Characteristics of the patients with E. faecium BSI
During the one-year study period, blood culture was performed for 87,399 patients with suspected BSI in eight sentinel hospitals, and 10,990 (12.6%) were positive for at least one bacterial or fungal pathogen. Among them, 308 cases (2.8% of positive blood culture) of E. faecium BSI were identified and included in this study (Table 1). The median age of the patients was 72.5 years, ranging from 61 to 80 years, and more than half (54.9%, 169/308) were male. Most patients were inpatients in general wards (51.6%, n = 159) and intensive care units (37.0%, n = 114); only 11.4% (n = 35) of them were outpatients. Almost four-fifths (77.9%, n = 240) of cases were hospital-originated infections. The most common underlying comorbidities were malignancies (28.2%, n = 87), followed by diabetes mellitus (16.9%, n = 52) and cardiovascular diseases (16.2%, n = 50).
Antimicrobial resistance phenotypes of E. faecium blood isolates
Among the 308 E. faecium blood isolates, 132 (42.9%) showed positive results in vanA PCR, with none being positive in vanB PCR. All vanA-positive isolates were resistant to ampicillin and ciprofloxacin. High-level resistance to gentamicin was also identified in 39.4% (52/132) of the isolates, but none of them showed high-level resistance to streptomycin. Three-quarters of the isolates (75.0%, 99/132) exhibited VanA phenotypes, i.e., high-level resistance to vancomycin (MIC, > 64 mg/L) and resistance to teicoplanin (MIC, ≥ 32 mg/L); 29 (22.0%) isolates exhibited VanD phenotypes, i.e., high-level resistance to vancomycin (MIC, > 64 mg/L) and intermediate resistance (n = 22; MIC = 16 mg/L) or reduced susceptibility (n = 7; MIC = 8 mg/L) to teicoplanin. The remaining four (3.0%) isolates were susceptible to both vancomycin (MIC, 0.5–1 mg/L) and teicoplanin (MIC, 0.12–0.25 mg/L), indicating vanA-positive but vancomycin-susceptible (vanA+VS) phenotypes.
Clinical outcome of patients with E. faecium BSI
Compared to those caused by vanA-negative E. faecium, BSIs caused by vanA-positive E. faecium occurred more frequently in inpatients (95.5% versus 85.2%; P value = 0.006) and in patients with a higher SOFA score (median value, 7.0 versus 4.0; P value = 0.001). The 30-day mortality rate was higher in patients with vanA-positive E. faecium BSI than in those with vanA-negative E. faecium BSI, but without statistical significance (36.4% versus 27.3%, P = 0.114). However, both the 60-day mortality (46.2% versus 30.1%; P value = 0.005) and in-hospital mortality (54.5% versus 33.5%; P value < 0.001) were significantly higher in vanA-positive E. faecium BSI patients than in vanA-negative E. faecium BSI patients (Figure S1).
Genetic characteristics of vanA-positive E. faecium blood isolates
Circularized chromosomes were obtained from 120/132 vanA-positive E. faecium isolates by whole-genome sequencing, and mean value of coverage depth was 267.3 ranging from 92 to 620. Median size of circularized chromosomes was found to be 2,881,288 bp, ranging from 2,446,701 bp to 3,003,360 bp. The isolates carried one to three plasmids, with Rep A_N family and Inc18 family plasmids most frequently being identified. The most common strain type of vanA-positive blood isolate was ST1421 (n = 52), followed by ST17 (n = 36), ST80 (n = 15), ST192 (n = 12), and ST252 (n = 5) (Fig. 1). By cgMLST, 37 different complex types (CTs) were identified, and CT6141-ST17 (n = 23) was the most common, followed by CT6552-ST1421 (n = 20), CT6555-ST1421 (n = 15), and CT6554-ST192 (n = 10). One (n = 123) or two (n = 9) copies of the vanA operon were identified in each isolate, regardless of its location on a plasmid (n = 127) and/or the chromosome (n = 7); 2/132 isolates carried the vanA operon both on the chromosome and on a plasmid. Other resistance genes frequently identified on chromosomes were aminoglycoside-modifying enzyme-encoding genes aac(6’)-Ii (97.7%, n = 129) and ant(9)-Ia (60.6%, n = 80) and macrolide resistance-related genes msr(c) (97.7%, n = 129) and erm(A) (62.9%, n = 83).
Phylogenetic tree based on cgMLST of vanA-positive E. faecium isolates
Structure of Tn1546-type transposons and glycopeptide resistance phenotypes
vanA operons were identified as a part of Tn1546-type transposons, and 128 blood isolates exhibiting a VanA or VanD phenotype carried one (n = 119) or two (n = 9) copies of the transposon classified as six structural variants both by deletion or truncation of vanY and vanZ and by insertion of ISEfa11 between vanX and IS1216 (Fig. 2A). The vanS gene of all 128 isolates showed nucleotide sequence variations resulting in three amino acid substitutions, L50V, E54Q, and Q69H, compared with the vanS gene of pIP501 [7, 30]. The remaining four isolates with vanA+VS phenotypes were found to carry a structurally impaired vanA operon, either lacking one or both regulatory genes vanR and vanS or having a truncated D-alanyl-D-alanine dipeptidase gene vanX, on a plasmid.
Structure of Tn1546-type transposons and teicoplanin MICs of E. faecium hosts according to the structure of vanY. Blue bars in the bar graph (C) indicate the clinical breakpoints according to the CLSI guideline, and red one indicates those according to the EUCAST
The 128 isolates with the VanA or VanD phenotype were grouped according to the vanA operon copy number and structure. Group 1 isolates (n = 24) showing the typical VanA phenotype carried a plasmid harboring a copy of a Tn1546-type transposon (Tn1546vanRSHAXYZ) with all seven component genes of the vanA operon, with (variant Ib, n = 2) or without (variant Ia, n = 22) insertion of ISEfa11. Group 2 isolates (n = 89) carried a copy of the Tn1546 variant with both truncation of vanY and deletion of vanZ by insertion of an IS1216 into vanY (Tn1546vanY::IS1216,ΔvanZ) with (variant IIb, n = 45) or without (variant IIa, n = 44) insertion of ISEfa11. The isolates exhibited VanA or VanD phenotype, and the Tn1546vanY::IS1216,ΔvanZ variants were found to be located either on a plasmid (n = 85) or on the chromosome (n = 4). The length of remnant vanY in Tn1546vanY::IS1216,ΔvanZ varied from 165 to 901 bp according to the insertion site, which did not show any correlation with teicoplanin MICs. Group 3 isolates (n = 6) exhibiting the VanD phenotype carried a copy of the Tn1546-type transposon with deletion of both vanY and vanZ [Tn1546Δ(vanY-vanZ)] with (variant IIIb, n = 2) or without (variant IIIa, n = 4) insertion of ISEfa11 on a plasmid. Nine isolates harbored two copies of Tn1546-type transposons on a plasmid and/or on the chromosome (Fig. 2B). Of them, five and one isolates exhibiting the VanA phenotype possessed one each copy of Tn1546vanRSHAXYZ and Tn1546Δ(vanY-vanZ) and Tn1546vanY::IS1216,ΔvanZ and Tn1546Δ(vanY-vanZ), respectively. The remaining three isolates carried two copies of Tn1546vanY::IS1216,ΔvanZ and showed the VanA (n = 2) or VanD (n = 1) phenotype (Fig. 2B).
Chromosomal vanA operon
One (n = 6) or two (n = 1) copies of the vanA operon were identified on the chromosome in seven isolates of CT6555-ST1421 (n = 4), CT6141-ST1421 (n = 1), CT6552-ST1421 (n = 1), and CT6557-ST78) (n = 1), and two of them carried an additional vanA operon-harboring plasmid (Fig. 3). The vanA operons found on the chromosome were always identified in an insertion unit with (n = 6) or without (n = 2) plasmid-originated components, resulting in variable sizes of the units, ranging from 9 to 43 kb. All eight insertion units were shown to be flanked by a pair of IS1216 of the same or opposite orientations and left (5‘-GGT TCT GTT GCA AAG TTT TAA ATC TAC TAT CAA ATA AGG TAG AAT AG-3‘) and right (5‘– GGT TCT GTT GCA AAG TTT TAA ATA AAG AAT AAA ATC CTT ACG GTA TCT AT-3‘) inverted repeats. The insertion events on the chromosome were shown to be neither site-specific nor nucleotide sequence-specific, occurring in coding regions (n = 5) or in intergenic regions (n = 3). Of note, both isolates C0019EM0016 and C0019EM0037 of CT6555-ST1421 recovered in a hospital with a 7-month gap shared an insertion unit at the same location on the chromosome, indicating a clone; however, the isolate C0019EM0037 carried another insertion unit at a different location on the chromosome, suggesting that another independent insertion event occurred.
Structure and location of the chromosomal vanA operon. The red bars in chromosome indicate the insertion sites of Tn1546-type transposon. Red arrows indicate the genes identified as resistance determinants, and blue arrows indicate the insertion sequences
VR-plasmids harboring the vanA operon
A total of 127 VR-plasmids carrying one (n = 122) or two (n = 5) copies of Tn1546-type transposons were identified. Circular plasmids of the Inc18 family (n = 96) were most common, followed by putative linear plasmids of the RepB family (n = 21); a circular plasmid of hybrid Inc18:RepA_N was also identified. The remaining nine plasmids were nontypeable due to failure in plasmid circularization (n = 5) or in identification (n = 4) of the plasmid replication origin.
All circular plasmids of the Inc18 family were found to share 14–15 kb-sized derivatives from the plasmid pRE25, including a rep2 replication origin, the erm(B) gene, and a zeta-epsilon toxin-antitoxin system, but to lack the probable conjugation regions (ORF25 to ORF39 of pRE25) [31], compatible with the unsuccessful results in conjugation experiments for all E. faecium isolates carrying the plasmids. The 96 circular plasmids were divided into types A (n = 50) and B (n = 46) according to both sequence similarity (> 60%) and Tn1546 variant type (Fig. 4A and B). Type A plasmids had a median size of 32,082 bp, ranging from 18,701 bp to 43,334 bp, and were most frequently identified in E. faecium isolates of ST17 (46%, 23/50). Type A plasmids possessed either Tn1546 variant IIa (n = 45) or variant IIIa (n = 3), except for two plasmids harboring a Tn1546-type transposon with a structurally impaired vanA operon, both identified in vanA+VS isolates (Table 2). Type B plasmids had a median size of 42,610 bp, ranging from 27,214 bp to 51,084 bp, and were mostly (80.4%, 37/46) identified in E. faecium isolates of ST1421. Tn1546 variants identified in type B plasmids always showed insertion of ISEfa11 between vanX and IS1216, and variant IIb (n = 41) was most common, followed by variant Ib (n = 3) and variant IIIb (n = 2). Most (97.8%, 45/46) type B plasmids also had an aminoglycoside resistance determinant aph(3′)-IIIa.
Structure of four types of plasmids containing Tn1546-type transposons
A putative replication origin of the RepB family was found in 21 putative linear plasmids with either one (type C, n = 16) or two (type D, n = 5) hairpin ends composed of inverted tandem repeat sequences of 2 kb with 5‘-TATA-3‘ hairpin loops (Fig. 4C and D). The plasmids exhibited homology of approximately 70% with the linear plasmid pELF1 [32], and they shared putative transfer-related components ftsK and parA. More than half (57.1%, 12/21, 10 type C and two type D) of the plasmids were successfully conjugated to the recipient E. faecium DSM13589. The type C plasmids had a median size of 106,938 bp, ranging from 72,306 to 112,181 bp. The single hairpin end of the plasmids were observed to harbor a copy of Tn1546vanRSHAXYZ (14/16), Tn1546Δ(vanY-vanZ) (n = 1), or a Tn1546-type transposon with a structurally impaired vanA operon (n = 1); other resistance determinants ant(9)-Ia (n = 6) and erm(A) (n = 5) were also identified at the opposite end of the plasmids. The type D plasmids had a median size of 118,791 bp, ranging from 102,459 to 161,239 bp, and harbored two copies of the Tn1546-type transposon, each copy of Tn1546vanRSHAXYZ and Tn1546Δ(vanY-vanZ) at each hairpin end. The plasmids also harbored the resistance determinants erm(B) and aph(3′)-IIIa.
Bacterial growth rate
The median growth rate [Max(∆lnOD600/s)] of vanA-negative ampicillin-resistant E. faecium blood isolates in MH broth was 0.195 (1st to 3rd interquartile range, 0.178 to 0.211), which was significantly higher than that of vanA-positive ampicillin-resistant isolates (median value, 0.178; 1st to 3rd interquartile range, 0.162 to 0.191; P < 0.001) (Fig. 5A).
Bacterial growth rates of E. faecium blood isolates. *** indicate P value < 0.001
The growth rates of four vanA+VS isolates (median value, 0.200; range, 0.156 to 0.220) were similar to those of vanA-negative isolates. Growth rates of vanA-positive isolates did not differ by strain type or R plasmid type carrying the vanA operon (Fig. 5B and C). Addition of vancomycin at concentrations of 4 mg/L and 16 mg/L to MH broth resulted in a significant decrease in the growth rates of vanA-positive E. faecium isolates, regardless of plasmid type (Fig. 5D–F). Notably, the growth rates of ST1421 and ST17 vanA-positive E. faecium isolates at a vancomycin concentration of 4 mg/L (median value, 0.154; 1st to 3rd interquartile range, 0.139–0.166) were significantly faster than those of vanA-positive isolates of other STs (median value, 0.143; 1st to 3rd interquartile range, 0.122–0.160; Fig. 5G). The difference between growth rates in MH broth without vancomycin and with vancomycin at a concentration of 4 mg/L was significantly lower for isolates carrying a type A or B circular plasmid than those carrying a type C or D linear plasmid (Fig. 5H; P = 0.020).
Add Comment