All VREfm isolates from the years 2008, 2013, 2015 and 2018 were analyzed and came to a total of 120 cases. We investigated patient and isolate details, including age, sex, LOS, LOS at the time of specimen collection, type of ward, and site of specimen collection. In single quarters of each year, a variety of different wards were represented. The VREfm cases did not differ among calendar years with regard to age (in total median across all calendar years 66 years, IQR 53-75, p=0.839), gender (in total male 52% and female 48%, p=0.223), LOS (in total median 33 days, IQR 14-64%, p=0.209), or the above mentioned ward type. Between 2008 and 2018, there was no difference in intensive care unit, hematology/oncology, surgery, or others (p=0.945, p=0.825, p=0.867 and p=0.729). In contrast, LOS at the time of specimen collection and site of specimen collection differed in various calendar years. In 2008, most of the samples were obtained as clinical cultures (n=25, 83.3%), while in 2013 (n=4, 13.3%), 2015 (n=6, 20%), and 2018 (n=7, 23.3%) screening cultures were more frequent. Urine accounted for the majority of clinical cultures. The LOS at the time of specimen collection decreased from 2008 (18.5 days; IQR 4-40) to 2018 (1.5 days; IQR 0-21) and is consequently of significance, p=0.011.
Characteristics of VREfm isolates
All 120 VREfm isolates were sequenced, with an average coverage ranging from 41- to 121-fold. The percentage of good targets based on the core genome ranged from 95.2% to 99.8% with an average of 99.0%. STs as well as CTs were determined for all strains. Regarding classification into ST, each consecutive year saw a higher percentage of isolates that were ST117, rising from 16.7% in 2008 to 56.7% in 2018 (p=0.012). In total, 43 (35.8%) of the 120 E. faecium isolates were classified as ST117. In addition to ST117 strains, we detected strains assigned to ST203 (11.7%), ST80 (7.5%), ST78 (9.2%), ST192 (8.3%), ST17 (6.7%), and others (each ≤ 5%) (Fig. 1). While ST117, ST203 and ST80 were identified in all years, the number of isolates belonging to ST203 and ST80 increased until 2015 but decreased subsequently from 2015 to 2018. Beside the increase in ST117, we also observed a rise in isolates assigned to ST78 between 2015 and 2018 as well as a higher diversity of STs in 2008 than in 2018 (Fig. 1).
When classifying the strains with a higher resolution into cgMLST, there was a clear shift of dominant CTs from CT164 in 2008 to CT71 in 2018. We did not find any CT71 isolates in 2008, 2013, or 2015 although there was a variety of other CTs such as CT24, CT36 and CT190 (Fig. 2). In contrast, 43% of isolates from 2018 were CT71 strains (13/30, p<0.0001). Of all 120 isolates of the years studied, CT71 (11%) was the most common CT, followed by CT36 and CT162 (both 8%), and CT164 and CT894 (both 6%).
Risk factors for the frequent occurrence of ST117 strains
Table 1 shows patient characteristics for ST117 and non-ST117 carriers in each year. Non-ST117 comprised all strains other than ST117, including those which could not be assigned to a known ST. Regarding ward type, 54% of all CT71 strains were collected in ICUs, which comprised five different ICUs located in different buildings across the city. Most samples were obtained from rectal swabs (n=76, 63%), urine samples (n=14, 12%) and stool samples (n=9, 7.5%). Regarding the rectal swabs, there is a great increase from 2008 (n=4) to 2013 (n=26), whereas the number remained approximately constant in the following years (2015 n=24 and 2018 n=22). Urine samples accounted for 12% of the total number of samples, including 6 samples (20%) in 2008, 1 sample (3%) in 2013, 3 samples (10%) in 2015, and 4 samples (13%) in 2018. Stool samples only occurred in 2008 (n=9) and no samples in 2013, 2015 or 2018. Most CT71 strains (n=9; 70%) were obtained from rectal swabs.
Multivariable risk factor analysis for ST117
The multivariable analysis supported a strong association of ST117 with the calendar year (Additional file 1: Table S1). Samples from 2018 were more than 9 times more likely to be type ST117 than in 2008 (OR 9.4, 95%CI 2.3-37.7, p=0.002). A similar association was found for urine as specimen collection site and ST117 strains (OR 10.6, 95%CI 1.4-82.5, p=0.024). CT71 was not an independent risk factor for ST117. Because CT71 did not appear until 2018, the calendar year could not be estimated in the model.
Antimicrobial resistance, virulence factors and pangenome analysis of ST117 strains
Resistance genes and virulence factors for all 43 ST117 strains were identified using ResFinder, VirulenceFinder, VFDB, and CARD, all of which indicated various resistance genes and virulence factors (Additional file 2: Table S2). Resistance genes for macrolides, lincosamides, and streptogramin B (msrC, erm(B) and efmA) as well as for trimethoprim (dfrF and dfrG) and aminoglycosides (aac(6')-aph(2''), aph(3')-III, ant(6)-Ia and aac(6')-Ii) were detected in all ST117 strains. Moreover, genes for ciprofloxacin resistance (gyrA, parC) and ampicillin resistance (pbp5) were identified in all ST117 strains, which are intrinsic genes and expected to be present in all isolates. While almost all non-CT71 strains had several genes conferring resistance to aminoglycosides, we found only one such gene (aac(6')-Ii) in CT71 strains, which is also an intrinsic gene. Some non-CT71 isolates featured resistances to chloramphenicol (cat) and tetracycline (tet(M)) which were absent in CT71 strains. Among all ST117 strains, 74% (32/43 isolates) displayed the vanB genotype and the vanA in 26% (11/43 isolates). There was a shift from vanA to vanB between 2008 and 2018, with 80% (4/5 isolates) and 67% (6/9 isolates) vanA in 2008 and 2013, to 100% (11/11 isolates) and 94% (16/17 isolates) vanB in 2015 and 2018, respectively (Additional file 3: Figure S1). All CT71 strains harbored vanB.
All ST117 strains carried the virulence factor acm, which encourages cell wall-anchored collagen adhesion and has characteristics typical of a microbial surface component recognizing adhesive matrix molecules (MSCRAMM) and sgrA to stimulate surface adhesion. Moreover, nearly all ST117 strains were characterized by the presence of ecbA (98%, 42/43 isolates), the E. faecium collagen binding protein A, as well as the enterococcal surface protein esp (95%, 41/43 isolates) to promote biofilm formation. The presence of a putative glycoside hydrolase hylEfm was detected in more than half of the ST117 strains (78%, 33/43 isolates). Occasionally, the virulence factor scm (14%, 6/43 isolates), the second collagen adhesin of E. faecium, occurred. All CT71 strains presented the same set of virulence factors: acm, esp, hylEfm, ecbA, and sgrA (30%, 13/43 isolates). This set of virulence factors was also detected in another 15 non-CT71 strains, a total of 28 isolates out of 43 (65%). These 15 strains belonged mainly to CT36 (21%, 9/43 isolates). The virulence factors of non-ST117 strains showed no striking difference from ST117 strains.
Regarding the alcohol tolerance assay, we tested five CT71 strains and one VREfm ATCC strain (ATCC 700221). At an isopropanol concentration of 23% and 60%, we could not detect any growth in any of the strains.
The phylogenetic tree with all ST117 strains revealed that the CT71 strains were phylogenetically separated from non-CT71 strains (Additional file 4: Figure S2). Additionally, a pangenome analysis was conducted with all ST117 strains to study the differences in gene content between CT71 strains and non-CT71 strains (n=43). The isolates shared a core genome of 2166 genes (42 <= strains <=43), a soft-core genome of 130 genes (40 <= strains <42), a shell genome of 1338 genes (6 <= strains <40) and 1618 cloud genes (strains <6). In summary, the core genome comprises 42% of the pangenome (5252 genes). The pangenome matrix (presence and absence of genes) revealed that the distribution of genes among strains was relatively similar. With regard to the accessory genome profile of the CT71 strains compared to the other CT strains, there seemed to be a set of genes more frequently present in CT71 strains than in the other strains (Additional file 5: Figure S3). Additionally, the hierarchical clustering based on presence and absence of genes clustered CT71 isolates distinctly from non-CT71 strains. The frequency chart shows that many genes were present in all genomes and that many gene were present only in single genomes, but we could not detect peaks for a set of isolates indicative for CT71 strains (n=13) (Additional file 5: Figure S3).