urinary tract and all but two from urinary catheters (Supplementary Material Table S1). All patients were females with age ranging from 49-90 years (median-79)

3.2. Antibiotic Susceptibility

Antimicrobial susceptibility results for the 20 P. mirabilis isolates are listed in Table 1. They all exhibited identical resistance profile with MICs of amoxycillin alone and combined with clavulanic acid, cefuroxime, cefotaxime, ceftriaxone, cefepime, gentamicin, amikacin and ciprofloxacin exceeding 128 mg/L in the majority of isolates. Regarding carbapenems, isolates displayed reduced susceptiblity towards imipenem with variable MIC values ranging from to 4 to 128 mg/L, but susceptiblity to meropenem and ertapenem. High resistance rate of 90% (18/20) was observed for ceftazidime. (Table 1) Piperacillin-tazobactam retained activity with all isolates being susceptible according to CLSI with MIC ≤ 16 mg/L . However, if EUCAST criteria were applied with lower resistance breakpoint (16 mg/L), 35% (n=7) of the isolates would be resistant. On the other hand, according to EUCAST there would be only four isolates resistant to imipenem (20%). All but one isolate showed susceptibility to cefiderocol. All isolates were classified as XDR. Meropenem, ertapenem and cefiderocol exerted high activity with MICs of meropenem below 0,5 mg/L.

3.3. Phenotypic Detection of β-Lactamases

Based on DDST using clavulanic acid and combined disk-test all isolates were ESBL positive. All isolates were susceptibile to cefoxitin, indicating the lack of p-AmpC. Regarding carbapenemases, Hodge test was negative in all but one isolate (95%) as no distortion of the inhibition zone was seen. Furthermore, CIM and e IM were positive in 80% (n=16), indicating production of an MBL, but only with imipenem disk. However, inhibitor based disk test with EDTA yielded positive result in 70% (n=14) again only with imipenem disk, rasing suspicion of an MBL.

3.4. Molecular Detection of Resistance Genes

PCR revealed bla_CTX-M cluster 1 and bla_VIM in all tested isolates. Sequencing of eight randomly selected VIM amplicons identified one bla_VIM-1, bla_VIM-4, and bla_VIM-78 genes, respectively and five bla_VIM-78 genes.All 11 CTX-M amplicons revealed CTX-M-15.

3.5. Genotpying by Interarray CarbaResist Kit

The genomes of all tested isolates possessed the identical resistance gene content including bla_TEM, bla_CTX-M-9 gene associated with the ISEcp insertion element and bla_VIM providing β-lactam resistance, aac(6)-Iic, aadA1 and aadA2 and armA coding enyzmes for aminoglycoside modification, sul1 and sul2 for sulphonamide resistance and dfrA1 for dihydrofolate reductase, rendering trimethoprim resistant. One isolate harboured dfrA15 allelic variant. All isolates harboured integrase genes for class 1 and class 2 integrons.

Table 2. Resistome of P. mirabilis isolates by Interarray CarbaResist Kit.

Isolate and Protocol Number	Res phenotype	β-Lactam	Aminoglycoside	sulphonami	trimethoprim	Integrase genes
PM 2	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1	dfrA1	Intl2
PM 3	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaCMY blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM4	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1 dfrA15	Intl1 Intl2
PM11	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM14	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM 19	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM 20	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2

Table 2. Resistome of P. mirabilis isolates by Interarray CarbaResist Kit.

Isolate and Protocol Number	Res phenotype	β-Lactam	Aminoglycoside	sulphonami	trimethoprim	Integrase genes
PM 2	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1	dfrA1	Intl2
PM 3	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaCMY blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM4	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1 dfrA15	Intl1 Intl2
PM11	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM14	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM 19	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2
PM 20	AMX, AMC, TZP, CAZ,CTX, CRO, FEP, GM,AMI CIP	ISEcpblaCTX-M-15 blaTEM blaVIM	aac(6′’)IIc aadA1 aadA2 armA	sul1 sul2	dfrA1	Intl1 Intl2

3.6. WGS

All four tested isolates presented identical antimicrobial resistance genes (AMR): bla_CTX-M-202 bla_TEM-2, bla_TEM-1A, bla_VIM-4 for β-lactam resistance, aac(3)-IId, aph(6)-Id, aph(3'')-Ib, aadA1, armA, aac(6')-IIc for aminoglycoside resistance, sul1 and sul2 for sulphonamide resistance, dfrA1 encoding dihydropholate reductase, cat for chloraphenicol acetyltransferase and tet(J) responsible for tetracycline resistance as shown in Table 3. Fluoroquinolone resistance genes were not detected.

3.7. Plasmid Analysis

There we not typable plasmids found by multiplex PCR but WGS detected IncQ1 plasmid in representative isolates.

3.8. Detection of Virulence Determinants

All isolates were positive for urease activity, hemolysis, and motility.

3.9. Genotyping

Genotyping demonstrated that four representative isolates were clonally related but distinct from ESBL and p-AmpC positive isolates from previous study as shown in Figure 1.

4. Discussion

The study described an outbreak of infections with VIM-1 positive P. mirabilis. We characterized resistance determinants and molecular epidemiology of the isolates. All isolates had identical resistance patterns and genes. The study showed development of resistance of this important hospital pathogen and its ability to accumulate β-lactam resistance determinants. The bibliographical references on MBLs in P. mirabilis are scarce and mostly reported from Far East [31]. In Europe, Greece and Bulgaria are focal points of VIM producing P. mirabilis [32,33,34,35]. Sporadic occurrence was recorded in Germany [36]. Our isolates were resistant or intermediate susceptible to imipenem, but susceptible to ertapenem and meropenem and thus represent a hidden reservoir of VIM-MBL in Enterobacterales.The use of molecular methods and immunochromatographic tests could help to uncover the underestimated carriage of VIM producing P. mirabilis since the isolates remain susceptible to two out of three carbapenems available in Croatia. VIM-4 and VIM-75 allelic variants are reported for the first time in Croatia. They were previously found among XDR P. mirabilis isolates from Germany [37]. Similarly, as in Germany two of our isolates harbored duplicate VIM carbapenemases. In contrast to the wide variety of VIM allelic variants all CTX-M amplicons encoded CTX-M-15, the most widespread allelic variant of CTX-M enyzmes. All isolates were categorized as XDR, as they were susceptible only to meropenem, ertapenem and cefiderocol. However, the resistance phenotype is dependent on which criteria were applied. According to CLSI all isolates were susceptible to piperacillin-tazobactam and resistant to imipenem, but if MICs were interpreted using EUCAST recommendations, all isolates would be resistant to piperacillin-tazobactam and the majority intermediately susceptible to imipenem (susceptible at increased exposure). The isolates exhibited variable level of resistance to expanded-spectrum cephalosporins and carbapenems with MICs of imipenem ranging from 4 to >128 mg/L. This could be due to variable expression of bla_VIM genes.

MBL encoding genes were probably embedded in class 1 or class 2 integrons which can be transferred and integrated in the plasmids or chromosomes and are responsible for the spread of gene cassetes with β-lactam, aminoglycosides and sulphonamide resistance genes, similarly as those from Germany [37]. CHDL previously identified only in Acinetobacter baumannii, are dominant in P. mirabilis in France, Belgium and Germany and similarly as MBLs also pose a diagnostic challenge due to high susceptibility to carbapenems [37,38,39]. In the past, Pseudomonas aeruginosa was a typical host for VIM carbapenemases, but later they were frequently detected among carbapenem-resistant Enterobacter cloacae isolates [40]. In contrast to other Enterobacterales, VIM positivity affected only imipenem susceptibility and rendered it resistant. Meropenem and ertapenem remained susceptible. Among Enterobacterales, K. pneumoniae and E. coli are typical hospital pathogens, bearing bla_CARB genes, associated with hospital outbreaks. However, P. mirabilis is gaining importance as nosocomial pathogen and outbreaks due to ESBL, p-AmpC and carbapenemase producing isolates occur all over the world. In Croatia, two hospital outbreaks were reported: one with TEM-52 producing P. mirabilis and the other with CMY-16 producing P. mirabilis, both from Split [8,9,11].

CTX-M-15 ESBL was detected as additional β-lactamase to VIM. It was conferring on the producing isolates high-level resistance to third and fourth generation cephalosporins and amoxicillin-clavulanate. This allelic variant has achieved worldwide spread and was described in E. coli and K. pneumoniae from Croatia [41,42]. This is in contrast with the recent study on ESBLs in P. mirabilis from Croatia which identified CTX-M-9 cluster to be dominant with CTX-M-14 as the most frequent variant [43]. ISEcp upstream of bla_CTX-M gene acts as promotor and increased the expression of the gene. In spite of very high MICs of aminoglycosides, there were no plasmid-born resistance genes and thus the resistance is most likely due the mutations of gyrA and parCgenes. High-level resistance to aminoglycosides is in line with a plethora of resistance genes encoding acetylases, adenylases and phosphorilases which modify aminoglycosides and render them inactive. The presence of sul, tet and cat genes is in concordance with sulphonamide, tetracycline and chloramphenicol resistance, respectively. Intrestingly, there was only one allelic variant of dfr gene coding for dihydropteroat syntethase, unlike previous studies on ESBLs in P. mirabilis which revealed several allelic variants [43].

Unfortunately, phenotypic tests based on carbapenem hydrolysis proved unreliable. MHT did not detect any of the isolates as carbapenemase producers and CIM exhibited positive results only with imipenem disk. This is due to low expression of bla_VIM genes and weak carbapenem hydrolysis. Phenotypic detection of MBLs is mostly based on carbapenem-EDTA synergy test. In the present study we performed two forms of tests: combined-disk test with EDTA and eCIM, but both exerted moderate sensitivity, visible only with imipenem disk. On the contrary, immunochromatographic test showed high concordance with PCR, WGS and Interarry CarbaResist Kit. All four methods detected VIM resistant determinant in all tested isolates. A similar problem was observed with OXA-23 and OXA-58 producing isolates in France and Belgium [38,39]. Apparently, expression of the carbapenemase encoding genes is very weak among Proteus spp. isolates, regardless of the carbapenemase type. Some researchers recommended the use of dipicolinate as metal chelator for Proteus spp. which provides better sensitivity compared to EDTA [43]. However, the compound is not available in the most routine laboratories.

Therapeutic options are very limited also due to the intrinsic resistance to tigecycline and colistin. Fortunately, cefiderocol susceptibility tested positive in all except one strain. There was no resistance recorded to meropenem and ertapenem. However, in case of carbapenem administration, there is a risk of development of mutants hyperproducing VIM β-lactamase under selection pressure.

The fact that all isolates possessed identical resistance genes points towards the common source. The described MDR clone successfully spread within two women's wards of a psychiatric hospital, causing urinary infections in catheterized patients. The hospital epidemic was stopped after the introduction of measures to prevent the spread of the VIM-positive P. mirabilis clone, but after that, urinary infections caused by this clone were detected in two nursing homes in Zagreb, because the patients were transferred from the hospital. Thanks to the implemented prevention measures, the clone did not spread in those two homes.

5. Conclusions

Analysis of the local antimicrobial resistance patterns is necessary in order to avoid hospital outbreaks with difficult to treat isolates. Hospital hygiene measures should be implemented to limit the spread of MDR P. mirabilis.